Automotive High Speed Communication Technologies: SerDes and Ethernet for Sensor and Display Applications 9783446470422, 3446470425

Citation preview

Kirsten Matheus Michael Kaindl

Automotive High Speed Communication Technologies SerDes and Ethernet for Sensor and Display Applications

Matheus/Kaindl Automotive High Speed Communication Technologies

Kirsten Matheus Michael Kaindl

Automotive High Speed Communication Technologies SerDes and Ethernet for Sensor and Display Applications

The Authors: Dr. Kirsten Matheus joined BMW in 2009, and is responsible for the timely availability of standardized communication technologies within BMW. Michael Kaindl joined BMW in 1990, and has worked with in-vehicle communication technologies ever since.

Distributed by: Carl Hanser Verlag Postfach 86 04 20, 81631 Munich, Germany Fax: +49 (89) 98 48 09 www.hanserpublications.com www.hanser-fachbuch.de

The use of general descriptive names, trademarks, etc., in this publication, even if the former are not especially identified, is not to be taken as a sign that such names, as understood by the Trade Marks and Merchandise Marks Act, may accordingly be used freely by anyone. While the advice and information in this book are believed to be true and accurate at the date of going to press, neither the authors nor the editors nor the ­publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. The final determination of the suitability of any information for the use contemplated for a given application ­remains the sole responsibility of the user. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying or by any information storage and retrieval system, without permission in writing from the publisher. © 2023 Carl Hanser Verlag, Munich Editor: Dipl.-Ing. Volker Herzberg Production Management: Melanie Zinsler Cover image: © shutterstock.com/Open Studio Cover concept: Marc Müller-Bremer, www.rebranding.de, Munich Cover design: Max Kostopoulos Typesetting: Eberl & Koesel Studio, Kempten Printed and bound by Druckerei Hubert & Co. GmbH und Co. KG BuchPartner, Göttingen Printed in Germany Print-ISBN: 978-3-446-46918-1 E-Book-ISBN: 978-3-446-47042-2

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII Timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI Abbreviations and Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XXVIII 1

Introduction and Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

1.1

The Distinctive Properties of High-Speed Sensor and Display Use Cases . . . . . . 3

1.2

Background to Automotive SerDes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.1 The Origin of “SerDes” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.2 Automotive SerDes Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.2.3 The Status of Automotive SerDes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.3

Background to Automotive Ethernet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.3.1 The Origin of “Ethernet” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.3.2 Ethernet in the Automotive Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.3.3 Introduction to High-Speed (HS) Automotive Ethernet . . . . . . . . . . . . . . . 17

1.4 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2

The Automotive Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.1 Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.1.1 A Brief History of Displays in Cars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.1.2 Display Basics and Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.1.3 Display Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.4 Typical Communication Related Requirements of Displays . . . . . . . . . . 32 2.2 Cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.2.1 A Brief History of Cameras in Cars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.2.2 Camera Basics and Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

VI

Contents

2.2.3 Camera Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 2.2.4 Camera Software, Safety, and Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.2.4.1 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.2.4.2 Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 2.2.4.3 Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 2.2.5 Typical Communication Related Requirements for Cameras . . . . . . . . . 45 2.3

Other Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.3.1 Relevant Sensor Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 2.3.1.1 Sound Navigation and Ranging (Sonar)/Ultrasonic Sensors . . 48 2.3.1.2 Radio Detection and Ranging (Radar) . . . . . . . . . . . . . . . . . . . . 50 2.3.1.3 Light Detection and Ranging (Lidar) . . . . . . . . . . . . . . . . . . . . . 52 2.3.1.4 Time of Flight (ToF) Cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 2.3.2 Overall Comparison and Architecture Considerations for ­Sensors . . . . . 55

2.4

Other Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

2.5 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

3

The Automotive Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.1

The Automotive Industry as such . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.1.1 The Automotive Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 3.1.2 The Automotive Development and Production . . . . . . . . . . . . . . . . . . . . . 69 3.1.2.1 Impact with Respect to the Supply Chain . . . . . . . . . . . . . . . . . 69 3.1.2.2 Impact with Respect to the Product Life Cycle (PLC) . . . . . . . . 73

3.2

General Automotive Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3.2.1 Use-related Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.2.2 Regulatory Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 3.2.2.1 Government Driven Requirements . . . . . . . . . . . . . . . . . . . . . . . 78 3.2.2.2 Insurance Driven Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 80

3.3

Automotive Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.3.1 Semiconductor Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.3.2 Semiconductor Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.3.3 Semiconductor Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3.4 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Contents

4

The Electromagnetic Environment in Cars . . . . . . . . . . . . . . . . . . . . . . 89

4.1

ElectroMagnetic Compatibility (EMC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.1.1 Basic Principle of Electromagnetic Interference . . . . . . . . . . . . . . . . . . . . 91 4.1.2 Relevant EMC Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 4.1.3 Overview on EMC Test Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 4.1.4 Impact of a Shield on EMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.1.4.1 EMC for Shielded Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.1.4.2 Shield Connection at the Case . . . . . . . . . . . . . . . . . . . . . . . . . . 102 4.1.4.3 Interrelation between Shield and Ground . . . . . . . . . . . . . . . . . 103

4.2

ElectroStatic Discharge (ESD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.2.1 Unpowered ESD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 4.2.1.1 ESD Protection in the Production Process . . . . . . . . . . . . . . . . . 106 4.2.1.2 ESD Protection Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.2.1.3 Transmission Line Pulse Measurement (TLP) . . . . . . . . . . . . . . 109 4.2.2 Powered ESD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 4.2.3 How to Achieve ESD Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

4.3 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

5

The Automotive Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

5.1

Channel Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

5.2

Channel Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.2.1 Impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 5.2.2 Scattering-Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 5.2.3 Channel Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 5.2.3.1 Transmission Related Impairments . . . . . . . . . . . . . . . . . . . . . . 126 5.2.3.2 Self-noise Parameters that Distort the Signal . . . . . . . . . . . . . . 128 5.2.3.3 EMC Related Channel Parameters and Other Noise . . . . . . . . . 131 5.2.3.4 Transmission Channel Interference Model . . . . . . . . . . . . . . . . 134

5.3

Cables and Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 5.3.1 Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 5.3.1.1 Unshielded Twisted Pair (UTP) Cables . . . . . . . . . . . . . . . . . . . . 139 5.3.1.2 Shielded Twisted Pair (STP) Cables . . . . . . . . . . . . . . . . . . . . . . 141 5.3.1.3 STar-Quad (STQ) Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 5.3.1.4 Shielded Parallel Pair (SPP) Cables . . . . . . . . . . . . . . . . . . . . . . 142 5.3.1.5 Coaxial Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

VII

VIII

Contents

5.3.1.6 Other Multi-port Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 5.3.1.7 Aging and Mechanical Stress of Cables . . . . . . . . . . . . . . . . . . . 148 5.3.2 Connectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 5.3.2.1 Connectors for UTP Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 5.3.2.2 Connectors for STQ Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.3.2.3 Connectors for SDP (STP and SPP) Cables . . . . . . . . . . . . . . . . . 152 5.3.2.4 Connectors for Coaxial Cables . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.3.3 Quo Vadis? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 5.4

Printed Circuit Boards (PCBs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

5.5 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

6 Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 6.1

Supplying Power with the ­Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 6.1.1 General Considerations for Power-over . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 6.1.1.1 The Bias-T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 6.1.1.2 Voltage Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 6.1.1.3 Failure Detection and Protection . . . . . . . . . . . . . . . . . . . . . . . . . 170 6.1.2 Power over Differential (PoD) Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 6.1.3 Power over Coaxial Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 6.1.3.1 Required Inductive Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 6.1.3.2 Bias-T Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 6.1.3.3 Power Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 6.1.3.4 PoC Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 6.1.3.5 PoC Compendium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

6.2

Power (Saving) Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 6.2.1 Transitioning between (Power) Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 6.2.2 Low Power Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 6.2.2.1 Deep Sleep (and Wake-up) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 6.2.2.2 Light Sleep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

6.3 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

7

Automotive SerDes Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

7.1

Analogue Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

7.2

Low Voltage Differential Signaling (LVDS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

7.3

Proprietary Automotive SerDes ­Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 7.3.1 TI’s Flat Panel Display (FPD) Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203

Contents

7.3.2 Analogue Device’s Gigabit Multimedia Serial Link (GMSL) . . . . . . . . . . . 207 7.3.3 Inova’s Automotive PIXel Link (APIX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 7.3.4 Sony’s Gigabit Video InterFace (GVIF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 7.4

MIPI A-PHY/IEEE 2977 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 7.4.1 A-PHY Overview and Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 7.4.2 A-PHY Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 7.4.3 A-PHY Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 7.4.4 A-PHY Data Link and Higher Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

7.5

ASA Motion Link (ASAML) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 7.5.1 ASAML Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 7.5.2 ASAML Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 7.5.3 ASAML Data Link Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 7.5.4 ASAML Safety and Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 7.5.5 ASAML Application Stream Encapsulation Protocols (ASEPs) . . . . . . . . 243

7.6 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245

8

High-Speed (HS) Automotive Ethernet . . . . . . . . . . . . . . . . . . . . . . . . . . 251

8.1

Physical (PHY) Layer Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 8.1.1 IEEE 802.3ch for 2.5, 5, and 10 Gbps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 8.1.1.1 The IEEE 802.3ch Channel(s) . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 8.1.1.2 Technical Description of IEEE 802.3ch . . . . . . . . . . . . . . . . . . . . 255 8.1.2 IEEE 802.3cy for 25 Gbps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 8.1.2.1 The IEEE 802.3cy Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 8.1.2.2 Technical Description of IEEE 802.3cy . . . . . . . . . . . . . . . . . . . . 262 8.1.3 IEEE 802.3cz for 2.5 to 50 Gbps over optical media . . . . . . . . . . . . . . . . . 263 8.1.3.1 The IEEE 802.3cz Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 8.1.3.2 Technical Description of IEEE 802.3cz . . . . . . . . . . . . . . . . . . . . 266 8.1.4 Asymmetric Ethernet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268

8.2

Related Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 8.2.1 Power Saving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 8.2.1.1 Energy Efficient Ethernet (EEE) . . . . . . . . . . . . . . . . . . . . . . . . . 273 8.2.1.2 Wake-up and Sleep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 8.2.2 Quality of Service (QoS) with Time Sensitive Networking (TSN) . . . . . . 276 8.2.2.1 IEEE 1722 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 8.2.2.2 Synchronization/Timing with IEEE 802.1AS . . . . . . . . . . . . . . 279 8.2.2.3 Traffic Shaping with IEEE 802.1Qav and Qcr . . . . . . . . . . . . . . 281

IX

X

Contents

8.2.2.4 Ingress Policing and Filtering with IEEE 802.1Qci . . . . . . . . . . 282 8.2.2.5 Other TSN Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 8.2.3 Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 8.3

Automotive Ethernet versus Automotive SerDes . . . . . . . . . . . . . . . . . . . . . . . . . . 287 8.3.1 Principle Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 8.3.2 Specific Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290

8.4 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295

9

Related Standards and Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

9.1

Color Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

9.2

Video Compression Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 9.2.1 (M)JPEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 9.2.2 MPEG alias H.26x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 9.2.2.1 H.262, MPEG-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 9.2.2.2 H.264, MPEG-4, AVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 9.2.2.3 H.265, MPEG-H Part 2, HEVC . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 9.2.3 VESA Display Compression Codecs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311

9.3

Content Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313

9.4

Audio Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315

9.5

Control Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316 9.5.1 General Purpose Input/Output (GPI/O) . . . . . . . . . . . . . . . . . . . . . . . . . . . 317 9.5.2 Serial Peripheral Interface (SPI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317 9.5.3 Inter-IC Bus (I2C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320 9.5.3.1 I2C Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320 9.5.3.2 I2C Channel Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322 9.5.3.3 I2C Extensions and Derivates . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 9.5.4 MIPI Improved Inter-IC Bus (I3C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 9.5.5 Using a Memory Map Instead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325

9.6

Camera Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 9.6.1 MIPI CSI-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330 9.6.2 MIPI D-PHY and C-PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 9.6.3 MIPI CCS and CSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333

9.7

Display Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 9.7.1 Open LVDS Digital Interface (OpenLDI) . . . . . . . . . . . . . . . . . . . . . . . . . . . 334 9.7.2 Digital Visual Interface (DVI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 9.7.3 High Definition Multimedia Interface (HDMI) . . . . . . . . . . . . . . . . . . . . . . 337

Contents

9.7.4 MIPI Display Serial Interface (DSI-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340 9.7.5 DisplayPort (DP) and Embedded DisplayPort (eDP) . . . . . . . . . . . . . . . . . 341 9.7.6 V-by-One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344 9.7.7 Universal Serial Bus (USB) – C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 9.7.8 Overview and Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 9.8 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348

10

Test and Qualification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355

10.1 Development Methods and Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356 10.1.1 Waterfall Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356 10.1.2 V-Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 10.1.3 Agile Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358 10.2 Designed for Testability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 10.2.1 Built-in Status Registers and Quality Indicators . . . . . . . . . . . . . . . . . . . . 360 10.2.2 Accessibility of Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361 10.2.3 Scanlines for Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 10.2.4 Loopback Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364 10.2.5 Built-In Self-Tests (BISTs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 10.3 Test Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 10.3.1 Test Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366 10.3.2 Devices Under Test (DUTs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366 10.4 Test Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367 10.4.1 ISO 9646 Conformance Testing Methodology . . . . . . . . . . . . . . . . . . . . . . 368 10.4.2 Test Specifications for Automotive Ethernet . . . . . . . . . . . . . . . . . . . . . . . 370 10.4.3 Test Specifications for Automotive SerDes . . . . . . . . . . . . . . . . . . . . . . . . . 371 10.5 Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372 10.5.1 Tools for Channel Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 10.5.2 Tools for Transmitter/Transceiver Tests . . . . . . . . . . . . . . . . . . . . . . . . . . 375 10.5.3 Tools for Evaluating the Data Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376 10.6 Experiences with Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 10.6.1 The Pitfall of Test Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 10.6.2 Unexpected System Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379 10.6.3 “No Error Found” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380 10.7 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383

XI

Preface

It is common knowledge that the amount of electronics and software in cars is continuously increasing. Not only are more and more mechanical functions replaced by electronics, but also more driving related functions typically performed by the drivers are substituted or supported by electronic systems, making room for more elaborate and connected infotainment offerings at the same time. From our perspective, one of the most important enabling infrastructure elements for all of this is the right set of powerful and robust, automotive suitable communication technologies, for which we are at the source. We, the authors of this book, both work at the central department within BMW that is responsible for the in-vehicle communication technologies. The responsibilities of our department thereby entail it all: the early anticipation and identification of communication requirements, the development and standardization of suitable technologies, the validation and qualification of respective semiconductor products, the writing of requirement specifications on how to deploy the technologies in Electronic ­Control Units (ECUs) and within the Electric and Electronic (EE) architecture of the cars, ensuring that there are tools and test specifications available, problem solving in case of unexpected errors in the field, and more. All major car manufacturers have similar departments with a similar set of tasks. Some car manufacturers, like BMW, get involved early. Others might get involved later. What they all have in common is the responsibility for the networking technologies that are used for the communication between many different ECUs, such as LIN, CAN, FlexRay (if used), and lately Automotive Ethernet (especially 100 Mbps). The result is a broad, public knowledge base for all these technologies in the industry and various standardization groups maintaining and advancing the knowhow. For very application specific technologies, for communication links that are not part of the network (often called “private” communication links), or for communication functions bought in closed systems from Tier 1 suppliers, the situation is not quite as straight forward. These technologies are often not handled in the central departments but within the groups responsible for the application. For these technologies there is little (and practically no public) information available in form of technical descriptions and enabling specifications (EMC, channel, system functions, tests, . . .). Industry consortia, driving the technologies forward with consolidated interests, are rare. The high-speed communication technologies for connecting cameras and displays used to be such application specific technologies. Often seen as private Point-to-Point (P2P) links

XIV

Preface

with limited numbers per car that are/were supplied in closed systems using proprietary (if not analogue) communication technologies, there has been little incentive for communal efforts in the industry; up to now. We see various reasons, why it is high time to take responsibility and to broaden the knowledge base in the industry. 1. The number of cameras and displays inside cars is increasing, as is the number of communication links connecting them. 2. The importance and safety criticality of the camera and display applications is increasing. Reliability requirements are higher for a camera image used for an autonomous driving function than for one used during a low-speed parking maneuver. Digital ­instrument cluster or wing mirror replacement displays are more safety critical than a display showing comparably slow changing map data. 3. The increasing data rates for camera and display links mean increased technical challenges in form of lower Signal-to-Noise-Ratio (SNR) margins and higher sensitivity to link impairments. This requires more specific knowledge on how to make it work. ­Additionally, not only cameras and displays are aiming for higher resolutions. Higher data rates are also in discussion for various types of sensors. 4. Responsibilities are shifting. Car manufacturers are starting to buy cameras and displays from different Tier 1 suppliers. With that, the systems are no longer closed and the responsibility for the communication technology moves from the Tier 1s to the car manufacturers. 5. EE-architectures are changing. Car manufacturers are exploring zonal architectures, which of now, exclude camera and display data transmission for the lack of sufficient data rates supported by suitable communication technologies. New technical developments for Automotive SerDes and Ethernet allow for architecture options with fewer restrictions. 6. The boundaries between Automotive SerDes and Automotive Ethernet are blurring. For future architectures, both technologies support enough data rate. With the right IC product designs, future SerDes can integrate into an Ethernet network and Ethernet can address camera and display applications. How to efficiently explore this, when both technologies are handled in different departments? 7. Automotive SerDes is being standardized and now actually provides an official framework for the respective work in the industry. These are good reasons, why some car manufacturers have already moved the responsibility for the camera and display links to the central in-vehicle communication technologies departments. In our case, for example, some of the responsibility for SerDes was moved to us, the authors of this book, as early as 2015. Since then, we have investigated, learned, driven, collected, and are now eager to share. This book is the result. It keeps it technical. We intend this book to support beginners as well as experts at all stages of the value chain in gaining a comprehensive overview on the High-Speed (HS) sensor and display communication technologies Automotive SerDes and HS Automotive Ethernet. We believe in sound technical reasoning and would like to support all interested parties in drawing their own conclusions.

Preface

This is the first edition of a new book with lots of new content. It would not have been ­possible to complete it in the same quality without the many colleagues who answered all the smaller or larger questions we had. We would like to thank (in alphabetical order) Heather Babcock (TI), Kristian Baumann (BMW), Bert Bergner (TE), Andreas Brösse (BMW), Vijay Ceekala (TI), Jim Conder (Socionext), Kamal Dalmia (Aviva Links), Mario Heid (­Omnivision), Stefan Holzknecht (BMW), Kilian Jacob (BMW), Ariel Lasry (Qualcomm), Balagopal Mayampurath (ADI), Andy McLean (ADI), Chanakya Metha (TI), Thorsten Meyer (Valeo), Roland Neumann (Inova), Takashi Nishimura (SONY), Jochen Schyma (NXP), ­Anton Sifferlinger (BMW), Luisma Torres (KDPOF), Dirk Waldhauser (BMW), Rick Wietfeldt (­Qualcomm), Conrad Zerna (Aviva Links), and George Zimmerman (CME Consulting). A very special thanks goes to Daniel Hopf (Continental) who reviewed and annotated the complete book and who made it more consistent and precise with his effort. We would also like to thank BMW for giving us the opportunity to make a difference. June 2022, Michael Kaindl and Kirsten Matheus

XV

Timeline

1886

Carl Benz is granted a patent on “Fahrzeug mit Gasmotorenbetrieb (engl.: gas powered vehicle)” [1] and starts building identical copies of cars. While several motor vehicles have been built prior to this and successful commercial exploitation still needs some years to come [2], 1886 can be seen as the ­starting year for commercial automobile production. Tachometers had been invented in 1817 and were first used in trains in 1840. It is unclear when they were first used in cars [3].

1892

First ever law on ElectroMagnetic Compatibility (EMC) is passed in Germany in the context of the upcoming telegraph and telephone business [4].

1902

German engineer Otto Schulze patents a technology using a magnet created eddy-current that translates the speed of rotation of the wheels to a dial. Until  well into the 1980s, almost all speedometers in cars were based on this ­technology. Speedometers are standard equipment in cars by 1910 [5].

1904

First patent on radar technology is filed at the German patent office for “a method to notify of the presence of metallic objects with help of electromagnetic waves” which can determine the objects distance [6].

1908

The first car produced in a moving assembly line is the Ford Model T [7].

1917

Invention of the fuel gauge [8].

1927

Germany passes the first law on the use and installation of high frequency radio transmitters, which, with adaptations, is in place until 1995 [4].

1930

Sale start of first commercially successful in-car radio [9].

1931

The Commission Internationale de l’Éclairage (CIE) defines with the Red Green Blue (RGB) color space the first quantitative relationship between the distribution of wavelengths in the visible spectrum and perceived colors [10].

1933

Founding of the Comité International Spécial des Perturbations Radioélectriques (CISPR) in order to develop guidelines on EMC in Europe [4].

1952

First sale of commercial Frequency Modulation (FM) in-car radios. Amplitude Modulation (AM) is dominantly used in the market at the time [9].

Timeline

1956

Advent of the first fully automated mobile telephone system, allowing making and receiving calls in cars using the public telephone network [11].

1956 Jan.

First showcase of a backup camera is presented at the General Motors ­Motorama in a Buick Centurion concept car [12].

1958 Dec.

Publication on laser marks its invention [13].

1973

Year of the invention of Ethernet. Ethernet is demonstrated for the first time at Xerox PARC in order to enable the transmission of data between Xerox’s ­personal computer workstations and laser printers [14] [15].

1973

The International Electrotechnical Commission (IEC) creates a special technical committee to specify EMC for different fields of use [4].

1974

First Charge-Coupled Device (CCD) image sensor goes into production [16].

1974 Dec.

First release of the “Specification of Internet Transmission Control Protocol (TCP)” [17].

1976 Sep. 9

The president of JVC presents the Video Home System (VHS) [18]. Other ­markets outside Japan receive the first products from 1977 on [19].

1979

Aston Martin presents its Lagonda with an elaborate array of LED screens [20].

1979 Jun.

The 7-layer Open Systems Interconnection (OSI) model is published at the International Organization for Standardization (ISO) [21]. The respective ­committee was formed in 1977 [22].

1980 Dec.

The Institute of Electrical and Electronics Engineers (IEEE) starts the 802.3 working group dedicated to CSMA/CD (Ethernet) [23].

1981

Release of the first commercially available in-car navigation system by Honda called “Electro Gyro-Cator” that provided guidance by tracking the distance and direction travelled from the start point [24].

1982

Philips Semiconductors (now NXP Semiconductors) develops the Inter-IC bus (I2C) [25].

1982 Jan. 26

As the first car manufacturer, Toyota offers a sonar-based backup parking ­system in a series production car [26].

1982 Oct.

First commercial CD-player is sold in Japan (by Sony) [27].

1983

Introduction of the analogue content protection technology from M ­ acrovision for VHS video cassettes [28].

1985

First factory-installed in-car CD player [9].

1986

Kodak develops the first digital camera to record 1.4 MPixels. It uses a CCD imager [29].

1986

The Buick Riviera is likely the first series production car with a touch screen [20].

1986 Feb.

First release of the Philips Semiconductors’ (now NXP Semiconductors’) I2S audio bus interface specification [30].

XVII

XVIII

Timeline

1987

Toyota sells its Royal Crown model with a color display for its CD-based ­navigation system [20].

1988

Establishment of the Moving Picture Experts Group (MPEG) for the development of standards for the coded representation of media data such as audio and video [31].

1988 Nov.

The Video Electronics Standards Association (today only using its abbreviation “VESA”) is founded on the initiative of NEC in order to standardize video display interfaces [32]. The organization is incorporated in July 1989 [33].

1989 Oct.

The TCP/IP Internet Protocol Suite is being published as “Requirements for Internet Hosts – Communication Layers”, RFC 1122 [34] and “­Requirements for Internet Hosts – Application and Support”, RFC 1123 [35].

1989/90

The World Wide Web (www) is invented at CERN [36].

1990

Development of the CMOS active pixel sensor [37].

1990

Mazda introduces its Eunos Cosmo with an in-dashboard color display as the first GPS-based navigation system [20].

1990 Sep.

IEEE 802.3 ratifies the Ethernet specification 10BASE-T [15], with which Ethernet allegedly won the battle over competing technologies [14].

1992

The first “smart” mobile phone with a touch screen, the IBM Simon, is commercially sold [38].

1992 Sep. 18 The International Telecommunication Union (ITU) releases the Recommendation T.81 for the Joint Photographic Expert Group (JPEG) compression format [39]. 1994 Jun.

First release of the AEC-Q100 specification on automotive quality for integrated circuits at the Automotive Electronic Council (AEC) [40].

1994

National Semiconductor (now TI) introduces the Low Voltage Differential ­Signaling (LVDS) technology [41], which is subsequently published as ANSI/TIA/EIA-644-1995 [42] and as IEEE 1596.3 in July 1996 [43]. The data rate the standard originally supports is 655 Mbps.

1995

The ISO/IEC publishes a backwards compatible MPEG-2 Audio specification (MPEG-2 Part 3) – commonly referred to as MP3 – with additional bit and ­sample rates [44].

1995 Dec. 8

Toshiba, Matsushita, Sony, Philips, Time Warner, Pioneer, JVC, Hitachi, and ­Mitsubishi Electric announce their agreement on a unified DVD format [45].

1996

The ISO/IEC publishes the MPEG-2 video (MPEG-2 Part 2) specification, which is used among other, for the DVD standard. The ITU publishes it as H.262 [46].

1996

National Semiconductor (now TI) develops the first FPD-link specification, which it publishes in order to achieve a large market acceptance [47].

Timeline

1996 Nov. 5

Hewlett-Packard and Microsoft propose the standard RGB (sRGB) color space for monitors, printers, and the www [48]. In 1999, the IEC published it as IEC 61966-2-1:1999 [49].

1997

Publication of IEEE 802.3x, which supports full-duplex operation for Ethernet [50].

1998

First Publication of IEEE 802.1Q, which adds – among other functions – the option of eight priority queues and Virtual LANs (VLANs) to Ethernet commu­nication [51].

1998

Daimler introduces the first radar based adaptive speed driver assist system into the market [52].

1998 Oct. 28

U.S. president Bill Clinton signs the Digital Millennium Copyright Act (DMCA), which provides the basis for the prosecution of copyright infringements on the Internet. It is subsequently adopted similarly in other countries and regions [53].

1999 May

Napster launches its “share it with all for free” platform. This is possible because of the combination of Internet and audio compression standards. It irreversibly changed the media industry and media consumption. It lasted until February 2001 [54].

1999 May

The first phone with an integrated camera, the Kyocera VP-210, is commercially sold to the general public [55].

1999 Apr. 2

Release of the Digital Visual Interface (DVI) by the Digital Display Working Group (DDWG), which focusses on providing a standardized connection between a computer and a displaying device [56].

1999 May 13 National Semiconductor (now TI) releases the Open LVDS Display Interface (­OpenLDI) Specification v. 0.95 as an open standard to complete the digital connection between video sources and displaying devices [57] as initiated with the LVDS technology. 2000

The Nissan Infinity Q45 is offering a series production rear view/backup ­camera. This is said to have initiated the backup camera market [12].

2000 Feb. 17 Intel releases version 1.0 of the High-bandwidth Digital Content Protection (HDCP) specification, which targets at preventing the recording and distribution of HD video content [58]. In the coming years, its support is mandated by many content providers. 2001 Nov.

Start of Production (SOP) of the BMW 7 series using a central, dashboard-­ mounted display for user information and interaction (plus “iDrive”) [59]. Often, credit is given to BMW for initiating that such a screen as a central hub for car interaction has become a standard feature [60]. The same car is also the first with a digital video link to connect a display: The FPD-link is used for connecting the Rear Seat Entertainment (RSE) display.

2002 Dec. 9

Announcement of the HDMI 1.0 connectivity standard by the seven founding members Hitachi, Matsushita, Philips, Silicon Image, Sony, Thomson, and Toshiba [61] (now Lattice, Maxell, Panasonic, Philips, Sony, Technicolor, and Toshiba [62]).

XIX

XX

Timeline

2003

First publication of ISO/IEC 14496-10, also known as MPEG-4 Advanced Video Codec (AVC) or ITU H.264 [63].

2003 Jul.

ARM, Nokia, STMicroelectronics, and Texas Instruments (TI) found the Mobile Industry Processor Interface (today only using its abbreviation “MIPI”) ­Alliance to define standards for cell phones that at the same time reduce ­complexity and costs while allowing flexibility [64].

2003 Aug. 4

As the first region in Germany, Berlin ends its analogue terrestrial TV broadcast in favor of digital broadcast with Digital Video Broadcasting Terrestrial (DVB-T) [65]. This is an exemplary date for a worldwide transition. By 2017, worldwide digital terrestrial TV broadcast was split over four major technologies: DVB-T2, Integrated Services Digital Broadcasting (ISDB), Advanced ­Television Systems Committee (ATSC), or Digital Terrestrial Media Broadcast (DTMB) [66].

2004 Jul.

IEEE kicks off the development of standards that allow adding Quality of ­Service (QoS) functions to Ethernet [67]. This is first called Audio Video ­Bridging (AVB), shifted to IEEE 802.1 in 2005 [68], and, in 2012, renamed Time Sensitive Networking (TSN) [69].

2005

First release of the MIPI CSI-2 and DSI-2 specifications [70] [71].

2006 May

Publication of the first VESA DisplayPort specification v1.0 [72] [73]. The first embedded Display Port (eDP) specification is published in December 2008 [74].

2007

Nissan introduces the first surround view camera system with its Infiniti EX35 [75].

2008 Oct.

SOP of the first series production car deploying Ethernet as a communication technology. The BMW 7 series uses 100BASE-TX Ethernet as a diag­ nostic interface with Unshielded Twisted Pair (UTP) cabling and for the communication between HeadUnit (HU) and Rear Seat Entertainment (RSE) with Shielded Twisted Pair (STP) cabling [76].

2010

IEEE 802.3 releases the 802.3az specification on Energy Efficient Ethernet (EEE) [50]. This is an important step to saving energy in a switched Ethernet system. Much later, applied separately to each communication direction, it is especially useful in case of highly asymmetric communication.

2011 Oct.

The HDMI founders create the HDMI Forum in order to allow all interested companies to be an integral part of the development process [77].

2013 Sep.

SOP of the first series production car deploying “Automotive Ethernet”. The BMW X5 uses BroadR-Reach with single UTP for connecting the cameras to the surround view system [76]. The technology is ratified by the IEEE as IEEE 802.3bw/100BASE-T1 Ethernet on October 26, 2015 [78].

2014 Sep. 30 Publication date of the ISO 17215 specification series “Road Vehicles  – Video Communication Interface for Cameras (VCIC)” defining Ethernet as the ­communication interface [79].

Timeline

2015 Dec. 12 The so-called Paris Agreement is adopted by the United Nations Framework Convention on Climate Change (UNFCCC). Its goal to limit the global warming to below 2 °C (ideally to 1.5 °C) above preindustrial level [80] leads to ­stringent CO2 targets for the car industry. 2016 Nov. 10 Call For Interest (CFI) is presented and approved that initiates the efforts to standardize what later becomes MultiGBASE-T1 Ethernet for 2.5, 5, and 10 Gbps data rate in automotive environments at IEEE 802.3 [81]. 2016 Dec.

IEEE 802.3 releases the IEEE 802.3bu specification on Power over Data Line (PoDL) [50]. While this specification targets the single pair 100 Mbps and 1 Gbps Automotive Ethernet technologies, it set an important starting point for higher data rates.

2017 Jan.

The MIPI Alliance concludes its I3C specification v1.0. A public version is released in December of the same year [82].

2018

The Audi A8 is the first series production car with a Lidar [83]. However, after having subsequently removed it [84], the XPeng P5 might have rightly claimed to be once more the first in December 2021 [85].

2018 Aug. 2

The MIPI Alliance announces the standardization of their A-PHY [86].

2019 May

Founding of the Automotive SerDes Alliance (ASA) [87].

2019 July

CFI for an automotive suitable multi-Gbps optical Ethernet PHY technology is presented and accepted at IEEE 802.3 [88].

2020 Jun

IEEE 802.3 releases the IEEE 802.3ch/MultiGBASE-T1 specification for 2.5, 5, and 10 Gbps transmission over a single twisted pair in an automotive ­environment [50]. While the specification allows for symmetric data rates only, the EEE function may be activated individually per direction.

2020 Jun. 24 The “Greater than 10 Gb/s Electrical Automotive Ethernet PHYs Task Force (TF)” holds its first meeting at IEEE 802.3 [89]. Being able to use the technology asymmetrically is one of the properties discussed but adhered to only by applying EEE asymmetrically. 2020 Jul. 14

The Multi-Gigabit Optical Automotive Ethernet TF holds its first TF meeting at IEEE 802.3 [90].

2020 Sep. 15 MIPI announces the release of their MIPI A-PHY specification 1.0 [91]. 2020 Oct. 13

The Automotive SerDes Alliance announces the finalization of their ASA ­Motion Link specification 1.01 [87].

XXI

XXII

Timeline

Bibliography [1]

C. Benz, “Fahrzeug mit Gasmotorenbetrieb”. Mannheim Patent 37435, 19 January 1886.

[2]

Daimler, “1886–1920. Anfänge des Automobils,” not known. [Online]. Available: https://www. daimler.com/konzern/tradition/geschichte/1886-1920.html. [Accessed 26 January 2021].

[3]

P. Fears, “The History of the Tachometer,” Cearbont Automotive Instruments, 17 February 2017. [Online]. Available: https://www.caigauge.com/blog/the-history-of-the-­tachometer. [Accessed 12 March 2021].

[4]

D. E. Möhr, “Was ist eigentlich EMV? - Eine Definition,” not known. [Online]. Available: http:// www.emtest.de/de/what_is/emv-emc-basics.php (no longer available). [Accessed 6 May 2020].

[5]

N. Berg, “Great Inventions: The Speedometer,” Hagerty, 6 July 2020. [Online]. Available: https:// www.hagerty.com/media/automotive-history/great-inventions-the-speedometer/. [Accessed 2 April 2021].

[6]

W. Holpp, “Geschichte des Radars,” 2004. [Online]. Available: https://www.100-jahre-radar.fraun hofer.de/index.html?/content_gdr1.html. [Accessed 2 February 2022].

[7]

History.com, “Model T,” 2 May 2019. [Online]. Available: https://www.history.com/­topics/inventions/model-t. [Accessed 7 February 2021].

[8]

K. Kupchik, “10 Inventions That Changed Cars Forever,” 2 July 2014. [Online]. Available: https:// www.zaloomsautorepair.com/blog/10-inventions-that-changed-cars-forever. [Accessed 3 April 2021].

[9]

J. Berkowitz, “The History of Car Radios,” Car And Driver, 25 October 2010. [Online]. Available: https://www.caranddriver.com/features/a15128476/the-history-of-car-­radios/. [Accessed 7 February 2021].

[10]

T. Smith and J. Guild, “The C. I.E Colorimetric Standards and their Use,” Transactions of the Optical Society, vol. 33, no. 3, pp. 73–134, 1931.

[11]

Not known, “Facts about the Mobile. A Journey through Time.,” about 2007. [Online]. Available: https://web.archive.org/web/20100813122017/; http://www.mobilen50ar.se/eng/FaktabladENGFi nal.pdf. [Accessed 7 February 2021].

[12]

N. Andreev, “A Brief History of Car Parking Technology,” Confused.com, 13 June 2018. [Online]. Available: https://www.confused.com/on-the-road/gadgets-tech/parking-technology-brief-history# (no longer available). [Accessed 13 March 2021].

[13]

A. Chodos and J. Ouellette, “December 1958: Invention of the Laser,” December 2003. [Online]. Available: https://www.aps.org/publications/apsnews/200312/history.cfm. [Accessed 3 February 2022].

[14]

R. M. Metcalfe, “The History of Ethernet,” 14 December 2006. [Online]. Available: https://www. youtube.com/watch?v=g5MezxMcRmk. [Accessed 6 May 2020].

[15]

C. E. Surgeon, Ethernet: The Definite Guide, Sebastopol, CA: O’Reilly, 2000, February.

[16]

Wikipedia, “Charge-Coupled Device,” 9 August 2021. [Online]. Available: https://en.wikipedia. org/wiki/Charge-coupled_device. [Accessed 22 October 2021].

[17]

Wikipedia, “Internet Protocol Suite,” 30 June 2013. [Online]. Available: http://en.wikipedia.org/ wiki/Internet_protocol_suite. [Accessed 5 July 2013].

[18]

Museum of Arts and Design, “Film Series/VHS,” 2012. [Online]. Available: https://www.mad museum.org/series/vhs. [Accessed 15 March 2022].

[19]

total rewind, “JVC HR-3300,” 1978. [Online]. Available: https://www.totalrewind.org/vhs/H_3300. htm. [Accessed 15 March 2022].

Timeline

[20]

M. Bubbers, “Shift From Huge In-car Screens May be Under Way -- but First They’ll get Bigger,” The Globe and Mail, 2 May 2018. [Online]. Available: https://www.theglobeandmail.com/drive/ culture/article-shift-from-huge-in-car-screens-may-be-under-way-but-first-theyll/. [Accessed 12 March 2021].

[21]

W. Stallings, “The Origin of OSI,” 1998. [Online]. Available: http://williamstallings.com/Extras/ OSI.html. [Accessed 6 May 2020].

[22]

A. L. Russel, “OSI: The Internet That Wasn’t,” 30 July 2013. [Online]. Available: http://spectrum. ieee.org/computing/networks/osi-the-internet-that-wasnt. [Accessed 6 May 2020].

[23]

U. v. Burg and M. Kenny, “Sponsors, Communities, and Standards: Ethernet vs. Token Ring in the Local Area Networking Business,” Industry and Innovation, vol. 10, no. 4, pp. 351–375, December 2003.

[24]

CNN, “Japanese Inventions that Changed the Way we Live,” 30 June 2017. [Online]. Available: https://edition.cnn.com/2017/06/13/world/gallery/japanese-inventions-changed-how-we-live/ index.html. [Accessed 13 March 2021].

[25]

Arrow, “What is I2C? How Inter-Integrated Circuits Work,” 4 February 2019. [Online]. Available: https://www.arrow.com/en/research-and-events/articles/what-is-i2c-how-inter-integrated-circuitswork. [Accessed 15 May 2022].

[26]

Toyota, “Toyota Corona Undergoes full Model Change,” 26 January 1982. [Online]. Available: https://global.toyota/en/detail/7675511. [Accessed 1 February 2022].

[27]

Sony, “Product & Technology Milestones, Home Audio,” 2022, continuously updated. [Online]. Available: https://www.sony.com/en/SonyInfo/CorporateInfo/History/­sonyhistory-a.html. [Accessed 15 March 2022].

[28]

Wikipedia, “Macrovision,” 17 Oktober 2019. [Online]. Available: https://de.wikipedia.org/wiki/ Macrovision. [Accessed 2 November 2021].

[29]

M. Bellis, “The History of the Digital Camera,” ThoughtCo, 29 March 2020. [Online]. Available: https://www.thoughtco.com/history-of-the-digital-camera-4070938. [Accessed 14 March 2021].

[30]

Philips Semiconductors, “I2S specification,” 5 June 1996. [Online]. Available: https://web.archive. org/web/20070102004400/http://www.nxp.com/acrobat_download/various/I2SBUS.pdf.  [Accessed 18 April 2022].

[31]

MPEG, “MPEG,” 2022, continuously updated. [Online]. Available: https://www.mpeg.org/. [Accessed 16 March 2022].

[32]

M. Brownstein, “NEC forms Video Standards Group,” 14 November 1988. [Online]. Available: https://books.google.de/books?id=wTsEAAAAMBAJ&pg=PT2. [Accessed 20 March 2022].

[33]

Wikipedia, “Video Electronics Standards Association,” 29 April 2022. [Online]. ­Available: https:// en.wikipedia.org/wiki/Video_Electronics_Standards_Association. [Accessed 30 April 2022].

[34]

R. Braden, “RFC 1122,” October 1989. [Online]. Available: http://tools.ietf.org/pdf/rfc1122.pdf. [Accessed 23 June 2013].

[35]

R. Braden, “RFC 1123,” October 1989. [Online]. Available: tools.ietf.org/pdf/rfc1123.pdf. [Accessed 26 June 2013].

[36]

P. Barford, “The World Wide Web,” University of Wisconsin, 11 September 2008. [Online]. Available: https://pages.cs.wisc.edu/~pb/640/web.ppt. [Accessed 6 May 2020].

[37]

Wikipedia, “Active Pixel Sensor,” 22 September 2021. [Online]. Available: https://de.wikipedia. org/wiki/Active_Pixel_Sensor. [Accessed 17 October 2021].

XXIII

XXIV

Timeline

[38]

Georgia Southern University, Colleague of Arts and Humanities, “Future Phonics, H ­ istory of Phones,” (not known). [Online]. Available: https://georgiasouthern.libguides.com/c.php?g= 612229&p=4545365. [Accessed 16 April 2022].

[39]

Joint Photographic Experts Group, “Information technology – Digital Compression and Coding of Continuous-tone Still Images – Requirements and Guidelines,” International Telecommunication Union, Geneva, 1992.

[40]

Automotive Electronics Council, “AEC History,” not known. [Online]. Available: http://www.ae council.com/AECHistory.html. [Accessed 6 May 2020].

[41]

M. Defossez, “Xilinx D-PHY Solutions,” 1 February 2021. [Online]. Available: https://docs.xilinx. com/v/u/en-US/xapp894-d-phy-solutions. [Accessed 2022 18 April].

[42]

L. Davis, “RS-644 Bus,” 1998. [Online]. Available: http://www.interfacebus.com/Design_Connec tor_RS644.html. [Accessed 30 December 2021].

[43]

IEEE, “1596.3-1996  – IEEE Standard for Low-Voltage Differential Signals (LVDS) for Scalable ­Coherent Interface (SCI),” IEEE, New York, 1996.

[44]

ISO, “ISO 13818-3:1995 – Information Technology - Generic Coding of Moving Pictures and Associated Audio Information – Part 3: Audio,” ISO, Geneva, 1995.

[45]

Toshiba, “DVD Format Unification,” 8 December 1995. [Online]. Available: https://web.archive. org/web/19970501192002/http://www.toshiba.co.jp/about/press/1995_12/pr0802.htm. [Accessed 15 March 2022].

[46]

Wikipedia, “MPEG-2,” 17 February 2022. [Online]. Available: https://en.wikipedia.org/wiki/ MPEG-2. [Accessed 15 March 2022].

[47]

Wikipedia, “FPD-Link,” 9 December 2020. [Online]. Available: https://en.wikipedia.org/wiki/ FPD-Link. [Accessed 27 March 2021].

[48]

M. Stokes, M. Anderson, S. Chandrasekar and R. Motta, “A Standard Default Color Space for the Internet  – sRGB,” 5 November 1996. [Online]. Available: https://www.w3.org/Graphics/Color/ sRGB.html. [Accessed 13 March 2022].

[49]

International Electrotechnical Commission, “IEC 61966-2-1:1999 Multimedia Systems and Equipment - Colour Measurement and Management – Part 2-1: Colour Management – Default RGB Colour Space – sRGB,” International Electrotechnical Commission, Geneva, 1999.

[50]

Wikipedia, “IEEE 802.3,” 29 December 2020. [Online]. Available: https://en.wikipedia.org/wiki/ IEEE_802.3. [Accessed 7 February 2021].

[51]

IEEE 802.1, “802.1Q- IEEE Standards for Local and Metropolitan Area Networks: Virtual Bridged Local Area Networks,” IEEE Communication Society, New York, 1998.

[52]

H. H. Meinel, “Evolving Automotive Radar – from the very Beginnings into the Future,” in: The 8th European Conference on Antennas and Propagation, The Hague, 2014.

[53]

Wikipedia, “Digital Millenium Copyright Act,” 28 March 2022. [Online]. Available: https:// en.wikipedia.org/wiki/Digital_Millennium_Copyright_Act. [Accessed 26 June 2022].

[54]

T. Lamont, “Napster: the Day the Music Was Set Free,” 24 February 2013. [Online]. Available: https://www.theguardian.com/music/2013/feb/24/napster-music-free-file-sharing.  [Accessed 6 May 2020].

[55]

J. Callaham, “The First Camera Phone was Sold 21 Years Ago, and it’s not What You Might ­Expect,” Android Authority, 1 June 2021. [Online]. Available: https://www.androidauthority.com/ first-camera-phone-anniversary-993492/. [Accessed 14 March 2021].

Timeline

[56]

Digital Display Working Group, “Digital Visual Interface DVI,” 2 April 1999. [Online]. Available: https://web.archive.org/web/20120813201146/http://www.ddwg.org/lib/dvi_10.pdf.  [Accessed 20 April 2022].

[57]

National Semiconductors, “Open LVDS Display Interface (OpenLDI) specification v0.95,” 13 May 1999. [Online]. Available: https://glenwing.github.io/docs/OpenLDI-0.95.pdf. [Accessed 16 April 2021].

[58]

Wikipedia, “High-bandwidth Digital Content Protection,” 12 June 2022. [Online]. ­Available: https://en.wikipedia.org/wiki/High-bandwidth_Digital_Content_Protection. [Accessed 27 June 2022].

[59]

7-forum.com, “Das iDrive System im neuen 7er,” 7 March 2008. [Online]. Available: https://www. 7-forum.com/modelle/e65/idrive.php. [Accessed 12 March 2021].

[60]

D. Homer, “The First Car With a Touchscreen Came Out in the ’80s,” MotorBiscuit, 1 January 2021. [Online]. Available: https://www.motorbiscuit.com/the-first-car-with-a-touchscreen-came-outin-the-80s/. [Accessed 12 March 2021].

[61]

R. L. Maestra, “HDMI - A Digital Interface Solution,” 26 July 2006. [Online]. Available: https:// web.archive.org/web/20160530220657/http://www.hdtvmagazine.com/articles/2006/07/hdmi_ part_1_-_a.php. [Accessed 27 December 2021].

[62]

HDMI, “HDMI Founders,” 2021. [Online]. Available: https://www.hdmi.org/adopter/founders. [Accessed 27 December 2021].

[63]

ISO/IEC, “ISO/IEC 14496-10:2004 Information Technology  – Coding of Audio Visual Objects  – Part 10: Advanced Video Coding,” 2004. [Online]. Available: https://www.iso.org/standard/40890. html. [Accessed 18 March 2022].

[64]

R. Merritt, “Mobile Chip Interface Gets Real,” 13 February 2006. [Online]. Available: https:// www.eetimes.com/mobile-chip-interface-gets-real/. [Accessed 28 November 2021].

[65]

N. Jurran, “Berlin: Die Antenne ist tot, es lebe die Antenne!,” 4 August 2003. [Online]. Available: https://www.heise.de/newsticker/meldung/Berlin-Die-Antenne-ist-tot-es-lebe-die-Antenne-82989. html. [Accessed 1 January 2022].

[66]

R. A. Trappe, “DTV Status,” 26 June 2017. [Online]. Available: http://es.dtvstatus.net/. [Accessed 04 January 2022].

[67]

R. Brand, S. Carlson, J. Gildred, S. Lim, D. Cavendish and O. Haran, “Residential Ethernet, IEEE 802.3 Call for Interest,” July 2004. [Online]. Available: https://grouper.ieee.org/groups/802/3/ re_study/public/200407/cfi_0704_1.pdf. [Accessed 6 May 2020].

[68]

IEEE 802.3, “IEEE 802.3 Residential Ethernet Study Group Homepage,” 10 January 2006. [­Online]. Available: https://grouper.ieee.org/groups/802/3/re_study/. [Accessed 6 May 2020].

[69]

IEEE 802.1, “802.1 Plenary –11/2012 San Antonio Closing,” November 2012. [Online]. Available: https://www.ieee802.org/1/files/public/minutes/2012-11-closing-plenary-slides.pdf.  [Accessed 6 May 2020].

[70]

Wikipedia, “Camera Serial Interface,” 8 December 2021. [Online]. Available: https://en.wiki pedia.org/wiki/Camera_Serial_Interface. [Accessed 14 April 2022].

[71]

MIPI Alliance, “An Overview of MIPI’s Standardized In-vehicle Connectivity Framework for High-Performance Sensors and Displays,” August 2021. [Online]. Available: https://www.mipi. org/mipi-white-paper-introductory-guide-mipi-automotive-serdes-­solutions-mass. [Accessed 14 April 2022].

[72]

Hewlett Packard, “An Overview of Current Display Interfaces,” March 2011. [Online]. Available: http://h10032.www1.hp.com/ctg/Manual/c01285675. [Accessed 28 December 2021].

XXV

XXVI

Timeline

[73]

Wikipedia, “DisplayPort,” 12 March 2022. [Online]. Available: https://en.wikipedia.org/wiki/­ DisplayPort#Display_Stream_Compression. [Accessed 20 March 2022].

[74]

C. Wiley, “eDP Embedded DisplayPort The New Generation Digital Display Interface for Embedded Applications,” 6 December 2010. [Online]. Available: https://www.vesa.org/wp-content/­ uploads/2010/12/DisplayPort-DevCon-Presentation-eDP-Dec-2010-v3.pdf. [Accessed 7 June 2022].

[75]

M. Jerome, “Nissan to Bring Around-View Monitor to Infiniti EX35,” Wired, 16 October 2007. [Online]. Available: https://www.wired.com/2007/10/nissan-to-bring/. [Accessed 3 April 2021].

[76]

K. Matheus and T. Königseder, Automotive Ethernet, Third Edition, Cambridge: C ­ ambidge University Press, 2021.

[77]

HDMI Forum, “Questions about the HDMI Forum: Why Create an Open Organization Now,” 2021. [Online]. Available: https://hdmiforum.org/about/faq/. [Accessed 27 December 2021].

[78]

IEEE Computer Society, “802.3bw-2015 – IEEE Standard for Ethernet Amendment 1 Physical Layer Specifications and Management Parameters for 100 Mb/s Operation over a Single Balanced Twisted Pair Cable (100BASE-T1),” IEEE-SA, New York, 2015.

[79]

ISO, “ISO 17215-(1-4):2014 Road Vehicles – Video Communication Interface for C ­ ameras (VCIC) Part 1–4,” ISO, Geneva, 2014.

[80]

Wikipedia, “Paris Agreement,” 30 April 2020. [Online]. Available: https://en.wikipedia.org/wiki/ Paris_Agreement. [Accessed 9 May 2020].

[81]

S. Carlson, H. Zinner, K. Matheus, N. Wienckowski and T. Hogenmüller, “CFI Multi-Gig Automotive Ethernet PHY,” 9 November 2016. [Online]. Available: https://www.ieee802.org/3/ad_hoc/ ngrates/public/16_11/20161108_CFI.pdf. [Accessed 6 May 2020].

[82]

R. Wilson, “MIPI Makes Market Push for I3C Sensor Interface,” Electronics Weekly, 14 December 2017. [Online]. Available: https://www.electronicsweekly.com/news/mipi-makes-market-push-i3csensor-interface-2017-12/. [Accessed 15 May 2022].

[83]

P. E. Ross, “The Audi A8: the World’s First Production Car to Achieve Level 3 Autonomy,” IEEE Spectrum, 11 July 2017. [Online]. Available: https://spectrum.ieee.org/the-audi-a8-the-worlds-firstproduction-car-to-achieve-level-3-autonomy. [Accessed 7 June 2022].

[84]

F. Greis, “1000 Meilen – wenig Säulen, Lidar wird nicht mehr benötigt,” golem.de, 5 September 2019. [Online]. Available: https://www.golem.de/news/langstreckentest-im-audi-e-tron-1-000-meilen -wenig-saeulen-1909-143640-5.html. [Accessed 25 March 2022].

[85]

Pandaily, “XPeng P5, the World’s First Production Lidar Car, Rolls off Assembly Line,” 20 October 2021. [Online]. Available: https://pandaily.com/xpeng-p5-the-worlds-first-production-lidar-carrolls-off-assembly-line/. [Accessed 3 February 2022].

[86]

MIPI Alliance, “MIPI Alliance to Advance Autonomous Driving, other Automotive Applications with New Data Interface Specifications at 12-24 Gbps and Beyond,” 2 August 2018. [Online]. Available: https://www.mipi.org/mipi-to-advance-autonomous-driving-­other-automotive-applications.  [Accessed 29 March 2021].

[87]

S. Brunner, “Automotive SerDes Alliance (ASA) Completes the First Automotive SerDes Standard with Integrated Security,” Automotive SerDes Alliance, 13 October 2020. [Online]. Available: https://auto-serdes.org/news/automotive-serdes-alliance-asa-­completes-the-first-automotive-serdesstandard-with-integrated-security-325/. [Accessed 28 March 2021].

[88]

C. Pardo, H. Goto, T. Nomura and B. Grow, “Automotive Optical Multi Gig Call For ­Interest Consensus Presentation,” July 2019. [Online]. Available: https://www.ieee802.org/3/cfi/0719_1/ CFI_01_0719.pdf. [Accessed 29 March 2021].

Timeline

[89]

IEEE 802.3, “Homepage of the IEEE 802.3 Greater than 10 Gb/s Electrical Automotive Ethernet PHYs Task Force,” 2020, continuously updated. [Online]. Available: https://www.ieee802.org/3/ cy/public/jun20/index.html. [Accessed 7 September 2020].

[90]

IEEE 802.3, “Homepage of the IEEE 802.3 Multi-Gigabit Optical Automotive Ethernet Task Force,” 2020, continuously updated. [Online]. Available: https://www.ieee802.org/3/cz/public/index. html. [Accessed 6 September 2020].

[91]

MIPI Alliance, “MIPI Alliance Releases A-PHY SerDes Interface for Automotive,” 15 September 2020. [Online]. Available: https://www.mipi.org/MIPI-Alliance-Releases-­A-PHY-SerDes-Interfacefor-Automotive. [Accessed 28 March 2021].

XXVII

Abbreviations and Glossary

Abbreviation

Explanation

1PPODL

One Pair Power Over Data Line

Study group name for IEEE 802.3bu

2D

2-Dimensional

3D

3-Dimensional

4PPoE

Four Pair Power over Ethernet

IEEE 802.3bt 2018, for cables ­consisting of four twisted pairs

8P8C

8 Positions 8 Contacts

Modular connector specified in IEC 60603-7

µ

micro

µC

MicroController

A

Ampere

A2B

Automotive Audio Bus

AC

Alternating Current

ACC

Adaptive Cruise Control

ACK

ACKnowledge

ACMD

A-PHY Control and Management ­Database

ACMP

A-PHY Control and Management ­Protocol

AD

Autonomous Driving

ADAS

Advanced Driver ASsist or Advanced Driver Assistance System

ADC

Analogue to Digital Converter or ­Conversion

ADI

Analog Devices

AEB

Automated Emergency Braking

AEC

Automotive Electronic Council

Standardization organization focusing on electronic part qualification for the automotive industry

AFDX

Avionics Full-Duplex Switched Ethernet

Ethernet protocol used in the ­aerospace industry

Communication interface from ADI

Abbreviations and Glossary

Abbreviation

Explanation

AGC

Adaptive Gain Control

AIAG

Automotive Industry Action Group

ALEI

Adaptation Layer Extended Information

Part of the IEEE 2977 DLL packet

ALSE

Absorber-Lined Shielded Enclosure

Described in ISO 11452-2

AM

Amplitude Modulation

Used for the reception of analogue radio in the short, medium and long wave bands

AMEC

Automatable Module Ethernet ­Connector

AML

ASA Motion Link

AMP

Amplifier

ANSI

American National Standards Institute

US SSO based in Washington D. C.

AOSC

Always-On Sentinel Conduit

Part of CSI-2 v4.0

APD

Avalanche Photo Diode

API

Application Programming Interface

APIX

Automotive PIXel link

APPI

A-PHY Protocol Interface

ARQ

Automatic Retransmission/Repeat reQuest

ASA

Automotive SerDes Alliance

Alliance for Automotive SerDes ­connectivity, home of the ASA Motion Link

ASAML

ASA Motion Link

Also AML

ASE

Application Stream Encapsulator

Part of the ASAML

ASEP

Application Stream Encapsulation ­Protocol

Protocol adaptation for the ASA Motion Link

A-shell

Automotive shell

Unified communication interface for the side-channel of APIX

ASIC

Application-Specific Integrated Circuit

ASIL

Automotive Safety Integrity Level

ASP

Abstract Service Primitive

ATCA

Advanced TeleComputing Architecture

I2C derivate

ATS

Asynchronous Traffic Shaping

Defined in IEEE 802.1Qcr-2020

ATSC

Advanced Television Systems ­Committee

US American set of digital television standards

AUTOSAR

AUTomotive Open System ­ARchitecture

AUTOSAR SecOC

AUTOSAR SECure Onboard ­Communication

AV

Audio/Video

Also ASAML

Inova’s name for their proprietary ­SerDes technology

Classification methodology for ­functional safety

XXIX

XXX

Abbreviations and Glossary

Abbreviation

Explanation

AVB

Audio Video Bridging

AVC

Advanced Video Coding

AVP

Autonomous Valet Parking

AWG

Arbitrary Waveform Generator or ­American Wire Gauge

AWGN

Additive White Gaussian Noise

B2B

Business-to-Business

BCI

Bulk Current Injection

BER

Bit Error Rate

B-frame

Bi-directional predictive coded picture or frame

Part of MPEG encoding

BGA

Ball Grid Array

Package type for semiconductors

BIST

Built-In Self-Test

BK

Binding Key

Part of the ASA security concept

BMCA

Best Master Clock Algorithm

Part of IEEE 802.1AS-2011

BNC

Bayonet Neill Concelman

Connector used also for CBVS video, named after their inventors

BOM

Bill of Material

bpp

bits per pixel

bps

bits per second

BSD

Blind Spot Detection

BTA

Bus TurnAround

B/W

Black & White

CAD

Command-Address-Data

CAN

Controller Area Network

CAN FD

CAN Flexible Data rate

CAT

CATegory

Used for data center cable standards

CBS

Credit Based Shaper

Used with IEEE 802.1Qav 2009

CCC

Capacitive Coupling Clamp method

For testing resistance to fast transient pulses

CCD

Charge-Coupled Device

Imager technology

CCS

Camera Command Set

Part of the MIPI CSI-2 interface ­building blocks

CD

Compact Disc or Collision/Contention Detection

CDE

Cable Discharge Event

Type of ESD test

CDM

Charged Device Model

Type of ESD test

CE

Consumer Electronics

CEC

Consumer Electronics Control

Part of MIPI C- and D-PHY Part of the ASAML OAM

Part of HDMI supporting remote ­control of HDMI connected display devices

Abbreviations and Glossary

Abbreviation

Explanation

CERN

Conseil Européen pour la Recherche Nucléaire (European Council for Nuclear Research)

CFA

Color Filter Array

CFI

Call For Interest

Project initiation item at IEEE 802.3

CFS

Clock Forwarding Service

Part of IEEE 2977

CiA

CAN in Automation

Organization that drives the CAN specification development for the automotive industry

CIA

Confidentiality, Integrity, and ­Availability

CIE

Commission Internationale de l’Éclairage (engl. International ­Commission on Illumination)

CIS

CMOS Image Sensor

CISPR

Comité International Spécial des ­Perturbations Radioélectriques

CLK

CLocK

CTL

ConTroL

CTLE

Continuous-Time Linear Equalizer

CMC

Common Mode Choke

CMD

Command

CML

Current Mode Logic or Channel ­Monitor Loop

CMOS

Complementary Metal-Oxide ­Semiconductor

CMYK

Cyan Magenta Yellow blacK

International authority on light, ­illumination, color, and color spaces seated in Vienna Sets standards for EMC in cars, now part of IEC

Subtractive color format used for ­printing

CO2

Carbon DiOxide

con.

connector

CPU

Central Processing Unit

CRC

Cyclic Redundancy Check

CRT

Cathode-Ray Tube

CS

Chip Select

CSE

Camera Service Extensions

MIPI protocol

CSI

Camera Serial Interface

MIPI protocol

CSMA/CD

Carrier Sense Multiple Access with ­Collision Detection

CTS

Conformance Test Specifications

CuMg

Copper-Magnesium

CuSn

Copper-Tin

General term but specifically used in MIPI

XXXI

XXXII

Abbreviations and Glossary

CVBS

Abbreviation

Explanation

Color, Video, Blanking, and Synchronization

Analogue signal for color TV

CW

Continuous Wave

D2B

Domestic Digital Bus

DAC

Digital to Analogue Converter/­ Conversion

DC

Direct Current

DCC

Direct Capacitive Coupling method

For testing resistance to fast and slow transient pulses

DCP

Digital Content Protection LLC

Organization that licenses HDCP

DCS

Display Command Set

MIPI specification

DCT

Discrete Cosine Transformation

DDC

Display Data Channel

VESA specification

DDWG

Digital Display Working Group

Organization responsible for DVI

DEC

Digital Equipment Corporation

DES

DESerializer

DFE

Decision Feedback Equalizer

DFP

Digital Flat Panel

Early VESA specification

DIN

Deutsches Institut für Normung

German SSO based in Berlin

DK

Device Key

Part of the ASAML security concept

DL

DownLink

Transmission direction with the higher data rate in an asymmetric communication system. Often synonymously used with DS.

DLL

Data Link Layer

Layer 2 of the ISO/OSI layering model

DM

Dieselhorst-Martin

Type of cable stranding

DMA

Direct Memory Access

DMCA

Digital Millennium Copyright Act

DoS

Denial of Service

Security attack that floods a node’s resources with so much data that it starts denying additional communication requests

DP

DisplayPort

Interface for display connectivity from VESA

DPCP

DisplayPort Content Protection

DPI

Dots Per Inch or Direct Power Injection

Like PPI or type of EMC measurement

DS

DownStream

Transmission direction with the higher data rate in an asymmetric communication system. Often synonymously used with DL.

DSC

Display Stream Compression

Video compression format ­standardized by VESA

Abbreviations and Glossary

Abbreviation

Explanation

DSE

Display Service Extensions

MIPI specification

DSI(-2)

Display Serial Interface

Protocol defined by the MIPI Alliance

DSI3

Distributed Systems Interface 3

Sensor interface of the DSI consortium

DSNU

Dark Signal Non-Uniformities

DSP

Digital Signal Processor

DTLS

Datagram TLS

UDP variant of TLS

DTMB

Digital Terrestrial Multimedia ­Broadcast

Standard for digital television ­transmission used in China

DUT

Device Under Test

DVB(-T)

Digital Video Broadcasting (for Terrestrial)

DVD

Digital Video/Versatile Disc

DVI

Digital Visual Interface

E

Electric (field)

ECIA

Electronic Components Industry ­Association

US-based SSO

ECU

Electronic Control Unit

Name for physical units containing electronics inside cars

EDID

Extended Display Identification Data

Display mode information format of VESA used in the DDS

eDP

embedded DisplayPort

Interface for display connectivity from VESA

EE

Electrics and Electronics

EEE

Energy Efficient Ethernet

Specified in IEEE 802.3az

EFM

Ethernet in the First Mile

Specified in IEEE 802.3ah

EIA

Electronics Industry Alliance

US- based SSO dissolved in 2011, now ECIA

ELFEXT

Equal Level Far End CrossTalk

EMC

ElectroMagnetic Compatibility

EME

ElectroMagnetic Emissions

EMI

ElectroMagnetic Immunity

sometimes, not in this book though, also used for ElectroMagnetic Inter­ ference

ENIS

End-Node-Interconnect-Structure

MDI network in the A-PHY specification

EOP

End Of Production

EPON

Ethernet Passive Optical Networks

EPROM

Erasable Programmable Read-Only Memory

ESD

ElectroStatic Discharge

ESR

Equivalent Series Resistance

ETSC

European Transport Safety Council

Standard for digital television trans­ mission, which originated in Europe

DC resistance of capacitors

XXXIII

XXXIV

Abbreviations and Glossary

Abbreviation

Explanation

EU

European Union

EuroNCAP

European New Car Assessment ­Program

F

Farad

Unit for capacitances

FAKRA

FAchausschuss KRAftfahrzeuge

Subgroup in DIN, often synonymously used for a specific coaxial connector

FB

Ferrite Bead

FBAS

Farb-Bild-Austast-Synchron-Signal

German for CVBS signal, colloquially “Farbfernsehsignal”

FCC

Federal Communications Commission

US government agency that regulates, among other, radio frequency use

FCS

Frame Check Sequence

FCW

Forward Collision Warning

FDD

Frequency Division Duplex

FEC

Forward Error Correction

FEXT

Far-End CrossTalk

FFE

Feed Forward Equalizer

FFT

Fast Fourier Transformation

FHD

Full High Definition

FM

Frequency Modulation

FMCW

Frequency Modulated Continuous Wave

FMVSS

Federal Motor Vehicle Safety ­Standards

FoFa

Forwarding Fabric

FOT

Fiber Optical Transmitter

FPD

Flat Panel Display

fps

frames per second

FR

Flame Retardant

PCB material type

FRC

Frame Rate Control

Method to emulate a high color ­resolution on a display than available

FRR

Front Range Radar

See LRR

FSED

Frame Service Extension Data

Part of MIPI CSE and DSE

G

Gear or Giga

Name of different data rate classes for the MIPI A-PHY or 109

Gbps

Gigabits per second

GI-POF

Graded Index POF

GMSL

Gigabit Multimedia Serial Link

Method to separate two data streams (in the same or opposite directions) on one channel

Used for the reception of analogue radio in the ultra short wave band

Part of the ASAML SerDes technology

Trade name for the proprietary SerDes technology of Maxim Integrated (now ADI)

Abbreviations and Glossary

Abbreviation

Explanation

GND

Ground

GOF

Glass Optical Fiber

GoP

Group of Pictures

GPIO or GPI/O

General Purpose Input/Output

Gpps

GigaPixels Per Second

GPS

Global Positioning System

gPTP

Generalized Precision Time Protocol

GPU

Graphics Processing Unit

GUI

Graphical User Interface

GVIF

Gigabit Video InterFace

Trade name for the proprietary SerDes technology of SONY

H

Henry

Physical unit for magnetic (field) strength

HBM

Human Body Model

Type of ESD test

HBR

High Bit Rate

Data rate class for DP

HD

High Definition

HDCP

High-bandwidth Digital Content ­Protection

HDMI

High Definition Multimedia Interface

HDR

High Dynamic Range or High(er) Data Rate

HDTV

High-Definition TeleVision

HEIF

High Efficiency Image File format

New format for digital images

HEVC

High Efficiency Video Coding

Also known as H.265/MPEG-H Part 2

HF

High Frequency

HFM

High-speed FAKRA Mini

HMI

Human Machine Interface

H-MTD

High-speed Modular Twisted-pair Data

HQ

HeadQuarter

Hres

Horizontal RESolution

HS

High Speed

HSB

Hue Saturation Brightness

Color format derived from RGB

HSD

High-Speed Data

Connector type for STQ cables

HSI

Hue Saturation Intensity

Color format derived from RGB

HSL

Hue Saturation Lightness

Color format derived from RGB

HSV

Hue Saturation Value

Color format derived from RGB

HSVL

High Speed Video Link

Early name for Automotive SerDes

Hsync

Horizontal SYNChronization

Related to horizontal blanking

Part of video compression

Protocol specified in IEEE 802.1AS2011

The latter is an I3C terminology

Connector type for coaxial cables Connector type for STP cables

XXXV

XXXVI

Abbreviations and Glossary

Abbreviation

Explanation

HU

Head Unit

Main ECU for infotainment functions inside cars

HW

HardWare

I2C

Inter-IC, also I2C or IIC

Serial communication bus invented by Philips in 1982

I2S

Inter-IC Sound, also I2S

Audio bus invented by Philips in 1986

I3C

Improved Inter-IC bus

MIPI protocol

IATF

International Automotive Task Force

Defines an automotive quality ­management system

IBG

InterBurst Gap

Part of the ASAML TDD scheme

IC

Integrated Circuit

ICC

Inductive Coupling Clamp method

ICMP

Internet Control Message Protocol

ICT

In-Circuit Testing

ICV

Integrity Check Value

Essential part of security mechanisms for authentication

IEC

International Electrotechnical ­Commission

SSO situated in Geneva, Switzerland

IEEE

Institute of Electrical and Electronics Engineers

“The world’s largest technical pro­ fessional organization for the advancement of technology (ieee.org)”. Among other, standardizes Ethernet.

IET

Interspersing Express Traffic

IEEE 802.3br-2016

IETF

Internet Engineering Task Force

US-based SSO seated in Wilmington, DE

I-frame

Intra frame or picture

Still image representation of MPEG

IF

InterFace

IL

Insertion Loss

Infotainment

Information and Entertainment

InGaAs

Indium Gallium Arsenide

Alloy used for IR image sensors

InSb

Indium Antimonide

Compound used for photovoltaic ­sensors reacting to IR light

INTB

Interrupt pin

As used for TI SerDes chips

I/O

Input/Output

IoT

Internet of Things

IP

Internet Protocol

IP Code

Ingress Protection Code or Inter­ national Protection Code

IEC (= EN) 60529 defines classes for mechanical protection for components in cars

IPMI

Intelligent Platform Management ­Interface

I2C derivate

For testing immunity against slow ­transients

Abbreviations and Glossary

Abbreviation

Explanation

IPsec

Internet Protocol SECurity

IR

InfraRed

Frequency spectrum just below the ­visible light

ISDB

Integrated Services Digital Broadcasting

Digital television standard that ­originated in Japan

ISI

Inter Symbol Interference

ISM

Industrial, Science, Medical

Identification of “open” frequency bands that may be used for these ­purposes

ISO

International Organization for ­Standardization

SSO seated in Geneva, Switzerland

ISP

Image Signal Processor

IT

Information Technology

ITU

International Telecommunication Union SSO seated in Geneva, Switzerland

IUT

Implementation Under Test

IVC

In-Vehicle Communication

IVI

In-Vehicle Infotainment

IVN

In-Vehicle Network(ing)

JAE

Japan Aviation Electronics industry Ltd.

JEIDA

Japan Electronic Industry Development Association

Japanese SSO, now JEITA

JEITA

Japan Electronics and Information ­Technology industries Association

Japanese SSO

JITC

Just-In-Time-Canceller

Retrain possibility of the A-PHY 1.0

JPEG

Joint Photographic Experts Group

JTAG

Joint Test Action Group

JVC

Japan Victor Company

Originator of VHS

k

kilo

103

LAN

Local Area Network

Laser

Light Amplification by Stimulated ­Emission of Radiation

LCA

Lane Center Assist

LCD

Liquid Crystal Display

LCL

Longitudinal Conversion Loss

LDF

LIN Description File

LED

Light Emitting Diode

LFLT

Line FauLT

Pin at GSML deserializer

Lidar

LIght Detection And Ranging

Sensor type

LIN

Local Interconnect Network

Physical communication network in cars, typically comprising several IVC technologies

XXXVII

XXXVIII Abbreviations and Glossary

Abbreviation

Explanation

LISN

Line Impedance Stabilization Network

LK

Link Key

LLC

Limited Liability Company

LNB

Low Noise Block (converter)

LOMMF

Laser Optimized MMF

LP

Low Power

LPI

Low Power Idle

LRR

Long Range Radar

LSB

Least Significant Bit

LSFR

Linear Shift Feedback Register

LT

Lower Tester

LTE

Long Term Evolution

LVCMOS

Low Voltage CMOS

LVDS

Low Voltage Differential Signaling

m

mandatory

M

Mega

M2M

Machine to Machine

MAC

Medium or Media Access Control

MASS

MIPI Automotive SerDes Solutions

MC

Message Counter, MultiCast, or Mode Conversion

MCM

MultiChip Modules

MCS

Manufacturer Command Set

Part of MIPI DSI

MDC

Management Data Clock

Used with Ethernet PHY management

MDI

Media Dependent Interface

Part of Ethernet physical layer ­definition

MDIO

Management Data Input/Output

MEMS

Micro Electro-Mechanical System (­module)

(x)MII

Any type of Media Independent ­Interface

MIMO

Multiple Input Multiple Output

MIPI

Original meaning: Mobile Industry ­Processor Interface, however, this meaning is no longer used.

Alliance developing technical specifi­ cations in the mobile eco-system (and also the MIPI A-PHY)

MJPEG

Motion Joint Photographic Experts Group

Video and audio compression formats

MM

Machine Model

Type of ESD test

MMF

MultiMode Fiber

Type of GOF

Part of the ASA security concept Part of satellite antenna systems to enable a low noise reception

Part of EEE

4G mobile phone standard Early principle behind serialization 106 Part of ISO/OSI DDL layer for Ethernet

Interface used between Ethernet PHYs and MAC

Abbreviations and Glossary

Abbreviation

Explanation

MMIC

Monolithic Microwave Integrated ­Circuits

ICs optimized for processes running between 300 MHz and 300 GHz

MOST

Media Oriented Systems Transport

Automotive communication system (being phased out)

MP3

MPEG-2 Part 3

Audio compression format

MPAA

Motion Picture Association of America

MPEG

Moving Pictures Experts Group

MPEG-LA

MPEG Licensing Administration

MQS

Micro Quadlock System

MRR

Mid-Range Radar

MSB

Most Significant Bit

MSE

Mean Square Error

MST

Multi-STream

DP terminology

MTD

Modular Twisted-pair Data

Connector type for UTP cables

MTP

Multi-stream Transport Packet

Part of MST/DP

NACK or nACK

Not ACKnowledged

NBI

Narrow Band Interference

NCAP

New Car Assessment Program

NCF

Node Capability File

NEXT

Near-End CrossTalk

NFC

Near Field Communication

NHTSA

National Highway Traffic Safety ­Administration

US administration body for safety of road vehicles

nMQS

Nano MQS

Connector type for UTP cables

NRZ

Non-Return to Zero

Modulation scheme with two voltage levels

nt

thermal noise

NTSC

National Television System Committee

NVM

Non-Volatile Memory

NZ

Neutral Zone

o

optional

OAM

Operation, Administration, ­Management channel

Side channel available with, for example, the ASAML and IEEE 802.3ch 2020 Ethernet

OB

Odd Bytes

Part of MIPI A-PHY/IEEE 2977

OFDM

Orthogonal Frequency Division ­Multiplexing

Important group for video compression algorithms Connector type for UTP cables

Part of LIN

Analogue television standard used especially in North America and Japan Area in which electromagnetic ­inter­ference is neutralized

XXXIX

XL

Abbreviations and Glossary

Abbreviation

Explanation

OLED

Organic Light-Emitting Diode

OPEN

One Pair EtherNet (Alliance)

OpenLDI

Open LVDS Display Interface

OSI

Open System Interconnection

OTA

Over The Air (Updates)

OTP

One-Time Programmable memory

P

Profile or Power

MIPI A-PHY terminology

P1/P2

Profile 1/Profile 2

Part of MIPI A-PHY

P2P

Point-to-Point

Communication that starts and ends within one physical link.

PA

Parking Assist

PAEB

Pedestrian AEB

PAL

Phase Alternation Line or Protocol Adaptation Layer

Alliance developing the enabling ­specifications for Automotive Ethernet

Analogue television standard used especially in Europe and China Connect between native protocols and the MIPI A-PHY

PAM

Pulse Amplitude Modulation

PCB

Printed Circuit Board

PCIe

Peripheral Component Interconnect express

High-speed serial computer expansion bus

PCLK

Pixel CLocK

Important in image sensors

PCM

Pulse Code Modulation

PCO

Point of Control and Observation

Part of ISO 9646

PCS

Physical Coding Sublayer

Part of the physical layer

PD

Powered Device

Device that receives power over the communication line

P&D

Plug & Display

First VESA display connectivity ­standard

PDU

Protocol Data Unit

PER

Packet Error Rate

PE-X

PolyEthylene (also XPE)

PFC

Priority-based Flow Control

Part of IEEE 802.1Qbb 2011

P-frame

Predictive coded Frame or picture

Part of MPEG encoding

PHD

PHY Header Data

Part of the IEEE 802.3cz PCS

PHY

PHYsical Layer

Lowest layer (layer 1) of the ISO/OSI layering model

PICS

Protocol Implementation Conformance Statements

PIN

P-type – Intrinsic region – N-type

Diode type with larger intrinsic region

Abbreviations and Glossary

Abbreviation

Explanation

PLC

Product Life Cycle or Power Line ­Communication

PLL

Phase Lock Loop

PLS

Physical Layer Signaling (service ­interface)

Communication between reconciliation and MAC layer in IEEE 802.3 specifications

PMA

Physical Medium Attachment

Part of the physical layer

PMBus

Power Management Bus

I2C derivate

PMD

Physical Medium Dependent

Part of the physical layer (used in A-PHY or IEEE 802.3 optical Ethernet transmission technologies)

PoC

Power Over Coaxial

PoD

Power Over Differential cables

PoDL

Power Over Data Line

Specified in IEEE 802.3bu 2016 for ­single pair (T1) Ethernet

PoE

Power Over Ethernet

Specified in IEEE 802.3af 2003 for two pair Ethernet versions

POF

Polymer/Plastic Optical Fiber

PP

PolyPropylene

p-p

Peak-to-Peak

PPI

Pixels Per Inch or PHY Protocol ­Interface

PPM

Parts Per Million

pps

Pixels Per Second

PRBS

Pseudo-Random Bit Sequence

Prio

Priority

Pro-AV

Professional Audio and Video

Prot.

Protocol

PS

PolyStyrene

PSAACRF

Power Sum Alien Attenuation to ­Crosstalk Ratio Far-end

PSANEXT

Power Sum Alien Near-End crossTalk

PSD

Power Spectral Density

PSE

Power Supply/Sourcing Equipment

Part that supplies the power in case power is supplied over the data line

PSI5

Peripheral Sensor Interface Five

Low speed sensor bus

PSNR

Peak Signal-to-Noise-Ratio

PSR

Panel Self Refresh

Part of DP/eDP

PTB

Precision Time Base

Part of the ASAML technology

PVC

PolyVinyl Chloride

PHY Protocol Interface is part of the MIPI C-PHY

Insulation material

XLI

XLII

Abbreviations and Glossary

Abbreviation

Explanation

PWM

Pulse-Width Modulation

Physical principle for simple data ­transmission

QAM

Quadrature Amplitude Modulation

QFN

Quad Flat No leads

Type of semiconductor housing

QM

Quality Management

Lowest functional safety level in ISO 26262

QoS

Quality of Service

R/W

Read/Write

Radar

RAdio Detection And Ranging

RAM

Random Access Memory

RBP

Reverse Battery Protection

RBR

Reduced Bit Rate

Data rate class for DP

RCA

Radio Corporation of America or Reverse Channel Audio

Connector used for CVBS video or part of the HDMI interface

RCCB

Red Clear Clear Blue

Alternative CFA for imagers

RCTA

Rear Cross Traffic Alert

RCW

Rear Collision Warning

RD

Running Disparity

RF

Radio Frequency

RFC

Request For Comments

RFFE

Radio Frequency Front End

RG

Radio Guide

RGB

Red Green Blue

RGGB

Red Green Green Blue

RJ

Registered Jack

RL

Return Loss

RMII

Reduced MII

ROM

Read Only Memory

RQ

ReQuest

RS-FEC

Reed Solomon FEC

RSE

Rear Seat Entertainment

RTP

Real-time Transport Protocol

RTS

ReTranSmission

RX or Rx

Receiver/receive

SA

Shield/Screening Attenuation

SAE

Society of Automotive Engineers

US-based SSO

SATA

Serial Advanced Technology ­Attachment

Computer bus interface connecting computing with storage

Part of the 8B10B encoding scheme Name for standard documents ­created by the IETF Old nomenclature for cables Name sometimes used for Bayer CFA

Type of FEC

Abbreviations and Glossary

Abbreviation

Explanation

SCART

Syndicat desConstructeurs d’Appareils Radiorécepteurs et Téléviseurs

Connector type for CVBS television

SCCP

Serial Communication Classification Protocol

Optional control protocol for the PoDL standard IEEE802.3bu 2016

SCI

Sub Constellation Index or Scalable Coherent Interface

Header field of the A-PHY or part of the LVDS standard

SCL

Serial CLock

Used for the I2C clock

SDA

Serial DAta

Used for the I2C data

SDI

Serial Data In

SPI terminology

SDL

Specification and Description ­Language

SDO

Serial Data Out

SPI terminology

SDP

Shielded Differential Pair

Comprises all shielded differential ­communication cables, STP and SPP

SDR

Standard Data Rate

I3C terminology

SecOC

Secure Onboard Communication

Part of AUTOSAR

SENT

Single Edge Nibble Transmission

Low speed sensor bus

SEooC

Safety Element out of Context

Part of ISO 26262

SEP

Service Extensions Packet

Part of the MIPI CSE protocol

SEPIC

Single-Ended Primary-Inductor ­Converter

Type of DC-DC converter

SER

SERializer

SerDes

SERializer/DESerializer

SFCW

Stepped Frequency Continuous Wave

SG

Speed Grade

SI-POF

Step Index POF

SMA

SubMiniature version A

Type of (none-automotive) coaxial ­connector

SMBus

System Management Bus

I2C derivate

SNR

Signal to Noise Ratio

SoC

System On Chip

SOME/IP

Scalable service-Oriented ­MiddlewarE over IP

Middleware used with Automotive ­Ethernet communication

Sonar

SOund Navigation And Ranging

Other name for ultrasonic sensors

SOP

Start Of Production

SOVS

System Operational Vector Space

SPAD

Single-Photon Avalanche Diode

S-param­eters Scattering parameters SPI

Serial Peripheral Interface

SPP

Shielded Parallel Pair cable

Name of different data rate classes in ASA

XLIII

XLIV

Abbreviations and Glossary

Abbreviation

Explanation

SQI

Signal Quality Indicator

sRGB

Standard RGB

SROI

Smart Region of Interest

SRP

Stream Reservation Protocol

SRR

Short Range Radar

SSL

Secure Sockets Layer

SSO

Standard Setting Organization

STP

Shielded Twisted Pair (cables)

STQ

STar-Quad/Shielded Twisted Quad (cables)

StVZO

STraßenVerkehrs-Zulassungs-Ordnung

SUV

Service or Sports Utility Vehicle

SVCD

Super Video Compact Disc

SVS

Surround View System

SW

SoftWare

sync

SYNChronization

TAS

Time Aware Shaper

TC

Technical Committee

TCL

Transverse Conversion Loss

TCP/IP

Transmission Control Protocol/Internet Protocol suite often used in conjuncProtocol tion with Ethernet, comprises also UDP and many other protocols

TCON

Timing CONtroller

Used in displays

TDD

Time Division Duplex or Test-Driven Development

Method to separate two data streams (in the same or opposite directions) on one channel or a type of agile development methodology

TDR

Time Domain Reflectometry

TEM

Transversal ElectroMagnetic

TF

Task Force

Nomenclature of the IEEE 802.1 and IEEE 802.3 groups developing the ­specifications

TFT

Thin Film Transistor

Type of LCD technology

TI

Texas Instruments

TIA

Telecommunications Industry ­Association

US-based SSO in Arlington, VA

TLIS

Transmission-Line-Interconnect-­ Structure

Link segment in A-PHY

TLP

Transmission-Line Pulse measurement

TLS

Transport Layer Security

Part of MIPI CSI-2 v3.0

Predecessor of TLS

Name of road traffic licensing ­regulations in Germany

IEEE 802.1Qbv 2015

Security protocol for TCP

Abbreviations and Glossary

Abbreviation

Explanation

TMDS

Transition-Minimized Differential ­Signaling

ToF

Time Of Flight

TP

Test Point

TRC

Three Repetition Code

TSN

Time Sensitive Networking

TTL

Transistor-Transistor Logic

TV

TeleVision

TVS

Transient Voltage Suppression

TX or Tx

Transmitter/transmit

UART

Universal Asynchronous Receiver – Transmitter

Serial interface

UDP

User Datagram Protocol

Transport protocol used in conjunction with Ethernet

UHBR

Ultra-High Bit Rate

Bit rate class for DP

UHD

Ultra-High Definition

UL

UpLink

UML

Unified Modelling Language

UNECE

United Nations Economic Commission for Europe

UNFCCC

United Nations Framework Convention on Climate Change

URL

Uniform Resource Locator

US

UpStream

USB

Universal Serial Bus

USD

United States Dollars

USGMII

Universal Serial Gigabit Media ­Independent Interface

USL

Unified Serial Link

USRR

Ultra Short Range Radar

USXGMII

Universal Serial 10 Gbps Ethernet Media Independent Interface

UT

Upper Tester

UTP

Unshielded Twisted Pair (cabling)

Camera type useable to create a 3D image

Various IEEE 802.1 standards ­supporting QoS over Ethernet.

Type of ESD protection

Transmission direction with the lower data rate in an asymmetric communication system. Often synonymously used with US.

Transmission direction with the lower data rate in an asymmetric communication system. Often synonymously used with UL.

Part of CSI-2 v3.0

XLV

XLVI

Abbreviations and Glossary

Abbreviation

Explanation

UUID

Universally Unique IDentifier

Used for the identity of an ASA Device

UWB

Ultra-Wide Band

V

Voltage or Volts

VCD

Video CD

VCIC

Video Communication Interface for Cameras

VCR

Video Cassette Recorder

VCSEL

Vertical Cavity Surface-Emitting Lasers

Light source for optical transmission systems

VDC-M

VESA Display Compression-M

Compression for mobile devices

VDE

Verband Deutscher Elektrotechniker

German SSO

VESA

Original meaning: Video Electronics Standards Association, however, it is no longer used

Alliance developing technical ­specifications in the realm of displays (including DCS and DP/eDP)

VGA

Video Graphics Array

Specific format for early video ­transmission

VHDL

Very high-speed integrated circuit ­Hardware Description Language

VHS

Video Home System

VLAN

Virtual Local Area Network

VNA

Vector Network Analyzer

Vres

Vertical RESolution

Vsync

Vertical SYNChronization

WG

Working Group

WOL

Wake-On LAN

www

World Wide Web

XAUI

10 Gbps Attachment Unit Interface

XFI or XIFI

No specifics given

XGMII

10 Gbps Media Independent Interface

XNOR

Exclusive Not OR

XOR

Exclusive OR

XPE

See also PE-X

XT or XTALK

Crosstalk

YANG

Yet Another Next Generation

YUV

Specified in ISO 17215 for Ethernet communication

Related to vertical blanking

Extension of XAUI, pronounced “ziffie”

Data modeling language for network management Name for a video color format, where Y is the luminance and U and V carry the chrominance information

1

Introduction and Background

Considering that cars have been developed and sold commercially since the end of the 19th century, high-speed sensors and displays are a comparably recent event. At the end of the 20th century, more than 100 years after the start of commercial car sales, high-speed sensors and displays were, if at all, presented in concept cars or sold with selected luxury models. However, since the turn of the 21st century, the number of sensors and displays has grown, with the market really just gaining momentum at the time of writing in 2021. While the exact number for the expected market growth differ, market research agrees on the trend: it is significant. In [1], for example, the number of cameras per car is expected to grow between 2020 and 2030 from five to 20 and the number of displays from three to 15. Displays and cameras are thereby not only growing in numbers, they are also growing in resolutions. Furthermore, thanks to the increasing adoption of Advanced Driver ASsist (ADAS) functions, the number of sensors other than cameras is also growing, as is the number of types of sensors. The race for being the first to successfully achieve the ultimate ADAS function where driver intervention is no longer required – level 4 or 5 Autonomous Driving (AD) [2] – is accelerating the trend in two different ways. First of all, more sensors are deployed in order to reduce the number of tasks drivers have to perform. Then, the drivers can use that freed capacity in order to focus more on information and entertainment (infotainment) on the displays provided. All these innovations are spurred by key technological inventions and developments. Next to the continuing empowerment and shrinking of digital processing technologies that are responsible for many amenities of modern life in general, more specific inventions are: high-resolution digital image sensor technologies, empowering (new types of) sensors for automotive use like Light Detection And Ranging (Lidar) sensors, digital video (­compression) formats, digital display technologies that are small, robust, and cost efficient enough to be commonly used inside cars, and modern user interaction methodologies proliferated by the use of smartphones (plus the mobile communication telecom infrastructure enabling it). One of the resulting key challenges for deploying all the sensors and displays inside cars is how to integrate them into the Electric and Electronic (EE-)architectures and, especially, how to realize their communication. When the adoption of (digital) cameras and displays in cars started at the beginning of the 21st century, the actual communication was analogue. However, analogue video transmission has severe limits with respect to resolution and quality, which prohibits the subsequent processing necessary to realize modern ADAS and infotainment functions. So nowadays, digital video data transmission drives the demand for data rates in the In-Vehicle Communication (IVC) systems, while the availability of suitable

2

1 Introduction and Background

high-speed communication technologies opens the door for innovations with respect to ­video-related customer functions and EE-architecture choices. Unfortunately, it is thereby generally not possible to simply reuse the communication technologies from the consumer and IT industries, which already support the required high video data rates in a mass market. It is one goal of this book to explain the additional constraints IVC technologies have to master with respect to robustness and costs, why automotive suitable physical layer developments are important, and why Automotive SerDes and Automotive Ethernet technologies are the available choices in this context. In order to support a profound understanding of the interrelations between the automotive environment, the high-speed sensor and display use cases, and the communication technologies, and to motivate the choices, this book is structured as follows: ƒ In the continuation of this introductory Chapter 1, Section 1.1 motivates the focus on sensor and display applications. It explains the differences between sensor and display applications and between them and other use cases inside cars. Section 1.2 introduces the terminology used in the context of SerDes communication and the background of Automotive SerDes. Section 1.3 provides information on the origin of Ethernet as such and on Ethernet used as an IVC technology. ƒ Innovations and their underlying technologies are rarely introduced for the sake of “using a new technology”. Normally, they serve a purpose. The three main reasons for innovations in the industrial Business-to-Business (B2B) environment are: first, to allow for new functionalities (and business), second, to save costs, and/or third, to fulfill new regulatory requirements. In order to provide the context, this book introduces first, in Chapter 2, the high-speed sensor and display use cases with respect to their history in the car industry as well as the underlying technical and architectural choices in more detail. ƒ Chapter 3 introduces the automotive environment, in which the use cases have to function reliably and safely. Cars are particularly complex products, because they have to provide a vast variety of functions under extremely different conditions, while needing to be attractive to customers in a very competitive market. The automotive environment impacts all technical choices made for cars and is therefore covered early in this book. ƒ One reason consumer and IT communication technologies are often not usable in cars, is their incapability to meet the automotive ElectroMagnetic Compatibility (EMC) requirements (at least not at reasonable costs). EMC is especially important for all electronics inside cars, and thus detailed in a separate Chapter 4. ƒ The cable harness is the third heaviest and third most expensive component inside cars [3]. Communication cables need to be robust, cost efficient, and light at the same time. One more reason why consumer grade products are generally unsuitable for in-vehicle use. Chapter 5 introduces general choices for the communication channel that have to be made for all IVC technologies. This includes options for cables and connectors. ƒ Power supply and power saving is another extremely important aspect in cars, independent of the actual technologies used. Aspects relevant for sensor and display use cases that impact the IVC technology in general are discussed in Chapter 6. ƒ Chapter 7 introduces the choices for Automotive SerDes technologies.

1.1 The Distinctive Properties of High-Speed Sensor and Display Use Cases

ƒ Chapter 8 introduces the High-Speed (HS) Automotive Ethernet technologies and provides a general comparison between HS Automotive Ethernet and SerDes standards. ƒ Both, Automotive SerDes and Automotive Ethernet are first of all use case independent physical and data link layer technologies. To deploy them for high-speed sensor and display use cases, quite a number of related higher layer ­standards and protocols are added, which might also affect or become part of specific SerDes or E ­ thernet products. Chapter 9 provides and overview and introduction to many related standards and protocols. These comprise color codes, control interfaces, video compression formats, content protection, as well as camera and display specific protocols. ƒ Last but not least, Chapter 10 looks at test, qualification, and tools. That they can be tested and serviced is an extremely important aspect for all use cases and technical solutions in cars. So, while this topic is addressed at the end of this books, to ensure testability for all system designs and new technologies is actually an important starting requirement. Note that, while the order of content and chapters is intended to be as logical and sequential as possible, a perfect order does not exist for a subject as complex as the one addressed in this book. There are many interrelations between chapters, so that the book contains many forward and backward references.

„ 1.1 The Distinctive Properties of High-Speed Sensor and Display Use Cases Displays and sensors in cars – including cameras as a special type of sensor – actually address quite distinct use cases. Displays have the sole purpose of relaying technical, entertainment, or other information to the car users. Especially when backed by touch functionality, voice recognition, or related dials and knobs, they serve as an important element of the Human Machine Interface (HMI), with which the customers can control various functions inside their cars. Sensors on the other hand, provide sensor specific, technical data that is, in its raw format, generally unusable to car occupants. Either the sensor data serves directly to control driving functions without the users ever being aware of their existence or it needs to be processed before it can be used for driver or passenger information or user interaction in ADAS functions. Camera images are the exception, as they might be used for machine vision/ processing as well as for human vision, for example in back-up camera systems. Table 1.1 lists additional properties that differ for display, camera, and other sensor use cases and that have some relevance for the architecture and other technical choices of the use cases discussed in the following chapters of this book, especially in Chapter 2. Table 1.1 also motivates why it makes sense to address cameras separately from other sensors. While there are some similarities between cameras and other sensors, there are also important differences.

3

4

1 Introduction and Background

Table 1.1 Comparison of distinct sensor and display use case properties Data recipient

Displays

Cameras

Other sensors

Human vision only

Human vision or machine processing

Machine preprocessing required

Quality of Service Human vision allows for (QoS) requirements some latency and losses

Machine processing requires low latencies and is ­sensitive to losses, compression or other

Size of housing

Generally large

Typically very small

Power requirement

Power hungry because of the display (depends on size)

Small housings easily accumulate heat, which can impede the sensing quality. Power dissipation should therefore be low

Location in car

Inside the cabin with ­stringent location requirements with respect to the occupants’ positions

Facing outside or to the driver or other occupants

Possible add-on functions

Might comprise microphones, Consumer Electronics (CE) connectivity (including auxiliary sockets), or even cameras, typically no speakers though

Generally singular, collects Might comprise InfraRed one type of data only (IR) Light Emitting Diodes (LEDs) for interior cameras and night vision, exterior cameras might comprise heating

ADAS sensors are typically on the body shell facing outside, other sensors might be anywhere including under the hood

There are, however, also aspects that unite the use cases. These are their requirement for highly asymmetric (high-speed) data communication and the related architectural choices. Both, sensor and display units, are generally located at the edge of a network as end nodes. Even if they are forwarding data in a type of display or sensor daisy chain – which happens seldom in any case – they can be designed such that they require no software-based processing, which might require frequent updates otherwise. These aspects not only unite the high-speed sensor and display use cases, it distinguishes them from (many) other Electronic Control Units (ECUs) inside the car. Figure 1.1 shows two, fundamentally different architecture options. In order to directly compare the sensor and display use cases, the examples depicted assume that the sensor data is – after having been processed accordingly – displayed on a screen to the user. In a real car, such direct link between one sensor and display is seldom. A display might also be used to present pre-stored entertainment data, or they show aggregated results from the evaluation of various sensors. Sensors outputs, on the other hand, might result in vehicle control without user interaction or with audible feedback only. The upper part of Figure 1.1 depicts the case in which sensor and display contain no video or sensor data processing themselves. The sensor data is transferred as collected (more or less) to the ECU, which processes the data, makes use of the result in its application, and then renders this into a video stream that is transferred to the display where it is presented on a screen. This could be the setup for a back-up camera. Colloquially, this scenario is ­often referred to as having “dumb” sensors and displays. The sensors and displays have no processing and thus “no intelligence”. While some might object to the exact wording, key is that the sensors and displays in this scenario do not run any software that might require regular updates or upgrades.

5

1.1 The Distinctive Properties of High-Speed Sensor and Display Use Cases

Camera/Sensor (Image) Sensor

ECU

IVC chip A

Use case specific protocol interface

IVC chip B

Use case independent IVC technology

Sensor data processing

(Image) Sensor

Use case specific protocol interface

IVC chip C

Protocol interface as supported by processor and IVC technology, can be the use case specific protocol or an IVC technology protocol

Camera/Sensor Sensor data processing

Application processing

Display Graphics & video processing

IVC chip B‘

IVC chip A‘

Use case independent IVC technology

IVC chip D

Use case independent IVC technology

Use case specific protocol interface

ECU

Display

Application processing

Graphics & video processing

IVC chip C‘

IVC chip D‘

Use case independent IVC technology

Screen

Screen

Use case specific protocol interface

Protocol interface as supported by processor and IVC chip, can be the use case specific protocol or an IVC technology protocol

Figure 1.1 Principle architecture options for sensor and display use cases In the lower part of Figure 1.1 the sensor as well as the display perform the major processing themselves. A typical example would be traffic sign recognition. The camera records the image, identifies the particular traffic sign in its processor, and then transfers only an ­identifier number to the ECU. The ECU would then perform a plausibility check in its application processing by comparing the identified traffic sign with its map data, before sending itself an identifier number to the display. The display then renders a picture of the sign that is displayed to the customer. Naturally, such a scenario makes the sensors and displays more complex. However, at the same time the amount of data that needs to be communicated is significantly smaller than in the case of sensors and displays without processing. The additional costs for the processing is potentially compensated for with a less expensive communication system that does no longer need to be “high-speed”. Note that in some cases the only processing that is being performed in the sensors and displays is data compression or decompression. This somewhat intermediate case is not depicted in Figure 1.1. The extra processing needed in the sensors and displays can often be realized in hardware. In general, hardware compression is faster and less power consuming than compression in software. With compression the data rate is decreased, but not as much as when just identifiers are transmitted, which would be the case after full processing. So, a scenario with compression would result in intermediate processing and inter­ mediate data rate. At the same time, the compression might have other impacts, such as compression losses or added processing latencies, which might not be acceptable (see also Table 1.1). For more details on the use cases, see Chapter 2. What is important in the context of this book: In both scenarios depicted in Figure 1.1, it is necessary to distinguish between the protocol interfaces that are used within the sensors or displays and the IVC technology. The protocol interfaces used for connecting the sensor and display chips are application specific, meaning that the imager interface technology inside a camera cannot be used for putting data onto the screen of a display and vice versa. At the same time, both camera and display might be connected to the ECU using the same IVC technology. Furthermore, the IVC chips used in both cases are not necessarily the same. This is why Figure 1.1, distinguishes between IVC chips with and without “ ’ ”. In the upper part of Figure 1.1, it is likely necessary to use an “IVC bridge” that bridges between the

6

1 Introduction and Background

use case agnostic IVC technology and the use case specific protocol. In the lower part of Figure 1.1, the interface combination used, it depends on the availability of interfaces in the processing and IVC.

„ 1.2 Background to Automotive SerDes The term “SerDes” is used for a number of different technologies in different use cases and scenarios. This section aims to clarify the ambiguity of the term at least for the use within this book. In order to do so, Section 1.2.1 starts with explaining the origin of the term “­SerDes”. Section 1.2.2 introduces the SerDes terminology common in the automotive industry and Section 1.2.3 outlines the status of Automotive SerDes in the car industry. The technical choices and properties of the Automotive SerDes technologies as such are discussed in Chapter 7.

1.2.1 The Origin of “SerDes” “SerDes” first of all describes a very basic physical principle. When two chips had to communicate in the early days, each output pin of one chip was simply directly connected to the input pins of the other chip and vice versa. When more than one information had to be ­exchanged, other sets of parallel pins and connections were added. For reasons explained in more detail further below, having more parallel data lines became impractical, and formerly parallel data was serialized before being transferred to other chips. There, it would be deserialized before being processed internally. Figure 1.2 shows this in a very simple example. To have this serializer-deserializer conversion of data at both ends of the communication then condensed into the term “SerDes”. TX

RX

Parallel  Serial 1

0

0

0

1

1

Serial  Parallel 0

0

Data

1

0

0

0

1

1

0

0

CLK

Figure 1.2 The basic principle of SERializer-DESerializer (SerDes) technologies There are three main reasons to favor serial data transfer over parallel transmission [4] [5]: 1. lower number of pins at the Integrated Circuits (ICs) 2. better synchronization and supported data rates 3. less interference, especially less crosstalk

1.2 Background to Automotive SerDes

Ad 1. Lower number of pins at the Integrated Circuits (ICs) Since their invention, the processing capabilities of IC’s made huge progress. Moore’s law observed that the transistor density has about doubled every two years [6]. At the same time, the packaging and pin density of ICs has not developed at the same pace, meaning that continued parallel data transmission would have resulted in prohibitively large ICs. This simply mandated using the existing pins more efficiently. Ad 2. Better synchronization and supported data rates Figure 1.3 shows a simple parallel transmission system consisting of one transmitter (TX), one receiver (RX), eight parallel data lines (D0 to D7), and one clock line (CLK). The clock line is important, because for the receiving unit it is essential that all eight lines are synchronized in order to be able to process the received data correctly. To the right of the TX – RX system shown in Figure 1.3, an example bit pattern is depicted as seen by the receiver. The upper part of Figure 1.3 shows the ideal situation. Here, the data of each data line is received in perfect synchronization. This might well be the case for low frequencies and short distances on well-designed Printed Circuit Board (PCB) layouts. The lower part of Figure 1.3 depicts – in a strongly simplified way – what can happen if the parallel data paths are not perfectly aligned. In this case, the receiver might not sample all bits in the same transmit slot.

TX

D0 D1 D2 D3 D4 D5 D6 D7

RX

CLK

D0 D1 D2 D3 D4 D5 D6 D7 CLK

Figure 1.3 Synchronization issue in case of parallel data transmission In this simplified figure, the data paths have unequal lengths. In real life such variations also depend on the chip process, voltage, and/or temperature. The higher the frequency, the more sensitive the system is to such delay variations, with the result that from a certain frequency on, it is not possible to reliably receive data transmitted on parallel lanes.

7

8

1 Introduction and Background

Naturally, transmitting over long cables increases the difficulties when compared with the transmission on a PCB. A serial system does not have such synchronization problems, even if it needs to transmit with an n-times higher data rate in order to achieve the same throughput, when compared with a transmission over n parallel lanes. Ad 3. Less interference, especially less crosstalk Another important aspect in parallel data communication is the reference potential of the signals, the signal ground. The parallel data transmission as depicted in Figure 1.3 is single-ended and not differential. Single-ended means that one lane or wire carries the varying voltage levels that represents the signal while the other lane or wire needed for the communication is, usually, the ground. Such a communication concept is quite susceptible to interference and would require a perfect signal ground to mitigate the effects of, for example, crosstalk. Crosstalk is the interference between adjacent data lines. The longer and closer the lines or cables and the higher the transmit frequency, the more severe the impact of crosstalk. In case of parallel data transmission, there are many adjacent lines per definition and the risk of crosstalk impairments is therefore high. To mitigate the impact of crosstalk, ground lines could be put between all parallel data lines on a PCB, meaning that at least the same number of signal ground lines are connected between the transmitter and the receiver. Serialized data allows easily for differential transmission. In case of differential trans­ mission the same signal is transmitted over two wires with opposite voltage levels. At the receiver of a differentially transmitted signal, the two signals are combined. This cancels out various noise sources. Serialized data with differential transmission thus has better interference robustness and avoids the impact of the signal ground on the signal integrity. High-speed SerDes has thus become the dominant form of input and output for (most) high-integration chips [4] and almost all modern communication technologies are based on the serialization/deserialization principle shown in Figure 1.2. The simple example of ­Figure 1.2 is single ended, it uses a dedicated clock line, and a dedicated voltage level for a single signal. Modern SerDes technologies are differential and do not need a dedicated clock line. The enhanced circuit technology can recover a stable and precise clock signal from the bit stream received. This further improves the robustness of the SerDes technologies as differences in transmission time between the clock and the data signal (“clock skew”) are eliminated. Furthermore, the available circuit technologies allow modulating and encoding the transmitted data prior to sending it. This means that with a single, physical voltage level, more than one bit can be transferred, and the data rate can be increased (see also Chapter 7 for more details on actual solutions).

1.2.2 Automotive SerDes Terminology The previous Section 1.2.1 explained, why the term “SerDes” might be used in different contexts for quite different communication technologies. “SerDes” as a physical principle does not distinguish whether the communication is on a Printed Circuit Board (PCB), across a wire, or even wireless. Often, even Ethernet is called a SerDes technology, simply because

1.2 Background to Automotive SerDes

it supports differential, serial transmission of data, while in this book Ethernet is treated as a different technology (see Section 1.3 or Chapter 8). One way to lessen the ambiguity around the term is to give what is being discussed as ­SerDes in this book a clear definition and a different name. The following thus defines “­Automotive SerDes” with listing the properties commonly associated with “SerDes” in the automotive industry. While it might not always be explicitly spelled out, apart from in the previous Section 1.2.1, “SerDes” or “Automotive SerDes” throughout this book has the characteristics as listed below. a) It drives a wire. b) It supports “asymmetric communication”, meaning high data rates in one communication direction (only). c) It supports Point-to-Point (P2P) communication (only). d) It supports the lowest two layers of the ISO/OSI communication model (only). Ad a) Automotive SerDes drives a wire. The electronics in cars are generally distributed. This is particularly true for sensors and displays, because they need to be at specific locations inside the car to fulfill their function. A lot of the sensing is done at the extremities of the body shell of a car, the displays need to be in alignment with the viewing positions from the seats. In contrast, processing units can be anywhere in the car where there is space and the right environment to put them. All units, however, need to communicate across copper or optical cables that can easily reach 10–15 m length. For installation in busses and trucks even 40 m are a typical requirement [7]. If a SerDes technology is used for sensors or displays, it thus has to be able to drive the respective cables, else it is not of interest for these use cases. Having cables and connectors available that support the high data rates in the challenging automotive environment, is therefore decisive for the success of the technology. See Chapter 5 for more ­details. Ad b) Automotive SerDes supports high data rates in one communication direction (only). SerDes communication is first of all unidirectional. The transmission direction goes from the serializing sender to the deserializing receiver. That SerDes allowed for unidirectional high data rates is how the technology was adopted in cars (see also Section 1.2.3); as it was usable for the one main transmit direction the sensor/video applications needed. For control, a separate, low data rate communication technology – for example the Local Interconnect Network (LIN) bus [8] – was used at the side to start with. It was then a matter of progress and cost reductions in semiconductor processing to optimize this set up. As a result, a bi-directional, low data rate control channel is now available with Automotive SerDes solutions. Naturally, the use cases would also work with symmetric highspeed communication. However, there is, generally, no need for the added complexity and costs, so Automotive SerDes solutions strived supporting high data rates in one transmit direction only. “High” data rates are thereby relative and a matter of perspective. When the first cameras in cars used digital transmission technologies, the imagers might have had a Video Graphics Array (VGA) resolution of 640 × 480 (see Section 2.1.2 for details). With 30 frames per second (fps) and 16 bits color, this lead to about 150 Mbps data rate. At the

9

10

1 Introduction and Background

time, this was considered a very high data rate for in-vehicle communication. When the Media Oriented Systems Transport (MOST) bus was introduced at about the same time, it supported 25 Mbps [9], which again was a huge leap from the Controller Area Network (CAN) bus [10] or LIN available before. In 2021 in the automotive industry (and therefore also in this book), data rates larger than 1 Gbps were considered high. Data rates larger than 10 Gbps were considered to be “very high”. In general, “very high” describes what is at the brink of feasibility at the time; also in this book. Ad c) Automotive SerDes supports P2P communication (only). At the physical layer, SerDes communication is P2P. This does not only mean that the SerDes link is not a bus, where more than two units would share the bandwidth, it also means that the complete SerDes communication starts at the one side of the communi­ cation and ends at the other, without extended networking capabilities. This suits especially camera and display use cases that only forward video data to or receive video data from the ECUs where the data is processed. Occasionally, Automotive SerDes architectures are discussed that envision a daisy chain of cameras or displays (see also Section 2.1.3). This is generally done to save hardware in the processing ECU and/or to reduce the needed cable length. On the physical layer ­anyway, but also on the Data Link Layer (DLL) the communication still typically remains P2P between each display/camera and the processing ECU. The cameras/displays do not ­communicate among each other as would be possible if the communication was truly networked. Ad d) Automotive SerDes supports the lowest two layers of the ISO/OSI communication model (only). As the Automotive SerDes communication is P2P, the respective technologies generally comprise the PHYsical layer (PHY) and some DLL functions. This means that of the seven different communication functions defined in the ISO/OSI layering model [11], Automotive SerDes only covers layer one and two. This in return means that Automotive SerDes technologies do not need communication-specific software. Any particular requirements that might affect the software are related to the handling of the application specific protocols, which might be part of the Automotive SerDes products or the application data transported across the SerDes link, but not the Automotive SerDes technology itself (see Section 9.6 and Section 9.7 for more details on the protocols). These are the general properties of “Automotive SerDes”. Yet another terminology with ­ambiguities refers to the actual chip products that are often just called “Serializer” and “Deserializer”. Figure 1.4 provides an overview. The term “Serializer (SER)” originally stands for the part that serializes and then transmits the data, the “Deserializer (DES)” for the part that receives and then deserializes the data. However, in modern Automotive SerDes technologies, the chip at the side of the communication that transmits the high data rate, also receives a smaller data rate for the control channel and the chip at the side that receives the high data rate also transmits a smaller data rate for control purposes. Both parts are, however, still called SER and DES. Furthermore, these now enhanced SERs and DESs can be integrated in a System on Chip (SoC) with the sensors, processing, or display control chips. They can also be part of stand-alone IVC bridge chips. In the automotive ­industry these bridge chips are also referred to as SER on the side that sends the high data rate and as DES on the side that receives the high data rate. This means, SER and DES might refer to three different sets of functionalities.

1.2 Background to Automotive SerDes

11

In order to reduce confusion, in this book, the bridge chip depicted in Figure 1.4 is called a “SerDes bridge”, a “SER-bridge”, or a “DES-bridge”, depending on the context. Just SER or DES, describes the function on one or the other side of the communication link discussed, including a potential control channel. When, in the following text, exceptionally the original meanings of SER and DES are relevant, it is explicitly mentioned. Note that SerDes bridge chips can come in a number of flavors. These depend on the application specific protocols they bridge into, and also on the number of SERs and/or DESs they incorporate. Among other possible combinations, dual and quad DES-bridges are particularly common. SER

DES

TX

RX

RX

RX

TX

TX

RX

RX

TX

TX

RX

RX

TX

DLL

DES

TX

Prot. Bridg x ing

DLL

Processing or display

DES integrated into a System on Chip (SoC)

DES bridge

DLL

DLL

SER bridge Sensor or processing

Functions of Automotive SerDes deserializers

SoC

DLL

“SER-bridge” often used synonymously with “serializer (SER)” in the automotive industry

Sensor or processing

(Prot.)

SoC SER integrated into a System on Chip (SoC)

(Prot.)

Adapt.

DLL

SER Functions of Automotive SerDes serializers

Original meaning of deserializer

Adapt.

Original meaning of serializer

Bridg Prot. ing y

Processing or display

“DES-bridge” often used synonymously with “deserializer (DES)” in the automotive industry

Figure 1.4 Different uses of the terms “SERializer (SER)” and “DESerializer (DES)” One last note on the terminology. The terms “SerDes” and “Automotive SerDes” are a relatively new phenomena in the automotive industry. The industry tried other names, such as “High Speed Video Links (HSVL)” [12], “pixel links” [13], or, most commonly, “LVDS”. Low Voltage Differential Signaling (LVDS) is a Serialization/Deserialization standard published in 1995 that combines low level signaling and differential communication (see also Section 7.2). It is often seen as the birthplace of SerDes and the early SerDes technologies used in the automotive industry were LVDS based. However, many modern Automotive SerDes technologies have nothing in common with the original LVDS. It is therefore no longer correct to use the term LVDS synonymously with Automotive SerDes. When the term “LVDS” is used within this book, it is used only when exactly LVDS is meant.

1.2.3 The Status of Automotive SerDes The first time a SerDes technology was used in a series production car was in 2001. In its new 7-series, BMW used SerDes to connect the center display to the main infotainment, where the graphic data to be displayed was being rendered. The sources of original video data, such as cameras or a TeleVision (TV) receiver, were designed to be transferrable over analogue transmission systems. The graphic data for navigation systems was a new type of data that did not automatically cater for analogue transmission but required a high resolution on top. The SerDes technology used was the first Flat Panel Display (FPD) SerDes technology from National Semiconductor (now Texas Instruments, TI). The overall transmission

12

1 Introduction and Background

rate was about 500 Mbps using four wire pairs (three for data and one for the clock, see also Section 7.3.1) and a separate CAN connection for the control data. Since then, the market has grown slowly but continuously. From 2005 on, Automotive ­SerDes solutions were even usable with dedicated, automotive suitable connectors; a fact not to be underestimated for the successful use of a communication technology (see Section 5.3.2 for more details). In 2021, the overall number of SerDes nodes in cars was expected to be about the same as the overall number of Ethernet nodes in cars [14]. The market growth had been accompanied by new features, such as higher data rates, integrated control ­channel, capabilities to transmit power with the data, support of coaxial cables and alike. Furthermore, more suppliers had entered the market, albeit offering their own non-in­ teroperable, proprietary versions of Automotive SerDes solutions (see also Section 7.3 for technical details). And while the original FPD-Link technology was opened to be used by other semiconductor vendors, all follow up versions were also proprietary. It is not so obvious, how the situation came about. After all, every technology used inside a car requires extra effort in terms of qualification (tools and test), logistics, and maintenance and that over many years (see also Section 3.1.2.2). If a car manufacturer decides to select just one supplier and technology to avoid multiplying the effort, the car manufacturer risks to be locked-in with a suboptimal technology down the road. This is because one vendor would need to supply the changing and growing portfolio alone, and it is unlikely that this one vendor will be the best choice for all chip variants needed. The monopolistic vendor might even lose the incentive to adapt and improve in the future. A living standard, for which a number of vendors is selling interoperable products, is the most desirable situation for a car manufacturer. It is likely optimized on various companies’ core competences, entails an eco-system for tools, tests, cables, and alike, and is bound to be developed further for future versions. So, why did this situation with various proprietary Automotive SerDes solution evolve? In the authors’ opinion, it is a combination of the following two aspects: first, fast advancements of camera and display technologies that swept into the automotive industry from outside, and second, the connectivity was (is) P2P at the edge of the IVC network outside strategic decision making. Furthermore, camera and display applications have always had a large car user visibility. Up to know, only proprietary technologies were able to support the new features, as fast as the automotive industry wanted to use them. At the same time, it did not matter as much when proprietary technologies were used, especially, when the two units at each end of the communication link were provided by the same Tier 1 supplier in a closed system. The Tier 1 supplier offers exactly what the car manufacturer requires and looks for a cost optimized solution in order to win the contract. The car manufacturer also wants the best possible features available for its customers. As long as the costs work out, the incentive to push for a standard in such a scenario is limited. While there might have been discussions on standardizing Automotive SerDes, until very recently though, they have not been followed through. There are a number of reasons, why the situation with respect to standardization has changed just now. First of all, it is a matter of sheer volume. The number of cameras and displays in cars is growing, while at the same time analogue connections for these applications are being phased out. Second, the car manufacturers are envisioning EE-architectures, in which c­ameras, high-speed sensors, and displays are bought from different Tier 1s than the ECU processing the data, potentially even with different time lines.

1.3 Background to Automotive Ethernet

With this, a) being more flexible with respect to the IVC supplier for units to be exchanged becomes more important, and b) it is now the car manufacturer who is fully responsible for the communication link. Third party support for compliance and interoperability is key. Having a standardized solution makes these things easier. The result is two Automotive SerDes standardization efforts. The MIPI Alliance (originally abbreviated from Mobile Industry Processor Interface Alliance) announced the development of their “A-PHY” in 2018 [15] and the release of its A-PHY 1.0 specification in September 2020 [16]. The Automotive SerDes Alliance (ASA) was founded in May 2019 and announced the completion of its “ASA Motion Link (ASAML)” specification in October 2020 [17]. Both standards support anticipated upcoming requirements on Automotive SerDes solutions in order to sustain changing requirements (for technical details see Sections 7.4 and 7.5). At the time of writing, both standards were too new to have been adopted in series production cars.

„ 1.3 Background to Automotive Ethernet Compared with when Ethernet was invented  – in 1973 [18]  – Automotive Ethernet is a fairly recent event. 2008 saw the first series production car with Ethernet communication, a BMW 7-series. In 2013 the first series production car using a dedicated Automotive ­Ethernet communication technology, a BMW X5, was put on the road. In contrast to the term “SerDes”, the term “Ethernet” is comparably unambiguously used. Section 1.3.1 ­therefore only provides an outline of the origin of Ethernet and what Ethernet comprises. Section  1.3.2 then explains the distinction of “Automotive Ethernet” before Section 1.3.3 introduces the HS Automotive Ethernet technologies discussed in more detail in Chapter 8.

1.3.1 The Origin of “Ethernet” Ethernet originated in 1973 at Xerox PARC, when Xerox PARC needed a communication technology that allowed to transmit data between its first personal computer workstations, its printers, and the early internet. Xerox PARC did not keep the invention to itself but published it in various steps, first as a paper in 1976 [19], then jointly with Digital Equipment Corporation (DEC) and Intel as the “DIX Standard” in 1980 [20], and then as 10BASE-5 for 10 Mbps over thick coax cable; the first Institute of Electrical and Electronics Engineers (IEEE) Ethernet standard of the newly founded group 802.3 in 1983 [21]. Since then, numerous Ethernet variants have been published especially, but not only, by IEEE 802.3 [21] and numerous industries have adopted Ethernet communication. The ­specifications cover different transmission modes (half-duplex and full-duplex), different data rates (from 10 Mbps to 400 Gbps), different transmission media (coaxial, twisted pair, twinax, variations of optical media, backplane), and added features such as flow control, auto negotiation, link aggregation, Power over Ethernet/DataLine (PoE, PoDL, see also Section 6.1.2) and Energy Efficient Ethernet (EEE, see also Section 8.2.1.1). The industries that have adopted Ethernet communication in their products comprise, next to the data centers

13

14

1 Introduction and Background

where Ethernet originated, telecommunication, industrial automation, aviation, building automation, Professional Audio and Video (Pro-AV) and more [22]. Ethernet has two core properties: It always only covers the lower 1½ layers of the ISO/OSI layering model (see Figure 1.5) and all Ethernet PHY variants use the same packet/frame format towards the Medium Access Control (MAC) in the DLL (see Figure 1.6). To always only cover the lower 1½ layers and, with that, to provide just a data container as inde­ pendent from the application as possible (so that anyone can decide for themselves what application data to use Ethernet for), is seen as a major contribution to the success of Ethernet [18]. Always having the same frame format means that the microcontrollers (μCs) that process the packets and the switches that forward the packets within the network always see (and always have seen) the same frame structure independent of the physical layer technology used. The difference the speed grade makes is that the MACs integrated into the μCs or switches have to be able to process the data at a speed that matches the one of the PHY. And of course, both PHY and connected IC need to support the same Media Independent Interface (MII) type (see Figure 1.5). The MAC always has the same function; what differs is how fast the functions are performed. 2.2.2 Bridging 2.2.1 MAC Control

7. Application

2.1.1 MAC

6. Presentation 2.3 Logical Link Control

4. Transport

2.2 Bridging & MAC Control

3. Network

2.1 Media Access Control (MAC)

2. Data Link

1.2 Physical Signalling

Media Dependent Interface (MDI)

1. PHY

1.1 Media Specification

1.1.1 Medium

IEEE coverage

802.3 medium access specification

Media Independent Interface (xMII)

5. Session

ISO/OSI layers

802.1, 1722, TSN/QoS specifications

1.2.3 Reconciliation 1.2.2 Physical Coding 1.2.1 Physical Media Attachment

802.3 physical layer specification

IEEE Ethernet specification work

Figure 1.5 ISO/OSI layers covered with IEEE Ethernet specifications Figure 1.6 shows the elements of an Ethernet packet. Every packet starts with a preamble and a start frame delimiter that have the task to alert and enable the receiver to receive the upcoming data. Next, each packet contains the MAC addresses of the recipient and the sender. The optional IEEE 802.1Q tag can be used to add Virtual Local Area Network (VLAN) information and priorities. Each recipient then expects the Ethertype. The 2-byte Ethertype is a core Ethernet element. It identifies the type of payload or header extensions (such as the 802.1Q tag or MACsec, see Section 8.2.3) to be expected. The list of Ethertypes is maintained by the IEEE [23]. The following payload then has a length of minimum 42 bytes (with IEEE 802.1Q tag) or 46 bytes (without IEEE 802.1Q tag) and a maximum length of 1500 bytes. The packets end with a Cyclic Redundancy Check (CRC) also called Frame Check Sequence (FCS), which causes a packet with a detected error to be dropped. Between packets there always is a gap of minimum twelve bytes length. The Ethernet packet format is decisive for the processing

1.3 Background to Automotive Ethernet

15

and forwarding of Ethernet on the MAC layer. A fundamental requirement for every new Ethernet PHY technology is thus, that it supports the frame format towards the MAC layer. On the physical channel, the packet format shown in Figure 1.6 is not visible.

Preamble

Start frame delimiter

Destination address

Source address

802.1Q tag (optional)

Ethertype

Payload

7 bytes

1 byte

6 bytes

6 bytes

4 bytes

2 bytes

42/46–1500 bytes

CRC/FCS Interpacket gap 4 bytes

Figure 1.6 Elements of an Automotive Ethernet packet (not to scale) Most of the management and QoS functions are realized in the “switch”, which is called “bridge” in IEEE specifications. Here the term “bridge” – “layer 2 bridge” to be more precise – covers the communication path between two different Ethernet PHYs on the ISO/OSI layer 2 (a layer 3 bridge would do so on IP level). In order not to confuse an Ethernet bridge with the IVC-bridge introduced in Section 1.1, this book uses the term “switch” for the IC that forwards packets in an Ethernet network on layer 2 based on the MAC addresses. It is not relevant whether the IC additionally has the PHYs integrated or not. The switch functionality stays separate. In data centers, an “Ethernet switch” is often a standalone box which allows to connect many Ethernet cables. Such a unit is unusual in an automotive network and would more likely be called a “switch box” in the car industry. The switch realizes most of the Ethernet Quality of Service (QoS) functions within an Ethernet network. The Ethernet QoS is defined by the IEEE 802.1/1722 Time Sensitive Networking (TSN) activity (see Section 8.2.2 for relevant details). It is important to point out that normally in an Ethernet network, the communication is just “Ethernet-based”. This means all other layers of the ISO/OSI layer model are addressed, too, with the use of, for example, the Transmission Control Protocol/Internet Protocol (TCP/IP) protocol suite. TCP/IP is not Ethernet, but commonly used in an Ethernet network. When the network functionality is not needed, as is the case for many P2P sensor and display applications, Ethernet might be used a lot closer to its core without the protocol overhead. The Success of Ethernet When Ethernet was invented, it was by no way obvious that Ethernet would persevere. Strictly sticking to the ISO/OSI layer separation and covering the lower 1½ layers only, was only one aspect of its success. Another important reason for the success of Ethernet was that it was supported by a broad community, without a single company dominating the standardization process [24]. Apparently, in the late 1970s/early 1980 not only the DIX group but also IBM realized, that it would be advantageous to have an open communication standard for Local Area Networks (LANs). The first intention was to agree to a common solution within IEEE. In 1980, when it became obvious that the group was “hopelessly divided”, the IEEE split the effort into three groups: 802.3 for Ethernet, 802.4 for Token Bus, and 802.5 for Token Ring [24]. In the following years, Ethernet competed especially with Token Ring.

Min. 12 bytes

16

1 Introduction and Background

To start with, Token Ring was seen as technically superior and had, at least at the beginning, a higher adoption rate. Its channel access was more deterministic with bounded latency, the topology was more reliable, and the network easier to manage and extend [24] [25]. Nevertheless, Ethernet succeeded. The main reason is seen in the dominance of IBM in the Token Ring technology. IBM did not only have more advanced products available (almost) before the new features had actually found their way into the standard. Also, some higher layer management protocols were dominated by proprietary IBM solutions. This made it difficult for other companies to compete with interoperable products and investments were too risky. For Ethernet, in contrast, the originators made efforts not to dominate the ­standardization, and the large number of (small) companies investing in products made, in the end, the difference [24]. 

1.3.2 Ethernet in the Automotive Industry The BMW 7-series that went into series production in 2008 used 100BASE-TX Ethernet, a technology released by the IEEE in 1995 for IT applications [21]. 100BASE-TX was introduced at BMW because there was the requirement to transmit packet data at higher speeds than were supported by other communication technologies available for automotive use at the time. The reason behind the increase in data rate was the map data required for the GPS-based navigation system. It was for the first time stored inside the main infotainment ECU, the Head Unit (HU), instead of using an external media. This caused a) a huge increase in the amount of data that needed to be flashed into the car in case of updates and b) the data was to be accessible also from the high-end Rear Seat Entertainment (RSE) system. At the time, duplicating the map data to a flash storage inside the RSE was simply too ­expensive. It was significantly more cost efficient to access the map data via Ethernet from the HU. The 100BASE-TX links served the purpose. However, using 100BASE-TX during the runtime of a car required the use of Shielded Twisted Pair (STP) cables with two wire pairs. This was not cost efficient enough to make 100BASE-TX a viable solution for extended incar use. It required finding an Ethernet solution that transmitted 100 Mbps data over a single pair of Unshielded Twisted Pair (UTP) cables to have Ethernet gain traction in the automotive industry. This technology, which was later called 100BASE-T1, went first into series production in a BMW X5 in 2013. Interestingly enough, in this introductory use case 100BASE-T1 connected the surround view cameras to the Surround View System (SVS) ECU. The predecessor system, introduced with the then new X5 in 2006, had actually used SerDes for the first time to connect cameras in a BMW car. In 2013, the camera data was compressed in order to fit over a 100 Mbps link. At the time, compressing the camera data and transmitting it over a single Unshielded Twisted Pair (UTP) cable per camera simply had a better business case than continuing to use an Automotive SerDes technology that, while not needing compression, required STP ­cables [22]. As a consequence, Automotive Ethernet and Automotive SerDes are often seen as competing technologies, even though there are actually very few use cases in which they might compete. Because these use cases, nevertheless, are the high-speed sensor and

1.3 Background to Automotive Ethernet

display applications discussed in this book, Section 8.3 details the relationship and competition between the two technologies. The Ethernet technologies requiring a single twisted pair cable for transmission are identified at IEEE with the suffix “T1”. Because these technologies have primarily been developed for automotive use and because they fulfil the automotive requirements with respect to link length and robustness, the T1-technologies are often summarized as “Automotive Ethernet”. In a broader sense, “Automotive Ethernet” might be and is also used instead of “Ethernet-based communication in cars”, which then comprises all the higher protocol layers as well. Because Ethernet was introduced into cars at 100 Mbps, 100BASE-T1 Ethernet was the most successful Automotive Ethernet technology at the time of writing. It had gone into ­series production at almost all major car manufacturers by 2021 [26]. For further Automotive Ethernet variants see Section 1.3.3. Technical details of the HS Automotive Ethernet technologies are provided in Chapter 8.

1.3.3 Introduction to High-Speed (HS) Automotive Ethernet Automotive Ethernet was introduced into the car industry with 100 Mbps data rate. However, it was immediately clear that in order for Ethernet to persevere in the car industry, the support of other – first of all higher – data rates needed to be enabled as well. As standardization efforts for a technology generally need to start eight to ten years ahead of an envisioned Start Of Production (SOP) (see also Section 3.1.2.2 or [27]), first discussions about the standardization of Gbps Ethernet started in 2011. This was two years before the first Automotive Ethernet technology for 100 Mbps Ethernet even had its SOP. Likewise, the Automotive Ethernet standardization effort for Multi-Gbps Ethernet was kicked off just after the 1 Gbps Automotive Ethernet specification had been completed in 2016, which again was years before the Gbps technology was actually brought onto the road, while the larger than 10 Gbps Automotive Ethernet standardization efforts were ongoing in 2021, years before any of the multi-Gbps Ethernet speeds had been adopted by any car manufacturer into a series production car. Table 1.2 provides an overview on the Automotive Ethernet physical layer technologies specified at IEEE. Because video is the main driver for data rate, all of the listed projects addressing data rates of 1 Gbps or more, mentioned camera or display use cases as motivations for the technology development [3] [27] [28] [29]. The transmission of aggregated sensor data within a network is also seen as a reason for supporting higher data rates. Naturally, with Ethernet being a networking technology other bandwidth hungry use cases have also been listed. These are: software updates and diagnosis, communication between high-performance processing and/or zonal ECUs, and communication to the antenna module connecting to the mobile network. The data rate for the communication between HU and antenna module is thereby also driven by video (and software updates).

17

18

1 Introduction and Background

Table 1.2 Overview on the availability of Automotive Ethernet speed grades [21] [26] Name

Speed

Cabling

Started

Completed

SOP

IEEE 802.3cg

10 Mbps, typically unidirectional (shared)

UTP

2016

2019

2025 (expected)

100 Mbps bi-directional

UTP

2013 (2008*))

2015 (2011*))

2013

1 Gbps bi-­ directional

Jacketed UTP or STP

2011

2016

2019

1 Gbps per fiber and ­direction

Two Plastic Optical Fibers (POF)

2012**)

2017

2019

2016

2020

2026 (expected)

10BASE-T1S IEEE 802.3bw 100BASE-T1 IEEE 802.3bp 1000BASE-T1 IEEE 802.3bv 1000BASE-RH IEEE 802.3ch MGBASE-T1 IEEE 802.3cz MGBASE-AU IEEE 802.3cy MGBASE-T1 or T2 or T4

2.5, 5, 10 Gbps STP bi-directional 2.5, 5, 10, 25, 50 Gbps per fiber and ­direction

MMF GOF

2019

2022 (expected)

open

25 Gbps ­bi-directional

STP (reduced length)

2018

2023 (expected)

open

 *) Numbers in brackets relate to the One Pair EtherNet (OPEN) Alliance, where the technology was published first as “BroadRReach”. **) The standardization efforts started first at the Verband Deutscher Elektrotechniker (VDE).

Table 1.2 also lists two Automotive Ethernet projects for optical data transmission, 1000BASE-RH and MGBASE-AU. At the time of writing, these projects seemed to play a minor role in the automotive industry. However, without optical data transmissions the world wide network we know today would not exist. Optical data transmission has distinct advantages, for example with respect to Electromagnetic Compatibility (EMC) and  – depending on the exact optical medium used – attenuation and link budget. The attenuation the signal experiences when being transmitted is significantly smaller over optical media. These aspects might play a more important role also in future IVC networks. Whether this will include sensor and display communication remains to be seen. Note that for the sensor and display use cases discussed in this book, Ethernet is used in three different ways: first, as a control channel, second, for compressed sensor/video data, third, for uncompressed sensor/video data. The first SerDes technology to support Ethernet as a control channel was Inova’s Automotive PIXel (APIX) technology (for details, see Section 7.3.3). When Ethernet is used for compressed sensor/video data at a lower data rate, this is a standard, almost application independent use case for Automotive Ethernet and the reader might consider referring to [22]. For details on the HS Automotive Ethernet technologies needed for uncompressed sensor/video data transmission, see Chapter 8.

1.4 Bibliography

„ 1.4 Bibliography   [1]

S. Brunner and C. Gauthier, “Automotive Ethernet and SerDes Technologies; Friends or Foes?,” in: Automotive Ethernet Congress, Munich, 2020.

 [2]

J. Shuttleworth, “SAE Standard News,” SAE International, 7 January 2019. [Online]. Available: https://www.sae.org/news/2019/01/sae-updates-j3016-automated-driving-graphic. [Accessed 11 August 2021].

 [3]

S. Carlson, K. Matheus, T. Hogenmüller, T. Streichert, D. Pannell and A. Abaye, “Reduced Twisted Pair Gigabit Ethernet PHY Call for Interest,” March 2012. [Online]. Available: https://www. ieee802.org/3/RTPGE/public/mar12/CFI_01_0312.pdf. [Accessed 29 March 2021].

 [4]

D. R. Stauffer, J. T. Mechler, M. Sorna, K. Dramstad, C. R. Ogilvie, A. Mohammad and J. Rockrohr, High Speed Serdes Devices and Applications, New York: Springer, 2008.

 [5]

F. Zhang, High-speed Serial Buses in Embedded Systems, Singapore: Springer Nature, 2020.

 [6]

Wikipedia, “Moore’s Law,” 20 March 2021. [Online]. Available: https://en.wikipedia.org/wiki/ Moore%27s_law. [Accessed 22 March 2021].

  [7]

S. Buntz, “Update on Required Cable Length,” 11 July 2012. [Online]. Available: https://grouper. ieee.org/groups/802/3/RTPGE/public/july12/buntz_02_0712.pdf. [Accessed 26 March 2021].

 [8]

CAN in Automation, “LIN standards and specifications,” 2021, continuously updated. [Online]. Available: https://lin-cia.org/standards/. [Accessed 4 April 2021].

 [9]

A. Gzemba, MOST, The Automotive Multimedia Network, Munich: Franzis, 2008.

[10]

W. Voss, A Comprehensible Guide to Controller Area Network, Greenfield: Copperhill Media Corporation, 2005.

[11]

W. Stallings, “The Origin of OSI,” 1998. [Online]. Available: http://williamstallings.com/Extras/ OSI.html. [Accessed 6 May 2020].

[12]

S. Buntz, T. Kibler, H. Leier and H. Weller, “Koaxialer High-Speed-Video-Link für Automobil­ anwendungen,” Elektronik Net, 7 May 2012. [Online]. Available: https://www.elektroniknet.de/ automotive/infotainment/koaxialer-high-speed-video-link-fuer-automobilanwendungen.88099.html. [Accessed 27 March 2021].

[13]

K. Matheus and T. Königseder, Automotive Ethernet, Second Edition, Cambridge: Cambridge University Press, 2017.

[14]

Strategy Analytics, “Data Extraction on Customer Request,” Strategy Analytics, Milton Keynes, UK, 2019.

[15]

MIPI Alliance, “MIPI Alliance to Advance Autonomous Driving, other Automotive Applications with New Data Interface Specifications at 12-24 Gbps and Beyond,” 2 August 2018. [Online]. Available: https://www.mipi.org/mipi-to-advance-autonomous-driving-other-automotive-applications.  https://www.mipi.org/mipi-to-advance-autonomous-driving-other-automotive-applications [Accessed 29 March 2021].

[16]

MIPI Alliance, “MIPI Alliance Releases A-PHY SerDes Interface for Automotive,” 15 September 2020. [Online]. Available: https://www.mipi.org/MIPI-Alliance-Releases-A-PHY-SerDes-Interfacefor-Automotive. [Accessed 28 March 2021].

[17]

S. Brunner, “Automotive SerDes Alliance (ASA) Completes the First Automotive SerDes Standard with Integrated Security,” Automotive SerDes Alliance, 13 October 2020. [Online]. Available: https://auto-serdes.org/news/automotive-serdes-alliance-asa-completes-the-first-automotive-serdesstandard-with-integrated-security-325/. [Accessed 28 March 2021].

19

20

1 Introduction and Background

[18]

C. E. Surgeon, Ethernet: The Definite Guide, Sebastopol, CA: O’Reilly, 2000, February.

[19]

D. Boggs and R. Metcalfe, “Ethernet: Distributed Packet Switching for Local Computer Networks,” Communications of the ACM, vol. 19, no. 7, pp. 395–405, July 1976.

[20]

Digital Equipment Corporation, Intel Corporation, Xerox Corporation, “The Ethernet, A Local Area Network. Data Link Layer and Physical Layer Specifications, Version 1.0,” 30 September 1980. [Online]. Available: http://ethernethistory.typepad.com/papers/EthernetSpec.pdf. [Accessed 6 May 2020].

[21]

Wikipedia, “IEEE 802.3,” Wikipedia, 3 March 2021. [Online]. Available: http://en.wikipedia.org/ wiki/IEEE_802.3. [Accessed 28 March 2021].

[22]

K. Matheus and T. Königseder, Automotive Ethernet, Third Edition, Cambridge: Cambridge University Press, 2021, pp. 75-108.

[23]

IEEE, “Ethertype,” constantly updated. [Online]. Available: http://standards.ieee.org/develop/­ regauth/ethertype/eth.txt. [Accessed 7 May 2020].

[24]

U. v. Burg and M. Kenny, “Sponsors, Communities, and Standards: Ethernet vs. Token Ring in the Local Area Networking Business,” Industry and Innovation, vol. 10, no. 4, pp. 351–375, 2003.

[25]

Wikipedia, “Token Ring,” 11 March 2022. [Online]. Available: https://en.wikipedia.org/wiki/­ Token_Ring. [Accessed 12 March 2022].

[26]

K. Matheus, “Automotive Ethernet Beyond Speed,” in: Automotive Ethernet Congress, Virtual event, 2021.

[27]

S. Carlson, H. Zinner, N. Wienckowski, K. Matheus and T. Hogenmüller, “CFI Multi-Gig Automotive Ethernet PHY,” November 2016. [Online]. Available: https://grouper.ieee.org/groups/802/3/ cfi/1116_1/CFI_01_1116.pdf. [Accessed 12 February 2021].

[28]

C. Pardo, H. Goto, T. Nomura and B. Grow, “Automotive Optical Multi Gig Call For Interest ­Consensus Presentation,” July 2019. [Online]. Available: https://www.ieee802.org/3/cfi/0719_1/ CFI_01_0719.pdf. [Accessed 29 March 2021].

[29]

S. Carlson, C. Mash, C. Wechsler, H. Zinner, O. Grau and N. Wienckowski, “10G+ Automotive Ethernet Electrical PHYs,” March 2019. [Online]. Available: https://www.ieee802.org/3/cfi/ 0319_1/CFI_01_0319.pdf. [Accessed 29 March 2021].

2

The Automotive Use Cases

Key for the use cases discussed in this book is the highly asymmetric nature of the required communication technology; high speed in one direction and low speed in the other, at the edge of the IVC network. In the following, this chapter first introduces display use cases, see Section 2.1, then camera use cases, see Section 2.2, then other high speed sensor types, see Section 2.3, while closing with an outlook on use cases outside of the three previous cate­gories, see Section 2.4.

„ 2.1 Displays An electronic display, as used inside cars, is a unit that has a flat screen on which characters or images are presented to the user for the visual perception of INFOrmation and enterTAINMENT (infotainment) data based on electrical input. Compared with other environments in which displays are used, the physical constraints inside cars are noteworthy. Not only are there special requirements on the mechanical robustness and visibility under all kinds of conditions, also the spaces in which displays can and should be used are limited. A key challenge for displays inside cars is thus the right location and technical realization that allows displayed content to be well visible when needed for the particular purpose without being distracting at all other times. In the context of this book, it is not relevant whether a display is single purpose and displays one type of content only, like a rear-view mirror replacement display, or used for a variety of different applications, like the center stack display. While it makes a huge difference for the graphics processor, it does not directly impact the communication. It is therefore not considered in the following discussion. Instead, the discussion comprises a brief ­history of display usage in cars in Section 2.1.1, the introduction of relevant technical terms in Section 2.1.2, and a display architecture overview in Section 2.1.3. Section 2.1.4 closes this chapter with a set of typical display requirements.

22

2 The Automotive Use Cases

2.1.1 A Brief History of Displays in Cars Introducing the display use case has two strands: one relates to the development of display technologies (usually happening outside the car industry), the other to the need to communicate information to the car users and especially the drivers. From early on, drivers of cars needed information related to the driving function. As early as 1910, speedometers were standard equipment. With help of a magnet that created an eddy-current from the rotation of the wheels, a respective dial was activated [1]. Tachometers that measure the rate of ­rotation of the engine’s crankshaft were an even earlier invention, which was useful in cars with manual gear shift [2]. Another extremely important piece of information was (and still is) the filling level of the fuel tank. The fuel gauge is also an early invention, made in 1917 [3]. These type of meters and gauges had to be placed such that they were well visible to the car driver. Combined in the “instrument cluster”, they were, and still are, often placed directly behind the steering wheel. For a very long time, and in some cars even up to today, the realization of such an instrument cluster was/is analogue. In 1976 the Aston Martin Lagonda was the first car in which the car’s speed was presented in digits. Instead of a magnet, a sensor was used to measure the speed. The displaying concept of the Lagonda [1] [4] – first using Light Emitting Diodes (LEDs) then a Cathode-Ray Tube (CRT) display, see Section 2.1.2 for more information on CRTs – did not persevere though. An LED display was too expensive at the time and CRTs cumbersome, too large and complex to handle. CRT is the same technology that had been used in TV sets since 1934 and that dominated also computer monitors until the year 2000 [5]. While, as is explained in more detail below, it required Liquid Crystal Displays (LCD) and Organic Light-Emitting Diodes (OLEDs) to make displays really successful in cars, measuring the speed with a sensor instead of a magnet, was adopted by the rest of the car industry almost immediately after its market intoduction with the Aston Martin. [1]. What had more impact on the use of displays in cars was the entertainment. The first in-car radios were sold as early as 1930 [6] and more than 60% of cars had in-car radios by 1963 [7]. It started in the 1980s that in-car radios were enhanced with small digital screens to display the radio frequency or equalizer outputs [4]. In a close interrelation with technical advancements of display technologies, it was possible to increase the size of the radio displays and car manufacturers added more information for the drivers, such as temperature or fuel consumption. The authors even remember early implementations of driver monitoring (recommendations for making a pause after 2½ hours driving in VW/Audi) or the graphic display of the wear of the brake pads (BMW). The invention that really spurred changes, was in-car navigation. It changed the demand not only concerning the size and use of the displays but also for other aspects such as processing power, amount of software and data, types and size of electronic storage capabilities, data rates to be supported in the in-­vehicle network. Honda was the first car manu­ facturer to release a commercially available in-car navigation system in 1981 [8], Toyota the first to include a color display in 1987, and Mazda the first to offer a Global Positioning System (GPS-)based navigation system in 1990 [4] [9]. The function of navigation is inherently linked to driving. Generally, when people use cars, they intend to reach a specific destination. Especially when in unfamiliar terrain, car navigation systems are a huge improvement over needing to have the right paper map at hand

2.1 Displays

or needing to find someone to ask [3]. In-car navigation thus immediately attracted customers. However, it also was extremely expensive to provide. One cost factor was the display. All the above-mentioned early implementations used cumbersome CRT displays, which ­inhibited proliferation [10]. In order to be more cost efficient, the majority of the early navigation systems (at least the ones sold outside Japan) thus started with turn-by-turn navigation systems instead of displaying the complete map. The arrows needed were fitted into the small displays available with the in-car radios. Thirty years later, today, the use of navigation systems with full map displays in cars is standard. One important development for this was the rapidly expanding laptop market in the mid-1990s, which drove the proliferation of flat panel displays, especially LCDs [11] and Thin-Film Transistors (TFT) displays (which are a specific type of LCD). With some delay these displays made it into the cars, just like LEDs and OLEDs, allowing for more elaborate graphics. Naturally, all display technologies require specific in-car adaptations with respect to visibility and achievable contrasts. In-car displays need to work in all kinds of uncon­ trollable exterior light conditions. Furthermore, the screens need to have physical characteristics that enhance their robustness in case of accidents and ensure that they do not sliver when smashed. Last but not least, the advent of smartphones has accustomed consumers to display-centric control. Tesla set the pace for incorporating this into cars by offering large enough displays to match the smart phone experience to a driving situation. And while there are legitimate discussions on the distraction of drivers caused by the content on a screen or by having to control it via touch [12], visual display of information is one of the best ways to relay information to humans fast [13]. With decreasing costs and advanced technological developments allowing for it, the number and size of in-vehicle displays can thus rightly be ­expected to continue to grow [14] (for example from 164 million in 2018 to 350 million in 2025 [15]). The most important displays are center-stack displays, rear seat displays, and head-up displays [16]. Naturally, also the instrument cluster is digitized (in many cars) and thus a display. The largest display, at the time of writing, had been announced by Daimler (now Mercedes). It is a 56-inch-wide display spanning the entire dashboard (including the instrument cluster) in Mercedes’ new electric sedan EQS [17]. One last note to the role of in-car TeleVision (TV) in the development of in-car displays. With TV having dominated the entertainment scene for decades, in-car TV was, of course, also offered to customers early on (see [18] for an example from the 1960’s). One of the core features of TVs is that they require a screen, and the use case might therefore have been relevant to the display discussion. However, the in-car television never dominated neither the in-­vehicle display development nor the in-vehicle design. First, drivers get distracted when a TV is running in their field of vision, which is why a respective TV screen has to be blanked when the car is moving or cannot be in the driver’s field of view (see for example [19]). As in most cases cars have only the driver as an occupant, see [20], who is then not able to watch, there is little incentive to pay a premium for enabling in-car TV r­ eception. Second, before TV was digitized and video formats allowed for compression, in-car TV ­required a CRT display that – as has been mentioned earlier in this section – is costly and space consuming. Plus, the reception quality for analogue TV broadcast services was not very good in a moving car. Thus, in-car TVs were sold only as options and as separate units. It was therefore not well possible to reuse the TV display for other information. The change in display technologies,

23

24

2 The Automotive Use Cases

digitized video formats, and digital TV broadcast standards like Digital Video Broadcasting for Terrestrial (DVB-T or DVB-T2), Integrated Services Digital Broadcasting (ISDB), or ­Advanced Television Systems Committee (ATSC), improved the quality of TV reception while driving [21] (see also Section 2.4). However, in-car TV seems to have missed its ­market window, even if all cars were autonomously driving today, Internet and its video/TV on demand services have taken over [22]. Displays dedicated to entertainment inside cars thus must cater for modern media consumption more than for TV-reception. This trend supports the market development of displays and the fact that displays are more than ever important element in modern cars. Today, the most typical location for a display inside a car is the center stack, from which the driver can control various infotainment and comfort functions (first realized in a BMW 7-series in 2001 [23] [24]). Another display might be found behind the steering wheel for information such as speed, fuel gauge and alike instead of the traditional instrument cluster with analogue gauges. Yet another display for the driver is the head-up display, which is located as much as possible in the driver’s field of vision ahead, without distracting it. The goal is that the driver can be visually informed on navigation directions or current speed limits, without having to turn the head to the center stack or instrument cluster display. Frequently, the head-up display is realized as a projection onto the windshield. In this case, the technological realization differs from more conventional in-vehicle displays, without changing the basic requirements. Another use case are rear-view mirror replacement systems. Their displays are generally part of the actual rear-view mirror, which either presents camera data or simply the real mirrored image. They have the purpose to show an obstacle-free (without rear seat passenger heads, large loads, dividers between passenger cabin and freight hold or trunk, and alike) and potentially enlarged view to the back of the car. The images of wing mirror ­replacement systems are typically displayed in the column to the driver’s side of the windshield. Wing mirror replacement systems can improve the wind resistance properties of a car and overcome any blind spots. Next, there can be additional displays for front or rear seat passenger entertainment. To serve the front passengers, car manufacturers occasionally offer dual-view displays in the center stack (see, for example, [25]). Dual-view displays use a parallax barrier technology to allow to show two graphics on the same screen, of which each can be viewed only from a specific viewing angle [26]. A driver would see, for example the navigation data, while the front passenger could watch a movie on the same display. The number of displays might be increased further with displays in car doors or roofs again for entertainment or user interaction.

2.1.2 Display Basics and Terminology A key property of an electronic display is its specific resolution, which represents the overall number of individually addressable and controllable PICture ELements (pixels) on the horizontal and vertical axes of its screen, available for the representation of the image. Two important materials used for in-car displays are liquid crystals and OLEDs. In LCDs, a layer of liquid crystal is sandwiched between transparent, conductive polarizers that activate the light modulating properties of the liquid crystal depending on the applied voltage. In an OLED, the material between the electrodes is an organic compound that emits light based

2.1 Displays

25

on the electric current. The main difference between the two possibilities is that LCDs ­require a backlight, whereas OLED displays inherently emit light. OLED displays can thus be thinner and lighter than LCDs [27]. If a head-up display does not use a projector or laser to project the information onto the specifically prepared windshield, generally also a (transparent) OLED display is used [28]. A pixel is the smallest unit an original image is divided into. Generally: The more pixels per area, the more accurate is the representation of the original image and the higher the resolution. Figure 2.1 shows the horizontal and vertical resolutions of a number of specific display formats. It can be seen that the for decades prevailing TV standards Phase Alternation Line (PAL) and National Television System Committee (NTSC), have tiny resolutions – 0.44 and 0.3 MPixels – compared with what we are seeing today. Also, the originally prevailing aspect ratio has changed from 4 : 3 to 16 : 9 or similar.

15

14

13

12

11

10

9

8

7

6

5

4

3

2

1 0.5

4:3

17:9

3000

Vertical resolution (Vres) [pixel]

MPixel 16

2500

16:9

2000

8000

7500

7000

6500

6000

5500

5000

4500 4000 3500 3000 Horizontal resolution (Hres) [pixel]

2500

2000

1000

VGA/NTSC 640 x 480

HD 1280 x 720

1500

PAL 768 x 575

1000

FHD 1920 x 1080

2K 2048 x 1080

UHD-1 3840 x 2160

4K 4096 x 2160

5K 5120 x 2880

2x UHD-1 7680 x 2160

1500

500

Figure 2.1 Different display resolutions of specific video formats [38] [39] While any display format and ratio is in theory possible (and is used as needed, also in cars), TV and video standards have dominated not only technological advancements of displays but also customer expectations. TV and video standards thus provide an important reference for display solutions. For example: The High-Definition (HD) resolution was first made available with digital TV standards like DVB in Europe, ISDB in Japan, and ATSC, in the US [29] [30] [31]. Within a relatively short time, the digital TV allowed to more than double the resolution by supporting up to Full HD (FHD). In 2021, 4K or Ultra HD (UHD) resolutions were being adopted fast with content offers from video-on-demand streaming services, while 8K (7680 × 4320 pixels) was shown to be feasible yet lacking market attraction [32]. It might be a little surprising, that such large resolutions are also discussed in the context of cars. After all, the space for displays inside cars is limited, especially when comparing it with a home scenario. Inside a car, it is hard to imagine how to fit a 42 inch screen – with exceptions, see the example of Mercedes in Section 2.1.1 or the new 8K panorama display of BMW [33] – even though 42 inches is at the lower end of what is common in homes [34].

500

26

2 The Automotive Use Cases

Additionally, car passengers are at a close, fixed viewing distance to the respective screens, which is contra productive to extra-large screens. One reason that large resolutions are discussed nevertheless for cars is that an increase in resolution does not necessarily mean an increase of the size of the screen. The size of the screen also depends on the Pixels Per Inch (PPI) or Dots Per Inch (DPI) value. 100 PPI would be a very high value for a home TV screen; values in the range of 50 PPI are more common. In cars, 300 PPI might just be perceived as a good image quality, while a good smart phone has 500 PPI and displays with over 10,000 PPI are already possible [35] [36] [37]. The color control of every pixel is normally based on the Red Green Blue (RGB) color model. RGB, because the largest part of visible colors can be derived from mixing red, green, and blue lights, each with the respective intensity needed [40]. In an electronic color display each pixel thus combines three – or four, if the display additionally uses white – separate but physically very close light sources that together give the impression to represent just one color [41]. The number of bits used to encode the pixel information is often called color resolution or depth. Monochrome displays often use eight bits, allowing for 256 different shades. For color displays 18 or 24 bits per pixel (for all three colors together) are common. This allows for 262,144 or 16,777,216 tones of color respectively [42]. 30 bits per pixel yields more than one billion tones. Inside the displays, the binary code words are converted into the respective voltages to drive the pixels. The more bits per color, the finer the color resolution. However, not all displays truly have 24 or 30 bits per pixel available. In such case, the image processing might apply dithering or Frame Rate Control (FRC) to emulate a higher color resolution than available. Dithering reproduces missing colors with help of specific pixel patterns of neighboring pixels using the available colors [43]. FRC alternates the available colors of a specific pixel with every frame such that the impression of a different color is given [44]. Note that RGB always requires some color management and calibration, as different devices might present the same RGB value differently. Many people have experienced this personally, when comparing the presentation of one and the same image on two different devices. The gamut, meaning the range of colors that can be displayed, is actually smaller for ­standard RGB (sRGB) [45] than for many other color systems [46]. However, RGB still is one of the most common (additive) color systems used and many other color models like YUV and its variants are derived from RGB (see also Section 7.1. and Section 9.1). Concerning the overall color management, it is important to know that the human eye has a better color sensitivity in dark tones than in light tones. This helps us to see over a broader range of light. If video was displayed using a linear color system, it would either need too much bandwidth to display all colors correctly or have too little variation in the dark tones. Therefore, the so-called “Gamma correction” is applied to digital (video) images. It redistributes tonal levels more in line with the human perception such that more bits are available for darker tones and fewer for lighter. Overall, this significantly reduces the number of bits needed for the images and is an important component in digital image processing [47]. To address the array of pixels in a display, a matrix is used to address each pixel individually. In this so-called “active matrix” this is done with help of a capacitor and a transistor per pixel. The transistor controls the voltage. The capacitor holds the pixel state while other pixels are being addressed. Often, the expression TFT display is used. It refers to the TFTbased, active matrix addressing of, for example, an LCD [48].

2.1 Displays

It is relatively intuitive that even in case of the comparably small NTSC resolution of 640 × 480 = 307,200 pixels, it is not possible to perform parallel processing and to connect each pixel/transistor with a dedicated wire. Serialization is needed. So, each individual frame of a moving image is read into the display row by row (which, by the way, was also done in case of analogue TV, see also Section 7.1). This leads to another important aspect of display technologies: blanking. Blanking originates from analogue TV or, more precisely, from the CRTs used for analogue TV. Within these CRT displays, (also) line and frame-wise moving beams of electrons generate the visible image on the screen. When reaching the end of a line, the beam needs to be switched off and repositioned at the beginning of the next line (see gray dashed line in Figure 2.2). The magnetic field that moves the beams of electrons inside the CRT, needs to perform a large change in order to achieve the repositioning. This cannot happen instantly. The required time is called horizontal blanking interval and during this time, no video i­ nformation is broadcasted. Likewise, when reaching the end of a frame the beam needs to be switched off and returned to the beginning of the next frame (see gray dash-dotted line in Figure 2.2), which causes the vertical blanking interval. LCDs or OLED displays do not require the same blanking intervals in order to display content correctly. However, because it was part of the TV standards, also presenting images on a digital display accommodates the blanking periods. This does not only have to do with backwards compatibility to analogue communication systems. The blanking periods were and are often used also in a digital environment to transmit other information like audio, test data, time codes, teletext, and more [49]. Display Line feed (visible) Line feed (not visible) Line return/change Image return/change Visible area

Activity

...

t

Figure 2.2 Actual line feed in displays showing the blanking areas The overall video data rate fed into a display depends on the horizontal and vertical resolutions, the overall bit depth (distributed over the three colors red, green and blue), the number of frames transmitted per second (fps), and the blanking overhead. Note that the size of the blanking overhead varies, depending on the exact transmission and screen formats ­selected. As the blanking period in modern digital displays is no longer needed to cover the time needed for readjusting the magnetic field inside the CRT but for transmitting additional information (especially audio in case of Consumer Electronics (CE) displays), it just needs to be ensured that the “blanking” gaps are long enough such that this information

27

28

2 The Automotive Use Cases

still fits. Table 2.1 gives some examples for the resolutions presented in Figure 2.1 plus adding the information on the 8K format. When looking at the resulting data rates, it is understandable that video transmission has always been a major driver for new communication standards in the consumer industry. Considering the importance of video for CE, it is also understandable that quite a few protocols have been developed to support the display use case (see Section 9.7 for details). Table 2.1 Example data rates in Gbps depending on the resolution, the fps, the blanking, and the bit color resolution1 Name

Hres

Vres

fps

Blanking 12 bits

18 bits

24 bits

30 bits

VGA/NTSC

 640

 480

60

16%

 0.26 Gbps

 0.38 Gbps

 0.51 Gbps

 0.64 Gbps

PAL

 768

 575*) 60

13%

 0.36 Gbps

 0.54 Gbps

 0.72 Gbps

 0.9 Gbps

HD

1280

 720

60

 9%

 0.73 Gbps

 1.09 Gbps

 1.45 Gbps

 1.81 Gbps

FHD

1920

1080

60

 6%

 1.58 Gbps

 2.37 Gbps

 3.16 Gbps

 3.95 Gbps

2K

2048

1080

60

 7%

 1.70 Gbps

 2.55 Gbps

 3.40 Gbps

 4.26 Gbps

UHD-1

3840

2160

60

 5%

 6.27 Gbps

 9.41 Gbps

12.54 Gbps

15.68 Gbps

4K

4096

2160

60

 5%

 6.68 Gbps

10.02 Gbps

13.36 Gbps

16.70 Gbps

5K

5120

2880

60

 4%

11.09 Gbps

16.63 Gbps

22.18 Gbps

27.72 Gbps

2x UHD-1

7680

2160

60

 4%

12.41 Gbps

18.62 Gbps

24.83 Gbps

31.04 Gbps

UHD-2

7680

4320

60

 4%

24.82 Gbps

37.24 Gbps

49.65 Gbps

62.06 Gbps

8K

8192

4320

60

 4%

26.46 Gbps

39.69 Gbps

52.92 Gbps

66.15 Gbps

*) Some publications also use 576.

In Table 2.1 a frame rate of 60 fps has been assumed for all formats. For some use cases, for example when emulating analogue dials from an instrument cluster on a d ­ igital display, 120 fps is being discussed. This roughly doubles the example data rates of Table 2.1 (roughly, because of the variations in the vertical blanking). A key invention that comes with digital video (transmission) is therefore video compression (see also Section 9.2). It allows to somewhat decouple the possible resolutions and bit depths from the data rates needed for the video transmission. Terrestrial, digital TV broadcast would be unthinkable without video compression. Also, Internet video services heavily rely on compression ­algorithms. Our entire modern media consumption would look completely different without them (see, for example, [50]). In cars, whether compression is possible and/or recommendable or not, generally depends on the requirements with respect to latency and potential compression losses. For display use cases, the displayed image quality is the most relevant, and the impact of compression needs to be investigated especially for lossy compression formats in detail. Section 9.2 provides more detailed information on frame-based compression formats like H.264 and H.265 as well as pixel/line-based compression formats like Display Stream Compression (DSC). Note that different horizontal resolutions are possible and common for NTSC and PAL. The resolution numbers have been selected based on [38]. To calculate the blanking overhead, a fixed number of 80-pixel clocks was used for the horizontal blanking and minimum 460 µs was used for the vertical blanking (the respective value in full lines), which, to the authors’ assessment seem to be the minimum values necessary. Naturally, other assumptions for the blanking (especially the horizontal blanking) yield different results.

1)

2.1 Displays

Latency is not as critical for the display use case as it might be for camera and other sensor applications (see also the following Section 2.2 and Section 2.3), because displays are for human perception only. In the authors’ opinion, compression is unavoidable for the extremely large data rates on the horizon. It will be a matter of feasibility costs (see also Chapters 7 and 8 for more details on the options for the communication technologies).

2.1.3 Display Architectures Figure 2.3 provides an overview on various functions inside a display that might affect the communication. Every display starts with a unit that provides the image(s) to be displayed; in the example shown, and common in cars, it is the Graphics Processing Unit (GPU). Based on the output from the GPU using one of the standard display protocols (for example, eDP or HDMI, see also Section 9.7), a Timing CONtroller (TCON) converts the video protocol to the protocol used by the panel and needed by the driver ICs. The display controller monitors the temperature, enables the respective adaptation of the Gamma sequences, controls the voltages, and watches the link status. This control communication typically uses ­Inter-IC (I2C) or Serial Peripheral Interface (SPI, see Section 9.5.2 and Section 9.5.3). In case of an LCD, the display requires a backlight (matrix, zone, or edge). The respective control signal may directly use Pulse-Width Modulation (PWM). In case the display has a touch screen, the touch information needs to be collected and reversed to the GPU (or a potential Central Processing Unit (CPU), depending on the set up and the commands observed). The IVC chips in Figure 2.3 are marked in grey dashed lines. The EE-architecture of the display, intelligent or without intelligence (see Figure 1.1 in Section 1.1), determines whether the (high-speed) IVC chips are needed or not. When the display is intelligent, the GPU is inside the same box. When the display is without GPU, a suitable communication technology needs to connect the display to the unit that hosts the GPU. The functions shown in solid line boxes are required independently of the architecture (with touch, marked in grey, only occasionally available), with the ones depicted on the right side in Figure 2.3 needed in close physical proximity to the panel though. When there is communication across a wire, it is simply important that it is as transparent to the overall system as possible. Figure 2.3 also shows, that in addition to the display protocol format, it might comprise several other protocols, so that different protocols are transported across the same transmission tech­nology. In addition, Figure 2.3 shows other functions that are not inherent as such in the display functionality but that might be added to the display because of a convenient and logical physical location of the display with respect to the use cases. First of all, there might be interfaces to connect consumer devices, for example via High Definition Multimedia Interface (HDMI) or Universal Serial Bus (USB). Customers might want to use the enlarged screen installed in the car instead of the small one of their mobile devices. Usually, the image would first be communicated to the GPU, where it is enhanced and adapted to fit the screen. Respective solutions were offered at the time of writing. Typically though, the socket to connect the mobile devices is not attached directly to the display but somewhere more ­convenient to safely keep the mobile device. However, having it directly in the close vicinity of the screen is an option.

29

30

2 The Automotive Use Cases

Video protocol

Graphics Various Processing Unit (GPU)

Display controller

PWM

Backlight controller (for LCDs)

Central Various Processing Unit (CPU)

IVC chip CE protocol

Display driver IC

x Touch controller

y

Display panel

Display driver IC

Consumer device interface

I2S

Microphone and ADC

I2S

Aux and DAC

Camera interface(s)

Display driver IC

... ... ...

I2C/SPI

I2C/SPI/ MII IVC chip

Timing Controller (TCON)

Camera chip(s)

Figure 2.3 Example of functions and interfaces inside a display Video conferencing via the car infrastructure adds another set of functions. It is a high-end feature, not only because video conferencing could also be realized using the mobile device and simply relaying the image to the in-car display as just discussed, but also because it requires passengers not needing to pay attention to driving (as rear-seat passengers or in autonomously driving cars). When video conferencing is supported, there might be microphones set with the display that collect spoken words or an auxiliary socket for attaching a headset (both typically using Inter-IC Sound, I2S, see also Section 9.4). Finally, a camera in the same location as the display completes the video conferencing function, with all the requirements cameras have on the communication, as discussed in Section 2.2. Both, camera(s) and microphone(s), might also be reused from other locations or functions inside the car. Also, this is a choice the designers must make. The one function that is generally NOT discussed to be integrated with displays in cars are speakers. Cars normally have a carefully crafted audio design and for whatever audio related function, these are reused. An important aspect with respect to video replay in such a distributed system – audio is processed at a different location than video – is lip synchronization. In order to be perceived well, the delay between sound and picture needs to be smaller than ±80 ms [51]. Means like timestamps and presentation times are useful means to achieve a sufficiently good synchronization. There are two functions that Figure 2.3 does not include: Compression and power supply. Compression might be added in a distributed architecture in order to limit the data rate ­required for the IVC technology. The compression thus relates to the IVC technology but is not part of it. It is performed separately; related to a specific display protocol or video format (see also Chapter 9 for options). In the authors’ opinion, compression is essential, especially for data rates beyond 10 Gbps.

2.1 Displays

Because of the large amount of power needed, displays are typically powered separately, and not with power over the IVC technology. In this case, there are no dependencies on the IVC technology, and the function is thus not included in Figure 2.3. However, power-over might be a possibility for future, very small displays that are used instead of dials and switches, for example, in doors. For details on power-over, see Section 6.1. One last note with respect to the use of consumer devices inside cars. Customers often wonder why car manufacturers hesitated so long to enable the reuse of the user interface and other functions from mobile devices. This had various reasons, the most important of which were safety related. For example, a user interface for a driver is simply very different from one for users sitting at a table. Even when enabling device integration for rear-seat passengers that do not need to pay attention to driving, just replaying the consumer interface is difficult for car manufacturers. In case of an instability of the functions running on the consumer’s device or in case the function is not well displayable on the in-car display, it is generally the car manufacturer who is considered responsible. Naturally, connecting and integrating consumer devices also affects the security system. In addition to the required link security for display applications, which is discussed in more detail with the camera use case in Section 2.2.4.3, content protection requirements need to be respected for some external content. Automotive displays presenting such content – independent whether its source is a passenger device or integrated in the car  – typically must support content ­protection protocols such as High-bandwidth Digital Content Protection (HDCP, for more details see Section 9.3). In the context of this book, the integration and support of mobile consumer devices does not add anything other than what has been discussed above, which is why this topic is not addressed beyond this point. Up to here, this section discussed the functions and architectures of a single display. Figure 2.4 now shows two simplified architecture options in case the same, remote graphics processor services two displays. The upper half of the picture shows two individual P2P connections to the displays in a star topology and the lower half shows a daisy-chain scenario. The depiction further distinguishes between two individual video streams, Stream 1 and Stream 2, that the displays receive and a single stream, Stream 1*, that is multicasted to both displays. To have two (or more) displays in a daisy-chain configuration is actually an expectable scenario, not only for the back seat passenger entertainment in mini vans, but also for the various displays in the front. It saves interfaces and connectors at the main ECU, which usually needs to handle many more connections and therefore profits from fewer connectors. It might also sit somewhere in a central location and further away from the displays so that cables can be saved. In case of two unicast streams in a daisy-chain scenario, the main ECU would need to support larger data rates on one link than is necessary in a star topology. In case of multicast, the topology makes no difference to the data rate needed. A key requirement in either ­scenario for multicast is, that the video display on both screens is highly synchronized so that there are no visible differences (including being synchronized to the audio stream if applicable). This indeed is a requirement that the communication technology must support with a good time synchronization mechanism that includes presentation times and time stamping.

31

2 The Automotive Use Cases

Display 1

ECU IVC chip A

Stream 1

Con.

Stream 1*

Con.

Display 2 IVC chip C

Stream 1* Con.

Stream 2

Con.

Stream 1*

IVC chip D

Stream 2

Display 1

ECU Stream 1+2 Stream 1*

Con. Con.

IVC chip B‘ IVC chip C‘

Stream 2

Con.

Stream 1*

IVC chip A‘

Graphic processor

Stream 1 Screen Stream 1*

IVC chip B

Stream 1

Stream mapping

Graphic processor

Stream mapping

32

Screen Stream 1*

Display 2

Con.

IVC chip D‘

Stream 2 Screen Stream 1*

Figure 2.4 Architecture options in case of two displays: P2P or daisy-chain In both depicted topologies, the communication is always just between the main ECU and each display. In case communication is required between the displays directly, this is again a different architecture that requires full networking capabilities of the communication technology. A use case might be gaming for rear seat passengers. However, for such a specific application again a different architecture with intelligent displays might be better to start with.

2.1.4 Typical Communication Related Requirements of Displays The following list provides an overview of important display requirements. ƒ Key to the displays is the resolution. For displays for 2025 and later, 60 Hz, 10 bits per color (meaning 30 bits color resolution), and 8K resolution is possible (see Figure 2.1 and Table 2.1), leading to a theoretical data rate of about 65 Gbps in the DownLink (DL), high speed direction. However, very large displays in cars are likely to be stitched together out of well synchronized smaller displays and data compression (see also ­Section 9.2) is another means to reduce the data rate transmitted. Also, for cost and usability reasons, many displays will remain smaller with 10 Gbps data rate giving ample room for usability.

2.1 Displays

ƒ For the lower data rate UpLink (UL) direction, it depends on the number of other ­applications that need to be supported by the display. Without any additional features 1 Mbps should be more than enough for displays also in the future. With touch, microphones, and compressed camera data 10–100 Mbps should suffice in the UL. The ­latency requirement of ≤ 8 ms between touch and reaction might be somewhat relaxed if some touch feedback is directly included in the display. Uncompressed camera data in an UL might require data rates up to multiple Gbps (see also Section 2.2). ƒ There is a variety of video and control formats a display might need to support. For the video data the display typically needs to support just one of the following: openLDI, MIPI DSI, HDMI, eDP/DP, V-by-one, HDMI (see Section 9.7 for details). Displays showing ­entertainment videos likely require content protection formats such as HDCP (see also Section 9.3) and Dolby Vision support as well as sufficient (lip) synchronization between the audio and video replay. For the control formats it also varies, and the ­display communication might have to support more than one technology. Options are Ethernet with 100 Mbps, I2C with 1 Mbps, SPI with 100 kbps to 10 Mbps, and General Purpose Input/Outputs (GPIO, see also Section 9.5.1). ƒ Visibility is key. This is not just to please the eye, but especially for safety reasons. Displays might have Automotive Safety Integration Level (ASIL) requirements from standard Quality Management (QM) to ASIL B (see also Section 3.2.2 or [52]). This ­requires, for example, a High Dynamic Range (HDR) and the use of Gamma sequences to ensure optimal visibility for all types of light environments. Another aspect of this is, that a driver is never blinded by the display’s brightness. It is important that frozen content is immediately identified, and counter measured. A display could, for example, stop showing anything in such a situation, as the user will then immediately realize that something is wrong. Furthermore, displays showing driving critical information like an instrument cluster display should have a back-up, for example a scaled down version, in case some more elaborate graphic content fails. The link technology as such will need to be able to identify and/or correct transmission errors with a CRC or FEC, do link quality assessments, and short/open notifications. More details on functional safety are discussed with the camera use case in Section 2.2.4.2. ƒ Security is increasingly important for all use cases inside cars, also for displays. The aspects of theft and counterfeit protection as well as the prevention of data manipulation are very similar for cameras. Because the risk of security incidents is higher for cameras – many of them are located at the outer shell of a car – more details are discussed with the camera use case in Section 2.2.4.3. ƒ Most displays, except for very small ones, will have a power supply separate from the communication. For power saving, a deep sleep functionality is important, allowing for wake-up based on touch or a button; this has to be also possible for individual nodes in the middle of a daisy chain set up. As displays are inside the cabin, their ambient temperature resilience is typically between –40 °C and +95 °C.

33

34

2 The Automotive Use Cases

„ 2.2 Cameras This section discusses the use of cameras in cars. For this, Section 2.2.1 starts with looking at the history of automotive camera systems. Section 2.2.2 describes important technical basics and terminologies for cameras. Section 2.2.3 looks at the different functions inside the camera architecture. Section 2.2.4 introduces software, safety, and security relevant aspects that are specifically important for camera use cases. Section 2.2.5 closes the camera discussion with a list of typical, communication related requirements.

2.2.1 A Brief History of Cameras in Cars When cameras were first introduced in cars, it was for supporting parking maneuvers with backup images [53]. The video recorded by the cameras of the scenery behind the cars was directly presented to the drivers. Various sources mention three cars with early backup camera systems: A Buick Centurion concept car presented in 1956 [53], a Volvo Experimental Safety Car presented in 1972 [54], and a Toyota Soarer/Lexus SC going into series ­production in 1991 [55]. None of these early systems seem to have been continued, so that Nissan is given the credit for initiating the backup camera market with its Nissan Infinity Q45 in 2000/2001 [53]. Nissan was also the first car manufacturer to sell cars with a 360-degree Surround View System (SVS) in 2007 [56] [57]. None of these early camera systems was fully digital. The first available imager sensors used video camera tubes, which, considering the general timeline, is what must have been used by Buick and Volvo for their backup camera systems. Toyota and Nissan already used the Charge-Coupled Device (CCD) imager technology that were available from 1974 on [58] and had completely replaced the video tubes by 1990 [59]. The first digital camera to record 1.4 MPixels at Kodak in 1986 also used a CCD imager [60]. The proliferation of digital cameras was, however, spurred by the development and improvement of Complementary Metal-Oxide Semiconductor (CMOS) image sensors (see also Section 2.2.2). At first, the CMOS image sensors produced a significantly inferior image quality compared with CCD sensors. As CMOS technology as such was becoming widely deployed in many other application fields though, its market adoption surged. As a result, also the imager quality received the necessary development attention and significantly improved [58]. Mobile phones had their share in proliferating digital cameras, with a likewise huge growth since the first mobile phone with an integrated, digital camera appeared in the year 1999 [61]. With digital camera systems, it is not only possible to improve the image quality for the human vision/viewing directly (compared with analogue systems). Additionally, the digital data allows to easily enhance the information in the image with trajectories or object identification. The camera use cases thus differentiate between human vision and machine vision, which is sometimes also referred to as viewing (human vision) and sensing (machine vision) [62]. Some functions mainly require human vision. These comprise simple backup/ surround-view cameras, wing or rear mirror replacement cameras [63] [64]. For the latter, Japan was one of the first countries to allow wing mirror replacement cameras on its roads [65]. Other functions require machine vision only, for example, traffic sign recognition, lane

2.2 Cameras

departure warning, blind spot warning, driver monitoring (for fatigue or distraction warning), and gesture recognition (for selected vehicle controls). Some functions, such as advanced backup/SVSs, use both. The exact use consequently affects the architectural choices for the camera systems (for more details see Section 2.2.3). The development of the communication technologies used to connect digital cameras is somewhat disjunct from the development of the imager technologies. To start with, all digital automotive cameras – and in some cases even new automotive camera systems sold at the time of writing in 2021 – used analogue communication; typically NTSC, as this used to be the common communication interface for the displays showing the images (see also Section 2.1.2 ). Only with increasing image quality requirements and the availability of automotive suitable SerDes technologies (see Chapter 7), did the communication of the digital cameras slowly shift towards digital communication. In 2021, 8 MPixel cameras were being planned for series production. Important drivers for the camera market in cars are the innovativeness and competitiveness of car manufacturers with the ultimate goal of fully AD cars. Also, several new laws drive the proliferation of ­automotive cameras. Since May 2018, the US only licenses new car models that provide backup cameras and video displays [66]. In Japan, Service/Sports and Utility Vehicles (SUVs) require a special mirror or a camera in order to be able to view a specific area on the passenger side of the car [67]. The European Union (EU) decided in 2019 that new cars, vans, trucks, and busses have to fitted with a number of new safety features starting from 2022 [68]. These include advanced emergency braking (also for pedestrians and cyclists), drowsiness and attention detection, as well as distraction recognition and prevention [69]. The EU does not mandate how exactly the functions need to be realized. All three of the listed functions may use cameras but may also be realized differently. The drowsiness and attention detection, for example, can be realized with steering pattern monitoring. Vehicle position in lane monitoring might also use other sensor technologies, like radars (see Section 2.3.1.2). Should distraction r­ ecognition be realized with help of driver eye and face monitoring however, the use of cameras is likely [70]. In any case, more growth in the automotive camera market can be expected from these regulations. With additional cameras installed for entertainment and video conferencing, the significant market projections (see, for example, [62]) seem realistic.

2.2.2 Camera Basics and Terminology The heart of automotive cameras is the image sensor or “imager”. This section focusses on automotive cameras in which image sensors translate dynamic scene brightness levels into electrical signals in order to recreate 2-dimensional (2D) images. Time of Flight (ToF) cameras, for example, create images by measuring the distances of objects based on the runtime of reflections of light pulses and can create 3D images (see Figure 2.10 for the basic principle). This would be “another sensor” discussed in Section 2.3. The distinction is irrespective of the use of the data, which might be for either human and/or machine vision. For many applications, automotive cameras use the visible light spectrum from 380 to 750 nm (see also Figure 2.11). However, there are also automotive cameras that use the not visible InfraRed (IR) light. Driver monitoring cameras, for example, use a wavelength in the

35

36

2 The Automotive Use Cases

“Near” IR spectrum (between 750 nm and 1.4 μm) plus a separate IR LED light to ensure that the driver is always visible to the imager irrespective of the light conditions; and that without ever distracting or blinding the driver. Also, night vision cameras, offered in some high-end cars as options, are based on IR light detection [71]. There are two variants of ­automotive night vision system, each using a different part of the IR spectrum. The active system sends out IR light pulses that are synchronized with the image capture process, like has just been described for the driver monitoring systems. The passive system measures thermal radiation for which an IR wavelength between 8 and 15 μm is optimum (see Figure 2.11). The following description is based on the most used CMOS image sensors, which can detect visible and IR light in the range between 300 nm and 1 μm. Like displays, also the imagers separate images into individual pixels, which, depending on their count, lead to the specific imager resolution. Figure 2.5 shows the technical principle behind an individual pixel ­inside the imager. The silicon surface functions as a photodiode, which generates, depending on pixel irradiance, a voltage that is amplified and creates the pixel output. The voltage values of every pixel are then passed through an Analogue to Digital Converter (ADC, not depicted in Figure 2.5) [72].

VCC Reset

Amplifier Photo element

Amplifier Amplifier

Read data Signal out

Light sensitive area

GND

Figure 2.5 Schematic of an active image sensor pixel [73] An image sensor consists of an array of independent pixel sensors, whose values are accessed via an active matrix of rows and columns (see Figure 2.6). Detecting the incoming photons is a physical process and it is easily imaginable that the surface size for one pixel matters. The smaller the size, the more pixels can be integrated into a certain space but the less light one pixel receives. For mobile phone cameras, a 64 MPixel resolution with 1 μm pixel size is not uncommon. Potential image quality issues, for example, because of reduced pixel sensitivity, can be mitigated by post processing. In automotive applications the visual image appearance is not necessarily the most important criterium, and the camera resolution is not as prominent a selling point as, for example, for mobile phones. The pixel size is therefore normally larger, for example, 4.2 μm. This allows for more light per pixel, meaning less effort for post processing, but fewer pixels in the same physical space. The resolution is thus smaller than in mobile phones. The expectation is that 3–8 MPixel cameras will remain the relevant resolution range for automotive cameras for a while. In any camera the lens, and the interrelation between lens and imager, is an essential ­element, as the lens is used to focus the image onto the active sensor area. The sensor area as such cannot distinguish between colors. Therefore, color filters are placed in front of ­every pixel. These filters determine what color(s) each active pixel element senses, typically

2.2 Cameras

e­ ither red, green, or blue light, as depicted in Figure 2.6. Figure 2.6 also shows that there are more green pixels than red and blue ones. This is because the human eye is more sensitive to green than to blue and red. The distribution as depicted is called the Bayer pattern filter, after its inventor Bryce Bayer [74], also referred to as “Red Green Green Blue (RGGB)”. Another Color Filter Array (CFA) used for automotive applications is Red Clear Clear Blue (RCCB) [75]. Instead of a green filter, every second pixel receives no filter. This has the ­advantage that the pixel receives more light (and thus might be smaller). However, while the green information may be calculated from the difference between clear, red, and blue, the color resolution is inferior to that of RGGB. Depending on the use case – for example, human or machine vision – yet different CFA patterns might be applied.

Red Green

Digital signal

Blue

Digital signal

Figure 2.6 Individual digital output of a CMOS imager [76] Image sensors in cars use one of the following two shutter types: Rolling shutter or global shutter. In case of a rolling shutter, more or less, pixel sensors are exposed to the light one row after the other and while one row is exposed, the data of the previous row is being transmitted. For the global shutter, all pixel sensors are exposed to light at the same time. Because not all data can be read out simultaneously, the global shutter imagers require a storage gate and more logic. They are therefore more costly and, despite their better ­suitability for moving images [77], only used in cars when necessary. One such case is, when the image capture needs to be synchronized with a pulsed LED light, like used for the aforementioned driver monitoring or active night vision cameras. One of the specific requirements for automotive camera systems is to cover the complete dynamic range of automotive scene conditions, meaning the darkest as well as the brightest area in a driving scene. Therefore, HDR imagers are required, which normally rely on some sort of multi-capture HDR scheme. This means that one scene is sampled in multiple im-

37

38

2 The Automotive Use Cases

ages or captures per video frame. In order to present only one single image to the application for post processing or human vision display – and also in order to save transmission bandwidth  – these multiple captures may be combined to a single HDR image. Further compression of this combined HDR video frame is possible by reducing the bits per pixel by taking advantage of the square-root-of-signal behavior of the photon shot noise. Because the photon shot noise dominates the mid- and high-end signal range, it allows applying a gamma curve [47] or a similar logarithmic shape companding scheme for lossless signal com­pression. Last but not least, imagers are very heat sensitive, meaning that the image quality reduces in case its temperature is too high. As automotive cameras are in small housings (see Figure 2.7 for an early example from 2010) and often located in exposed positions with solar loading (like wing mirrors or directly behind the windshield), an overall low power consumption is essential for the cameras as not to additionally increase their temperature. Thus, also thermal simulations are very important during the development of cameras and imagers. This shall ensure that the overall ambient temperature in the camera does not rise above a certain critical value and that no thermal hot spots in the imagers will cause (Dark) Signal Non-Uniformities (DSNU) on the pixel array [78].

Figure 2.7 Example of an automotive camera (Photo: Michael Singer, 2010) The image sensor itself normally generates and outputs only so-called “raw” image data. For computer vision applications, this raw image data may be used with only minimal additional processing. However, in case of human vision applications extensive further pro­ cessing, typically performed in an Image Signal Processor (ISP), is needed to convert the raw data into a fully processed data format like YUV for a display (see also Section 2.1.2 or Section 9.1). The ISP processing unit may be located anywhere in the path between image sensor and GPU/System on Chip (SoC, see also Section 2.2.3). The main ISP functions are as follows [27]:

2.2 Cameras

ƒ De-mosaicing calculates a red, green, and blue pixel value from the Bayer pattern CFA for each pixel. It interpolates the missing color channels for a given pixel by interpolation from surrounding pixels. ƒ The color correction matrix is used to calculate the correct color values with minimal color error for all color channels. ƒ Image enhancement capabilities support features such as defect pixel correction, noise filtering, and image sharpening. ƒ For HDR image sensors also specific functions like tone mapping (similar to Gamma) and histogram equalization are included in order to fit the large dynamic range of the sensor output into the limited digital range of the displays.

2.2.3 Camera Architectures Figure 2.8 provides an overview on the different elements that might be found inside an automotive camera. In the following the different elements are explained in more detail.

Lens

Heat- EPROM ing Imager

Clea- LED light ning

ISP

Processing/compression

Com. (bridge) Power (over)

Figure 2.8  Possible camera elements; dashed lines indicate optional items

Key for the function and quality of digital cameras are the lens and imager and their interrelation. There are different lens types being used: tele lenses, wide-angle lenses, fish-eye lenses (with a view angle of at least 180 degrees), to name a few. Earlier camera implementations used an Erasable Programmable Read-Only Memory (EPROM) to store the camera specific, intrinsic calibration data, which needed to be readable by the ECU processing the data. Modern implementations might use a One-Time Programmable (OTP) memory for the calibration data, which would also need to be accessible from the ECU for configuration purposes. Additionally, automotive cameras might contain a lens heating and/or cleaning function. To start with, automotive imagers used classic parallel interfaces as output. At the time of writing, lower data rate products with such interfaces were still being sold. With increasing data rates, however, this became impractical (see also Section 1.2.1) and the automotive industry widely adopted the MIPI CSI-2 interface for the higher data rate imagers, most commonly in combination with the MIPI D-PHY (for details see Section 9.6). Figure 2.8 shows the ISP as an optional feature and standalone block. If an ISP is needed because of a human vision function, it is, however, rather uncommon to implement it as a separate IC inside the camera. The ISP might be integrated with the imager or the processing chip (if available) in the camera, or the ISP is part of the ECU the camera is connected to. In the ECU, the ISP might be a separate chip, or it is part of the SoC or GPU processing the video data.

39

40

2 The Automotive Use Cases

As soon as compression is included in the camera, not only hardware acceleration, but also some software and the updatability of that software is required. Software always contains errors, especially in complex functions such as compression. The first automotive Ethernet cameras used Motion Joint Photographic Experts Group (MJPEG, see also Section 9.2.1) compression, because at the time hardware acceleration was not available for other suitable compression methods. A software-only implementation would have yielded too much latency [79]. However, once hardware acceleration became available, implementations switched to H.264 and H.265 compression (see Section 9.2.2 for more details on these). Mind you, even with hardware acceleration, compression causes some, if small, delays and, more important, may result in compression losses (depending on the exact algorithms). Therefore, using compression in automotive cameras is often limited to human vision applications. Other processing done in automotive cameras might include object recognition or driving trajectory calculations. The communication chip and technology selected for the camera highly depends on the ­architecture items discussed so far. A camera with compression will use a different tech­ nology than a camera that consists of lens/imager and communication bridge chip only. Chapter 7 and Chapter 8 present high data rate communication technologies that might be used in such IVC bridge chips. For cameras that require active lighting the respective LEDs are also included. As a control channel for the cameras typically I2C or SPI are used (see Section 9.5 for more details). Last but not least, Figure 2.8 also includes the power supply. As has been mentioned with the discussion of the imagers in Section 2.2.2 low power consumption is essential for ­cameras. This is a significant difference to display applications, where power supply faces no particular restrictions in terms of delivered power nor space, and therefore was not ­explicitly identified in Figure 2.3 in Section 2.1.3. The camera’s power supply might need to ­output various power levels, depending on the requirements of the used components. The power consumption as such depends on the features included. A camera with lens heating and or LED light consumes significantly more power than a camera without these features. To give an idea: At the time of writing and from the experience of the authors an 8 MPixel/30 fps camera would run below 2 W, while a lens heating could easily add 3 W. An LED light would go beyond what was deliverable via Power over Coaxial (PoC) at the time. Especially for cameras connected via coaxial cables, it is, however, of interest to transmit the power with the data communication, as this significantly simplifies the connector design (see also Section 6.1.3). The amount of power that can be transmitted over the data line is an interrelation between the used inductors (and the space available for them, see also Section 6.1.4) and the Media Dependent Interface (MDI) Return Loss (RL) requirements of the used communication technology (see Section 8.3 for a comparison). Note that the inclusion of security in the design does not only increase the need for pro­ cessing in at least one of the ICs used, but also impacts the requirements on software and updateability in the cameras. Also, it requires the use of a secure storage that cannot be accessed externally. For more details, see Section 2.2.4.3. Finally, to obtain depth information with cameras, 2D stereo cameras may be used (another option is to use one 3D ToF camera, see Section 2.3.1.4). In case of 2D stereo cameras, typically an Application-Specific Integrated Circuit (ASIC) would preprocess and combine the camera data before it is forwarded to a single communication chip.

2.2 Cameras

2.2.4 Camera Software, Safety, and Security This section presents some relevant considerations with respect to software, safety and security. The addressed aspects are most prominent for camera applications. However, they are in parts equally relevant for display or other sensor applications, as is noticed in the respective subsection. Section 2.2.4.1 discusses specific software items, Section 2.2.4.2 looks at safety considerations, and Section 2.2.4.3 presents architectural considerations for security solutions.

2.2.4.1 Software In the discussions presented in this chapter so far, the focus was on hardware related ­requirements. However, the seamless introduction of electronic components in cars also requires software. For cameras the most obvious software packages concern the image ­processing, for example, for object detection and objection classification in ADAS applications – independent from where the processing is located. To achieve the best results, specific hardware accelerated processors are designed for a specific task, with again specifically designed software completing the system. Such dedicated hardware and software developments often come from one company that offers the complete package (for example, Mobileye/Intel [80], NVidia [81], Qualcomm [82]). Without these products, the car industry could not envision to enable autonomous driving. The respective suppliers providing these products are in consequence often the hidden project leaders of complex ADAS developments. For more general projects, automotive system and software designers have to consider the AUTomotive Open System ARchitecture (AUTOSAR), which has the main purpose to de­ couple the software from the hardware on which it is used [83]. In order to achieve this, AUTOSAR provides a set of Application Programming Interfaces (APIs) that specifically fit the automotive requirements and are very scalable; AUTOSAR can run on tiny 8-bit processors as well as on very large state-of-the-art systems. AUTOSAR offers two types: AUTOSAR classic and AUTOSAR adaptive. AUTOSAR classic caters for deeply embedded systems with static configuration and monolithic updates. It is typical for body electronics and small ECUs. AUTOSAR adaptive is the platform for more complex, high-performance ECUs and typically used for ADAS functions that process ­camera and sensor data. It allows for run-time configuration as well as partial updates and upgrades. AUTOSAR support is a prerequisite for the development of automotive ECUs that contain software and thus needs to be considered also for sensors and cameras with intelligence/software. AUTOSAR as an operating system supports the core functionalities of ECUs and the required interfaces like I2C, SPI, and GPIO. It also supports Ethernet communication, but, at the time of writing, not SerDes. Further efforts that affect the software development for camera applications address the control of the cameras and/or the video communication interface. For this, the MIPI Alliance defined the MIPI Camera Command Set (CCS), which is part of the building blocks around the MIPI CSI-2 interface [84] (see also Section 9.6). ISO specified in ISO 17215 a Video Communication Interface for Cameras (VCIC) for Ethernet connected cameras [85]. ISO 17215 uses a subset of the Scalable service-Oriented MiddlewarE over IP (SOME/IP) [86] used with Automotive Ethernet for the camera configuration [79].

41

42

2 The Automotive Use Cases

2.2.4.2 Safety Safety in cars is of utmost importance. Continuously improving the safety of passengers and the environment is a concern to all car manufacturers. Offering new ADAS functions is part of the measures to achieve this. In the context of developing sensor and display applications, safety is typically discussed as “functional safety”, which aims at systematically protecting car users from an unacceptable risk of injury because of malfunctioning electronics [87]. The framework for functional safety is described in ISO 26262 [88]. Key for the measures that have to be provided is the respective Automotive Safety Integration Level (ASIL) classification for the use case. The more often a function is in use or relevant in a potentially critical situation, the more severe the consequences if it fails, and the more ­difficult to control and prevent the situation, the higher the ASIL classification and the more stringent the requirements that need to be fulfilled [52]. The highest safety goal is ASIL D, the lowest is ASIL A, and if no particular functional safety measures, but just the usual quality measures need to be taken, the classification is “Quality Management (QM)”. The functional safety applies to the top-level function and is always end-to-end including the complete application and considering all relevant elements of the respective implementation. In the discussed sensor and display applications, these elements comprise the sensor/imager/display chips, potential additional processing (ICs) in the sensor/imager/display, the communication chips and technology on either end of the link, the link itself, the processing in the main ECU, and any notifications/actions that might be a result of the ­application. In ISO 26262, a communication technology as such is typically a “Safety Element out of Context (SEooC)”, as it is developed without knowing in what implementation it is going to be used and exactly how. It thus has to be developed based on assumptions with respect to its later use. ISO 26262 makes no statement concerning the means to achieve the safety goals but defines different levels of target values and provides recommendations. The failure modes to be considered for communication technologies are: loss of communication peer, corrupted message, message unacceptably delayed, lost message, unintended message repetition, incorrect message sequence, inserted message, masqueraded message, and incorrectly addressed message. Typically, communication systems are thus developed with the goal to guarantee a certain Bit Error Rate (BER) and latency. From a typical worst case bit error rate of 10–10 for legacy communication technologies the requirement was increased to a BER 3 years

EOP 5...12 years

3 years

2 years

1 year

Production

Production volume

Spare parts

5...12 years

10...20 years

Figure 3.2 Example lifecycle of a vehicle platform and corresponding semiconductor timeline For consumers, such timeframes are extensive. The consumer perception of actuality is dominated by their experience with smart phones and their apps, whose development and release takes three to nine months [34]. In consequence, the automotive industry has been targeting shorter development cycles for years [35]. It is actively pursuing strategies for product updates and upgrades after SOP, beyond the customary facelift that usually happens halfway between SOP and End Of Production (EOP). One measure to more actuality are software updates at the dealer or, as the latest trend, Over The Air (OTA) updates [36]. However, the paradigm changes it would entail to truly develop and sell cars differently, would merit a book in itself. We will therefore focus in the following on the general availability needs for electronic parts after SOP. For a period from five to twelve years, different variants of cars are produced based on a specific platform. Accordingly, components are needed in high volumes during this time. After EOP, customers can normally buy original spare parts for many more years. Map data for navigation systems is an example of a product customers often buy even after EOP (and that customers would buy subscriptions for with the car). The number of years car manu­ facturers provide spare parts after EOP varies. BMW, for example, guarantees original spare parts for 17 years after EOP. For Rolls-Royce spare parts are offered for 50 years, as 65% of all Rolls-Royce ever built are still on the road [37]. Aftermarket support for a specific model thus does not end at EOP. It does not require to know Moore’s law [38] to imagine that semiconductors needed 20 years after their first release are no longer state of the art. Generally, even the semi­ conductor process used for the original part is no longer available. To ensure supply during the complete PLC of a platform either the vendors of the electronic components put a contractually agreed number of spare parts in storage, and/or the Tier 1 does the same for a contractually agreed number of ECUs. If the forecast of pieces needed later was too small, the car manufacturer pays for the development of making a later ECU backwards compat­ ible, or, in the worst case, for rebuilding an old assembly line. None of such aspects have to be taken into consideration in the CE or IT industries in the CE or IT industries.

3.2 General Automotive Requirements

Personal experience with developing electronics for cars Before I, Michael Kaindl, started in the automotive industry as an engineer, I was first ­employed as a hard- and software engineer at a major computer manufacturer and then at a small company creating professional, digital audio equipment. In the first company, I was in touch with big main frames, PC programming, and embedded control. In the second company, I oversaw the hardware and the control software of a multiprocessor system for audio-processing. This audio system consisted of a 32-bit microcontroller, a Digital Signal Processor (DSP), and an 8-bit microprocessor as Input/Output (I/O) controller. For this project, I designed the hardware, created the first prototype by wire-wrap and wrote software for the 32-bit processor and 8-bit microprocessor. Later, one of my first projects in automotive dealt with an 8-bit micro with 4 kByte Read Only Memory (ROM) and 176 Byte Radom Access Memory (RAM). A simple and easy job for a few days, I thought. Creating the hard- and software for the first prototypes needed several weeks from the first sketch on a sheet of paper to a running device. This was within expectations from my ­experience from the projects I had done before. To finalize the project and start the series production took about three years: doing all the fine-tuning, fix all errors, enable the 1st and 2nd Tier suppliers to deliver the required quality, fix all mechanical issues. At the first pre-series production, the version developed for right hand drive cars caused a stop in production for two days. I received quite some management attention during these two days. I learned fast that automotive business is different to that of other industries and the ­complexity of even a simple project can increase quickly. A small project or tiny product does not mean that it is not important. The output of my first project was in production for one year. Then the functionality was integrated into another module, which saved about 25 € per car. The integrated module was in production for two different car types in about 14 different mechanical versions and the series production was active for about twelve years. For the second, integrated version a redesign was made about 13 years after EOP in order to maintain the supply for spare parts. The originally used microprocessor was no longer available on the market. 

„ 3.2 General Automotive Requirements The description of the product “car” in Section 3.1.1 already provided an outlook into the diverse requirements cars must fulfill. The following subsection represent two different concepts on how to cluster these requirements. Section 3.2.1 describes the requirements based on physics and the ways cars are used. Section 3.2.2 provides an overview on respective standards and regulations. The focus in the following descriptions is on those requirements that affect the choices for electric and electronic components.

75

76

3 The Automotive Environment

3.2.1 Use-related Requirements Consumers want reliable and robust cars with minimal maintenance. Considering how old cars get – in Germany, for example, over 40% of the registered cars are ten years or older and 7.5% are 20 years or older [39] – this is a huge task. In no other major industry, the products face such severe physical strain when in use, while being expected to last and be safe to use for as long as it is the case in the automotive industry. At the core of the resulting requirements for electronics is the aging effect. It is particularly important, because the aging of the electric and electronic components used for many years anyway is accelerated under the strenuous circumstances described in more detail below. In consequence, the components need to be selected and produced such that they can optimally withstand the environmental impacts. Some requirements vary depending on the exact location of a component inside the car. Car manufacturers often have an intricate system to determine into which location class a certain position falls. The resulting class then determines the capabilities and tests the components have to comply with. Specifying the thresholds for every requirement and ­subsequently deriving the class is done differently at every car manufacturer. It might therefore differ from the example values given below. The general outline, however, is very similar for all car manufacturers. First, electric and electronic components need to be robust against a large range of different temperatures and many temperature changes. Typical values are –40 °C to +85 °C inside the cabin, –40 °C to +105 °C for exterior components like the wing mirror, and –40 °C to +105 °C/125 °C in the engine compartment or gearbox [40]. During the assumed standard lifetime of 15 years, cars face about 10,500 significant temperature changes [41]. Then, there is the physical stress that results from driving a car in different conditions. All parts need to withstand damage from continuous vibration. One of the parts most vulner­ able to vibrations are connectors. They must be mountable with reasonable effort while showing no danger of dropping off because of not having been properly connected during production or material fatigue. Furthermore, depending on the exact location of a com­ ponent – for example inside, outside, in the engine compartment – the components need to be physically robust, in varying degrees, against water, dust, accidental contact, and alike. Furthermore, lenses, seals, and plastic materials are particularly vulnerable to chemical stress from fuel, oil, lubricants, salt detergents, mixed gases etc. For components that pass into doors or the tailgate, like cables, there are additional requirements to be robust against moving cycles: 100,000 open/close cycles for a door, 20,000 open/close cycles for a trunk lid (exact numbers might vary depending on the car manufacturer). A typical car is driven 300,000 km in 15 years [41]. In the past, this resulted in a required lifetime of automotive electronics of 3000 hours power on, because electric and electronic parts were assumed to be needed only, when the ignition was active. This is changing. An increasing number of components is not turned off when the car is parked. Examples are the access control, the anti-theft system, the almanac of the navigation system, some components of the infotainment system, some Advanced Driver Assistance Systems (ADAS), and the electronic switch boxes, where traditional fuses are replaced by intelligent semiconductor switches. These parts remain in low power or stand-by mode when the car is parked. Other ECUs in the car can be powered down. They might not need as much power,

3.2 General Automotive Requirements

but must sustain several cold-starts a day, which also affects the electronics, but not as much as the continuous power-on. Those ECUs that are continuously on, face, in 15 years, a steady power-on time of more than 130,000 hours. There are various consequences to the continuous powering. One is the discussed acceleration of aging of the electric and electronic components. Especially critical are the connectors, the wiring harness, components like electrolytic capacitors, and the programmable devices that even at an advanced age need to continue to retain the data. Next, it affects the power needed when the car is parked. Customers of CE devices are quite prepared to find the batteries of their CE devices drained, if they have not used them for a while. They ­simply change the batteries or connect the units to electric outlets. Customers of cars are not amused when their car does not start because of a drained battery following a longer period of inactivity. It is therefore essential to minimize the power consumption of ECUs when the ignition is off. Those ECUs not needed, when the car is parked, might simply be disconnected from the battery completely using smart fuses, so as not to drain it. For the increasing number of ECUs that remains connected, it is essential to limit the quiescence current. A typical value is a maximum allowed quiescent current of a few 100 µA per ECU; with the communication interface being allowed 10–20 µA [42]. For CE devices this is not nearly as stringent. Also when the car is in use, low power (fuel or battery) consumption is desirable to the customers. This is not so different from other industries. All users would like small electricity bills, long battery life, or stand-by times. From this perspective, a car is simply a more complex product in which to achieve this. However, the power consumption of a car has other implications. For cars with a combustion engine, a car additionally has to fulfill regulatory requirements in terms of Carbon DiOxide (CO2) reductions; the latter having become even more important since the so-called Paris Agreement of the United Nations Framework Convention on Climate Change (UNFCCC) signed in 2015 [43]. The smaller the power consumption of the electronic components, the less fuel is needed and the less CO2 is emitted. In case of electric cars, a smaller power consumption means extended reach. So, the power consumption of every single electronic component in cars is very important, when in use and when not in use. One more important aspect with respect to the power inside cars is the quality and availability of the power supply. Electronics often require to be constantly and evenly powered. In cars this is more difficult to achieve and the electronics inside a car therefore have to be able to cope with larger variations in the voltage levels than in many other industries. The variations might be due to batteries being in a low charge level, batteries being old, because the outside temperature is very low, or because all power is needed elsewhere, for example, to start the engine. This can cause ECUs to reset, which in return requires fast recovery mechanisms. Last but not least, with the vast number of variants of a single car model (see Section 3.1.1) and with the actual use or not-use of functions by the customer, more or fewer ECUs might require power from the supply network simultaneously, while the battery is likely always the same model. The varying number of active ECUs thus can also lead to different power levels in the supply network. The fast recovery leads yet to another automotive requirement: Fast start-up and availability of functions. This is required during the start of a car as well as when ECUs have been put to sleep for the power saving reasons discussed above. At start-up, it is required that the vehicle network is available and ready for communication 100–200 ms after power-on [42],

77

78

3 The Automotive Environment

in order not to cause any noticeable delay in the wake up of the systems beyond. Even to just start the engine, the in-vehicle network must be available beforehand so that the immobilizer system can exchange and calculate the required certificates upfront. Concerning the parking aid, it is expected that two seconds after entering the car, a customer can receive the required warning signals [44]. While it might be surprising to someone designing, for example, mobile phones, space is also a concern inside cars. Even if the car is many times larger than a phone, it is still not obvious where to place all the ECUs plus the wiring for power and communication. Cameras are a good example, as they need to fit small spaces in the side mirrors, the dashboard or alike. When Automotive Ethernet was introduced in 2013 with the BMW X5 surround view system to connect the cameras with the processing ECU, being able to meet the spatial restrictions was one important requirement that had to be met [28]. The wiring can be a nightmare when needing to fit through small openings, for example, between the main cabin and the doors, or the trunk. The wiring also directly affects the weight of a car, which in return affects the energy needed when driving [41] and the CO2 emissions or reach. The wiring harness is the third heaviest and the third most expensive component inside a car, following engine and chassis [45]. Keeping weight limits and achieving weight reductions can thus also be important requirements and the wiring harness is particularly scrutinized for this. The communication technologies selected, as well as various choices within the EE-architecture directly impact the weight of the wiring harness. Example choices include centralized versus distributed processing or zonal versus domain architecture. Other higher order requirements that are interrelated with the EE-architecture are updateability and, more recently, also upgradeability of cars (see also Section 3.1.2.2), which in return makes security extremely important for car manufacturers. The possible architectural choices for displays, cameras, and other sensors and their implications on aspects like flexibility, scalability, and have been discussed in Chapter 2.

3.2.2 Regulatory Requirements A car is not only a complex product, but also a product that can potentially endanger its users and others. While exhaust fumes and pollution play some role in regulatory requirements, the key focus is on safety. To be able to use cars, drivers thus do not only have to have a valid driver’s license and the cars need to be registered and insured. Before driving a car on a public road can even be considered, the car model is subject to severe regulations that the car manufacturers must ensure are met. These are based on legislative provisions (see Section 3.2.2.1) and those encouraged by insurance companies (see Section 3.2.2.2).

3.2.2.1 Government Driven Requirements First of all, the exact requirements a car has to fulfill can vary depending on the country it is supposed to be used in. One reason among many, might be specific climate conditions. This section can only give an outline of the commonalities relevant for the use cases discussed in this book. Whoever needs to know the exact details for a specific case, has to check the latest publications at the respective issuing organizations.

3.2 General Automotive Requirements

Nevertheless, in order to limit the diversity, there are efforts to harmonize the regulations. The most prominent example is the World Forum for the Harmonization of Vehicle Regulations, which is hosted by the United Nations Economic Commission for Europe (UNECE). 54 countries especially from the EU, but also leading economies from Asia like Japan and South Korea, agree here on common technical requirements especially for motor vehicles with the goal of mutual acceptance and homologation. At the time of writing, the latest published vehicle regulations addressed cyber security, over the air updates, and lane keeping [46]. In the U. S., the counterpart to the UNECE is the National Highway Traffic Safety Administration (NHTSA), which writes and enforces the respective Federal Motor Vehicle Safety Standards (FMVSS) [47], which again substantially overlap with the rules applicable in Canada [48]. Such regulations set in most cases a high-level framework for the construction and operation of cars. These regulations are translated into legislation in every country – in Germany, the superordinate requirements are in the Straßenverkehrs-Zulassungs-Ordnung (StVZO) [49]. Their fulfillment requires even more detailed requirements and test standards, many of which are specified at Standard Setting Organizations (SSOs) like the International ­Organization for Standardization (ISO), International Electrotechnical Commission (IEC), or the Society of Automotive Engineers (SAE), to name a few. The ISO 20653 and IEC 60529, specify, for example, the Ingress Protection Codes (IP Codes). These are used to determine the protection of mechanical and electrical parts against water, dust, etc. [50] [51]. The use of IP Codes is not limited to cars, but particularly ­important for them. Components located on the exterior of a car, like wing mirrors, or in the engine compartment are obvious candidates needing to withstand all kinds of environmental impacts. Another very important specification is the ISO 26262 series derived from IEC 61508. It defines the functional safety requirements for electric and electronic systems in cars and a methodology for classification called Automotive Safety Integration Level (ASIL). The more likely a failure of a system is, the more severe the consequences if it does fail, and the more difficult to control and prevent the situation, the more stringent the requirements (see, for example, [52]). ADAS systems generally have particularly high ratings. As ADAS systems often comprise cameras and other sensors, the impact of ISO 26262 and functional safety is touched in Section 2.2.4.2. Then, there are many specifications and requirements for EMC. These are an extremely important factor, when designing and/or qualifying communication technologies for cars, as the many communication lines inside cars are prime recipients and emitters of electromagnetic waves (for more details on EMC, see Section 4.1). Naturally, also a connected CE or IT device must comply with EMC rules. However, in cars the consequences of disruptions can be more severe, while at the same time, the cost and weight of the wiring needs to be smaller than what is generally affordable in the IT or consumer industries. EMC is a typical example for high level requirements defined by the UNECE [53], while many specifics are detailed at IEC or ISO (see also Sections 4.1.2 and 1.2.1). Furthermore, car manufacturers tailor the requirements further with their own (more stringent) limit lines and tests (see, for example, [54]). Last but not least, the breakdown of requirements affects every semiconductor used inside cars, where again, a number of specifications apply (while every car manufacturer might add its own rules, see Section 3.3). Furthermore, the production process and the manufacturing sites have to submit to automotive standards. The Automotive Industry Action Group

79

80

3 The Automotive Environment

(AIAG) publishes the respective ISO/TS 16949 that provides harmonized, technical requirements for quality in the supply chain and assembly process based on ISO 9001 [55]. The mentioned specifications represent only a few of the most important ones. Just the list of all specifications touched when developing and building a car would be a book in itself. The amount of different quality specifications is an expression of how deep the quality aspects are in place for the automotive products. Every component must simply fit perfectly during the assembly and must operate as specified within the overall system for the long lifetime of the vehicle without undesired behavior.

3.2.2.2 Insurance Driven Requirements With the car being a valuable product that can, at the same time, cause a lot of damage, ­insurance is a relevant economic component when discussing cars. Being able to insure a car at all and at what rate, are important considerations for customers. Therefore, also car manufacturers have to take the implications for insurability into account when designing a car. Even if insurance companies do not have the legal force of governments, they have the power to prevent the marketing of a car or a car technology, as without an insurance available for that car, the real operation is nearly impossible. The insurance companies enforce their interests in order to control their economic risks (with improved accident prevention being also in the interest of the customers). This may be done directly by formulating requirements on the technical construction of a car or indirectly, with the height of the insurance rates for end customers. In order to be able to give qualified assessments, insurance companies have significant ­automotive knowledge in-house. The Allianz Zentrum für Technik in Germany, for example, investigates innovations in the automotive industry at an early stage in order to assess their impact on road safety in general and potential insurance claims specifically [56]. It is obvious, that the responsibilities in case of accidents of level four or five autonomous driving cars are a huge topic of interest to insurance companies, car manufacturers, and governments alike. The technologies investigated do not need to be as all-encompassing as ­autonomous driving though, to be relevant for insurance companies. When, for example, insurance companies notice that a large amount of damage occurs during parking or maneuvering in small spaces at small speeds and that this can be decreased if automated braking systems include object detection by sensors to the side or back of the car [57], it is a matter of time until this will find its way into cars via the insurances. Thatcham Research is a UK based organization in which a number of major insurance companies consolidate their efforts to investigate insurance risks like physical injuries, material damage, or theft. Their requirements for safe repair of ADAS equipped vehicles show the importance of ADAS and the impact insurance companies can have on new developments for cars [58]. Last but not least, there are worldwide New Car Assessment Programs (NCAP). The Euro NCAP, for example, is a voluntary car safety assessment program backed by various ministries of transport as well as automobile associations and insurance companies. While it is a voluntary program, its star rating is an important marketing instrument for car manu­ facturers. Designing and developing cars with Euro NCAP tests in mind has thus become customary. This explains why even low-end cars often offer high-end safety features that would go beyond the means of most of its customers. The marketing impact of the NCAP rating outweighs the costs for the development effort [59].

3.3 Automotive Semiconductors

„ 3.3 Automotive Semiconductors Semiconductors are what electronics and software function with. Without them modern cars and comforts would not exist, and innovations would look different. The performance and quality of modern cars are fundamentally linked to the performance and quality of the used semiconductors. Semiconductors are needed early in the value chain. From the semiconductors, ECUs are built and from the ECUs, cars are built. The later in the process faults originating in malfunctioning semiconductors are found, the more expensive it is to resolve them [60]. Car manufacturers thus have stringent quality requirements and for 1st and 2nd Tier suppliers having qualified parts and all the necessary release documents, is more ­important than the SOP itself. As a result of semiconductor development and production, the requirements for semiconductors go beyond the pure quality (see Section 3.3.1). Additionally, the semiconductors have to meet certain performance requirements (see Section 3.3.2) and have to meet requirements in the supply chain (see Section 3.3.3).

3.3.1 Semiconductor Quality An important quality criterion for semiconductors is the statistical defect rate for one million parts, which is measured in Parts Per Million (PPM). The lower the PPM value, the better of course, but the more intense the necessary testing for the semiconductors and the more expensive the part in consequence. In cars, the large number of parts makes a low PPM essential, though. Figure 3.3 shows a simple example: When a car manufacturer that produces one million cars per year with 60 ECUs consisting of 100 electronic parts each targets a PPM value of 30 – which means very good quality in the CE industry [61] – and performs no additional quality checks, that car manufacturer would deliver 50,000 cars with defects because of faulty semiconductors per year. If the same car manufacturer ­applied severe testing to ECUs and to the car, missing only 10% of the defect ECUs first and then the defect cars, this would still result in 500 defect cars per year, and five cars if 1% of the defects were missed. Note, that not all semiconductor errors in ECUs can be detected immediately. With faulty semiconductors, some errors might occur sporadically or only at a certain temperature or after a short time in use or alike. The key is thus to ensure semi­ conductors of good quality to start with and to target the lowest possible PPM values [60]. Zero defects are desirable, but difficult to achieve. As can be seen in Figure 3.3, a target PPM of 0.1 plus performing the relevant testing of ECUs and cars, reduces the probability of delivering a defective car because of a faulty semiconductor significantly. In the automotive industry testing is thus a relevant cost factor (even if done early in the process). The PPM difference between 30 and, for example, 0.1 explains why automotive semiconductors and parts are more expensive than those of the CE industry. As a first approximation, the testing might well represent a third of the costs for automotive semiconductors. The logic in the chip area, the package with the number of pins, and some potential licensing fees are responsible for the other two thirds.

81

3 The Automotive Environment

Probability of a defective car depending on the PPM and the errors missed during ECU and car quality checks 0,1

One car defect for 1 million produced

0.1% missed

1,0E+00 1,0E-01 1 1,0E-02 1,0E-03 1,0E-04 1,0E-05 1,0E-06 1,0E-07 1,0E-08 1,0E-09 1,0E-10 1,0E-11 1,0E-12 1,0E-13 1,0E-14 1,0E-15 PPM 1% missed 10% missed

10

100

Good CE quality

0,01

Probability

82

No quality control

Figure 3.3 Probability of delivering a defect car with respect to the semiconductor quality (PPM) and the effort spent on testing per ECU and per car (assuming the same value for both). In the example it is assumed that one million cars are produced per year, each containing 60 ECUs with 100 parts each [22]. Some of the key requirements for automotive semiconductors are specified by the Automotive Electronics Council (AEC). The AEC was founded based on an initiative of US car­ makers in 1992, who thought that it would benefit all (including the customer), if car manu­ facturers harmonized their requirements, instead of semiconductor vendors having to qualify their products to a variety of different rules [62]. Today, the AEC-Q100 qualification for Integrated Circuits (ICs) represents the minimum requirement ICs must fulfill when used in cars. At the time of writing the AEC had 37 active specifications covering additional requirements for discrete components in AEC-Q101, for optoelectronic components in AEC-Q102, for Micro Electro-Mechanical Systems (MEMS) in AEC-Q103, for MultiChip Modules (MCM) in AEC-Q104, and for passive components in AEC-Q200, with 18 major Tier 1s listed as the decision makers [63]. One of the most used terminologies from AEC-Q100 are the temperature grades. For example, AEC-Q100 grade 1 defines a temperature range for components from –40 °C to +125 °C ambient temperature. AEC-Q100 grade 2 defines the range from  –40 °C to +105 °C [40]. Another characteristic of automotive semiconductor qualification is the test of different, non-consecutive wafer lots, in order to optimize the chance of defect discovery with minimum effort. Because of the small number of samples used, the tests generally must pass with zero fails. Any fail disqualifies the complete lot. The AEC quality specifications are important for automotive semiconductor quality, but not comprehensive. Many car manufacturers have additional requirements, some of which vary depending on the location of a semiconductor inside a car. Is the ECU in a so-called “wet area” or “dry area” or are there additional temperature requirements that must be met? Some car manufacturers closely monitor  – at the semiconductor vendor  – aspects like ­quality management, design rules, test coverage, test strategy, and process technologies for the different stages of development as well as the ramp up of serial production for key semi-

3.3 Automotive Semiconductors

conductors. Sometimes, the respective process control is a prerequisite for the selection and release of a semiconductor for series production. The semiconductor vendors do not have to realize the complete automotive qualified production process in-house. Those semiconductor vendors that are fabless anyway (meaning, they have no production facility on their own) or those that focus in their production on other industries, can outsource the production of their parts to companies that offer semiconductor production following the automotive qualified process. Many of these so-called “foundries” are well accepted by the car manufacturers. Note though that the production itself is only one aspect to achieve automotive qualification. Designing the chips with sufficient test coverage to start with is also important. Another important aspect is the selection of the housing/package of the chip. The size and type of housing is obviously important for the heat dissipation and temperature grade it can meet. But there is more to it. For semiconductors with a small package and many pins, often Quad Flat No leads (QFN) packages are used. This package type is soldered onto the surface of a PCB and connects without the use of through holes. An important feature are the socalled “wettable flanks”, which show at the sides of the QFN packages [64]. This allows for the automated optical inspection of the connection after the soldering onto the PCBs with standard camera equipment, a test feature so important in the automotive industry that wettable flanks originated there [65]. Ball Grid Array (BGA) packages, for example, require in most cases expensive X-rays to inspect the soldering. To the authors’ knowledge, this is the reason why they are not the preferred choice in the automotive industry. Generally, automatic In-Circuit Testing (ICT) is required with full coverage for correct placement and connection of the flanks. ICT can also be used to perform digital adjustment of parameters or to program the used components. During ICT or another test step, it might be desirable to program individual series numbers or encryption keys, in order to be able to track all boards and components. The package, the form, and any other characteristic for the selected components shall allow for an automated assembly process with no manual interaction at any step and unnecessary impact on the throughput and production capacity. Any manual manufacturing or manual adjustment or trimming of any parameter is extremely critical and has to be avoided, because it generally results in a large variation of quality unacceptable to the automotive industry. A component for which this is particularly relevant are inductors, which, even at the time of writing, were still often wound and soldered by hand before being cased into the housing. In principle, the wires are isolated, and it should not matter if they are wound ­irregularly. However, when soldered onto the PCB, irregular wires running unintentionally close to pins can cause shorts when the isolation is destroyed in the soldering process. When introducing Automotive Ethernet, finding a suitable Common Mode Choke (CMC) that allowed for its production to be automated, was a real concern [22].

3.3.2 Semiconductor Performance Having to discuss the performance of a semiconductor might seem surprising. Of course, all semiconductors are selected for their functionality and performance to start with. However, it makes a difference what type of functionality the semiconductors are used for, in

83

84

3 The Automotive Environment

particular, if they have a more or less standalone function or if they are used as transmitters or receivers for in-vehicle communication. In the latter case, semiconductor vendors do not only have to produce EMC test results for their parts, but also – provided the communication technology allows for a multivendor environment – proof of compliance and interoperability. It is one of the advantages of using a communication technology standard that the e­ nabling test specifications are generally developed in co-operations among car manu­facturers, Tier 1s, semiconductor vendors, and specialized test houses. For Ethernet, the enabling speci­ fications are developed in the One Pair EtherNet (OPEN) Alliance [66]. For the SerDes ­technologies, the Automotive SerDes Alliance (ASA) [67] as well as the MIPI Alliance [68] are responsible for the enabling specification of their automotive SerDes ­technologies. For communication technologies additionally testing, monitoring, logging, and tracing of the data traffic must be possible during the development and when the technology is in ­series production. The investment in tools and the maintenance of a communication technology is a cost and logistics factor and part of the selection process for communication technologies. It is part of a successful standard that tool vendors and test houses support the process.

3.3.3 Semiconductor Supply While the performance and quality of semiconductors rightfully receive a large amount of attention, they are not the only important aspect with respect to semiconductors. Complex semiconductors need specific planning to secure the required volumes along the supply chain, also and especially in case of natural disasters, union strikes, or other hazards like a fire destroying a manufacturing site of a specific subcomponent. Therefore, carmakers ­prefer an active sourcing from at least two independent sources for components, either as a direct pin-to-pin replacement or a functional replacement. In any case, the semiconductor vendor has to ensure that no aspect of its production chain relies on the availability of a single site. For key components at least, car manufacturers often have specific audit departments that assess and confirm the expected quality of the manufacturing process of the semiconductor suppliers and their capability to deliver the required volume. In order to keep and maintain the technical and quality progress, an adequate balance between a conservative care of existing products and processes and continuous innovation of products is needed. The long lifetime of cars seems to be contradicting this requirement. However, without adequate improvements, semiconductor products will disappear from the market sooner or later, because they will have lost their competitiveness. Innovation and improvement are thus important, also for once successful products. And this innovation is more likely in a market with competition than in a market without. The supply and prospect of supply of new generations of products, either with cost reduction or new functionalities for the same price, is also important. With this background the selection of a single manufacturer with a specific product is not the most efficient way for a long-term business.

3.4 Bibliography

„ 3.4 Bibliography   [1]

IBISWorld, “Global Biggest Industries by Revenue in 2021,” IBISWorld, 2021. [Online]. Available: https://www.ibisworld.com/global/industry-trends/biggest-industries-by-revenue/ (no longer available). [Accessed 26 January 2021].

 [2]

iggrabs, “Die umsatzstärksten Branchen,” 21 November 2019. [Online]. Available: https://iggrabs. ch/die-umsatzstarksten-branchen/. [Accessed 26 January 2021].

 [3]

I. Wagner, “Worldwide Automobile Production through 2019,” statista, 9 November 2020. [­Online]. Available: https://www.statista.com/statistics/262747/worldwide-automobile-productionsince-2000/. [Accessed 26 January 2021].

 [4]

A. Chesterton, “How Many Cars are There in the World?,” 6 August 2018. [Online]. Available: https://www.carsguide.com.au/car-advice/how-many-cars-are-there-in-the-world-70629. [Accessed 26 January 2021].

 [5]

I. Wagner, “Revenue – Automotive Industry Worldwide 2017 – 2030,” statista, 17 September 2020. [Online]. Available: https://www.statista.com/statistics/574151/global-automotive-industry-reve nue/ (no longer available). [Accessed 26 January 2021].

 [6]

Toyota, “How Many Parts is Each Car Made off?,” Toyota Motor Corporation, 1995. [Online]. Available: https://www.toyota.co.jp/en/kids/faq/d/01/04/ (no longer available). [Accessed 26 January 2021].

  [7]

T. Hess, “How Many Parts are There on a Modern Car?,” Quora, 2020. [Online]. Available: https:// www.quora.com/How-many-parts-are-there-on-a-modern-car. [Accessed 3 March 2021].

 [8]

Mercedes-Benz, “1886–1920. Anfänge des Automobils,” not known. [Online]. Available: https:// group.mercedes-benz.com/unternehmen/tradition/geschichte/1886-1920.html. [Accessed 26 January 2021].

 [9]

Zukunftsinstitut, “Die unlogische Zukunft des Automobils,” Zukunftsinstitut, Frankfurt, ~2013.

[10]

Green Car Congress, “IHS Markit: Average Age of Cars and Light Trucks in US Rises Again in 2019 to 11.8 Years,” 28 June 2019. [Online]. Available: https://www.greencarcongress.com/ 2019/06/20190628-ihsmarkit.html. [Accessed 25 April 2020].

[11]

Grandview Research, “Used Car Market,” Grandview Research, San Francisco, 2020.

[12]

S. Haj-Assaad, “What is Resale Value and What You Can Do to Maintain it,” AutoGuide.com, 16 May 2016. [Online]. Available: https://www.autoguide.com/auto-news/2016/05/what-is-resalevalue-and-why-should-you-care-.html. [Accessed 7 February 2021].

[13]

Infinity Research, “Global Car Leasing Market 2020–2024,” Infinity Research, Toronto, 2020.

[14]

European Alternative Fuels Observatory, “Homepage of the European Alternative Fuels Observatory,” European Commission, 2021, continuously updated. [Online]. Available: https://www. eafo.eu/. [Accessed 12 February 2021].

[15]

Wikipedia, “Government Incentives for Plug-in Electric Vehicles,” 2 February 2021. [Online]. Available: https://en.wikipedia.org/wiki/Government_incentives_for_plug-in_electric_vehicles. [Accessed 12 February 2021].

[16]

Amatech, “OEMs, Tier 1, 2 & 3 – The Automotive Industry Supply Chain Explained,” 29 September 2017. [Online]. Available: https://www.amatechinc.com/resources/blog/returnable-packaging /tier-1-2-3-automotive-industry-supply-chain-explained. [Accessed 9 April 2022].

85

86

3 The Automotive Environment

[17]

M. Pesce, “Software Takes on More Tasks in Today’s Cars,” Wired, 25 April 2011. [Online]. Available: https://www.wired.com/2011/04/the-growing-role-of-software-in-our-cars/. [Accessed 9 February 2021].

[18]

O. Burkacky, J. Deichmann and J. P. Stein, “Mapping the Automotive Software and Electronics Landscape Through 2030,” Mc Kinsey, Düsseldorf, 2019.

[19]

WikiChips, “Tesla FSD Chip,” 28 December 2020. [Online]. Available: https://en.wikichip.org/ wiki/tesla_(car_company)/fsd_chip. [Accessed 9 February 2021].

[20]

BMW, Intel, Mobileye, “BMW Group, Intel and Mobileye Team Up to Bring Fully Autonomous Driving to Streets by 2021,” 1 July 2016. [Online]. Available: http://www.bem-ev.de/wp/wp-con tent/uploads/2016/08/BMW-Intel-Mobileye-Press-Release.pdf. [Accessed 12 February 2021].

[21]

I. Riches, “Powertrain, Body, Chassis & Safety Service,” Strategy Analytics, Boston, 2021.

[22]

K. Matheus and T. Königseder, Automotive Ethernet, Third Edition, Cambridge: Cambridge University Press, 2021.

[23]

J. Scobie, “The Future of Mobility: A Centralized Vehicle Architecture,” Electronic Design, 9 December 2020. [Online]. Available: https://www.electronicdesign.com/markets/automotive/article /21149210/arm-the-future-of-mobility-a-centralized-vehicle-architecture. [Accessed 2 February 2021].

[24]

S. Burghardt, S. Choi and F. Weig, “Mobility Trends: What’s Ahead for Automotive Semiconductors,” April 2017. [Online]. Available: https://www.mckinsey.de/~/media/McKinsey/Industries/ Semiconductors/Our%20Insights/Mobility%20trends%20Whats%20ahead%20for%20automotive% 20semiconductors/Mobility-trends-Whats-ahead-for-automotive-semiconductors.pdf.  [Accessed 26 January 2021].

[25]

P. Amsrud, “Global Automotive Semiconductor Revenue in 2020,” IHS Markit, 5 November 2020. [Online]. Available: https://ihsmarkit.com/research-analysis/global-automotive-semiconductor-­ revenue-in-2020.html. [Accessed 26 January 2021].

[26]

Fortune Business Insights, “Big Data in Manufacturing Industry,” Fortune Business Insights, Pune, India, 2020, March.

[27]

TCgen, “Time To Market: What it is, Why it’s important, and Five Ways to Reduce it,” not known. [Online]. Available: https://www.tcgen.com/time-to-market/. [Accessed 9 April 2022].

[28]

K. Matheus, “Ethernet @ BMW; From Camera Link to System Bus,” in: Nikkei Electronics Symposium “Car-mounted Ethernet to Shape the Future of Automobile”, Tokyo, 2013.

[29]

J. Link, “The Top Three Things Customers Look for When Buying a Car,” AutoFi, 16 July 2019. [Online]. Available: https://blog.autofi.com/the-top-three-things-customers-look-for-when-buying-acar/. [Accessed 31 January 2021].

[30]

Volkswagen Group, “Wer mit wem? Die Verflechtungen der Autobranche,” Volkswagen, Wolfsburg, 2012.

[31]

S. Cerchez, “Ungleiche Zwillinge,” eurotransport.de, 25 June 2012. [Online]. Available: https:// www.eurotransport.de/artikel/mercedes-sprinter-vw-crafter-ungleiche-zwillinge-625122.html.  ­[Accessed 9 February 2021].

[32]

M. Andrusio, “Toyota-Supra Chefingenieur: Wie die Zusammenarbeit mit BMW lief,” motor.at, 26 May 2019. [Online]. Available: https://motor.at/technik/toyota-supra-chefingenieur-wie-diezusammenarbeit-mit-bmw-lief/400504426. [Accessed 9 February 2021].

[33]

S. Carlson, H. Zinner, N. Wienckowski, K. Matheus and T. Hogenmüller, “CFI Multi-Gig Automotive Ethernet PHY,” November 2016. [Online]. Available: https://grouper.ieee.org/groups/802/3/ cfi/1116_1/CFI_01_1116.pdf. [Accessed 12 February 2021].

3.4 Bibliography

[34]

E. Barrow, “How Long Does it Take to Build an App?,” 3 Sided Cube, 22 May 2019. [Online]. Available: https://3sidedcube.com/how-long-does-it-take-to-build-an-app/. [Accessed 12 February 2021].

[35]

T. Grünweg, “Modellzyklen der Autohersteller. Eine Industrie kommt auf Speed,” Spiegel Online, 10 February 2013. [Online]. Available: https://www.spiegel.de/auto/aktuell/warum-lange-entwick lungszyklen-fuer-autohersteller-zum-problem-werden-a-881990.html. [Accessed 12 February 2021].

[36]

A. Agarwal, “Understanding Automotive OTA (Over-the-Air Update),” 26 June 2020. [Online]. Available: https://www.pathpartnertech.com/understanding-automotive-ota-over-the-air-update/. [Accessed 12 February 2021].

[37]

V. Vivek, “65% of Rolls-Royce Cars Ever Made are Still on Road,” inshorts, 27 August 2016. [­Online]. Available: https://inshorts.com/en/news/65-of-rollsroyce-cars-ever-made-are-still-on-road1472301358191. [Accessed 12 February 2021].

[38]

Wikipedia, “Moore’s Law,” 10 February 2021. [Online]. Available: https://en.wikipedia.org/wiki/ Moore%27s_law. [Accessed 12 February 2021].

[39]

Kraftfahrt Bundesamt, “Bestand an Kraftfahrzeugen und Kraftfahrzeuganhängern,” January 2022. [Online]. Available: https://www.kba.de/SharedDocs/Downloads/DE/Statistik/Fahrzeuge/ FZ15/fz15_2022.pdf?__blob=publicationFile&v=5. [Accessed 15 February 2021].

[40]

Automotive Electronics Council, “AEC-Q100 Failure Mechanism Based Stress Test Qualification,” 11 September 2014. [Online]. Available: http://www.aecouncil.com/Documents/AEC_Q100_Rev_ H_Base_Document.pdf. [Accessed 8 April 2022].

[41]

VDE Verband der Elektrotechnik Elektronik Informationstechnik e. V., “ITG-Positionspapier: Kfz-Anforderungen an Elektronik-Bauteile,” about 2005. [Online]. Available: https://docplayer. org/12978134-Itg-positionspapier-kfz-anforderungen-an-elektronik-bauelemente.html. [Accessed 22 May 2020].

[42]

H. Zinner, K. Matheus, S. Buntz and T. Hogenmüller, “Requirements Update for RTPGE,” July 2012. [Online]. Available: https://grouper.ieee.org/groups/802/3/RTPGE/public/july12/zinner_ 02_0712.pdf. [Accessed 21 February 2021].

[43]

Wikipedia, “Paris Agreement,” 30 April 2020. [Online]. Available: https://en.wikipedia.org/wiki/ Paris_Agreement. [Accessed 9 May 2020].

[44]

M. Kicherer (Turner) and T. Königseder, “BMW Proposal for an AVB Gen 2 Automotive Profile,” BMW White Paper, Munich, 2013.

[45]

S. Carlson, “Reduced Twisted Pair Gigabit Ethernt PHY Call For Interest,” March 2012. [Online]. Available: https://www.ieee802.org/3/RTPGE/public/mar12/CFI_01_0312.pdf. [Accessed  21 Februrary 2021].

[46]

United Nations Economic Comission for Europe, “Three Landmark UN Vehicle Regulations Enter into Force,” 5 February 2021. [Online]. Available: https://unece.org/sustainable-development/ press/three-landmark-un-vehicle-regulations-enter-force. [Accessed 25 February 2021].

[47]

NHTSA, “Regulations,” 2022 (continuously updated). [Online]. Available: https://www.nhtsa.gov/ laws-regulations/fmvss. [Accessed 9 April 2022].

[48]

Wikipedia, “Federal Motor Vehicle Safety Standards,” 14 January 2021. [Online]. Available: https://en.wikipedia.org/wiki/Federal_Motor_Vehicle_Safety_Standards. [Accessed 24 February 2021].

[49]

Bundesministerium für Justiz und Verbraucherschutz, “Straßenverkehrs-Zulassungs-Ordnung (StVZO),” 26 April 2012. [Online]. Available: https://www.gesetze-im-internet.de/stvzo_2012/ BJNR067910012.html. [Accessed 24 February 2021].

[50]

IEC, “IEC 60529:1989/AMD2:2013/COR1:2019 Corrigendum 1 – Amendment 2 – Degrees of Protection Provided by Enclosures (IP Code),” IEC, Geneva, 2019.

87

88

3 The Automotive Environment

[51]

ISO, “ISO 20653:2013 Road Vehicles — Degrees of Protection (IP Code) — Protection of Electrical Equipment Against Foreign Objects, Water and Access,” ISO, Geneva, 2013.

[52]

R. Bellairs, “What is ISO 26262? Overview and ASIL,” Perforce, 3 January 2019. [Online]. Available: https://www.perforce.com/blog/qac/what-is-iso-26262. [Accessed 24 February 2021].

[53]

Economic Commission for Europe of the United Nations, “Regulation No 10 of the Economic Commission for Europe of the United Nations (UN/ECE) — Uniform Provisions Concerning the Approval of Vehicles with Regard to Electromagnetic Compatibility,” EUR-Lex, 26 July 2012. [­Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:42012X0920(01)&from=DE. [Accessed 25 February 2021].

[54]

Ford Motor Company, “Component EMC Specifications FMC1278 & FMC1280,” 27 July 2019. [Online]. Available: http://www.fordemc.com/docs/requirements.htm. [Accessed 26 February 2021].

[55]

Automotive Industry Action Group, “IATF 16949:2016,” Automotive Industry Action Group, Southfield, 2016.

[56]

Allianz Zentrum für Technik, “Kompetenzzentrum der Allianz für Automobiltechnologie,” 2021, continuously updated. [Online]. Available: https://azt-automotive.com/. [Accessed 26 February 2021].

[57]

Allianz Zentrum für Technik (AZT), “Themen aus dem AZT: Einige Highlights aus dem Jahr 2020,” 2021. [Online]. Available: https://azt-automotive.com/de/themen/Jahresrueckblick2020. [Accessed 26 February 2021].

[58]

Thatcham Research, “Insurance Industry Requirements (IIR) for the Safe Repair of ADAS Equipped Vehicles,” July 2020. [Online]. Available: https://www.thatcham.org/wp-content/up loads/2020/07/006-IIR-Requirements-March-2020-v2.pdf. [Accessed 27 February 2021].

[59]

Wikipedia, “Euro NCAP,” 4 January 2021. [Online]. Available: https://en.wikipedia.org/wiki/ Euro_NCAP. [Accessed 27 February 2021].

[60]

Infineon, “Living Automotive Excellence, on the Way to Zero Defect Products and Services,” 13 September 2010. [Online]. Available: https://www.infineon.com/export/sites/default/cn/prod uct/promopages/ATV_Symposium/China_ATV_Symposium_ATV_Automotive_Excellence.pdf. [Accessed 5 March 2021].

[61]

C. Kymal and P. Patiyasevi, “Semiconductor Quality Initiatives,” April 2006. [Online]. Available: https://www.qualitydigest.com/april06/articles/05_article.shtml. [Accessed 5 March 2021].

[62]

Automotive Electronics Council, “AEC History,” not known. [Online]. Available: http://www. aecouncil.com/AECHistory.html. [Accessed 6 May 2020].

[63]

Automotive Electronics Council, “AEC Documents,” 2021, continuously updated. [Online]. Available: http://www.aecouncil.com/AECDocuments.html. [Accessed 5 March 2021].

[64]

NXP, “AN1902 Assembly Guidelines for QFN (Quad Flat No-lead) and SON (Small Outline Nolead) Packages,” 6 February 2018. [Online]. Available: https://community.nxp.com/pwmxy87654/ attachments/pwmxy87654/nxp-designs%40tkb/204/5/AN1902.pdf. [Accessed 6 March 2021].

[65]

R. Fey, “What is Surface Mount, Wettable Flank?,” TechForum, March 2019. [Online]. Available: https://forum.digikey.com/t/what-is-surface-mount-wettable-flank/3055. [Accessed 6 March 2021].

[66]

OPEN Alliance, “We Enable Wide-scale Adoption of Ethernet-based Automotive Connectivity,” 2021, continuously updated. [Online]. Available: https://opensig.org/. [Accessed 3 March 2021].

[67]

Automotive SerDes Alliance, “Automotive SerDes Alliance Homepage,” 2021, continuously updated. [Online]. Available: https://auto-serdes.org/. [Accessed 3 March 2021].

[68]

MIPI Alliance, “MIPI Alliance Homepage,” 2021, continuously updated. [Online]. Available: https://www.mipi.org/. [Accessed 3 March 2021].

4

The Electromagnetic Environment in Cars

When electronic engineers from other industries are confronted with the automotive industry for the first time, they typically hear early in the discussion that the electromagnetic requirements are particularly stringent, especially in the context of communication technologies. This chapter thus provides an overview on ElectroMagnetic Compatibility (EMC) in Section 4.1 and ElectroStatic Discharge (ESD) in Section 4.2.

„ 4.1 ElectroMagnetic Compatibility (EMC) EMC is related to various effects caused by electric, magnetic, or electromagnetic fields. Every electric circuit creates such fields. In our daily lives, these effects are generally not that noticeable though; also, because there are government regulations in place that r­ equire device makers to limit the ElectroMagnetic Emissions (EME) of their devices. Additionally, next to limiting EME, devices have to be robust against electromagnetic interference with ­sufficient ElectroMagnetic Immunity (EMI) in order to be sellable to customers [1]. EMC has a long history. Already in the early years of the telegraph and telephone business, it was evident that physically close telegraph or telephone lines would interfere with each other’s transmissions. In 1892, Germany was the first country to pass a law to limit such effects. In 1927, Germany also was the first country to pass a law specifically on the use and installation of high-frequency radio transmitters, which, with adaptations, was in place until 1995 [1]. A number of international organizations followed suite to develop guidelines for the respective regions, like the Comité International Spécial des Perturbations Radioélectriques (CISPR) for Europe [2] or the American National Standards Institute (ANSI) and Federal Communications Commission (FCC) for the U. S. [3]. With the invention and spread of the transistor, the need arose to regulate EMC on an even broader scale. In 1973 the International Electrotechnical Commission (IEC) created a special technical committee with the purpose of handling EMC topics [1]. The IEC thus publishes relevant EMC related specifications not only for cars as such but also for the communication technologies used within. The negligence or non-observance of EMC can have dramatic consequences. In July 1984, for example, electromagnetic interference caused a “Tornado” fighter airplane to crash. The pilot lost control over the jet in the vicinity of a radio transmitter station near Munich, Germany. The high electrical field strength of the radio station, which was known to operate at

90

4 The Electromagnetic Environment in Cars

a high transmission power in the AM and short-wave bands, interfered with the insufficiently protected flight control and steering systems. It caused the jet to change course and crash to the ground causing the death of both occupants [4] [5]. This is a very dramatic example of the lack of EMC. However, it shows that EMC in cars is more than preventing crackling noises in the car stereo. It is a safety issue, even more so considering the increasing amount of electronics taking over originally mechanical driving functions, functions for which cameras and other sensors often play an important role. This section provides an overview on important EMC topics related to high-speed communication systems in cars. Section 4.1.1 gives a brief introduction into relevant types of electromagnetic interference. Section 4.1.2 describes the different levels at which EMC needs to be considered for cars. Section 4.1.3 introduces important test methods. Section 4.1.4 discusses the relevance and importance of shields and ground connections for high-speed communication systems. EMC is a complex topic, with close interrelations also to the transmission channel described in Chapter 5, where specific channel related EMC information can be found. For a deep dive, we recommend [6] [7] [8] [9] [10]. What makes EMC in cars so special? After all, all electronic products of all industries need to be safe and EMC compliant. There are various aspects to this. The most obvious challenge is the electric ground (GND) of cars. In the past, the GND was the steel chassis and it was predominantly used as the return path for the power supply. Today, the chassis-GND is needed increasingly also as a return path for communication systems and/or as a GND for their shields, while at the same time, cars might be built of nonconductive compound materials [11]. Furthermore, cars drive around on rubber tires, meaning that the chassis is insulated from the real ground. Per defi­nition, cars constantly change their position. This is not the case for stereos, desktop computers, or industrial production plants. Then, cars can be a significant source of electromagnetic interference themselves. In early times, the ignition of combustion engine cars with its archaic generation of the ignition spark was a heavy EME source. Today, the power generation and converters for the motors of electric vehicles generate a significant amount of EME in the frequency range between 50 kHz and 100 MHz. Furthermore, the 12 V power supply is not very stable, significantly less so than the 230 V mains power supply in houses. Then, the regulation sets higher requirements for the car industry than in other industries. The ECE-R-10, for example, requires in general an immunity up to 23 V/m for automotive homologation. 3 V/m is a typical requirement for office or industrial environments. 200 V/m is required by car manufacturers, such as BMW and Ford [12] [13], for their cars. Requirements for even higher values exist in the car industry. Note that already 200 V/m is at a level, which might be well n ­ oticeable (and critical!) in the human body, if applied. The level is nevertheless required for vehicle testing, because cars must operate safely in all possible, not predetermined environments! Then there are spatial and economic restrictions (see also Section 3.2.1). The space restrictions cause the industry to bundle power supply and communication wires close together in the wiring harness. The economic restrictions emphasize the use of cost efficient and light material. Extra shields, for example, are good for EMC but add to hardware costs and weight. 

4.1 ElectroMagnetic Compatibility (EMC)

4.1.1 Basic Principle of Electromagnetic Interference Electromagnetic interference happens when an electric current is unintentionally induced/ coupled into a system by an external source. When looking at the physics, there are four different ways in which this can happen. The first distinction is whether the coupling is conducted or caused by a field. A second distinction is whether the field is near field or far field. Near field means that the interfering source is less than 1/6th of the wavelength away [14]; at a frequency of 1 GHz this means about less than 5 cm. Near-field coupling result in an interfering current either from a magnetic field (inductive) or from an electric one (­capacitive) [8]. Table 4.1 provides an overview on the different variants. Table 4.1 Coupling mechanisms of electromagnetic interference [8] Type

Source of ­interference

Physical mechanism Countermeasure

Conducted coupling

Can couple common Inside the same device, e. g. un­itended or differential signal HF signal leaves a unit via power supply cable

Far-field coupling (also called Radio Frequency (RF) interference)

Systems that are ­further away, such as mobile phone base stations, terrestrial broadcast stations, airport radars

Filtering, good ground connections and ECU design

Electromagnetic energy from Trans­ versal ElectroMagnetic (TEM) waves, energy is generally below the defined limits

Generally small, ­coupled energy, robustness against interference in the communication ­system (see also ­Section  4.1.4)

Capacitive near-field Systems that are coupling nearby with high impedance such as high voltage power lines, ignition systems, communication ­transceivers

Changing electric field of high impedance systems, crosstalk (XTALK)

Shielded cable (see also Section 4.1.4.1), system margin

Inductive near-field coupling

Shielded cable (see Changing magnetic field of low impedance also Section 4.1.4.1), systems, interference system margin increases with ­nearness, frequency, power, XTALK

Systems that are nearby with low impedance such as highway control ­transmitters, wireless stations and radio frequency transmitters

All four coupling mechanisms can occur simultaneously. It is therefore important to identify the possible sources of interference in order to provide the right counter measures. In all cases, the transmission frequency, the transmit power, and the actual distance of the interferer are decisive criteria.

91

4 The Electromagnetic Environment in Cars

4.1.2 Relevant EMC Levels With ADAS functionality and plans for fully autonomous driving cars, the requirements for EMC are ever more important. Up until recently, safety-critical systems like the steering always had a fully mechanical fallback for the driver. In the future, this is likely no longer the case, which stresses the importance of EMI. Figure 4.1 depicts the different levels at which EMC can occur in cars. The highest order is the electromagnetic interrelation between the car and its environment. It is an important aspect of the homologation of a car model, that it does not cause electromagnetic interference to electric devices outside of it. Furthermore, as the example of the Tornado fighter showed so vividly, the car needs to be immune against a reasonably high and expectable level of electromagnetic radiation existing outside the car. Examples for possible sources of electromagnetic interference are heavy machinery at construction sites, mobile ham-radio stations, or the mobile phone brought into the car by the customer. For car manufacturers, this is the most important level of EMC and the level at which the car manufacturers perform a significant number of tests themselves. The specification ECE-R-10-7 [15] defines the valid legal requirements for the European Union, which are harmonized with the requirements worldwide. However, to fulfil these legal requirements is only the absolute minimum for car manufacturers. As a rule, car makers exceed them and create company specific, generally proprietary specifications (see [12] [13] for examples of published versions). These company specific specifications are based on relevant ISO and IEC norms (see Table 4.2 for examples). As EMC is safety relevant and subject to the same laws of physics, to the authors’ knowledge, most car manufacturers have very similar requirements.

Communication cable

Environment

Environment

Environment

Environment

92

Figure 4.1 Different levels of EMC from the environment down to IC-level [16] The government regulates the EMC behavior of cars to the outside world. Inside, however, the car is not one single unit but consists of various electronic systems represented by the ECUs and their communication links. The risk of electromagnetic interference thus does not only have to be controlled between the car and its environment but also inside the car, between the individual systems, so that they do not disturb each other but perform as intended.

4.1 ElectroMagnetic Compatibility (EMC)

The most sensitive systems inside the car are those that rely on antennas per definition, like GPS, (digital) radio, TV, in-car mobile communication, WiFi-systems etc. Crackling noises in the radio and TV-systems have a noticeable negative impact on the user experience (some might remember the early days of mobile phones, when the localization signal from base stations resonated in nearby speaker systems). With future self-driving cars, the impact of a disrupted GPS or wireless communication can be significantly worse. In order to ensure EMC inside cars, car manufacturers require respective tests from their Tier 1 and component suppliers that are in line with their requirement and test specifications (such as the aforementioned [12] [13]). Car manufacturers perform regular reviews on the test plans and test results of the systems and ECUs. The requirements for systems and ECUs are set such that the combination of all units in the car – knowledge gained with generations of cars – will result in a “pass” for the final car test. The last level depicted in Figure 4.1 is the IC and communication level. For most ICs within a car, car manufacturers themselves perform only a very limited number of respective EMC tests. For some categories of IC types, different sets of EMC requirements exist for the semiconductor ­manufacturers [17]. Here, it is the responsibility of the Tier 1 suppliers to select the right products. The communication technologies are particularly important for the EMC behavior of a system. After all, every communication wire is a type of antenna and particularly susceptible to collecting as well as emitting in electromagnetic interference. For most standardized communication technologies specific EMC and ESD test specification have been developed. For widely used in-car communication technologies like LIN, CAN, and Ethernet, harmonized requirements and tests exist [18] [19] [20] [21]. Semiconductor vendors are responsible for testing their communication ICs and components. The test results are an important selection criterion for Tier 1 suppliers. Car manufacturers might use the test results in order to decide whether a specific IC can make it onto their recommendation list. Furthermore, for new communication technologies, car manufacturers might perform their own tests in order to decide whether to use a specific technology at all. The Automotive SerDes technologies used in cars at the time of writing were all proprietary and for those technologies no harmonized definition of IC or ECU EMC tests existed. This is one of the issues when using proprietary communication technologies: There is little incentive to develop respective specifications, also because the IC makers often have little interest to make the necessary technical details known to a wider audience and because the market coverage of proprietary solutions is generally too small to make it worthwhile for third parties. The standardization of Automotive SerDes technologies changed this, and at the time of writing, the respective (EMC) test specifications were being developed. So, while the narration along Figure 4.1 described the different levels of EMC top down, the tests themselves are traversed bottom up. The semiconductors – especially for the communication links – must show a good EMC behavior first. Then the ECU itself, into which the ICs are combined, and the ECU connectivity need to comply with the EMC limits, before ­final tests are preformed inside the car. Only when every element is EMC compliant is it possible that also the sum of elements fulfills the requirements. Table 4.2 shows an example hierarchy for EMC measurement methods used for automotive communication systems. The TEM cell tests are no longer commonly used at car manufacturers. It is therefore not listed and detailed. The exact tests required may vary from car manufacturer to car manufacturer. Selected test methods are explained in more detail in the next Section 4.1.3.

93

94

4 The Electromagnetic Environment in Cars

Table 4.2 Example for EMC measurement methodology for IVC technologies. Semiconductor

ECU

Vehicle

Main responsibility of ­semiconductor vendor

Main responsibility of Tier 1 supplier

Main responsibility of car ­manufacturer

ISO 11452-5: Stripline

ISO 11451-3, CISPR12/EN55012:

Emissions IEC 61967-4: 150 Ohm method

On-board transmitter, protection ­off-board receiver ISO 11452-2: Antenna ­measurements in absorber lined chambers

CISPR25/EN55025:

ISO 11452–4: Bulk Current Injection (BCI)

ISO 11451-4: Bulk-Current-­Injection

ISO 11452-2: Antenna ­measurements in absorber lined chambers

ISO 11451-2: Antenna measurements in absorber lined chambers

ISO 7637-3: (Direct, inductive, or capacitive) coupling

ISO 11452-5 with OEM adaptations for large cars: Stripline

Measurements with vehicle on-board antennas/receiver, ­measurements in absorber lined chambers

Immunity IEC 62132-4: Direct Power Injection (DPI)

4.1.3 Overview on EMC Test Methods The following list provides a short description of the test methods seen as most relevant to the authors. ƒ 150 Ohm method The 150 Ohm method measures the emissions of a communication technology by adding a defined coupling network to the communication link (see left side of Figure 4.2, the right side shows the setup for testing individual I/O pins or the power supply). The coupling network has a load impedance of about 150 Ω. The values for the capacitors depend on the frequency of the communication systems under investigation. For 100 Mbps Ethernet, for example, the values for C1 and C2 are 470 pF. For higher frequencies the values need to be adapted to respective smaller values. Setup for differential communication

DUT

RF analyzer

Ri 50 Ω

C1

Rm 50 Ω

C2

Setup for I/O pins

Setup for power supply pins Vx

RF analyzer

R1 120 Ω R2 120 Ω

Ri 50 Ω DUT

Figure 4.2 Test setups for the 150 Ohm method

Rx Cx 120 Ω 6.8 nF Rm 50 Ω

Lx 4.7μH Pin Vx Pin X

DUT

95

4.1 ElectroMagnetic Compatibility (EMC)

ƒ Direct Power Injection (DPI) method The DPI method “inverses” the process performed in the 150 Ohm method. As can be seen in Figure 4.3 the RF power is coupled directly into the DUT by a network of passive components. Setup for differentiell communication

RF generator & amplifier Ri 50 Ω

DUT

C1

C2

Setup for /IO pins

RF generator & amplifier

R1 120 Ω

Ri 50 Ω

R2 120 Ω

Cx 6.8 nF

Pin X

DUT

Figure 4.3 Test setups for DPI method ƒ Stripline measurements For the stripline measurement the DUT is placed in the middle between a so-called “septum” and a GND plate (see Figure 4.4). When applying RF power to the stripline setup, an electrical field is created between the septum and the GND, which allows to test the signal integrity/immunity of the DUT. The septum and the GND form a transmission line with a controlled impedance, typically of 90 Ω. The RF power test value is specified in V/m. Typical test values are from 50 V/m to 400 V/m. The setup has the best coupling properties between 30 MHz and 500 MHz, while it is not uncommon to use the setup up to 1 GHz.

RF generator & amplifier Ri 50 Ω

GND Septum

DUT Impedance matching

E-field

DUT

RLoad

Figure 4.4 Stripline test setup The impedance of the measurement setup is maintained by keeping the relation between the mechanical parameters constant. The width of the septum “b” and its height over GND “a” have the same relation as the inner “d” to outer “D” diameter of the ­coaxial cable feeding the stripline test setup (see Figure 4.5).

DUT

4 The Electromagnetic Environment in Cars

Stripline

Septum

Coax cable a

96

Impedance = Z

R Load

GND

Impedance = Z

Figure 4.5 Stripline setup with respect to the impedance Z Stripline works for emissions as well as signal integrity measurements. For emissions tests an RF test receiver replaces generator and amplifier. ƒ Bulk Current Injection (BCI) testing In the BCI testing, an inductive coupling clamp is applied to the wiring harness that induces currents into the wires of the harness with help of the magnetic field generated by the clamp (see Figure 4.6). A second sensing clamp is used for monitoring the induced current, measured in mA. The frequency range in which to perform BCI tests depend on the coupling clamps used. In most cases the clamps work from very low frequencies like 100 kHz up to 400 MHz, but versions that can be used for frequencies up to 1 GHz are available, too. There are two ways to perform BCI tests: the closed loop method and the substitution method.

RF generator

BCI testing Inductive coupling clamp

RF Power Cntl

DUT

DUT Sensing clamp

RF amplifier LISN

GND plane

Figure 4.6 BCI test setup (the Line Impedance Stabilization Network (LISN) emulates the ­wiring harness of the power supply in a car) In the closed loop method, the sensing clamp measures the current induced in the wiring harness and controls the amplifier of the coupling clamp to apply the RF power necessary in order to achieve the defined current level in the sensing clamp. Closed loop measurements are in general the correct way for tests with high-speed communication technologies such as Automotive SerDes or Ethernet.

4.1 ElectroMagnetic Compatibility (EMC)

The substitution method is used in order to have harmonized test conditions for real wiring harnesses, which do not have a defined load impedance. An example is 100BASE-T1 with UTP wires, which has a very high impedance. It might also be specific actuators, which have a very low impedance. In the substitution method the RF amplifier is controlled via prerecorded test values that are such that specific currents are realized in the sensor clamp. ƒ Antenna tests Tests, which use antennas, can cover the widest frequency range. The frequency range intended to be tested determines the form of the antenna. A rod antenna, for example, can be used for frequencies up to about 100 MHz. A periodic broadband antenna (­LogPer antenna) is used in the range of up to 1 to 3 GHz. For higher frequencies horn antennas are used. LogPer antennas and horn antennas have a polarization characteristic. The tests have to be done with both vertical and horizontal polarization of the antennas. Above 100 MHz antenna tests have a directional sensitivity. Therefore, the position of the ­antenna towards the DUT is essential and part of the test variation. Antenna tests generate a radiated electro-magnetic field. The field strength is defined in V/m. Antenna tests are also used for testing the emissions of a DUT. Then an RF test receiver replaces the generator and amplifier behind the antenna.

RF generator

Antenna testing

RF amplifier

DUT

LISN

DUT

GND plane

Figure 4.7 Setup for antenna tests Antenna tests for signal integrity on system level are defined in ISO 11452 [22] for test setups of system components within an Absorber-Lined Shielded Enclosure (ALSE). This means the setup is tested in a chamber which absorbs RF power at the walls in order to simulate an open field. Ferrite tiles or RF absorbing foam cones are installed and absorb RF energy at the walls. There are no reflections from the walls. Radiated power in the chamber comes only from the according source or defined test setup, not from the environment.

97

98

4 The Electromagnetic Environment in Cars

ƒ Coupling methods Clamp coupling methods are used to test the immunity of communication systems to transient bursts. The definition for transient bursts is given in ISO 7637-3 [23] as electrical transmission by capacitive and inductive coupling to the communication lines by bursty interference. ISO 7637-3 defines two basic classes of bursts: Slow repetitive pulses with a duration of 0.05 ms and fast repetitive pulses with a duration of 0.15 µs. The pulses are defined with both polarities. For the tests, several test levels are defined. Depending on the test method, the required test level transient voltages peak from +/–6 V to +75 V or –110 V. There are three different clamp coupling methods. The Capacitive Clamp Coupling method (CCC method) is used to test with fast transient pulses. The Direct Capacitive Coupling method (DCC method) is used to test with fast and slow transient pulses. The Inductive Coupling (or Current) Clamp method (ICC method) is used is to test with slow transient pulses. See Figure 4.8 for details on the respective test setups. Communication system

Communication system Capacitive coupling clamp

DUT communication link

50 Ohm

Pulse generator Communication system

Communication system

Direct capacitive coupling

DUT communication link

Pulse generator Communication system

Communication system Inductive coupling clamp

DUT communication link

Pulse generator

Figure 4.8 Different coupling setups

4.1 ElectroMagnetic Compatibility (EMC)

4.1.4 Impact of a Shield on EMC Generally, in electromagnetically challenging situations various forms of metal shields are used to limit the electromagnetic emissions and to protect from electromagnetic interference. A shield might be a metal cage to protect a specific chip on a PCB, it might be the metal cask of a complete ECU, or it is part of the cables and connectors used for the communication link. When using shielded cables, not only the right cable but also the complete ECU design and ground situation in the vehicle needs to be considered in order to achieve the desired protection. This sections details the impacts by discussing in Section 4.1.4.1 the electromagnetic impact of shielded cables, in Section 4.1.4.2 the relevance of the connection between the cable shield and the ECU case, and in Section 4.1.4.3 the necessary ground connection.

4.1.4.1 EMC for Shielded Cables For transmission systems working at high frequencies it is generally understood that a shielded cable has to be used in order to ensure the required transmission quality. The ­basic reason behind is simple and depicted in Figure 4.9. Every communication system is designed with a maximum transmitter output and a minimum receiver input in mind. If the transmitter output is higher, the risk that the immunity of other systems would be impaired is too high. If the receiver input is lower, the risk that noise would be mistaken for a valid signal is too high. Between the transmitter output and the receiver input is the transmission channel, which attenuates the signal strength of the transmitted signal. With increasing frequency, the attenuation increases. Unshielded channel Power level

Power level

Immunity limit of others reached

Unshielded channel Power level

Immunity gap

Immunity gap

Shielded channel

Immunity gap

Maximum transmitter output Attenuation / IL

Increased frequency

Attenuation / IL

Attenuation / IL

Minimum receiver input Link budget / SNR / emissions gap Interfering noise floor

Link budget / SNR / emissions gap

Link budget / SNR / emissions gap

Figure 4.9 Principal effect of shielding on the power levels of a communication channel [6]

99

100

4 The Electromagnetic Environment in Cars

In order to still achieve a specific, required Signal to Noise Ratio (SNR) thus either the signal strength at the transmitter needs to be increased and/or the receiver has to be able to function with less input strength. Without a shield, neither is generally possible though (see middle part of Figure 4.9). Increasing the transmit power would harm neighboring systems. To reduce the input level would mean the receiver can no longer distinguish between signal and noise. A shield helps in both directions (see right part of Figure 4.9). It ­allows to increase the transmit power, because the shield reduces the emissions seen by others (meaning the level at which the immunity of others is harmed is increased). At the same time, it shields off some of the interfering noise, thus reducing the noise level at the receiver. The following text explains in more details how exactly the effect of the shield is achieved in case of coaxial and Shielded Twisted Pair (STP) cables. In the latter case, additionally the positive effect of the differential transmission over the twisted wires is explained. To start with, Figure 4.10 depicts the electromagnetic situation for coaxial cables. outer H-field

E-field

Transmission loop

inner H-field Ri

Zcable

RL

GND loop

Zshield Gen Zgnd_loop Zgnd2

Zgnd1 GND

Figure 4.10 Electromagnetically relevant fields for coaxial cables On the left side is the sender, shown as a generator with the source resistance Ri. Then f­ollows the transmission channel: the coaxial cable with the impedance Zcable. On the right side the receiver is represented by the load resistance RL. The shield of the coaxial cable is connected to a common ground (GND). In most cases – but not always – the ground con­ nection is done at both ends, as this is seen as the best method to get an optimum EMC behavior. The generator, the cable, and the load resistance form an electrical circuit called the “transmission loop”, with a current flowing in the inner conductor and returning in the outer conductor/shield. Figure 4.10 depicts an electric (E) and a magnetic (H) field in an around the conductor. In lower frequencies the magnetic field, and in higher frequencies the ­electric field is dominant. As can be seen, the electric field is between the inner conductor and the shield, which blocks the radiation from leaving the wire. For this reason, a shield is an effective method to suppress EME. Because the magnetic fields are induced by currents in opposite directions, Figure 4.10 shows magnetic fields also in opposite directions. Due to the different geometries of the ­inner conductor and outer shield, the two fields do not annihilate each other completely. Because normally the shield of coaxial cables is made of a non-ferromagnetic material, the shield does not block magnetic fields either, so some magnetic field strength remains,

4.1 ElectroMagnetic Compatibility (EMC)

which effects the lower frequencies in the range of maximum several hundred kHz. Overall, the effect of the shield is measured in the so-called “screening attenuation”. Figure 4.10 shows another effect that is a result of the GND connection. In a real set up, when the car’s body is made of steel or aluminum and used as GND, there are real or at least parasitic connection impedances, Zgnd1, Zgnd2, and Zgnd_loop in Figure 4.10. This results in a GND loop, which is also exposed to EMC effects. In case of coaxial cables, where the shield is also the outer conductor, the transmission and the GND loops are coupled. This means the electromagnetic interference in the GND loop can affect the data transmission. This explains why a coaxial cable does not solve all EMC issues and why the right GND connection is so important. Traditionally in cars, the body is used as the return path for minimum all DC currents as well as for some, generally low speed, single-ended commu­ nication technologies like LIN, J1850 [24], Pulse-Width Modulation (PWM) [25], Peripheral Sensor Interface five (PSI5) [26], Single Edge Nibble Transmission (SENT) [27], so the ­effects are not negligible. The situation is somewhat different for differential signaling, when a pair of wires is used to transmit the signal in a closed electric circuit, independent from the environment or a ­common electrical GND system. The following explains why differential signaling has a very positive effect on EMC behavior and noise immunity even in case Unshielded Twisted Pair (UTP) cables are used. Adding a shield with STP further improves the EMC characteristics. Figure 4.11 depicts the situation for STP cables. Common mode transmission loop Idiff

Inner H-fields

E-fields

Differential transmission loop

Zdim Ri

RL

Zshield

Zcim

GND Loop Transfer common mode to GND loop

Zgnd_loop Zgnd1

Zgnd2

GND

Figure 4.11 Electromagnetically relevant fields for STP cables The STP channel, first, also has a generator with a source resistance Ri, a receiver with the load resistance RL and a cable impedance for the twisted pair of Zdim, which form the (differential) transmission loop. The key difference to the coaxial cable is that the two wires of the twisted pair are subject to the same electromagnetic field. The coupling therefore is the same on both wires, which means that any common mode interference is eliminated, when the differential signal is recombined at the receiver. The more symmetric/balanced the two twisted wires are, the better. This also affects the generated E- and H-fields. A differential signal is driven by a differential current Idiff, which has opposite directions in the used wires and generates a magnetic field in the opposite direction. In an ideally symmetric setup, the superposition of the H-fields extinguishes them to zero. The same applies to the electrical fields in short distance from the individual wires.

101

102

4 The Electromagnetic Environment in Cars

However, if the setup is not ideal – for example because the two used wires are somewhat different in terms of diameter, length, conductive or insulation material  – the difference between both wires is seen as a common mode signal. The common mode current generates a parasitic common mode transmission loop. The effect of the common mode transmission loop is the same as for the inner wire of a coaxial cable; however, with a significantly smaller amplitude, as it comes from asymmetries only. In addition to the shield attenuation, STP cables have the positive effect of the differential transmission and unbalance attenuation in their favor (see also Figure 5.7). In contrast, the coupling attenuation of coaxial cables is only determined by the shield attenuation. This effect is not only theory but can be seen in real EMC tests. An additional benefit of STP cables is that any (interfering) current induced into the shield, for example by the GND loop, does not affect data signals on the twisted pair wires, as the transmission wires are independent from the shield system.

4.1.4.2 Shield Connection at the Case As described in the previous section, a shield is an effective method to improve the EMC behavior of the wiring harness. This also applies to a case shield, which is an appropriate method to improve the EMC behavior of a complete ECU or sensor. As a rule, the shield of a cable should be connected to the shielded case by a low impedance, 360 degrees connection (see left part of Figure 4.12). This method has two effects: First, when the case itself is ­either directly connected to a ground plate or by a low impedance ground connection, the ground loop is almost eliminated as the impedance is close to zero Ohm. Second, the 360-degree shielding enclosure has no gaps where RF radiation can pass into the case or exit from the case. Case

Case EMC radiation

EMC radiation ConnecConnector tor

IC

EMC radiation

EMC radiation

EMC radiation EMC radiation IC

Connector EMC radiation

Figure 4.12 Shielded connectors When communication systems use frequencies in the GHz range, mechanical dimensions – for example of length of PCB traces or cables – may cause resonances. Mechanical geometries are also relevant for gaps around connectors, when the connector is not an integral part of the case. Reasons might be the mechanical design or the fact that no direct connection of the cable shield to the case ground is possible. In this situation special care is needed to avoid that high frequent electromagnetic radiation enters or leaves the case. One possibility to block the direct entry or exit of high frequency, EMC relevant radiation is to use a meander structure as shown in the right part of Figure 4.12.

4.1 ElectroMagnetic Compatibility (EMC)

103

4.1.4.3 Interrelation between Shield and Ground In Figure 4.10 for coaxial and Figure 4.11 for STP cables ground loops are depicted. One method to eliminate the effect of a ground loop is to ensure a low impedance between the shield and the GND at both ends. When having Zgnd1 and Zgnd2 at a low value and the resistance of the shield itself at low value, the ground loop has a small effect only. However, in contrast to general expectation, the steel body of a car is nowhere near to being an ideal ground. The steel body is a combination of separate elements. Some might be welded together, others are mounted by screws, yet other parts – like doors, the hood, or trunk lid – are connected as moveable parts using hinges. In consequence, the steel body has a non-negligible resistance. Figure 4.13 shows the effect by considering two connected GND areas, whose connection results in a non-negligible resistance RGND and a resulting ground shift voltage of UGND_shift . I2 Fuse Z shield ECU1

ECU2

Coaxial cable

RLoad

Ri S1 Battery

R1 GND1

I2

Z GND1

Z shield

I2

CCoupling

I2

Z GND2

R GND I2 UGND_shift

GND2

Figure 4.13 Potential risk of GND loops in case of coaxial cables In case an ECU2 on GND2 has a high electrical load, the high load current I2 flows from the battery to the ECU2 and returns not only via GND2 and the ground resistance RGND to GND1 (I2G in Figure 4.13) but also via the shield of the coaxial cable (I2S in Figure 4.13). The lower the impedance coupling of ZGND1/2 of each GND area to the respective end of the shield in relation to RGND, the higher the interfering current I2S. To avoid this unwanted effect of the ground shift, a capacitor – Ccoupling instead of the dotted line in Figure 4.13 – is used to prevent the impact of these common, low frequency ground shifts caused by, for example, actuators. Car manufacturers normally define a GND shift that electronic systems need to be able to cope with. At BMW, for example, a ground offset of +/–1 V is defined [28]. Until quite recently, only a few coaxial and almost no shielded cables were used in cars. With more telecommunication services and more high-speed data communication the number of shielded cables used in cars is increasing. With the knowledge on GND loops and GND shifts, a detailed planning of the ground connections of shielded systems is essential. This means, it needs to be decided where a DC coupled ground connection can be used or where stronger means are needed and where capacitors have to interrupt a potential DC or low frequency current.

S2

R2

104

4 The Electromagnetic Environment in Cars

Learning about EMC During my, Michael Kaindl’s, education at the University of Applied Sciences the lectures and exams addressing high frequency technologies where not particularly popular with the students. For me this was no different. After passing the respective exams, I was looking forward to a future in which I would not have to deal with the subject again. I could not have been more wrong. The absence of high frequency topics from my engineering life lasted only for a very short time; just about into my diploma thesis, which I wrote at the university on a system finding the geographical position of RF sources in the 100 MHz range. Since I started working in the automotive industry, EMC tests accompanied my tasks almost from the beginning. I learned fast that EMC in theory – the electrical and physical principles and respective design rules – is one thing, while real implementations can be something quite different. I learned, by doing as well as trial and error, that the solution to some problems would be the complete opposite, of what I originally expected when applying the theory. Looking at the door of the BMW EMC chamber (see Figure 4.14), I am not the only one thinking this way. For me, the most important success factor in EMC is – besides knowing the theory – being curious about the effects and to proof and rebut explanations until the source of the effects is identified and understood as well as it is possible for the particular case. Building radios myself was a good training, even if the receiver I had intended to use in my thesis at uni­ versity, came to life only 25 years later. I would also like to thank Fritz Stadler, a former EMC technician working with me at BMW, for the many discussions we had on the topic. The list of recommended EMC books ([6] [7] [8] [9] [10]) is a result of the exchange with him.

Figure 4.14 Door at the EMC chamber at BMW Munich in 2021 (Photo: Michael Kaindl) 

4.2 ElectroStatic Discharge (ESD)

„ 4.2 ElectroStatic Discharge (ESD) ESD occurs when, following the build-up of static electricity, a sudden flow of electricity occurs between two differently charged objects coming sufficiently close. The main effect responsible for the build-up of static electricity is tribology, a word originating from the Greek τρίβειν (tribein = to rub) and λόγος (lógos = knowledge of), which means science of friction, mainly in the context of mechanical engineering [29]. What happens is that by friction and/or movement of specific materials, a separation of electrons happens at their surface. The result is that the normal equilibrium of electrons is disturbed, which causes a situation in which electrons are exchanged at the next possible chance to reestablish a (­better) equilibrium. The triboelectric series lists different materials showing the polarity of charges generated with the triboelectric effect. Figure 4.15 shows an extract of the lists [30]. The distance from the neutral position defines the strength of the effect; the further away, the stronger. + Postive charge

Neutral

Hair Nylon Glass Cat fur

Wool Steel

Negative charge Silicon rubber Teflon (PTFE) PVC Polypropylene

Figure 4.15 Polarity of charges generated with the triboelectric effect [30] With this basic knowledge, the common sources generating electrostatic charges are easy to understand: each movement of a person on a surface, rolling wheels and bars, the ­handling of devices, packing and unpacking, or tearing off adhesive tapes holds the risk of generating electrostatic charges. Most people have experienced this in their everyday lives, for example, when touching a door handle after walking on specific synthetic carpet with rubber soled shoes. Electricity always seeks the way of least resistance. All insulating materials have a specific dielectric strength, which defines the capability to maintain the insulation without an electrical short. For air the dielectric strength is about 30 kV/cm [31]. This means that when a spark is seen at about 1 cm, the discharge was at least 30 kV. For silicon dioxide, the most common insulation material in semiconductors, the dielectric strength is in the range of about 8 to 10 MV/cm [32]. This seems to be a very high value. However, active structures of semiconductors are in a sub-micron area. The ­latest announcements for automotive semiconductor technologies are below 10 nm for high end microcontrollers and integrated circuits. With this, the effective dielectric strength of semiconductors is in the area of a few volts. If we put this value into relation with an ESD of 30 kV, it can be quite a challenge for semiconductor devices and their insulation structures to sustain such high voltages intact. Generally in semiconductors, an ESD in form of an avalanche of electrons thus causes an electrical breakdown of the semiconductors. This is temporal and reversible, provided the semiconductor does not overheat at the same time. The absence of further thermal effects is thus essential. If, for example, an ESD in a very small semiconductor generates too much heat in a certain chip area, the thermal breakdown then leads to irreversible shorts in the

105

106

4 The Electromagnetic Environment in Cars

circuitry and destroys the chip for good. If the ESD discharge faces a large enough chip area, the electrical breakdown will be reversible without thermal damage of the respective silicon structure. Following this introduction that helps to understand the effects and the risks of ESD for modern semiconductors in general, the following subsections will detail all relevant aspects for the automotive use cases discussed in this book. Section 4.2.1 and Section 4.2.2 first distinguish between handling ESD for unpowered and ESD for powered devices. Section 4.2.3 then discusses ESD protection devices.

4.2.1 Unpowered ESD Unpowered ESD means that ESD happens to an IC or device, which is in a non-operational, unpowered state. This type of ESD is in the automotive industry particularly likely during manufacturing, transportation, and handling of components and ECUs. Manufacturing comprises next to car manufacturing, the manufacturing of ECUs and ICs. The handling comprises packing and unpacking, mounting, and assembly of devices to a system. Un­ powered ESD handling thus covers two aspects: Exclude/limit the risk of ESD during the different process steps and test the ESD protection of the devices itself.

4.2.1.1 ESD Protection in the Production Process The DIN EN 61340-5-1 is a norm for handling ESD sensitive devices during manufacturing, transportation, and assembly. The focus is on avoiding electrostatic charging of the personnel involved with a guided process guide to prevent ESD events. The specification covers a broad spectrum of handling issues with respect to ESD sensitive devices. Examples are ­requirements for using machines, for floor covers for workspaces, for handling humidity, for the type of packaging material to use for transportation, for clothes and shoes of operators, and a control process to ensure the measures are applied correctly. Table 4.3 provides some examples for electrostatic charge levels of potentially workspace related activities. Table 4.3 Examples for electrostatic charge levels [33] Type of activity

10–25% relative air ­humidity

64–80% relative air ­humidity

Walking over a carpet

35 kV

1500 V

Walking over a vinyl floor

12 kV

 250 V

Taking off a pullover

22 kV

1000 V

Standing-up from chair

18 kV

1500 V

Tearing paper out of plastic foil

 8 kV

 600 V

Working at a table

 6 kV

 150 V

To enable an effective ESD protection process, consultant services are available as well as a broad range of respective products. Examples of such products are conductive ESD wristband, conductive clothes and conductive shoes, conductive floor covers and ESD mats, conductive tools, conductive card boxes and plastic trays, conductive furniture and workplaces,

4.2 ElectroStatic Discharge (ESD)

107

air humidifier and air ionizers, ESD tester. Examples of what should be avoided are non-conductive insulation materials, adhesive tapes, and metal tools. Metal tools are a problem, because metal is a low ohmic conductive material. In case of ESD discharge, the low resistance of the metal might lead a high current into an ESD sensitive device. Conductive in case of ESD protection means an optimal resistance value in the range of kΩs for a defined discharge of the electrostatic voltage without excessive current. In a laboratory, care should be taken when ESD specific tools are used beside normal tools or specific tools for high voltages. The ESD tools are intentionally not insulated and should therefore not be used with high voltages.

4.2.1.2 ESD Protection Tests In a general form, the tests for ESD protection are specified in IEC 61000-4-2 [34]. The norm ISO 10605 [35] is derived from IEC 61000-4-2 and details the automotive requirements for components and systems in case of an ESD of a human person. The test voltages are significant. Table 4.4 provides an extract from the test voltages specified in the two norms. As can be seen, ISO 10605 details the test voltages not only depending on the type of discharge, but there are also different categories within each group. Car manufacturers may again ­define different, specific test voltages and severity levels. Table 4.4 Example for ESD test voltages [34] [35] IEC 61000-4-2

ISO 10605

Contact Air Contact discharge discharge ­discharge Severity level

Test ­voltage

Test ­voltage

Test ­voltage Cat 1

Test ­voltage Cat 2

Air discharge (direct) Test ­voltage Cat 3

Test ­voltage Cat 1

Test ­voltage Cat 2

Test ­voltage Cat 3

1

±2 kV

 ±2 kV

±2 kV

±2 kV

 ±4 kV

 ±2 kV

 ±4 kV

 ±6 kV

2

±4 kV

 ±4 kV

±4 kV

±4 kV

 ±6 kV

 ±4 kV

 ±6 kV

 ±8 kV

3

±6 kV

 ±8 kV

±6 kV

±8 kV

 ±8 kV

 ±8 kV

 ±8 kV

±15 kV

4

±8 kV

±15 kV

±8 kV

±8 kV

±15 kV

±15 kV

±15 kV

±25 kV

Depending on the mechanism that causes the ESD, different test scenarios are defined. The following description details the tests for the Machine Model (MM), the Charged Device Model (CDM), the Human Body Model (HBM), and the Cable Discharge Event (CDE). ƒ Machine Model (MM) The MM reflects an ESD event caused when machines or other mechanical equipment handles chips and/or assembles PCBs during the manufacturing process. For example, moving or rotating machine parts can generate electrostatic energy, which might discharge during the manufacturing process. The basic test setup and the wave form of an MM ESD event are shown in Figure 4.16. As there is almost no resistive element in the discharge pass – only the resistive part of the inductor – the current peak can be up to 10-times higher than in the HBM (even though the applied voltage is significantly smaller). The oscillation is in the range of 20 to 100 MHz. If a chip cannot sustain the MM, the chip is not suitable for mass production. The MM therefore applies very early in a production process.

4 The Electromagnetic Environment in Cars

R Charge = 1 MΩ

HighVoltage Supply

Charge

L Discharge = 0.5 µH

Discharge

Current (A)

108

DUT

C 200pF

3 0

0

1000 ns

Figure 4.16 MM test setup [36] ƒ Charged Device Model (CDM) The CDM reflects an ESD event caused by connecting charged components, like ECUs, during the assembly process. During the transport of ECUs to the assembly line, the devices might get charged, each with a different electric voltage level. During the ­assembly, either when an ECU electrically connects to the metal car body, or when the wiring harness is attached, different charge levels equalize and an ESD event can ­happen. This means that in the first production step of a car, each component has to individually sustain the MM at the suppliers and then, in the next step, the CDM at the car manufacturer. The basic test setup and the wave form of a CDM ESD event are shown in Figure 4.17. Discharge

I

RCharge = 100 MΩ

DUT

Charging plate

Highvoltage supply

RDischarge =

0

1Ω

0

5

ns

Figure 4.17 CDM test setup [36] ƒ Human Body Model (HBM) The HBM reflects an ESD event caused by an electrostatically charged person that handles or touches a chip or device. This can happen anytime during a production process. Tests against the HBM are a necessary requirement for automotive ECUs and sensors, as these devices are handled by humans during manufacturing, transport, car assembly, as well as in service and maintenance. In some cases, even the customers might touch ECUs and sensors. The basic test setup and the wave form of an HBM ESD event as described in IEC 61000-4-2 is shown in Figure 4.18. ISO 10605 uses the same basic setup with some modifications.

4.2 ElectroStatic Discharge (ESD)

Highvoltage supply

Charge

RDischarge = 1.5 kΩ

Discharge

C 100pF

DUT

Current (A)

RCharge = 1 MΩ

1

0

0

200

400

ns

Figure 4.18 HBM test setup as defined in IEC 61000-4-2 [36] ƒ Cable Discharge Event (CDE) The CDE is like a charged device ESD with some minor differences. The CDE is not referenced as a test model and is specific for cables. Cables that are not terminated might charge in the same manner as devices during the assembly process, for example when unreeling from the delivered cable reel and rubbing against a surface. Because cables have a high-quality insulation, the charge can stay on the metallic conductor of the ­cable for a long time. Voltages seen at a cable discharge event are in the range of about 20 V. However, the cable capacitance is proportional to the cable length and can be quite high, in the range of several hundred pF [37]. While CDE is an effect occurring mainly during the cable installation for communication networks in buildings, it should also be considered for Automotive SerDes and Ethernet. For communication without power over data line, the signal bus is decoupled from the chip by capacitors, for both STP and coaxial cables. That means that the conductors in the cable are decoupled from any ground and might charge up to almost any voltage level. For SerDes, Ethernet or other bus systems with differential, galvanically insulated bus ­signals, it is therefore good practice to have a resistor to GND in the range of several hundred kΩs, to prevent an unwanted charging and discharging. CDE should also be kept in mind during tests with a TDR. In order to serve its purpose, the high-performance TDR needs to create and sample fast pulses in the picosecond range. The input ports of the test equipment therefore cannot have any ESD protection and the maximum voltage at these ports should not exceed the maximum level. In case of a TDR with a 20 ps rise time, the maximum voltage is, even for the shortest pulses, only 3 V. Exceeding this small voltage destroys the input stage of the test equipment.

4.2.1.3 Transmission Line Pulse Measurement (TLP) TLP is a test with goal to determine the current characteristic and timing of ESD events defined in ANSI/ESD STM5.5.1 [38]. The main components of the TLP setup are a transmission line of length L, a switch, and a high voltage power supply. The main benefit of TLP is the defined coupling between the setup and the DUT, which ensures that the test results can be reproduced with only minor deviations. TLP uses rise times of 1 ns and a pulse width of 100 ns. The pulse width is determined by the length L of the transmission line TL1 (see also Figure 4.19). With suitable test e­ quipment, for example a fast oscilloscope and suitable voltage and current probes, the main benefit of TLP is the possibility to observe the dynamic behavior during an ESD event [39].

109

110

4 The Electromagnetic Environment in Cars

L

TL2 , Z0

TL1, Z0

Discharge

HighVoltage Supply

DUT

Figure 4.19 Setup of a TLP measurement to observe the dynamic behavior of an ESD [39] TLP is currently not used as a test method in the automotive industry. It is nevertheless explained, because it is a well-defined method with detailed pulse shapes and the authors expect that it will be important in the automotive industry in the future.

4.2.2 Powered ESD In the previous Section 4.2.1, ESD was discussed for ICs and devices, when not connected to a power supply. When powered the effects discussed in that section are still valid and there is no fundamental difference between an ESD event for a powered or unpowered ­device. Also in a powered stage, an ESD event stresses the electronic elements connected, potentially beyond the input levels supported. When testing ESD for powered devices, the goal though is not only to ensure that units are sufficiently protected against destruction, but also to test against temporary mal functions. The rationale is the following: With decreasing silicon structures the internal voltages within the chips decrease at almost the same ratio. Today, in HS communication chips (which means that this is a comparably new phenomena to be considered), the internal logic structures are designed for high speed and operate with voltages below 1 V, with the internal thresholds at about half of the operation voltage. In case of a powered ESD event, the ESD protection circuit of the device causes, for a short time, a high current to flow in the circuit towards ground. The resulting current can be as high as 30 A. If the ground path has a resistance of 20 mΩ, the voltage drop to GND will be 0.6 V, which is above the threshold of the logic. An ESD event, in this case, does not destroy the input structure, but causes a potential status change of the logic. As many ECUs are permanently powered, a change in the logic can have a permanent and undesired impact on the intended f­ unctionality. Figure 4.20 shows a test setup for such a scenario. It tests the effects of ESD on a cable ­bundle with ongoing communication. During the test the DUT is powered, and a load box contains the communication partner. The test levels are adapted from human contact discharge as defined in ISO 10605 [40]. The functional status is checked with error flags. The setup is derived as a model for wiring along the A-columns of car (A-columns are the car body parts to the left and right of the windshield). The A-columns hold the cables that ­connect the ­devices in the car roof. The discharge may occur when an electrostatically charged driver touches the A-column from the inside and discharges against the chassis.

Metal plate Load box

Battery

4.2 ElectroStatic Discharge (ESD)

Insulation DUT

Wiring harness

Copper band

Discharge position

Discharge position

Discharge position

ESD test generator

GND

Figure 4.20 ESD test setup for powered ESD [35] Figure 4.21 shows ESD affecting the shield of an STP cable. In the example depicted, the bus lines are protected by an ESD protection device. The shield is connected to the ground of the PCB at the front-end. An ESD event affecting the shield will not trigger the ESD protection diode and it will not destroy the input structure. However, in such a case, a voltage drop is created and a discharge current traverses the ECU.

Power supply

µC

Vdrop Idiscarge

PHY Vthres

Cap

ESD

R

Figure 4.21 Potential ESD discharge path in case of an unfavorable ground connection of the ECU The effects can be reduced, when the ground of the device is close to the connection of the shield (see Figure 4.22). The same applies to the ground connection of the ESD protection circuit. When this is done, the voltage drop is smaller and does not affect the logic of the communication PHYs and other electronics.

111

112

4 The Electromagnetic Environment in Cars

Power supply

µC

PHY R

Cap

ESD Vdrop

Idiscarge

Figure 4.22 ESD discharge path in case of a recommended ECU GND location To fully test such effects of powered ESD, a test of all logic states and the complete function would be necessary. This cannot be done easily within a reasonable time. Instead, the functionality of the respective applications is monitored for potential faults or error states, during and after the test. It is significantly easier to tests the effect of an ESD on I/O characteristics than to test impacts on the internal logic structure of an IC.

4.2.3 How to Achieve ESD Protection The main goal of ESD protection for communication lines is to actively limit the resulting voltage at the pins of ICs or other components, to a value that is not critical for them. There are several ESD protection technologies available. ƒ ESD or Transient Voltage Suppression (TVS) Diodes These devices use the characteristics of diodes in the reverse direction. In the forward direction, the current increases with a low voltage drop, after the voltage reaches a specific level between 0.3 and 1 V. In the reverse direction, the diode blocks a current to the very low leakage current of a few micro ampere until the reverse breakdown. Then the current increases, too. The exact value of the reverse breakdown voltage ­depends on the geometry of a diode and the semiconductor technology used to produce it. The range of breakdown voltages is from a few volts up to several hundred volts. When in reverse direction the current or time is limited to short pulses there is no thermal, meaning permanent, damage and the effect is reversible. Figure 4.23 shows the characteristics of two typical uses of ESD diodes, a u ­ nidirectional and a bidirectional application. Depending on the needed protection – against positive, negative, or oscillating discharges  – and the signal form on a communication link, other combinations of diodes might be used. Additionally, there are diodes with so-called snapback characteristics. After a certain trigger value is reached, the resistance reduces in high steps and every time the diode characteristic is shifted to different values, which allows for a more efficient ESD protection. In Figure 4.23, this characteristic is shown with the dashed line on the uni­ directional side.

4.2 ElectroStatic Discharge (ESD)

113

Bidirectional

Unidirectional I Vbus

I

0V

+Vbus

Signal

0V

-Vbus

Signal

Headroom VC

VRB VM

-V

-VReverseBreakdown

IT

IR VFW

V+

-V

+VReverseBreakdown

IPP -I

-I

Better protection

Figure 4.23 ESD protection with uni- or bidirectional diodes [41] ƒ Varistors/polymer ESD protection element. Varistors are electronic components, which change their resistance depending on the applied voltage. For ESD protection these elements are based on polycrystalline ­semiconductors, mainly zinc oxide, silicon carbide, or silicon dioxide. Varistors have a bidirectional characteristic. Due to the polycrystalline structure, the in-current flows are distributed over multiple semiconducting junctions. Varistors are inexpensive and robust ESD elements. Table 4.5 shows the main parameters for the selection of ESD protection devices. Table 4.5 Relevant parameters of ESD protection devices (see also Figure 4.23) Abbreviation Meaning VC

Clamping voltage @Ipp, nominal protection voltage

IC

Pulse current

VM

Maximum voltage for normal operation

IR

Leakage current @VM

VRB

Reverse breakdown voltage

IT

Test current

VFW

Forward voltage

CESD

Capacitance of ESD protection component

The most important parameter for selecting the ESD protection is the clamping voltage VC, which can have a value between 3 to 100 V, depending on the use case. The second most important parameter in case of multi-gigabit data communication is the capacitance CESD of

V+

114

4 The Electromagnetic Environment in Cars

the ESD component. It needs to be in the range of 1 pF or below, as it otherwise has an impact on the transmission behavior of the communication technology. Most ICs provide an internal structure at their input pins that functions as ESD protection. The structure consists of two diodes with defined breakdown characteristics (see Figure 4.23). The diode towards GND is, in most silicon process technologies, inherent in the bulk substrate form of the IC and is always present at each I/O pin. Due to the size of the ­substrate, the forward voltage is low, generally lower than for any external diode. The diode towards the I/O voltage is included specifically for ESD protection. With these two diodes the requirements of a MM can be fulfilled, which allows for a standardized, mass production handling of these ICs during the PCB assembly process. For the HBM and CBM it depends. For communication technologies like CAN and FlexRay mixed mode bipolar-CMOS processes are used for the transceiver ICs. With these processes sufficient internal ESD protection can be achieved for automotive use cases without additional external ESD protection devices. For automotive high speed communication chips, semiconductor technologies below 65 nm are used. With such processes, a sufficient chip internal ESD protection for CBM and HBM cannot be sensibly realized, as such a protection would require a larger silicon size. In the case this is not possible, external ESD protection is needed in order to fulfill the automotive requirements. The external ESD protection works in conjunction with the internal ESD protection. As a basic rule, the external protection should be placed close to the ECU pin, as there the ESD event is most likely. For an effective and economic ESD protection the complete electronic circuit of the ECU or sensor should be analyzed, as other passive components also help to reduce the impact of an ESD event (see, for example, the gray elements in Figure 4.24). U+ ESD protection with external diodes

UI/O ESD protection with internal diodes Integrated circuit

Logic

Figure 4.24 ESD protection with IC internal and IC external diodes Most test specs for ESD protection require only a low number of tests. Sometimes, as few as ten discharges are part of the ESD test. Please be aware of two effects: First, it is impossible to say how often an ESD happened to an unpowered unit before it is put into operation for its intended purpose. In case of a perfect ESD handling process, no ESD might have hap-

4.2 ElectroStatic Discharge (ESD)

pened. In a less perfect world, hundreds of ESDs might have happened. When the unit had sufficient ESD protection, it would still work without any noticeable impairments. Second, as stated before, the ESD is, when protected, in principle, a reversible process. At first glance, this is true. At a second glance and closer look, an ESD has almost always some irreversible impact on the involved components. After several ESD events the parameters of the ESD protection parts deteriorate slowly, without this being visible. ESD testing should reflect this, for example by testing more often or by testing until the destruction of the part, in order to understand the head room. Note that there is an interrelation between ESD and EMC and within the EMC between EME and EMI. What is good to improve the EME is generally bad for the EMI and vice versa. ESD is now an additional, adverse aspect (visualized in Figure 4.25) which needs to be considered. Out of these, ESD protection must be ensured first, because it is a fundamental requirement, without which EME and EMI are of limited value. At the same time, the ESD protection may affect EME and EMI. The EME may change, when ESD protection components add imbalance to the communication links. This might be, because of the parasitic parameters of the ESD components or because they add echoes and impedance steps at discontinuities. Potential places for these discontinuities are the pads for the pins of the ESD protection component on the PCB. The additional capacitance may add to the frequency dependent attenuation, which decreases the signal strength and thus impact the immunity. EMC emission

ESD protection

EMC signal integrity

Figure 4.25  Conflict between ESD, EME (emissions) and EMI (signal integrity)

The signal integrity may additionally be impacted when a voltage is coupled into the bus line by electromagnetic noise. BCI tests, for example, may induce such noise into the ­communication link. Most ESD components for commercial applications have their clamping voltages below 10 V. BCI tests may induce up to 100 V. A voltage, which is too high, may cause ESD components to get activated and be conductive. ESD components in cars must be able to handle several kV and peak currents of several Amperes. This is way beyond the induced 100 V and respective currents of tens of mA. However, the ESD components can handle very large voltages only for a very short period. During BCI tests, the test runs for seconds or even minutes. The overall power dissipation in case of a BCI test is therefore much higher than for an ESD event. BCI or other signal integrity tests can therefore destroy the ESD protection and related circuitry of an ECU or sensor.

115

116

4 The Electromagnetic Environment in Cars

For comprehensive testing, the EMC tests should therefore be applied before and after the ESD test of an ECU or sensor, in order to ensure that the ESD components do not change the EMC in an unacceptable manner after ESD events. For further reading, we recommend the following references [36] [41] [42]. Personal Experience with ESD My, Michael Kaindl’s, first experience with ESD was during the work on my diploma theses. It was a CD4027 that was directly and obviously damaged by ESD. In my professional work, an ESD damage occurred during the series production of a sunroof electronics I had designed, and was responsible for. The ECU was one of the first examples of mechatronic. The electronics was integrated in the actuator, a DC-motor. The complete mechanical frame with the mounted actuator was delivered to the plant. After months of production without any errors, a sudden increase of the same ESD-based error was reported. The nastiest effect of this error was that all functions were present, except that the sunroof ignored the go-to-sleep command, which kept the car active, in the worst case until the battery was drained. In a worst case ranking of errors this is right at the top. In the search for the root-cause of the ESD event, the complete process chain over the five involved process partners including the BMW plant was tracked. At the last station, the mounting of the frame into the car the source of the ESD damage was found: From the frame of about two square meters in size, with its interior textile finish, the plastic foil from the transport protection was removed in a single, fast move. This was not a problem until the winter period, with dry air. The removing of the foil charged the frame to above 50 kV. When connecting the electronics, the ESD happened via the communication bus and in some cases the bus IC was partially destroyed to the effect that the operation and quiescent currents were increased. Now, the sunroof ECU had a “safety function” that reacted on a certain current level. If that current level was exceeded in sleep mode, a reset was initiated, which caused a wake-up of the car. With the ESD-damaged units, the sleep currents did exceed this limit, with the mentioned result of drained batteries. The corrective action was to change the foil used for the transportation of the frames by paper. This was a classic form of a CDM ESD event, caused by different circumstances and requirements along the process chain for automotive manufacturing. When investigating the mounting station and the container with the disposed foil, it was observed that the foils were charged so much, that it was possible to drag them out of the container with the ­fingertips, without actually gripping them. 

„ 4.3 Bibliography [1]

D. E. Möhr, “Was ist eigentlich EMV? – Eine Definition,” not known. [Online]. Available: http:// www.emtest.de/de/what_is/emv-emc-basics.php (no longer available). [Accessed 6 May 2020].

[2]

Wikipedia, “CISPR,” 21 April 2022. [Online]. Available: https://en.wikipedia.org/wiki/CISPR. [Accessed 10 June 2022].

4.3 Bibliography

 [3]

Federal Communications Commission, “About the FCC,” Continuously updated. [Online]. Available: https://www.fcc.gov/about/overview. [Accessed 9 July 2021].

 [4]

Der Spiegel, “Absolut sicher,” 22 July 1984. [Online]. Available: https://www.spiegel.de/politik/ absolut-sicher-a-7d6913b5-0002-0001-0000-000013509655?context=issue. [Accessed 2 July 2021].

 [5]

Der Spiegel, “An die Nieren,” 10 August 1986. [Online]. Available: https://www.spiegel.de/politik/ an-die-nieren-a-9c3771bd-0002-0001-0000-000013518844?context=issue. [Accessed 2 July 2021].

 [6]

T. Williams, EMC for Product Designes, Oxford: Newnes, 2016.

  [7]

J. J. Goedbloed, Electromagnetic Compatibility, Hoboken, NJ: Prentice-Hall, 1993.

 [8]

R. Schmitt, Electromagnetics Explained, Amsterdam: Newnes, 2002.

 [9]

A. Kohling, EMV von Gebäuden, Anlagen und Geräten, Berlin: VDE-Verlag, 1998.

[10]

A. Schwab and W. Kürner, Elektromagnetische Verträglichkeit, 6. Auflage, Berlin: Springer, 2011.

[11]

D. Alexander, “9 Reasons Composite Materials Are Used Just About Everywhere Including Your Car,” 12 February 2019. [Online]. Available: https://interestingengineering.com/9-reasons-compos ite-materials-are-used-just-about-everywhere-including-your-car. [Accessed 5 August 2021].

[12]

Ford Motor Company, “Component and Subsystem Electromagnetic Compatibility Worldwide Requirements and Test Procedures,” 10 October 2003. [Online]. Available: http://www.jaguar landrover.com/emc/docs/download/ES-XW7T-1A278-AC.pdf. [Accessed 3 July 2021].

[13]

BMW AG, “BMW Group Standard 95002,” October 2004. [Online]. Available: https://www.yumpu. com/de/document/read/9890334/bmw-group-standard-gs-95002-elektromagnetische-xenona.  [Accessed 3 July 2021].

[14]

Williamson Labs, “EMC,” 1999-2011. [Online]. Available: http://www.williamson-labs.com/ltoc/ glencoe-emc-11.htm (no longer available). [Accessed 26 December 2013].

[15]

UN/ECE, “Regulation No 10 of the Economic Commission for Europe of the United Nations (UN/ ECE) — Uniform Provisions Concerning the Approval of Vehicles with Regard to Electromagnetic Compatibility,” 26 July 2012. [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/ HTML/?uri=CELEX:42012X0920(01)&from=DE. [Accessed 2 July 2021].

[16]

K. Lamedschwandner, “EMV in der KFZ-Technik,” 15 November 2013. [Online]. Available: https:// docplayer.org/7364551-Emv-in-der-kfz-technik.html. [Accessed 6 August 2021].

[17]

T. Steinecke, M. Bischoff, F. Brandl, C. Hermann, F. Klotz, F. Mueller, W. Pfaff and M. Unger, “Generic IC EMC Test Specification,” in Asia-Pacific Symposium on Electromagnetic Compatibility, Singapore, 2012.

[18]

ISO, “ISO 17987-6:2016: Road Vehicles - Local Interconnect Network (LIN) – Part 6: Protocol Conformance Test Specification,” ISO, Geneva, 2016.

[19]

IEC, “IEC 62228-3 Integrated Circuits – EMC Evaluation of Transceivers – Part 3: CAN Trans­ ceivers,” IEC, Geneva, 2019.

[20]

B. Körber, “IEEE 100BASE-T1 EMC Test Specification for Transceivers v1.0,” OPEN Alliance, ­Delaware, 2017.

[21]

B. Körber, “IEEE 1000BASE-T1 EMC Test Specification for Transceivers v1.0,” OPEN Alliance, Delaware, 2017.

[22]

ISO, “ISO 11452: Road Vehicles  – Component Test Methods for Electrical Disturbances from ­Narrowband Radiated Electromagnetic Energy Part 1 to 11,” ISO, Geneva, 2002–2015.

[23]

ISO, “ISO 7637: Road Vehicles – Electrical Disturbances from Conduction and Coupling – Part 3: Electrical Transient Transmission by Capacitive and Inductive Coupling via Lines other than Supply Lines,” ISO, Geneva, 2016.

117

118

4 The Electromagnetic Environment in Cars

[24]

A. Grzemba, MOST, the Automotive Multimedia Network; from MOST 25 to MOST 150, Poing: Franzis Verlag GmbH, 2011.

[25]

Wikipedia, “Pulse-width Modulation,” 3 August 2021. [Online]. Available: https://en.wikipedia. org/wiki/Pulse-width_modulation. [Accessed 12 August 2021].

[26]

Veoneer, Bosch, and Continental, “Overview,” not known. [Online]. Available: https://www.psi5. org/overview. [Accessed 12 August 2021].

[27]

Wikipedia, “SENT (Protocol),” 3 July 2021. [Online]. Available: https://en.wikipedia.org/wiki/ SENT_(protocol). [Accessed 12 August 2021].

[28]

BMW AG, “GS 95024-2-1 Electric and Electronic Components in Motor Vehicles, Electrical Requirements and Testing,” January 2010. [Online]. Available: https://www.doc88.com/p­ 819912711273.html. [Accessed 9 August 2021].

[29]

Wikipedia, “Tribology,” 7 April 2020. [Online]. Available: https://en.wikipedia.org/wiki/Tribo logy. [Accessed 10 May 2021].

[30]

Wikipedia, “Triboelectric Effect/Triboelectric Series,” 4 May 2021. [Online]. Available: https:// en.wikipedia.org/wiki/Triboelectric_effect#Triboelectric_series. [Accessed 10 May 2021].

[31]

Wikipedia, “Dielectric Strength,” 28 January 2021. [Online]. Available: https://en.wikipedia.org/ wiki/Dielectric_strength. [Accessed 10 May 2021].

[32]

TU Vienna, “Silicon Dioxide Properties,” not known. [Online]. Available: https://www.iue.tuwien. ac.at/phd/filipovic/node26.html. [Accessed 10 May 2021].

[33]

esd-berater.de, “Änderungen in der Normenreihe für den ESD-Schutz,” esd-berater, 2017. [Online]. Available: https://www.esd-berater.de/esd-infos/aenderungen-in-der-normenreihe-fuer-denesd-schutz. [Accessed 13 May 2021].

[34]

Wikipedia, “IEC 61000-4-2,” 29 March 2021. [Online]. Available: https://en.wikipedia.org/wiki/ IEC_61000-4-2. [Accessed 12 May 2021].

[35]

ISO, “ISO 10605-2008: Road Vehicles – Test Methods for Electrical Disturbances from Electro­ static Discharge,” ISO, Geneva, 2008.

[36]

Toshiba, “Basics of ESD Protection,” not known. [Online]. Available: https://toshiba.semicon-­ storage.com/eu/semiconductor/product/diodes/tvs-diodes-esd-protection-diodes.html#Documents. [Accessed 13 May 2021].

[37]

Texas Instruments, “AN-1511 Cable Discharge Event,” Texas Instruments, April 2013. [Online]. Available: https://www.ti.com/lit/an/snla087a/snla087a.pdf. [Accessed 13 May 2021].

[38]

ANSI, “For Electrostatic Discharge Sensitivity Testing: Transmission Line Pulse (TLP) – Device Level ANSI/ESD STM5.5.1-2016,” ANSI, Rome, NY, 2016.

[39]

ESDEMC.com, “Introduction of Transmisson Line Pulse (TLP) Testing for ESD Analysis – Device Level,” 17 April 2015. [Online]. Available: https://www.esdemc.com/public/docs/TechnicalSlides/ ESDEMC_TS001.pdf. [Accessed 16 May 2021].

[40]

S. Frei and J. Edenhofer, “KFZ-Komponenteprüfverfahren für die Sicherstellung der Störfestig­ keit gegen indirekte ESD,” 2005. [Online]. Available: http://www.bordsysteme.tu-dortmund.de/ publications/2005_ESD_Forum_Kfz-Komponentenpr%C3%BCfverfahren%20f%C3%BCr%20die%20 Sicherstellung%20der.pdf. [Accessed 17 May 2021].

[41]

Nexperia, ESD Application Handbook, self-published: Hamburg, 2018.

[42]

Apogeeweb Semiconductor Electronic, “ESD Protection Circuit Tutorial,” Apogeeweb Semiconductor Electronic, 8 January 2018. [Online]. Available: https://www.apogeeweb.net/article/34. html. [Accessed 13 May 2021].

5

The Automotive Channel

When new communication systems are being designed, its channel is normally selected first. This starts with very basic properties. Is it a copper cable, an optical link, or air? Over what distance does the communication need to function? It is easily imaginable that the communication technology used over ten meters of copper wire is quite different from a wireless communication technology needing to cover several kilometers or one that targets 10 cm distance on a PCB. However, even when the use case is narrowed down to multi Gbps communication over up to 15 m copper cables in cars – as required for the use cases discussed in this book – there are several choices concerning the transmission channel. As is discussed in more detail in the entirety of this chapter, the cable selected  – Unshielded Twisted Pair (UTP), Shielded Twisted Pair (STP), Star Quad (STQ), Shielded Parallel Pair (SPP), or coaxial – represents only one element. To make things even more complicated, these choices interrelate also with the communi­ cation mechanisms of the physical layer. More complex physical layers may allow for somewhat lower quality channels than less complex physical layers, which might require higher quality channels to yield the same overall transmission performance. The ­design of the physical layer is thus tightly coupled to the channel parameters. Communication technology standards always define the communication channel in detail and this – normally – at the beginning of the project. Not only does this provide a clear guideline with respect to the performance of the physical layer (PHY) design. Additionally, there are generally fewer and less granular choices for the channel than for the PHY. Naturally, also the proprietary SerDes solutions have a target channel. However, their parameters are not published, and customers receive information on usable cabling – and not the channel limit lines – as part of the product specification or application notes. The key performance parameter for digital communication systems is the Bit Error Rate (BER). After all, the utmost goal of any communication system is to relay data correctly. The release process of a communication system thus always includes extensive longtime BER tests. These need to confirm that a BER better than, for example, 10E-10 or 10E-12 can be achieved under various noise conditions. A BER of 10E-12 means that a single bit error is received for 10E12 transmitted bits. At 1 Gbps this translates into one bit error every 1000  seconds or 16.6 minutes. At 10 Gbps this means an average of one bit error every 100 seconds or 1.66 minutes. While this seems frequent, a target BER of 10E-12 is a typical value for automotive high-speed communication systems (see, for example, [1]). The BER represents the end result. The key design question is, how can it be achieved? Specific receiver designs yield a certain BER depending on aspects like modulation scheme,

120

5 The Automotive Channel

Nyquist frequency, and Signal-to-Noise Ratio (SNR) at the receiver entrance [2]. For Non-­ Return to Zero (NRZ)/Pulse Amplitude Modulation (PAM) 2, for example, an SNR of 18 dB or better at the receiver achieves a BER better than 10E-10. Depending on the transmission channel and frequency selected, a given value of transmit power determines the SNR at the receiver. So, how exactly does the channel impact the SNR? Commonly known parameters for copper cables are wire gauge, type of insulation used, number of braids, etc. From these the Direct Current (DC) resistance and the thermal behavior can be derived. For cables used in digital communication systems such parameters are only of minor importance though. In the following, the choices and channel parameters relevant for automotive HS communication systems are discussed in detail. As the channel is more than just the cables, Section 5.1 first defines the automotive communication channel as such. Section 5.2 details the impact and parameters relevant for high data rates. Section 5.3 provides insights on different types of cables and connectors in general, while further information on specific channel choices can be found with the respective commu­ nication technologies in Chapters 7 and 8. Section 5.4 addresses the impact some of the choices made for the Printed Circuit Boards (PCB) have on the transmission channel.

„ 5.1 Channel Definition While the cable is the most obvious part of the channel, it is generally not the only part the channel consists of. Figure 5.1 provides an overview of the different channel elements and the respective responsibilities. For the standardized Automotive SerDes technologies, the channel covers everything between Test Point (TP) 1 and TP2, which are located at the pins of the PHY chips in the transmitting and the receiving units. In detail, this means that the channel entails the cable with all end and inline connectors as well as all elements that ­affect the communication path on the PCB between the pins of the PHY chip and the ECU connector. These elements might be filters (for spectral shaping), AC-coupling capacitors (for DC suppression), Bias-T (for setting the DC bias of components without disturbing ­others), ElectroStatic Discharge (ESD) protection diodes, Common Mode Chokes (CMCs, for common mode suppression), etc. These latter, PCB related elements are in the following referred to as the “Media Dependent Interface (MDI) network”, with the MDI itself being where the media changes from PCB to cable connector (marked as reference points “Ref1” and “Ref2” in Figure 5.1). The so-called “(Ethernet) link segment” is identified as the section between the MDIs or “Ref1” and “Ref2” respectively. As testing an unmated connector is not really possible, the reference points Ref1 and Ref2 are, if anything, virtual test points. Figure 5.1 therefore identifies two further test points on either side of the mated ECU connectors: TP5 and TP3 on the side of ECU 1 and TP4 and TP6 on the side of ECU 2. Modern tools allow different methods to mathematically eliminate the difference between the real TP and the virtual reference point by de-embedding, meaning isolating the performance of the connector parts, and extracting it from the measurement [3].

5.1 Channel Definition

Chip vendor Harness supplier / car manufacturer

ECU 1 supplier

TP5

TP1

Ref1, MDI

TP3

ECU 2 supplier

TP4 TP6 Ref2, MDI

TP2

ECU 1

PHY transceiver

ECU 2

Filter ESD PCB

Filter ESD PCB

PHY transceiver

Inline connectors

ECU con. „MDI network“

ECU con. „Link segment“

PCB „MDI network“

Overall (SerDes) channel

Figure 5.1 Elements and responsibilities with respect to the communication channel The effect of the PCB, and the different components placed on the PCB can be simulated or tested from TP1 to TP5 and from TP2 to TP6. The cable within the transmission channel is the portion between TP3 and TP4. This is the part of the wiring harness without the ECU connector. The connector is defined by TP5 to TP3 and by TP4 to TP6. Figure 5.1 also identifies the responsibilities for the different parts of the channel. The semiconductor vendors providing the PHY chips are responsible for achieving the required performance for the complete channel between TP1 and TP2. For the stretch between TP1 and Ref1 or TP2 and Ref2, it is the Tier 1 suppliers that must guarantee that the ECUs, ­sensors, or displays meet the limit lines defined in the standards or application notes. Standards provide the Power Spectral Density (PSD) mask and/or MDI return loss as respective references for the Tier 1s (for details see Section 5.4). The portion of the communication channel between the two reference points Ref1 and Ref2 is the wiring harness. It is the car manufacturers who define the physical location of the units that are to be connected by the harness. It is their responsibility to use the right ­cables, connectors, and that they or their harness suppliers use the right manufacturing processes to meet the limit lines. Only the car manufacturer can make sure that the maximum link lengths the standards support are respected. The average link length for high-speed data transmission inside a car is normally not that dramatic. In an early study on Gigabit communication, it was estimated to be just above 3 m [4]. However, the actual spread of link lengths is large, with long link lengths for sensors that tend to be at the edges of the car and rather short link lengths for displays that might be uncharacteristically close to the unit rendering the display data. Link lengths of 10 m for cameras in limousines and 15 m in mini vans are realistic [5], but more difficult to achieve the higher the targeted data rate (see, for example, [6]). Note though: Very short link lengths

121

122

5 The Automotive Channel

can also pose a problem for high-speed data communication, because of the impact of reflections (see Section 5.2.3.2 for more details on the return loss). It is important to understand that the wiring harness generally represents the most challenging part of the communication channel. When looking at how the different sections of the overall channel impact the insertion loss (see also Section 5.2.3.1 for more details on the insertion loss), Table 5.1 shows that the cable accounts for up to 95% of the loss. This is important to remember. When SerDes technologies were introduced in the IT industry, they needed to cover a channel  10 Gbps Automotive Ethernet standardization. ƒ Inter-pair skew: When a communication system relies on multiple lanes or cables to realize a desired data rate, inter-pair skew may happen. Inter-pair skew is caused by differences in the signal propagation on the different lanes, for example due to small differences in length of the lanes, in the insulation, or in the connection. The effect is that two or more ­signals transmitted from a sender, will arrive at the receiver differently, for example at different times. In high-speed communication systems this may cause a corrupted reception of the ­signals in the receiver, when the effect is not compensated. After all, when assuming a baud rate of 5 Gbps, each symbol is transmitted 0.2 ns. As discussed with the latency in Section 5.2.3.1, a signal travels about 4 cm in 0.2 ns so just a relatively short mismatch of cable lengths might prevent combining the signal in a correct manner, when the effect is not compensated for (see also Figure 1.3 in Section 1.2.1). Example technologies that rely on multiple cables are the first version of the FPD link or 1000BASE-T Ethernet [17]. While with 1000BASE-T Ethernet means are implemented to compensate for a certain amount of inter-pair skew, it was a real issue with the original version of FPD-Link (see also Section 7.3.1).

5.2 Channel Description

5.2.3.3 EMC Related Channel Parameters and Other Noise ƒ Mode Conversion (MC) loss: MC describes the conversion of a transmission signal from single-ended to differential and back. It therefore does not apply to communication links relying on single-ended communication only, for example those using coaxial cables. In a perfectly symmetrical, differential system, common mode interference is cancelled out when the differential signal is re-combined at the receiver. However, perfectly symmetrical systems do not exist, so every differential communication system has to be able to sustain a certain amount of common mode interference that remains in the signal because of asymmetries in the MC. MC is thus often synonymously used with the terms “symmetry” or “balance”. Asymmetries affect the signal integrity, because of remaining common mode interference. The MC loss limit line thus tells PHY designers, with how much remaining CM interference their design has to be able to cope with. The cable manufacturers on the other hand need to provide cables that are symmetric enough to, at least, meet the limit lines. Note, that asymmetries also result in EME, because with asymmetries also the emissions cannot cancel each other out completely. The MC loss limit line thus also limits the EME caused by MC. There are various ways to determine the MC losses based on S-parameters [23]. The Transverse and Longitudinal Conversion Losses (TCL and LCL) are functions of Scd11 /Scd22 and Sdc11 /Sdc22 respectively. They measure the echoes at the near end of the communication. The Transverse and Longitudinal Conversion Transmission Losses (TCTL and LCTL) are functions of Scd12 /Scd21 and Sdc12 /Sdc21 respectively and measure the attenuation at the far end of the communication. LCL and TCL as well as LCTL and TCTL provide the same technical information concerning the symmetry, which is why standards g­ enerally select only either or for the MC loss. Another way of describing the latter effect is intra-pair skew delay (not to be confused with inter-pair skew as discussed in Section 5.2.3.2, where the transmission is parallelized over multiple cable pairs). A non-symmetric implementation of the two wires of the differential cable, of the insulation, or different length of the two wires cause a ­different signal propagation in each single wire. Originally, intra-pair skew delay represented the parameter specified for the release test of cables. However, as the effect is asymmetry/MC loss, testing for an appropriate level of the S-parameters for MC is ­replacing the explicit test for intra-pair skew. The MC is a parameter used especially for limit lines of UTP cables. For describing shielded cables, generally, the parameters ­explained in the following bullet points are used, especially the coupling attenuation. Note that before the automotive industry started with the developments of 100 Mbps Automotive Ethernet, it had paid significantly less attention to the impact of symmetry on data communication. The knowledge gained with developing 100BASE-T1 Ethernet can be seen as a contribution to the industry that can now be applied also to other, even higher data rate communication technologies [24]. ƒ Coupling attenuation, screening/shield and unbalance attenuation: The MC, as discussed with the previous parameter, addresses the symmetry of a differential system as an important property for the EMC behavior. But, as has been shown in Section 5.2.3.1, the higher the data rates/frequencies, the larger the attenuation and

131

132

5 The Automotive Channel

the lower the signal strength at the receiver. At some point, the resulting SNR of an unshielded cable is too small, even with very good symmetry, to combat the also-­ increased EMC noise and susceptibility. In this case, the communication system generally mandates the use of shielded cables, as the shield provides an additional EMC-margin with respect to emissions and immunity (see also Figure 4.9). The coupling attenuation is the parameter that describes the overall effectiveness of a channel to improve the EMC behavior. In its function, the coupling attenuation comprises the effects of the symmetry, also called “unbalance attenuation” (with a value comparable to the MC parameter LCTL, at least up to 2 GHz [24]) as well as the effect of the shield, also called “Screening Attenuation (SA)” [25] [26] (see also Figure 5.7). Note that a shield might actually somewhat reduce the symmetry of an STP cable when ­compared with a UTP cable, because of variations in the distance between cable wires and shield. For coaxial cables the coupling attenuation and the SA are the same, as coaxial cables rely on the shield only to combat electromagnetic interference. UTP cables, in contrast, rely on the balance only. STP or SPP cables can leverage both, symmetry – STP better than SPP – and shield and are thus most robust. While a shield is efficient to improve the EMC behavior, note that this is only the case with an appropriate connection between shield and ground. Because of the use of new compound materials and different voltage supply levels (12 V and 42 V) among others, serious attention needs to be paid to ensure an undisturbed ground connection (see also Section 4.1). IEEE 802.3ch, the MGBASE-T1 specification, lists both the coupling as well as the SA as channel parameters [27]. Coupling attenuation Unbalance attenuation

STP

Screening attenuation

UTP

Figure 5.7  Principle relation between screening, Coaxial ­unbalance, and coupling attenuation [26]

ƒ Alien crosstalk (XTALK): XTALK happens when cables are in close proximity to each other, and their emissions directly couple into these adjacent cables. The root cause of crosstalk is either parasitic resistive (via GND) or inductive or capacitive near-field coupling from neighboring ­transmission lines onto the victim transmission channel. Which of these types it is, depends on the operating frequencies of the involved communication technologies. As a rule of thumb: below 100 MHz, XTALK has only a minor impact. For frequencies from 100 MHz to 1 GHz XTALK increases with a factor of ∼ 10 dB per decade. Above 1 GHz, XTALK increases with 15 to 20-dB per decade. XTALK thus is relevant for the communication technologies discussed in this book and needs careful handling.

5.2 Channel Description

XTALK can be caused by an external/alien or system internal source of interference, depending on the number of wire(pair)s/lanes used for the transmission system. For the severity of the XTALK as such, it cannot be said whether the interference is worse from a wire belonging to the same communication system or from a wire of an alien system. However, for combating the impact, it does make a difference. When transmitter and receiver know what data is being transmitted on the interfering wire, like with 1000BASE-T Ethernet or the original FPD-Link, some of its effects can be compensated for by respective signal processing. In any case, in a car wiring harness, generally many cables are bundled together in proximity. Often, only a few of these will transmit data in a similar frequency range and therefore have the potential to cause crosstalk that interferes with the original signal. However, with the vast number of harness variations in cars, the possibility of alien crosstalk cannot be neglected and needs to be considered in the respective communication standards. XTALK distinguishes between Near-End CrossTalk (NEXT), when the interfering source and affected receiver are at the same end of the transmission channel, and Far-End CrossTalk (FEXT), when interfering source and affected receiver are at opposite ends of the same transmission channel. The impact of FEXT is smaller, due to the attenuation, and sometimes even neglected. For MGBASE-T1 Automotive Ethernet or the ASA ­Motion Link, the standards define the crosstalk limits for the near end with the Power Sum Alien Near-End crossTalk (PSANEXT) and the alien crosstalk ­limits on the far-end with the Power Sum Alien Attenuation to Crosstalk Ratio Far-end (PSAACRF) [27] [28]. Another parameter used might be the Equal Level Far-End ­crossTalk (ELFEXT). The exact details for each communication technology are discussed with the standards in Chapters 7 and 8. ƒ Thermal noise: Thermal noise (nt) is present in every electrical system. It is random and non-deterministic and its power is proportional to the temperature and relevant frequency range. Equation 5.5 shows how the thermal noise level is calculated. Figure 5.8 shows that, for example, at 100 kΩ resistance, the effective thermal noise voltage is about 1.3 mV at 1 GHz and about 4 mV at 10 GHz. (5.5) with k the Bolzmann constant T the temperature R the resistance B the bandwidth of the affected system

133

134

5 The Automotive Channel

Figure 5.8 Thermal noise for different bandwidths and resistance values

5.2.3.4 Transmission Channel Interference Model Figure 5.9 summarizes different types of impacts – inherent in the system or by external sources – on the communication channel in a comprehensive model and identifies the related parameters communication system developers have to take into account. Power level

Immunity limit of others reached

Immunity gap

Maximum transmitter output

Attenuation / IL Minimum receiver input

Transient burst

RF ingress

Attenuation

ECU Echoes/ RL

Link budget / SNR / emissions gap

Interfering noise floor

Noise

Power ripple

ISI

ANEXT AFEXT

ECU NEXT FEXT

Thermal noise

Figure 5.9 Interference model [29] The most important transceiver design criteria is the minimum receiver input level (see left side of Figure 5.9). When a signal is received with a strength below the minimum r­ eceiver input level, the receiver cannot detect a signal at all or only with too many errors. Any ­signal detected above the minimum receiver input level is interpreted as a valid signal. In order to realize a functional communication system, the minimum receiver input level does not only need to be above the natural noise floor. It needs to have sufficient margin to the noise floor (“emissions gap”) in order to be able to handle all the additional internal and

5.2 Channel Description

external interference sources shown on the right side of Figure 5.9. How well a system can handle interference, meaning how large the margin needs to be, is actually a result/part of the system design and generally an important part of the debates and decisions taken during a standardization process. Between the receiver input (of one device) and the transmitter output (of another device) is the communication channel (see Figure 5.1). A key parameter for the channel is its attenuation/IL (see also left side of Figure 5.9), which determines how much of the transmitted signal strength arrives at the receiver. Typically, the attenuation is more than 20 dB for a 10 m cable at 6 GHz (see, for example, Figure 5.4). This means that only 1% of the transmitted power arrives at the receiving station or 10% of the voltage respectively (see Table 5.5), and this needs to be above the minimum receiver input level. A higher transmitter output ensures a higher receiver input. However, the maximum transmitter output must stay below a certain limit (see also the left side of Figure 5.9), else it would create too much interference to other systems and impact their immunity. The maximum transmitter output is therefore another important system parameter. The lower the maximum transmitter output the less interference is generated to other systems and the better it is for their immunity. To allow for additional interference from other sources, also on the transmit side, there needs to be a margin between the maximum transmitter output and limit at which the immunity of other systems is impeded. This is called “immunity gap”. In the right part of Figure 5.9 different types or interfering noises are listed that might ­impact the signal transmission and distort the signal’s amplitude and potentially its timing behavior. At the base of all noises is the thermal noise present at any electrical system. It impacts only the receiver. In the middle are the types of interference the system causes in itself. It is the RL and the ISI. Communication technologies that rely on parallelized data transmission also have to consider the system internal XTALK in form of NEXT and FEXT. The distinct external types of interference are listed at the top right of Figure 5.9. First are transient pulses. The car industry distinguishes between slow and fast transient pulses (see also Section 4.1.3). Slow transient pulses are generated when inductive loads, like relays and valves, in the car are switched off. Loads controlled by switching the power supply on and off, as is the case for PWM [30] controlled loads, generate fast transients. The engine and the power supply of electrical vehicles are one of the main sources of fast transients induced in to all nearby systems connected with wires. Radio Frequency (RF) ingress is another potential source of interference. It is caused by RF sources like terrestrial broadcast stations, whose potential impact was so vividly demonstrated with the example given in the introduction of Section 4.1. All electronics including those in cars have to be able to handle this noise without relevant impairments of the functionality (see also [31]). Furthermore, the ripple at the power supply is source of electromagnetic interference. It is a common effect for all ECUs inside cars and has the following origin: Today, in combustion engine cars, an electric generator provides multiphase AC voltages. These multiple voltages are rectified and superimposed, so that the voltage that feeds the car’s battery or supplies power to different ECUs is almost constant. The battery normally dampens variations and

135

136

5 The Automotive Channel

especially peaks in this rectified AC voltage. It also minimizes voltage drops when load switches occur that would otherwise be caused by the sudden increase of current. However, depending on the age of the battery and the severity of the load switches, a voltage ripple on the car’s power supply remains. The share of the power ripple that originates in the rectification correlates with the revolution speed of the electric generator. Power ripple as such causes a low frequency electromagnetic interference in form of conducted coupling (see also Table 4.1 in Section 4.1.1). Electric vehicles use DC/DC converters instead of electric generators. In these cars, it is the switching characteristic of the DC/DC converter that generates a ripple (sometimes also referred to as “switching noise”). This general description of power ripple is reflected in the requirements for ECUs and the according test specifications defined by the car manufacturers. German carmakers generated a harmonized specification LV 124, which each car manufacturer adopted into their proprietary specifications (for example [32] [33]). Next to the power ripple, the following other effects related to the power supply are considered: Overvoltage, undervoltage, jump start, load dump, super-imposed alternating voltages from the electric generator, short ­interruptions, start pulses. The in-car systems, including their communication, have to be able to cope with all of these, even if not all are relevant for the EMC behavior. For Automotive SerDes, the general form of power ripple is replaced by a variation that occurs in case power is transmitted over the data carrying coaxial cable (Power over Coaxial, PoC). In case of PoC, the power ripple originates in the DC/DC power supply of the unit that supplies the power, called Power Supply/Sourcing Equipment (PSE). Load changes in the unit that receives the power over communication cable, the Powered Device (PD), generate additional power ripples. In a camera, for example, the load change could occur during the blanking period, when no data is transmitted (see also Section 2.1.2). In this case the power ripple as defined in the LV 124 is no longer relevant, because the DC/DC converter in the PSE eliminates the effects in PoC systems. Instead, the particular power ripple of the PoC system needs to be considered. The last external type of interference incorporated in the model is the alien XTALK, at the  near-end ANEXT, or at the far-end AFEXT. The higher the transmission frequency the more relevant interference from XTALK becomes. Counter measures against crosstalk are increasing the distance between the transmission channels or connectors, using a jacket, a shield (see also Section 4.1.4), or twisted pair cables with different lay lengths. Like the other external sources of interference, XTALK as such cannot be avoided. It is up to the system design and selected parameters – such as link margin, cable type, modulation, error correction methods, bit mapping  – to ensure a functioning communication is possible. Table 5.4 summarizes the discussed interference types and presents some measures on how to address them. Which measures were selected for the different high-speed communication technologies, is detailed with the respective technologies in Chapters 7 and 8.

5.3 Cables and Connectors

137

Table 5.4 Overview on interference types and counter measures Effect

Type

Addressed with

Attenuation/IL

Internal

Channel length, cable type

Echoes/RL

Internal

Echo canceller (interference is known), system design (for example Time Division Duplex, TDD instead of Frequency Division Duplex, FDD)

ISI

Internal

Signal processing (interference is known), system design (for example symbol duration)

NEXT/FEXT

Internal

Signal processing (interference is known), system design (no parallel lanes/carriers)

Thermal noise

External

Minimum receiver input level, link margin

Transient bursts

External

Link margin (modulation scheme, shield), FEC, retransmissions

Power ripple

External

Link margin (modulation scheme, shield), FEC, retransmissions

RF ingress

External

Link margin (modulation scheme, shield), FEC, retransmissions

Alien XTALK

External

Link margin (modulation scheme, shield), FEC, retransmissions, increased physical distance to potential interferer

Table 5.5 Conversion of dB into a voltage (V) or power (P) ratio dB –30

–6

–3

0

3

6

10

20

30

40

50

60

V

0.0316 0.1

–20

0.316 0.5

0.7

1.0

1.4

2

3.1

10

31.6

100

316

1k

P

0.001

0.1

0.5

1.0

2.0

4

10

100

1k

10k

100k 1M

0.01

–10

0.25

„ 5.3 Cables and Connectors When standardizing a communication technology, the target channel is an important starting point. It is not uncommon to select the desired cables and connectors first and then to develop the physical layer around it. The twisted pair Ethernet technologies developed at IEEE for data center use are a good example. For all but the very early BASE-T projects, the channel (specification) to meet was defined with the project start. These BASE-T channels generally comprise a 8 Positions 8 Contacts (8P8C) modular connectors as specified in International Electrotechnical Commission (IEC) 60603-7 (often incorrectly called Registered Jack (RJ-)45 [34]) as part of a Category (CAT) cable/link segment (as specified by the committee “TR-42” in the American National Standards Institute/Telecommunications Industry Association (ANSI/TIA)-568 [35] or ISO/IEC 11801 versions [36]). Table 5.6 provides an overview the different BASE-T PHY technologies and their cables/channel used for IT applications.

138

5 The Automotive Channel

Table 5.6 Data center Ethernet PHYs and their cables [37] PHY name

IEEE Publication EIA/TIA number number year of PHY

Publication year of channel

10BASE-T

802.3i

1990

CAT 3 (telephone wiring)

1991

100BASE-TX

802.3u

1995

CAT 5

1995

1000BASE-T

802.3ab

1999

CAT 5 or better

1995

2.5/5GBASE-T 802.3bz

2018

CAT 5e or better

1999/2002

10GBASE-T

802.3an

2006

CAT 6 (shorter distances), CAT 6a

2002

40GBASE-T

802.3bq

2016

CAT 7 or better

2002

Car manufacturers traditionally insist on their specific cables and connectors. In consequence, it was/is uncommon to define a complete channel with cable and connectors to use for an automotive communication technology. In automotive communication standard ­projects, it is more common to select a cable class (for example, UTP, STP, coaxial), measure the S-parameters of several different products with the right length and targeted number of inline connectors and then, based on these measured results, decide on the limit lines the channel has to meet. Of course, to finally adopt these limit lines, experts from the physical layer side need to confirm that they anticipate physical layer mechanisms to be available that result in a functioning and economically viable communication technology using these limit lines. In case of the wiring harness in cars costs are particularly relevant: The wiring harness is the third heaviest and third most expensive part of a car (after engine and chassis, see e.g. [38] [39]). The make of the cable and connectors thereby does not only affect the costs in terms of hardware but also in terms of energy efficiency/CO2 footprint. One of the main reasons Ethernet was successfully introduced in cars was that a physical layer technology was found – 100BASE-T1, called BroadR-Reach at the time – that met the automotive EMC requirement while requiring only a single UTP cable (and UTP connectors) [24]. When standardizing Automotive Gbps Ethernet, 1000BASE-T1, a lengthy discussion ensued whether a ­single or two pairs should be used for the transmission channel [40] [41]. Weight and cost concerns led to the selection of a single pair channel. In consequence, also for the high-speed data communication discussed with the use cases of this book, single pair ­cables are being preferred, though it is/was not always possible to use them (see Sections 7.1. and 7.2 for details). An important parameter for weight and costs, but also for the DC resistance of a cable is the diameter/cross section of the conducting material used. As different nomenclatures are used in different parts of the world, Table 5.7 shows the conversion between the American Wire Gauge (AWG), inch, and metric descriptions for typically used diameters for auto­ motive data cables. As the descriptions provided later in this section will use metric nomenclatures only, Table 5.7 allows for a quick reference for those readers used to other systems. The following subsections provide an overview on available cable (Section 5.3.1) and connector (Section 5.3.2) solutions. Section 5.3.3 addresses some future developments planned or recommended in the field of cables and connectors.

5.3 Cables and Connectors

Table 5.7 Conversion table for cable diameter nomenclatures [42] AWG

Diameter in inch

Diameter in mm

Cross section in mm2

21

0.0285

0.723

0.411

22

0.0253

0.644

0.326

23

0.0226

0.573

0.258

24

0.0201

0.511

0.205

25

0.0179

0.455

0.163

26

0.0159

0.405

0.129

27

0.0142

0.361

0.102

28

0.0126

0.321

0.081

5.3.1 Cables 5.3.1.1 Unshielded Twisted Pair (UTP) Cables UTP cables consist of two insulated wires that are uniformly twisted. The cables are produced in a continuous stranding process by the cable manufacturer, who needs to tightly control the manufacturing parameters in order to achieve a homogeneous and symmetrical twist as well as a constant lay length. Still, UTP cables generally present a cost (and weight) efficient solution and are thus preferred for communication systems in cars. However, ­because the signal attenuation increases with the transmission frequency, the usability of UTP cables for the high data rates required for modern sensor and display applications is limited. When 100BASE-T1 Ethernet was being introduced into the car industry, it was seen as inconceivable that a data rate as high as 100 Mbps would allow the use of UTP cables. That it was possible to use UTP cables for 100BASE-T1 is an important reason for its success. However, even at its comparably low Nyquist frequency of 33.33 MHz specific attention needs to be paid not only to the symmetry of the two wires but also to its twist rate, to the length of the untwist area in and before connectors, to the pinning in case multi-pin connectors are used, the wire gauge, and the insulation material of the wires [43]. The insulation material of data cables impacts its dielectric behavior. Not all insulation ­materials have absolutely stable and always reproducible dielectric coefficients, especially not in case of temperature changes. PolyVinyl Chloride (PVC) is an example for a very cost-efficient insulation material, which works for lower speed communication cables as needed for CAN, but which is not suited for higher speeds; in BMW´s opinion, not even those of CAN-FD. Its quality can vary significantly depending on the manufacturer or even between lots of the same manufacturer. One reason for its varying quality is that additional compound materials are used to achieve the required mechanical characteristics. Plasticizer and filling materials are used to make the PVC suitable for insulation. These additional materials cause a wider spread of the dielectric characteristics and have a negative impact on the stability against environmental influence. When plaster is used as a filler, the dielectric characteristic changes in case of humidity. PolyPropylene (PP) or crosslink PolyEthylene (PE-X or XPE) are better suited insulation materials, as they show a stable and

139

140

5 The Automotive Channel

defined dielectric behavior over the complete operational temperature range. Table 5.8 provides an overview on different dielectric materials used for automotive cables. Table 5.8 Properties of different insulation materials used for automotive cables [44] [45] Abbreviation Name

er

Temperature Comment range [°C]

PVC

Polyvinyl chloride

4–6

–40 to +105

PP

Cellular polypropylene

2.3

–40 to +125

PE-X or XPE

Crosslink polyethylene

3–4

–40 to +125/150

Maximum temperature depends on variant

FEP

TetraFluorEthylen HexafluorPropylen

2.1

–60 to +210

Not flexible, stiff

PS

PolyStyrene

1.05 –10 to +70

er changes with temperature

Currently not used in auto­ motive, shown as a reference for low er

Typical wire gauges are between 0.13 and 0.5 mm2. A gauge of 0.13 mm2 for the inner wires indicates in most cases the use of alloys like copper-tin (CuSn) or copper-magnesium (CuMg). With copper alloys of 0.13 mm2 similar robustness concerning the mechanical ­tensile strength can be fulfilled as with 0.35 mm2 pure copper. The wire gauge 0.14 mm2 generally indicates pure copper as conductor material (which is a simple observation of the products available on the market). When looking at the use of UTP cables for higher data rates inside cars, the situation is the following: 1000BASE-T1 Ethernet has a Nyquist frequency of 375 MHz. During the standardization it was seen as just feasible to use UTP cabling at this frequency. However, to ensure sufficient robustness against alien crosstalk, the UTP had to be jacketed in order to increase the physical distance to any neighboring wires. To the authors’ knowledge, all ­actual implementations of 1000BASE-T1 Ethernet in series production cars at the time of writing nevertheless use STP cables for risk mitigation. The price benefit of UTP has apparently been given a lower priority than the risk assessment of long-term efforts to control any EMC impact. A technology that has successfully gone into series production with UTP cables at a ­Nyquist frequency of 250 MHz is HDBASE-T [46] (though the authors do not know at which cable length). When used with a PAM 16 modulation, it allows for a gross data rate of 2 Gbps. A link adaptation mechanism reduces the modulation level in case the link budget of PAM 16 is not sufficient to combat the interference. This reduces the data rate but ensures continued communication. Likewise, during the early discussions of MGBASE-T1 Ethernet, UTP was also discussed for the 2.5 Gbps speed grade at least [47]. However, it was dismissed and STP was selected for 2.5GBASE-T using PAM 4 at a Nyquist Frequency of about 800 MHz [27]. These are very good examples that show that, within certain limits, the ­solution space allows for a trade-off between effort put into the channel and effort put into the PHY implementation.

5.3 Cables and Connectors

5.3.1.2 Shielded Twisted Pair (STP) Cables STP cables consist of a pair of twisted wires surrounded by a shield (see Figure 5.10). The shield has no function in the data communication as such, but solely the purpose to improve the EMC behavior by reducing emissions from the communication wire to the outside and shielding the communication from interference from the outside; in both cases by draining off any current induced into the shield to ground. The depiction in Figure 5.10 shows a shield consisting of a foil and a braid, as this is optimum for high-frequency data transmission. However, STP cables do exist that use either a braided or a foil shield (plus drain wire). With conductors in the range of 0.13 mm2 to 0.35 mm2 and common insulation, a typical impedance is 100 Ω, though impedance values of 90 Ω, 110 Ω, or 120 Ω also occur (see also Table 5.2). Compared with coaxial cables of the same outer diameter, STP cables have about 50% more insertion loss, because in order to achieve the same outer diameter, the conductors need to be thinner. 1 2 3

a F

F a

a F 1 2 3

Twisted pair system a Filler (optional) Outer cable jacket Braided shield Foil shield

Figure 5.10  Profile of an STP cable

STP cables have been used in cars occasionally not only for proprietary SerDes technologies used to connect cameras or displays, but also for microphones or for connecting low frequency antennas as needed for Near Field Communication (NFC) applications. It is also not uncommon to use STP for bus systems like CAN or FlexRay, when for example, a CAN wire has to run particularly close to a Radio Frequency (RF) antenna. For these uses, either a concrete cable type or just the geometrical data and the presence of a shield are specified by the car manufacturers. Note that 1000BASE-T1 Ethernet is the first automotive communication technology for which an automotive STP channel was fully specified [48], which emphasizes the comparable newness of the topic in the industry. Actual STP cables show a resonance effect in the IL parameter, called “suck-out”. These suck-out areas limit the frequency at which STP cables can still be used for data transmission. Different STP cables have the first suck-out areas at different frequencies. At the time of writing, improved manufacturing processes had shifted the first suck-out to above 5 GHz. Figure 5.12 in Section 5.3.1.4 shows the principal difference including the IL suck-out behavior (marked “1” and “2”).

5.3.1.3 STar-Quad (STQ) Cables STQ cables are a special form of STP cables. For STQ cables, instead of one pair of twisted wires, four wires are continuously twisted together within one shield. Figure 5.11 shows the principal set-up, in which always the opposite wires are assigned to one pair. This set up allows that, ideally, each pair is always in the neutral zone of the other.

141

142

5 The Automotive Channel

Such a two pair set-up allows for a simple form of parallelization and either the throughput can (roughly) be doubled or the IL can be improved by halving the transmission frequency per pair for the same throughput. In consequence, especially some of the earlier, proprietary SerDes communication systems require the use of STQ cables and the use of STQ ­cables in cars is common; also for USB 2.0 or 100BASE-TX Ethernet. Naturally, an STQ cable requires more material and is thus more costly than a simple STP cable. At the same time, it is more cost efficient than two STP cables – albeit with lower link margin and throughput – and thus preferred over the latter. Additionally, one STQ connector (see also Section 5.3.2.2) requires less space than two connectors for simple STP. In case of the communication technology requires only a single pair, the other two wires might be used for power and/or wake-up; with wake-up requiring only one wire. 1 2 3 4

a b

b F

a

a b F 1 2 3 4

Twisted pair system a Twisted pair system b Filler (optional) Outer cable jacket Braided shield Foil shield Inner jacket or filler (optional)

Figure 5.11  Profile of an STQ cable

Because STQ cables have been used in the car industry for a while, processing the harness, meaning attaching the respective connectors to the STQ cable, is done in a fully automated process. Once one wire is identified, the position of all other wires is defined. The arrangement of the individual wires is clockwise at one end and counter-clockwise at the other end. In order to identify the direction an “A” is printed in a short distance from the connector at one end and a “B” at the other end. Naturally, the application is not affected by this as the pinning is mapped 1:1 with no cross-over. Currently, the STQ cables loose importance for high-speed data communication. Due to manufacturing tolerances, the needed symmetry between the wires cannot easily be maintained and economically achieved for operating frequencies above 3 GHz (see inter-cable skew in Section 5.2.3.2). Star-Quad cables are available with a typical wire gauge from 4 × 0.14 mm2 to 4 × 0.5 mm2.

5.3.1.4 Shielded Parallel Pair (SPP) Cables SPP cables are another variant of shielded cables with one pair of wires. Other than for STP cables though the wires are not twisted but kept parallel throughout the complete length of the cable. This means that for the same length of cable, the length of the wires inside is shorter for SPP cables than for STP cables and SPP cables have a somewhat improved IL over STP cables of the same length. More important though, as a result of the very long, ideally indefinite lay length of SPP cables, the first suck-out visible for STP cables can ­almost be eliminated for SPP cables. Their first suck-out happens at a higher frequency. Additionally, it is smaller and caused mainly by the braid structure of the shield (similar as for coaxial cables). Figure 5.12 visualizes the difference by comparing typical IL curves for STP and SPP cables.

5.3 Cables and Connectors

f (log)

SPP

STP

1

2

IL / S12 (not to scale)

Figure 5.12 Principle comparison of suck-outs for STP and SPP cables (see, for example, [49]) Naturally, SPP cables require a very tight process control, as the mechanical stabilization of the stranding as well as the positive effect for EMC caused by the mutual change of the polarity of the wires is missing. Therefore, perfect symmetry of the conductors is an inevitable requirement for SPP cables. At the time of writing, SPP cables were available as prototypes only and not yet deployed in  series production cars. Nor has the use of SPP cables be foreseen for use with the ­MGBASE-T1 Ethernet or just released SerDes specifications. However, as SPP cables allow for the use of higher system frequencies they might be reconsidered for upcoming, even higher frequency systems [49]. Note, that in [49] the term Shielded Differential Pair (SDP) is used instead of SPP. Apparently, the term “parallel” had led to the expectation that the two conductors are necessarily in fixed straight lines and that the cable would therefore bend preferably in one direction [50]. This is not necessarily the case though. The conductors in the cables considered might simply not be twisted like in an STP cable. We think that using SDP in that sense is unfortunate, as also in an STP cable the data communication is differential. We therefore prefer to stay with the term SPP for all shielded conductor pair cables where the conductors are not twisted, whether they are in fixed positions or not, and within the shield. We prefer to use SDP as an umbrella term for SPP and STP cables.

5.3.1.5 Coaxial Cables Coaxial cables consist of an inner conductor, which is surrounded by a dielectric insulation, which is surrounded by a concentric shield, which is surrounded by a protecting jacket (see also Figure 5.2). The shield also serves as the back channel and is generally braided, as foil shield are difficult to connect. In the consumer world coaxial cables are best known for linking radios or TVs with their respective antenna systems [12]. Being used as antenna cables also represent the first use cases for coaxial cables in cars, meaning that the coaxial cables were and often still are used for analog signal transmission. 10BASE-T2 and 10BASE-T5 Ethernet are some of the most well-known technologies to use coaxial cables for

143

144

5 The Automotive Channel

digital data transmission. Their standardization was completed in 1985 and 1983 respectively. Note though, that data centers were happy to replace the coaxial cables with UTP, when this became possible with 10BASE-T in 1990. Being able to use UTP instead of coaxial is seen as an important element for the success of Ethernet [51]. In cars, the use of coaxial cables for digital data transmission is a very recent event and, to the knowledge of the authors, focusing on SerDes communication links. As is discussed in more detail with the physical layer technologies in Section 7.3.3, it was an important new feature when the proprietary SerDes technologies started supporting coaxial cabling in ­addition to STP. This was quickly followed by the possibility to also transmit power over the coaxial link. This allowed the use of less costly cables and connectors for the data trans­ mission – coaxial cables and connectors are generally less costly than STP cables and connectors – while additionally not having to provide for a separate power supply harness. For the camera use case, this put SerDes communication back into the arena, having just lost parts of the camera market to Ethernet [24]. Coaxial cables attenuate the signal least at an impedance of about 77 Ω [12]. This is why coaxial cables with 75 Ω are used for systems which transport low energy signals, such as antenna cables. For most general purposes including digital data communication over ­SerDes, coaxial cables with 50 Ω impedance are used. 50 Ω was chosen because it provides the best impedance match between source and load, and because it is almost in the middle between the 75 Ω with low attenuation and the 33–40 Ω, which exhibit maximum power handling capabilities [12] [52]. The attenuation for 75 Ω cables is about 10% less than for 50 Ω ones. The 50 Ω coaxial system is a good complement to the 100 Ω STP cables. Using only one wire of the differential system thus corresponds to a single ended coaxial system and vice versa. Another important coaxial cable parameter is the outer diameter. The most commonly used coaxial cables in cars have an outer diameter of 3.2 mm. Among automotive engineers, they are often called “low-loss” cables and are used for static cable routing (because they have less attenuation than the industrial standard RG174 with an outer diameter of 2.8 mm, see also Figure 5.13). Static cable routing means that, once installed, the cable position is not changed. The bending capabilities of such coaxial cables, especially when mechanically strained regularly as would be required when used between chassis and door, are limited. In a static placement the bending radius is about five times the diameter, when needing to bend and unbend dynamically the bending radius should not be smaller than ten times the diameter. Figure 5.13 shows exemplary IL curves for different coaxial cables. It can be seen that the smaller the diameter, the larger the attenuation. Furthermore, when coaxial cables are needed in cars at positions which require dynamic bending, a different built is needed. The only cable in Figure 5.13 that allows for frequent bending is the RG174. It is smaller in ­diameter and its inner core consists of copper cladded steel braids. It shows the largest ­attenuation. Care should be taken when reading the datasheet. The presented data might reflect typical values or the worst-case data within different lots. Note that the “Radio Guide (RG)” nomenclature was developed in the Second World War and was removed from the respective standard in 1985. Cables that use RG in the naming today thus do not necessarily comply with the original specification [10].

5.3 Cables and Connectors

Figure 5.13 Principal difference in IL/S12 of different coaxial cables [44] [53]

5.3.1.6 Other Multi-port Cables This section discusses a number of additional cables used for high-speed data communi­ cation. The selection presented is to highlight specific important aspects when selecting ­cables suitable for automotive high-speed communication technologies. Figure 5.14 shows the first example, the Dieselhorst-Martin (DM) stranding. The DM stranding is two or more twisted pair cables put aside each other within the same shield system and/or the same outer jacket. This type of cable is frequently used for telecom applications, where often more than two twisted conductor pairs are bundled together in one jacket. 1 2

a

b

3

a

b

a b 1 2 3

Twisted pair system a Twisted pair system b Outer cable jacket Braided shield (optional) Foil shield (optional)

Figure 5.14 Profile of a Dieselhorst Martin stranding cable Figure 5.15 visualizes the principal difference between a DM cable with two wire pairs and a STQ cable. When the same material is used for both cables and in both cases each pair is used for the differential transmission of data, similar performance results could be expected. In Figure 5.15, the decisive difference between the two cable types when used for high-speed data communication can be seen. In case of the STQ cable, the magnetic fields H of each wire pair is neutral at the position of the other wire pair. For the DM stranding, this is not the case and each wire pair interferes with the respective other pair with XTALK.

145

146

5 The Automotive Channel

STQ

Dieselhorst-Martin NZ

H

a

b

b

a

NZ

Ha

a H b H

H b is in the NZ of a

b

Hb a

b

a

b

NZ

a

Ha

a is in the NZ of b

Neutral Zone (NZ) with counterclockwise H fields

Hb

No NZ between a and b causing potential interference

Figure 5.15 Comparison between impacts of interfering magnetic fields for STQ and ­Dieselhorst-Martin stranding cables An example for a common data cable that uses the DM stranding is the CAT 5e cable, which is typically used for 1000BASE-T Ethernet. Figure 5.16 shows the profile of a CAT5e cable with shield. It comprises four wire pairs and a foil shield plus drain wire. XTALK between individual wire pairs is thus an effect 1000BASE-T needs to handle and NEXT compensation is generally part of the receiver design. Furthermore, the cable foresees different lay lengths for each twisted pair. This reduces areas with common coupling, which is always a good measure for similar cable constructions. Note that such cable solution can be used for 1000BASE-T, because 1000BASE-T works at a Nyquist frequency of 62.5 MHz and because the data that causes the XTALK is known. For systems working with multiple wire pairs at significantly higher frequencies, even with known XTALK, using a cable that shields the individual pairs might be necessary; a solution available, for example, with CAT 6a cables [54].

1 2

b

a

b

a 3

c

d

c d

a b c d 1 2 3

Twisted pair system a Twisted pair system b Twisted pair system c Twisted pair system d Outer cable jacket Foil shield Drain wire

Figure 5.16 Profile of a shielded CAT 5e cable as used for 1000BASE-T Ethernet One of the most well-known and widespread communication interfaces of the CE industry is the Universal Serial Bus (USB). USB supports large data rates – from 12 Mbps for USB 1.0 to 40 Gbps for USB 4.0 [55] –, power distribution, and P2P communication. Consequently, a regularly asked question is: Why is USB not used more inside cars? Next to the too short length that USB cables generally support – 5 m –, there is the complexity of the cable itself. Figure 5.17 shows the profile of a USB 3.0 cable, which is specified for speeds up to 5 Gbps. There are three wire pairs for the data communication: one unshielded one for standard USB 1.0 and 2.0 and two with individual shield and drain wire that allow for the increase in data rate. The individual shields for these two pairs are needed to mitigate the effects of XTALK. For car manufacturers, using USB with such cables is a cost and weight issue, and, if lengths beyond 5m are needed, a technical problem.

5.3 Cables and Connectors

U S1 S2 P G F D 1 2 3

F

1

F

2

S1 S1

P D F

3

U

G

U D

S2

F F

S2 3

USB (1.0/2.0) Super speed system 1 Super speed system 2 V+ supply GND Filler Drain wire (not insulated) Outer cable jacket Braided shield Foil shield of subsystems

Figure 5.17 Profile of a USB 3.0 cable [56] Figure 5.18 shows an example for a hybrid cable as presented by the cable manufacturer Kromberg & Schubert on their website [57]. The example contains one STP and two coaxial cables within one jacket. Such constructions might be useful, when antennas and (highspeed) data communication run side by side. The main disadvantage of such cable constructions is the, at least partially, manual assembly process when connecting the cables with the respective connectors. This is a quality risk for car manufacturers who have to ensure that thousands of parts all function perfectly without deviations between different cars (see also Section 3.3.1). The reason for the manual handling is that during the process of connecting cable and ­connector, different shields and fillers need to be handled. The detection and correct connection of the different wires, shields, and fillers is too complex for automation. Fillers help to stabilize a respective wire position, which simplifies their extrusion. Furthermore, fillers ensure an almost circular outer diameter of the cable. The circular geometry of a cable is important, for example, when the cable-connector interface needs to be sealed (or simply for an acceptable appearance to the user). 2

1 2

F a a

C1 D F

F

F C2

2

a C1 C2 D F 1 2

Figure 5.18 Example of a hybrid cable [57]

Twisted pair system a Coaxial system C1 Coaxial system C2 Drain wire Filler Outer cable jacket Individual jacket and shield

147

148

5 The Automotive Channel

5.3.1.7 Aging and Mechanical Stress of Cables Cars can be on the road for a long time; 20 years is no exception. The cables used in such cars experience a vast amount of temperature changes and mechanical stress. For many cables this stress mainly consists of vibration. However, there are some locations in the car, where the cable is regularly bent end unbent, like when bridging into a door or the trunk lid. Figure 5.19 shows IL and impedance measurements made with Time Domain Reflecto­ metry (TDR) of a coaxial cable before and after it had experienced mechanical stress of bending and unbending 15,000 times.

Figure 5.19 IL and TDR-based impedance measurements of coaxial cables before (left side) and after (right side) having experienced severe mechanical stress (Source: MD-Elektronik) After having undergone severe mechanical stress, the IL parameter increased up to about 50%. The TDR measurements also show an area with a strong change of the impedance. One specific cable shows even stronger impairments than others. This cable was X-rayed with results shown in Figure 5.20. As can be seen, the conductor braid of the cable is severely damaged with individual braid strands broken. Neither the insulation, nor the foil shield, nor the inner conductor are visible in the X-ray. However, further investigations showed impact also on these elements. The larger the mechanical deviation, the stronger the deviation of the S- and cable parameters from their initial values; including RL and SA, which are not visualized in Figure 5.19. It is important to note that also the cables with no such obviously visible impacts showed changes in the electrical parameters. A purely visual inspection is thus not sufficient to detect any impact. It is thus important to investigate each cable (type) used for its specific resilience in different mechanical stress situations considering its targeted location. Table 5.9 provides a ­general overview on different stress types and how they impact the different electrical parameters of cables. When looking at the details, every car manufacturer follows a company specific release process, which needs to be adhered to.

5.3 Cables and Connectors

149

Figure 5.20 X-ray of a coaxial cable before and after 1500 bending cycles (Source: MD-Elektronik) Table 5.9 Different cable stress types (see, for example, [33], the impacts are empirical observations made by the authors); room temperature tr = 22–25 °C, tmin = –40 °C, tmax can have different values depending on the use case, typical are 105 °C Stress type Typical release test

Mainly impacted Change of parameter ­parameter

Reversibility

Temperature @tr, tmin, tmax

IL

+~20% @tmax

Reversible

DC resistance

+~25% @tmax

Comment

Accelerated 3000 hours @tmax IL aging by high SA temperature

+~10% –~5 dB Non-reversible

Humidity

3000 hours

IL

+~10%

Reversible

Bending

30,000 cycles in 6 groups of:

IL

–~1 dB/m +/–~5 Ω

Non-reversible Only with flexible cables

4000 @tr

Impedance at bend location

600 @–25 °C

UTP/STP MC

Effect depends on insulation material

observable

Bending radius = 5 × cable diameter.

–~1 dB/m

Non-reversible Torsion seems to be one of the most impacting mechanical stress sources for data cables

400 @tmax Torsion/ bending

> 100,000 cycles: IL 60,000 @+40 °C SA 20,000 @–20 °C Impedance 20,000 @+85 °C 5000 @+40 °C

Pressure

Application ­specific

10–20 dB less +/–5 Ω

UTP/STP MC

observable

Impedance

+/–5 Ω

Depending on the type of pressure

Comes from tapes, fixtures, seals, GND clamps

It is desirable to accelerate some of the tests. A typical means to do so is to perform the tests at a higher temperature than the one specified. For this the Arrhenius equation [58] is applied to the highest ambient temperature foreseen. This utilizes the temperature increased reactivity of the material – about double per 10 °C – that accelerates the aging. For cables

150

5 The Automotive Channel

used, for example, for power supply, this method works well and delivers comparable results for aging in 240 hours rather than in 3000 hours. For data cables, and especially for those intended to be used for high data rates, accelerating the aging by a short-term temperature increase, is not as suitable. Because the shortterm aging somewhat changes the geometry of the cables, temperature induced short-term aging results in worse S-parameter changes than regular aging would. While for power ­cables or data cables for lower data rates this is uncritical, it does make a difference for the tight margins encountered with high-speed data communication. Worth knowing about general cable types What is the main different between a normal, general purpose cable and a dedicated data cable? When asked, an expert working for a cable manufacturer replied that the main difference is the speed of the production process. It can take ten times longer to produce data cables than general purpose cables in order to have a controlled manufacturing process at all stages, to keep the mechanical deviations within tight tolerances, and therefore to obtain a stable electrical characteristic. 

5.3.2 Connectors Connectors are of particular importance to car manufacturers. They are the part of the harness that is handled by the workers during the production process, who connect the wiring harness and ECUs. If the connector is too fiddly, the connecting process might take too long. If the connection is too stiff, too many connectors might not get plugged in correctly. If the coding of the connector is not easily identifiable, the wrong connector might be attached to the wrong socket or it might be connected the wrong way around. In the experience of the authors, most in-the-field communication problems are due to connectors; either they were not plugged-in correctly, or the connection was not properly secured, or the connector broke in the process, or worse, it partially broke during the process. Partially breaking is worse, because it is often not immediately apparent and more easily missed, but sure to cause problems later when the car is with the customer. A large number of options exist for connectors. The most important criteria, which has been used to structure the description in the following subsections, is the type of cabling intended to be used with the connector (shielded, unshielded, coaxial, etc.). Also very important is the impact of the physical environment. It decides on the particular variant of the connector (sealed, unsealed, straight, bent, single port, multi-port, hybrid, etc.). However, there is a vast number of other parameters, car manufacturers base their individual decisions on. These can be: The number of connecting cycles (for normal operation usually 25, for diagnosis 100), insertion force, extraction force, electrical resistance, maximum current and voltage, insulation resistance, leakage current limit, IL, RL, XTALK, SA, temperature resilience, size, mechanical robustness, possibility to interlock individual contacts or the complete connector, resilience against vibration, prevention of tilted plug-ins, color coding, length of crimp area, and, last but not least, whether the manufacturing process is manual or automated.

5.3 Cables and Connectors

The following provides an overview on the most common connector types relevant for the discussed use cases. The description cannot be complete but reflects the European preferences for specific connector types (as Europe is the location of the authors).

5.3.2.1 Connectors for UTP Cables For earlier communication systems using UTP cables like LIN, CAN, or FlexRay, it was not necessary to pay particular attention to the electric parameters of the used connectors. The requirements of these technologies were/are moderate, and quite a few different connectors and UTP cables may be used. The situation changed with the advent of 100BASE-T1 Ethernet. The goal was to use unshielded cables also at this higher data rate. To be able to realize this, it required to look at the electrical parameters of the cables and the connectors in more detail, especially with respect to parameters like impedance and MC. The impedance of a connector is defined by the ratio between the diameter/width of a pin and the distance between them (see also Figure 5.2). For the MC requirement of 100BASE-T1 a symmetric pin arrangement is needed, meaning that the two differential communication wires need to be attached to pins of the same length (double-row connectors generally have pins in two different lengths) avoiding asymmetric impact from neighboring pins. It turned out that connector systems with a pin distance of 2.54 mm and a width of 0.63 mm fulfill the requirement for a system impedance of 100 Ω. One example for such a connector is the Micro Quadlok System (MQS). The nano MQS (nMQS), which was developed later with a smaller form factor is also suitable. Both originated at TE Connectivity. However, based on that knowledge several companies have developed dedicated connectors fulfilling the same requirements. Table 5.10 lists a few of these connectors. Note that there are other companies in the US and Asia that have developed similar connector systems. Table 5.10 UTP connector types with 100 Ω impedance suitable for higher speed ­communication Connector name Crimp contact system Company

Cable type

MateNet

Nano MQS

TE Connectivity

2 × 0.13 mm2 to 2 × 0.35 mm2, also supports jacketed and shielded cables

AMEC

Nano MQS

Aptiv

2 × 0.13 mm2 to 2 × 0.35 mm2, also supports jacketed and shielded cables

H-MTD, H-MTDe

Round contact

Rosenberger

2 × 0.13 mm2 to 2 × 0.22 mm2, also supports jacketed cables

With enabling 100 Mbps Automotive Ethernet for UTP cables and connectors the ­auto­motive industry went through a huge learning curve. The learnings were applied also during the development of 1 Gbps Automotive Ethernet 1000BASE-T1. As a result, also 1000BASE-T1 Ethernet was specified for the use of UTP cables, albeit with jacket, which explains why the connectors listed in Table 5.10 support jacketed cables. Additionally, the larger impact of XTALK because of the higher frequency of 1000BASE-T1 became evident in the multiport

151

152

5 The Automotive Channel

PCB header and inline connectors. MateNet and AMEC thus introduced an additional shield between individual pin pairs that also serves as the shield contact for the later developed shielded version of these connectors. As mentioned in Section 5.3.1.1, actual 1000BASE-T1 implementations use shielded cables and connectors.

5.3.2.2 Connectors for STQ Cables In 2006 the company Rosenberger developed the so-called High-Speed Data (HSD) con­ nector connecting four wires like used in STQ cables [59]. At the time of writing, the HSD connector was available in a number of variants (also unshielded) from different connector suppliers and specified for up to 3 GHz Nyquist frequency supporting wire gauges from 4 × 0.14 mm2 to 4 × 0.5 mm2 with a system impedance of 100 Ω. Its development was an important milestone for the use of SerDes in cars, as the HSD connector allowed the transmission of high data rates while using automotive suitable and qualified connectors. The connectors used for Automotive SerDes before the advent of HSD were adapted from other industries and came with respective quality problems and concerns (see also the information box at the end of Section 5.3.2.4). For a long time, the HSD connector was the connector used when a shielded connector was required inside the car, especially with increased data rates. As a result, HSD connectors were and are deployed not only for SerDes, but also for CAN, 100BASE-TX Ethernet, USB 1.0/2.0, IEEE 1394 Firewire, and MOST [60] [61]. When only two pins are needed for the communication, the other two pins might be used for power and/or wake-up.

5.3.2.3 Connectors for SDP (STP and SPP) Cables It might seem surprising, that STQ connectors are discussed before the ones for SDP cables. This simply has the reason, that automotive suitable versions of STQ connectors were ­available before the ones for STP/SDP. With the originally comparably few uses of such high-speed data cables in cars, it took until 2017 for Rosenberger to start promoting a new connector for SDP cables called High-speed Modular Twisted-pair Data (H-MTD) connector. It exists in several variants and is licensed, for example, to Aptiv [62]. The H-MTD connector can connect shielded cables with a wire gauge of 0.14 mm2 up to 0.22 mm2 and a maximum outer diameter of about 4.5 mm. The good performance of the connector is due to a restriction of the mechanical tolerances. The H-MTD connector is specified for fre­quencies up to 20 GHz. The company Rosenberger also announced a hybrid multiport version that would connect a specific number of SDP and coaxial cables and an H-MTDe version for ­unshielded cables, to be used for example for 100BASE-T1 Ethernet. Such a set-up allows to use the same PCB header for either unshielded 100BASE-T1 or shielded 1000BASE-T1 Ethernet [63].

5.3.2.4 Connectors for Coaxial Cables In the year 2000, the interest group “FAchausschuß KRAraftfahrzeuge (FAKRA)” of the German SSO Deutsches Institut für Normung (DIN) defined a connector for coaxial cables in DIN72594-1 (adopted in the US as USCAR-18). Although, FAKRA addressed a number of different topics, the name “FAKRA connector” is often – also in this book – synonymously used for this specific type of coaxial connector that has become a standard connector used almost worldwide [64].

5.3 Cables and Connectors

The FAKRA connector was defined for applications up to 6 GHz and coaxial cables from 2.8 mm to 4.95 mm diameter. It is available from a number of suppliers in various variants with either 50 Ω or 75 Ω system impedance. FAKRA is also part of some multi-port hybrid ­connectors. For example, the connector for car doors might additionally be equipped with a FAKRA coaxial option. Care must be taken, with the actual realization of a particular FAKRA connector. As the product definition is quite old and, in the beginning, only frequencies up to 1 GHz were required, many implementations from non-automotive suppliers are offered with this frequency restrictions. Generally, FAKRA connectors are available as single-port and dual-port versions. Higher number of ports are application-specific designs and not in general use, because the space needed for these connectors is significant. As a result, “Mini-coax” connectors were developed. A relevant amount of space can be saved with the Mini-coax systems, especially for multi-port connectors. A 4-port Mini-coax connector requires only about the same space as a single-port FAKRA connector. The gain in geometry is about 70% when a single-port ­Mini-coax connector is compared to a single-port FAKRA connector. As a drawback, the Mini-coax connectors are often somewhat inferior (by up to 0.4 dB) to the FAKRA connector with respect to the IL according to the manufacturers’ data sheets. One of the companies to develop a Mini-coax connector is Rosenberger, whose prototypes of their “High-speed FAKRA Mini (HFM)” connector appeared in 2014. The HFM connector is currently promoted for frequencies up to 20 GHz [65] and Rosenberger has licensed the HFM design to other connector manufacturers like Molex [66]. Another Mini-coax connector is the “MateAX” from TE Connectivity. The development started in about 2013 and the MateAX connector is currently promoted for frequencies up to 9 GHz in its standard version and for frequencies up to 15 GHz in the optimized version. TE also licenses the MateAX design to selected other connector manufacturers like IMS Connector Systems [67]. The HFM and MateAX connectors have approximately the same size of the final product and the same basic mechanical size of the contact system. They target the same automotive high-speed data applications and are thus available for 50 Ω system impedance only. However, the two systems are not interoperable, as the contact system of the outer/shield contact has a different design layout in both cases. Compared with the FAKRA, both types have an advantage in the RL parameter. This is the result of less mechanical tolerances for the new design. At the same time, the connectable cable diameter is smaller. The Mini-coax connectors support cable diameters of 2.8 mm to 3.5 mm. Within a series production car, generally either HFM or MateAX are used, potentially with some FAKRA connectors. Cables with a FAKRA connector on one end and a Mini-coax connector at the other end are common practice. Cameras, for example, often have a FAKRA connector, whereas the ECUs at the other end of the communication holds a multi-port Mini-coax connector. While HFM and MateAX cannot be directly connected, they can be used each at one end of the cable. Figure 5.21 and Figure 5.22 show two examples of multiport FAKRA connectors for different use cases.

153

154

5 The Automotive Channel

Figure 5.21 Example of a multiport FAKRA connector for a central unit. The right side shows several SerDes chips close to the connector and SoC on the PCB. The left side shows how the connector provides eight mini-coax and one traditional FAKRA connection in one housing. (Photo: Michael Kaindl)

Figure 5.22 Example of a multiport FAKRA use in a surround-view camera system (Photo: Michael Kaindl)

5.3 Cables and Connectors

Having the right connector Digital cameras and displays started in the consumer and IT industries. Their advantages in terms of quality and ease to pre- and post-process the digital video data were immediately apparent; also to players in the car industry. Hence, there was a desire to use the tech­ nologies for cars, before any specific, automotive suitable transmission technologies or connectors existed that supported the suddenly very high data rates, not needed anywhere else in the car. In one case known to the authors, a Tier 1 supplier selected for an early, digital camera system a slightly modified FireWire [68] connector to transmit the FPD Link I video data, LIN control data, and power. This was accepted by the car maker and the solution fulfilled all the requirements with respect to EME and EMI. However, after SOP, the solution nevertheless caused serious problems. The selected connector was asymmetric, but not asymmetric enough. With some added force, it was possible to plug the connector 180 degrees turned, which then happened at regular intervals. Naturally, it caused a complete system failure. As a consequence, the used connectors needed to be amended such that they could no longer be connected the wrong way around. This solved the problem. However, as is well known, such late changes are not only extremely annoying but also very expensive. This shows how important it is to consider not only electrical performance values for electronic systems in cars, but also any impact they might have during the assembly and – naturally  – during runtime. As the assembly and mechanical strain are so different in the ­automotive industry from the consumer and IT industries, reuse of any of their defined SerDes or Ethernet cables is inadvisable. It can thus be assumed that Rosenberger’s HSD connector system (or that of other suppliers) played an important role in the adoption of SerDes links for the transmission of uncompressed video data in cars. 

5.3.3 Quo Vadis? When designing a communication system, there is a trade-off, between the system’s ­Nyquist frequency with the resulting IL and the modulation mode selected. Especially for Nyquist frequencies in the GHz range the IL is large, as has been shown in many of the previous sections of this chapter. Higher modulation modes allow to pack more bits into one symbol. They achieve the same bit rate at a reduced Nyquist frequency and thus a reduced IL. This is good for the SNR/link margin. However, higher modulation modes also reduce the amplitudes between signal levels, which at the same time reduces the SNR/link margin and thus countermeasures the gained SNR of the reduced IL. Normally, HS communication systems are designed for the sweet spots, meaning the maximum SNR/link margin that can be achieved with a particular IL/modulation rate combination. Because, in principle, the higher the SNR/link margin the more robust the system. This is particularly relevant for SerDes links used for ADAS sensors at the outer periphery of the car that have to fulfil functional safety requirements. So, with ever higher data rates and increasing Nyquist frequencies, what means are available on the channel side to push out the boundaries? One of the first parameters to look at

155

156

5 The Automotive Channel

is to reduce the attenuation/IL of the cables. We see four parameters, which might be changed as to improve the IL: 1. The diameter of the conductors: With larger diameters and thus increased size of the conductors the attenuation can be improved. 2. Improved conductor materials: The so-called skin-effect causes high-frequency ­currents to flow only at the surface of the conductors [69]. If the roughness of the surface is ­reduced, this improves the conductivity. This can be achieved, for example, with silver cladded conductors. Solid conductors also have a smoother surface than the stranded wires made of the braids that are currently used in the automotive industry. However, cables with solid conductors are difficult to crimp and thus unsuitable for use in the mechanically strenuous automotive environment. 3. Insulation material with a lower dielectric loss: Current automotive twisted pair data and coaxial cables use mainly PP or PE-X (see also Table 5.8), which are well suited for the use in automotive environment, but do not have the lowest dielectric values. 4. Shorten the length: For example by using SPP instead of STP cables. Having no twist makes the electrical length shorter and therefore cause less attenuation. Another parameter that needs to be considered is the crosstalk. With the higher speed grades and the use of higher symbol frequencies, the impact of crosstalk increases. This might require better shielding, but also more attention in the connector designs to mitigate the effects. Currently, the automotive industry has not classified the used communication cables, like the IT industry did with their “Cat” cables (see Table 5.6). The OPEN Alliance and Automotive SerDes Alliance provide general channel test specifications for their PHY technologies and the, at the time of writing ongoing, ISO/AWI 8092-6 and 8092-7 projects refine the ­requirements and tests for automotive (high-speed) connectors.

„ 5.4 Printed Circuit Boards (PCBs) The communication path on the PCB between MDI connector and pin of the communication chip is part of the overall communication channel. 5 to 20% of the IL of a communication channel are attributed to the so-called MDI network (see also Table 5.1). In consequence the specifications for the automotive high-speed communication standards define limit lines for the MDI RL and, at least for the SerDes standards, also for the MDI IL (see Figure 5.23). The MDI RL and IL are measured into the MDI connector, between MDI connector and pins of the communication chip respectively. For the proprietary SerDes technologies, the suppliers do not provide MDI IL or RL limit lines but recommend an IL budget for the PCB as well as a reference connector in the application notes and design guidelines.

5.4 Printed Circuit Boards (PCBs)

Figure 5.23 MDI IL and RL limit lines for different automotive high-speed communication technologies For the impact of the PCB on the IL, the transmission frequency makes a decisive difference. Table 5.11 shows the relationship between frequency and wavelength for some ­selected frequencies, based on Equation 5.6. As a rule of thumb, the speed c of an electric signal in a wire can be calculated as 2/3 of the speed of light in vacuum co [70]. (5.6) It can be seen in Table 5.11 that at lower frequencies, for example 30 MHz, the wavelength 6.67 m is way longer than what can expected to be the length of a PCB trace in an auto­ motive ECU. However, in the GHz frequency range used by many automotive HS communication technologies, the wavelengths are in the same range as the length of the traces. Considering that electrical effects like resonance or extinction are observable at multiples of l /4, PCB traces of a few centimeters are sufficiently long to comprise these undesirable effects. Table 5.11 Relationship between frequency and wavelength Frequency [MHz]

Wavelength lco in vacuum [m]

Wavelength lc in conductor [m]

  10

30

20

  30

10

 6.67

 100

 3

 2

 300

 1

 0.67

1000

 0.3

 0.2

3000

 0.1

 0.07

There is more to the impact of the PCB design on the communication channel [71]. One ­aspect to consider is the PCB material. In general, PCB materials are selected for their ­mechanical, electrical, and thermal parameters. The most popular material is the Flame Retardant (FR)-4 material, a glass-reinforced, epoxy laminated material [72], which is quite cost efficient. When using FR-4 as a cost reference, the cheapest material – laminated ­paper, FR-2 – is at 30%. High-performance Teflon or ceramic based material is eight times more expensive [73] (see also Table 5.12).

157

158

5 The Automotive Channel

However, FR-4 is most suitable for low frequency applications. Whether it can be used for high frequency applications depends strongly on the supplier of the FR-4 material; more particularly, on the construction of weave of the glass-fiber and the according quality process of the manufacturer. Low-speed FR-4 materials have a dielectric constant, which is strongly frequency dependent. The higher the frequency, the larger the dielectric loss and the larger the IL (see Table 5.12). Low-speed FR-4 material can be used up to a speed of a few GHz. For higher frequencies, different PCB materials are better suited. Very low loss materials can be used for frequencies up to 20 GHz. Note, that the respective PCB traces typically have a length of 2 to 5 cm. Table 5.12 Comparison of the impact of different PCB materials on the IL [71] [73] PCB material

IL/cm @2 GHz IL/cm @5 GHz IL/cm @10 GHz Relative cost

FR-2

unsuitable

unsuitable

unsuitable

0.3

Low speed FR-4

0.07

0.24

0.43

1

Medium speed materials 0.016

0.08

0.16

2

Teflon/ceramic based for (very) high speeds

0.01

0.02

8

0.003

For the design process the selection of a PCB material is one of the first steps. Another decision has to be taken with respect to the PCB traces: microstrip and/or stripline. In case of the microstrip, the PCB traces are on top of the dielectric material with the ground plane on the other side of it. In case of stripline, the PCB traces are embedded in the dielectric material with reference planes on either side (see also Figure 5.24). Microstrip

Stripline Reference plane 1

Trace PCB dielectric

GND plane

PCB dielectric Trace

Reference plane 2

Figure 5.24 Microstrip versus stripline PCBs [71] Microstrip routing has better signal characteristics and is easier to manufacture. The implementation of stripline designs is more complex but allows for a higher density of signal lines. Additionally, stripline designs are better protected against mechanical damages [74]. [75] shows how the thickness of the dielectric material impacts the IL. Thinner, high-­ density PCBs for microstrip set-ups have a higher IL than thicker, standard materials. In the example given, a 20% thinner board has about 20% more IL. Naturally, microstrips are only possible on the two outer layers of the PCB. Multilayer PCBs need to realize a stripline design between layers. While the number of connections between layers, the allowed transit time, avoiding accessibility for security reasons, and the acceptable attenuation do play a role for the decision on the number of PC layers to use, key is ­often the space that is available to realize a certain functionality.

5.5 Bibliography

In today’s PCB design tools build-in functions support the designer for the correct implementation of PCB traces in terms of required system impedance and calculated channel parameters are available as part of the layout process.

„ 5.5 Bibliography   [1]

S. Carlson, “IEEE 802.3 Greater than 10 Gb/s Electrical Automotive Ethernet PHYs (P802.3cy) Workgroup Approved Objectives,” 21 May 2020. [Online]. Available: https://www.ieee802.org/3/ cy/P802d3cy_OBJ_WG_0520.pdf. [Accessed 14 June 2021].

 [2]

N. Patwari, “ECE 5520: Digital Communications Lecture Notes,” 2009. [Online]. Available: https://my.ece.utah.edu/~npatwari/ece5520/lectureAll.pdf. [Accessed 4 August 2021].

 [3]

A. Henkel, “What is De-embedding?,” Everything RF, 6 August 2019. [Online]. Available: https:// www.everythingrf.com/community/what-is-de-embedding. [Accessed 4 August 2021].

 [4]

K. Matheus, M. Tazebay, S. Buntz and S. Korzin, “RTPGE Alien XTALK Scenarios,” 24 January 2013. [Online]. Available: https://www.ieee802.org/3/bp/public/jan13/matheus_3bp_02_0113.pdf. [Accessed 14 May 2021].

 [5]

S. Buntz, “Update on Required Cable Length,” 11 July 2012. [Online]. Available: https://grouper. ieee.org/groups/802/3/RTPGE/public/july12/buntz_02_0712.pdf. [Accessed 26 March 2021].

 [6]

B. Bergner, E. Cuesta and E. DiBiaso, “Link Segment Measurements,” 19 January 2021. [Online]. Available: https://www.ieee802.org/3/cy/public/adhoc/BergnerCuestaDiBiaso_3cy_01a_01_19_21. pdf. [Accessed 12 May 2021].

  [7]

Superuser, “What is the Maximum Length of a PCI-Express Flexible Extension Cable?,” 4 March 2015. [Online]. Available: https://superuser.com/questions/885232/what-is-the-maximum-lengthof-a-pci-express-flexible-extension-cable. [Accessed 4 August 2021].

 [8]

Seagate, “Serial ATA (SATA) Data Cable Lengths,” not known. [Online]. Available: https://www. seagate.com/de/de/support/kb/serial-ata-sata-data-cable-lengths-182453en/. [Accessed 4 August 2021].

 [9]

R. Chen, “Microspeak: Impedance Mismatch,” 23 January 2018. [Online]. Available: https:// devblogs.microsoft.com/oldnewthing/20180123-00/?p=97865. [Accessed 14 May 2021].

[10]

Wikipedia, “Koaxialkabel,” 21 March 2021. [Online]. Available: https://de.wikipedia.org/wiki/­ Koaxialkabel. [Accessed 14 May 2021].

[11]

Wikipedia, “Wellenwiderstand,” 5 December 2020. [Online]. Available: https://de.wikipedia.org/ wiki/Wellenwiderstand#Leitungswellenwiderstand. [Accessed 14 May 2021].

[12]

rfcables, “Coaxial RF Cables Classification Guide,” 2005. [Online]. Available: https://www.­ rfcables.org/articles/18.html (no longer available). [Accessed 14 May 2021].

[13]

USB.org, “Managing Connector and Cable Assembly for USB SuperSpeed,” 1 February 2013. [Online]. Available: https://www.usb.org/sites/default/files/USB_SuperSpeed_CabCon_Whitepaper. pdf. [Accessed 02 April 2021].

[14]

ISO, “ISO 11898-2:2016 Road vehicles — Controller Area Network (CAN) — Part 2: High-speed Medium Access Unit,” ISO, Geneva, 2016.

[15]

M. Hiebel, Grundlagen vektorieller Netzwerkanalyse, München: Rohde&Schwarz GmbH, 2011.

159

160

5 The Automotive Channel

[16]

Wikipedia, “Scattering Parameters,” 28 February 2021. [Online]. Available: https://en.wikipedia. org/wiki/Scattering_parameters. [Accessed 04 April 2021].

[17]

B. Noseworthy, “Gigabit Ethernet 1000BASE-T, update,” 29 March 2000. [Online]. Available: https://www.iol.unh.edu/services/testing/ge/knowledgebase/pcs.pdf. [Accessed 14 May 2020].

[18]

P. A. Rizzi, Microwave Engineering: Passive Circuits, Upper Saddle River: Prentice Hall, 1988.

[19]

C. DiMinico and H. Kadry, “802.3cy Link Segment IL Baseline Proposal,” 21 July 2021. [Online]. Available: https://www.ieee802.org/3/cy/public/jul21/diminico_kadry_3cy_01_06_22_21.pdf. [Accessed 2 August 2021].

[20]

E. Cuesta, E. DiBiaso and T. Müller, “IEEE 802.3cy Return Loss Limit Proposal,” 22 June 2021. [Online]. Available: https://www.ieee802.org/3/cy/public/jul21/CuestaDiBiasoMuller_3cy_01_06 _22_21.pdf. [Accessed 2 August 2021].

[21]

Cablefax, “Broadband: What is a Micro-reflection,” 1 November 2009. [Online]. Available: https:// www.cablefax.com/archives/broadband-what-is-a-micro-reflection. [Accessed 4 August 2021].

[22]

R. Jonsson and R. Farjadrad, “Method for Restricting Micro-reflections,” 14 October 2020. [Online]. Available: https://grouper.ieee.org/groups/802/3/cy/public/adhoc/jonsson_3cy_01a_10_ 14_20.pdf. [Accessed 4 August 2021].

[23]

C. DiMinico, “802.3bp Cabling Parameters to S-parameter Naming,” September 2014. [Online]. Available: https://www.ieee802.org/3/RTPGE/email/pdfDOLXj6Te3J.pdf. [Accessed 14 May 2020].

[24]

K. Matheus and T. Königseder, Automotive Ethernet, Cambridge: Cambridge University Press, 2021.

[25]

B. Mund and C. Pfeiler, “Balunless Measurement of Coupling Attenuation of Screened Balanced Calbels up to 2 GHz,” 6–8 October 2015. [Online]. Available: http://www.bmund.de/pdf-03/ IWCS_2015-Balunless_measurement_of%20coupling_attenuation.pdf. [Accessed 8 January 2020].

[26]

E. DiBiaso and B. Bergner, “Media Considerations - Insertion Loss and EMC,” May 2017. [Online]. Available: https://www.ieee802.org/3/ch/public/may17/DiBiaso_3NGAUTO_01_0517.pdf. [Accessed 15 January 2020].

[27]

IEEE Computer Society 1, “802.3ch-2020 - IEEE Standard for Ethernet Amendment: Physical ­Layer Specifications and Management Parameters for 2.5 Gb/s, 5 Gb/s, and 10 Gb/s Automotive Electrical Ethernet,” IEEE-SA, New York, 2020.

[28]

Automotive SerDes Alliance, “ASA Transceiver Specification Version 1.01,” Automotive SerDes Alliance, Munich, 2020.

[29]

T. Williams, EMC for Product Designers, Oxford: Newnes, 2016.

[30]

Wikipedia, “Pulse-width Modulation,” 3 August 2021. [Online]. Available: https://en.wikipedia. org/wiki/Pulse-width_modulation. [Accessed 12 August 2021].

[31]

UN/ECE, “Regulation No 10 of the Economic Commission for Europe of the United Nations (UN/ ECE) — Uniform Provisions Concerning the Approval of Vehicles with Regard to Electromagnetic Compatibility,” 26 July 2012. [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/ HTML/?uri=CELEX:42012X0920(01)&from=DE. [Accessed 2 July 2021].

[32]

German car makers (Volkswagen version), “LV124: Elektrische und elektronische Komponenten in Kraftfahrzeugen bis 3,5 t, allgemeine Anforderungen, Prüfbedingungen und Prüfungen,” June 2013. [Online]. Available: https://docplayer.org/27074440-Elektrische-und-elektronischekomponenten-in-kraftfahrzeugen-bis-3-5-t-allgemeine-anforderungen-pruefbedingungen-und-­ pruefungen.html. [Accessed 11 July 2021].

[33]

German car makers (BMW version), “LV124: Elektrische und elektronische Komponenten in Kraft­fahrzeugen bis 3,5 t, allgemeine Anforderungen, Prüfbedingungen und Prüfungen,” J­ anuary 2010. [Online]. Available: https://www.doc88.com/p-819912711273.html. [Accessed 11 July 2021].

5.5 Bibliography

[34]

J. Trulove, LAN Wiring, Third Edition, New York City: McGraw-Hill Professional, 2005.

[35]

Wikipedia, “TIA-568A/B,” 8 May 2020. [Online]. Available: https://de.wikipedia.org/wiki/TIA568A/B. [Accessed 5 April 2021].

[36]

ISO, “ISO/IEC 11801 2017: Information Technology – Generic Cabling for Customer Premises Part 1-6,” ISO, Geneva, 2017.

[37]

J. Rech, “7.2 Cable Types,” in: Ethernet, Heise, 2007, p. 550.

[38]

S. Carlson, T. Hogenmüller, K. Matheus, T. Streichert, D. Pannell and A. Abaye, “Reduced Twisted Pair Gigabit Ethernet Call For Interest,” 15 March 2012. [Online]. Available: https://www. ieee802.org/3/RTPGE/public/mar12/CFI_01_0312.pdf. [Accessed 6 May 2020].

[39]

T. Hogenmüller, H. Zinner and S. Buntz, “Tutorial for Lifetime Requirements and Physical ­Testing of Automotive Electronic Control Units (ECUs),” July 2012. [Online]. Available: https:// grouper.ieee.org/groups/802/3/RTPGE/public/july12/hoganmuller_01a_0712.pdf. [Accessed 21 February 2021].

[40]

K. Matheus, M. Kaindl, S. Korzin, D. Goncalvez, J. Leslie, Y. Okuno, N. Kitajima, M. Gardner, R. Orosz, M. Jaenecke, D. Kim, R. Mei and D. v. Knorre, “1 Pair or 2 Pairs for RTPGE: Impact on System Other than the PHY Part 1: Weight & Space,” January 2013. [Online]. Available: https:// www.ieee802.org/3/bp/public/jan13/matheus_3bp_01_0113.pdf. [Accessed 16 May 2021].

[41]

K. Matheus et al, “1 Pair or 2 Pairs for RTPGE: Impact on System Other than the PHY Part 2: Relative Costs,” May 2013. [Online]. Available: https://www.ieee802.org/3/bp/public/may13/ matheus_3bp_01a_0513.pdf. [Accessed 16 May 2021].

[42]

ASTM International, “Standard Specification for Standard Nominal Diameters and Cross-Sectional Areas of AWG Sizes of Solid Round Wires Used as Electrical Conductors,” ASTM International, West Conshohocken, PA, 2014.

[43]

B. Körber, S. Buntz, M. Kaindl, D. Hartmann and J. Wülfing, “BroadR-Reach® Physical Layer Definitions for Communication Channel, v1.0,” OPEN Alliance, Irvine, CA, 2013.

[44]

Leoni, “Leoni Automotive Cables,” not known. [Online]. Available: https://publications.leoni.com /fileadmin/automotive_cables/publications/catalogues/leoni_fahrzeugleitungen.pdf?. [Accessed 02 April 2021].

[45]

Wikipedia, “Permitivity; PS,” 25 February 2021. [Online]. Available: https://de.wikipedia.org/ wiki/Permittivit%C3%A4t. [Accessed 02 April 2021].

[46]

Valens, “Valens and Daimler Partner to Optimize In-car Connectivity,” 7 November 2016. [Online]. Available: https://www.valens.com/press-releases/valens-and-daimler-partner-to-optimize-incar-connectivity. [Accessed 2 August 2021].

[47]

L. Cohen and R. Shirani, “Initial RF Ingress Measurements for Coaxial and UTP Cables from Automotive BCI Test,” May 2017. [Online]. Available: https://www.ieee802.org/3/NGAUTO/pub lic/may17/cohen_shirani_3ch_01_0517.pdf. [Accessed 2 August 2021].

[48]

B. Körber, B. Bergner, D. Marinac, D. Dorner, F. Bauer, H. Patel, J. Razafiarivelo, M. Dörndl, M. Kaindl, M. Rucks, M. Nikfal, P. Gowravajhala, T. Müller, T. Wunderlich, V. Raman, W. Mir and Y. Bouri, “Channel and Component Requirements for 1000BASE-T1 Link Segment Type A (STP), v1.0,” OPEN Alliance, Irvine, CA, 2019.

[49]

T. Müller, “Automotive STP and SDP Cable Measurement Results,” 1 December 2020. [Online]. Available: https://www.ieee802.org/3/cy/public/adhoc/mueller_3cy_01_12_01_20.pdf. [Accessed 13 June 2021].

[50]

T. Müller, Interviewee, SDP or SPP. [Interview]. 14 July 2021.

161

162

5 The Automotive Channel

[51]

R. M. Metcalfe, “The History of Ethernet,” 14 December 2006. [Online]. Available: https://www. youtube.com/watch?v=g5MezxMcRmk. [Accessed 6 May 2020].

[52]

Microwaves101, “Why Fifty Ohms?,” not known. [Online]. Available: https://www.microwaves101. com/encyclopedias/why-fifty-ohms. [Accessed 17 May 2021].

[53]

Rosenberger, “Rosenberger Cable Groups,” Rosenberger, not known. [Online]. Available: https:// www.rosenberger.com/0_documents/de/codes/codes_cablegroups.pdf. [Accessed 08 April 2021].

[54]

Moris, “Shielded versus Unshielded CAT6a: How to Choose?,” FS community, 2 November 2016. [Online]. Available: https://community.fs.com/blog/shielded-or-unshielded-which-to-choose-for-cat6a-cabling.html. [Accessed 3 August 2021].

[55]

Wikipedia, “USB,” 3 August 2021. [Online]. Available: https://en.wikipedia.org/wiki/USB. [Accessed 3 August 2021].

[56]

L. Davis, “USB 3.0 Cable Interface,” 7 January 2012. [Online]. Available: http://www.interfacebus. com/usb-cable-diagram-30.html. [Accessed 5 April 2021].

[57]

KroSchu – Krombert & Schubert, “Coax Cable,” KroSchu-Cable, not known. [Online]. Available: https://www.kroschu-cable.de/de/automotive/multimedia/koax-kabel.html. [Accessed 5 April 2021].

[58]

D. Wiechoczek, “Wie hängt die Geschwindigkeitskonstante von der Temperatur ab?,” 29 November 2004. [Online]. Available: https://www.chemieunterricht.de/dc2/rk/rk-arrhe.htm. [Accessed 1 August 2021].

[59]

Rosenberger, “Rosenberger HSD High Speed Data Connections,” not known. [Online]. Available: https://www.rosenberger.com/us_en/automotive/hsd. [Accessed 15 April 2021, no longer available].

[60]

Rosenberger, “Auto HSD Technical Reference,” [Online]. Available: https://www.rosenberger. com/0_documents/de/catalogs/ba_automotive/AUTO_HSD_TechnicalReferences.pdf. [Accessed 15 April 2021].

[61]

DATACOM Buchverlag, “HSD (High Speed Data),” 2021. [Online]. Available: https://www.it wissen.info/HSD-high-speed-data-HSD-Stecker.html. [Accessed 15 April 2021].

[62]

M. Arpe and C. Ensley, “The Future of Automotive Data Connectivity,” 2021. [Online]. Available: https://www.aptiv.com/docs/default-source/white-papers/2021_aptiv_whitepaper_futureofauto motivedataconnectivity.pdf?Status=Master&sfvrsn=343df53d_3. [Accessed 1 August 2021].

[63]

Rosenberger, “H-MTD Modular Twisted-Pair Data,” not known. [Online]. Available: https://www. rosenberger.com/de/produkt/h-mtd/. [Accessed 15 April 2021].

[64]

Wikipedia, “Koaxialstecker,” 12 April 2021. [Online]. Available: https://de.wikipedia.org/wiki/ Koaxialstecker. [Accessed 14 April 2021].

[65]

Rosenberger, “HFM, High-speed-Fakra-Mini,” not known. [Online]. Available: https://www.rosen berger.com/product/hfm-high-speed-fakra-mini/. [Accessed 14 April 2021].

[66]

A. Zierler, “Dual-Sourcing-Vertrag für Automotive,” elektroniknet, 15 May 2018. [Online]. Available: https://www.elektroniknet.de/e-mechanik-passive/dual-sourcing-vertrag-fuer-automotive.153 594.html. [Accessed 1 August 2021].

[67]

IMS Connector Systems, “MCA – Mini Coax Automotive – Miniaturized Multipole RF- & High Speed Data Solution,” 2018. [Online]. Available: https://www.imscs.com/en/mca-mini-coax-auto motive/?cn-reloaded=1. [Accessed 1 August 2021].

[68]

M. J. Teener, “What is FireWire?,” 23 March 2015. [Online]. Available: https://www.johasteener. com/what-is-firewire.html. [Accessed 31 July 2021].

5.5 Bibliography

[69]

Wikipedia, “Skin-effect,” 17 July 2021. [Online]. Available: https://en.wikipedia.org/wiki/Skin_­ effect. [Accessed 31 July 2021].

[70]

Wikipedia, “Speed of Electricity,” 29 July 2021. [Online]. Available: https://en.wikipedia.org/ wiki/Speed_of_electricity. [Accessed 31 July 2021].

[71]

Sierra Circuits, “High-Speed PCB Design Guide,” 2020. [Online]. Available: https://pages.protoexpress.com/rs/727-TSC-367/images/High-Speed%20PCB%20Design%20Guide.pdf. [Accessed 11 April 2021].

[72]

Wikipedia, “FR-4,” 20 March 2021. [Online]. Available: https://en.wikipedia.org/wiki/FR-4. [Accessed 11 April 2021].

[73]

Wikipedia, “Leiterplatte,” 16 March 2021. [Online]. Available: https://de.wikipedia.org/wiki/­ Leiterplatte. [Accessed 11 April 2021].

[74]

Altium, “Stripline versus Microstrip,” not known. [Online]. Available: https://medium.com/ @Altium/stripline-vs-microstrip-understanding-their-differences-and-their-pcb-routing-guidelines9bad77303d2f. [Accessed 18 April 2021, no longer available].

[75]

H. Kadry, “PCB Insertion Loss Material Comparison,” 1 March 2021. [Online]. Available: https:// www.ieee802.org/3/cy/public/adhoc/Kadry_3cy_01a_03_01_21.pdf. [Accessed 25 June 2021].

163

6

Power

Modern cars contain a significant amount of electronics, without which we would neither have the comfort and safety available in cars today, nor would we envision any vehicular innovations for the future. A prerequisite for using electronics in cars is that they are powered with a stable and reliable power supply. This needs to be ensured. At the same time, the power consumption of the electronics adds to the energy consumption of a car. Both customers and legislation are becoming increasingly aware of this, and car manufacturers are interested in keeping the power consumption of their vehicles low; the growing market for electric vehicles enhancing this effect. In principle, car manufacturers have already been successful in reducing the energy consumption and CO2 emissions per vehicle. While, for example, the number of worldwide registered cars about doubled between 1991 and 2012, the overall CO2 emissions of road vehicles ­increased by “only” 66% in that same timeframe [1] [2]. In Germany, the average fuel c­ onsumption per car was reduced by about 20% [3] and that considering more powerful engines and faster driving cars. Despite these successes, it is essential to scrutinize all elements in cars that add to the energy needed and to continue the effort to reduce the power consumption of cars as much as possible. That includes sensor and display applications as well as their communication technologies. In general, there are three factors car manufacturers can influence with respect to the power consumption of their cars: The (type of) engine, the (amount and type of) electronics, and the overall weight of a car. For sensor and display applications, their power consumption as well as their weight come into play. This chapter thereby cannot discuss the power consumption of particular semiconductors or semiconductor technologies; it is up to the semiconductor vendor to offer power optimized products. This chapter discusses structural possibilities of power savings. Section 6.1 investigates the possibilities to transmit power with the data communication instead of using separate wires. This saves weight in the wiring harness and is a worthy consideration. After all, the wiring harness is the third heaviest component inside cars, after chassis and engine [4]. Section 6.2 looks at different operation and power modes. Having low-power or sleep modes available for units currently not in use, is essential. This Chapter 6 discusses the general power related considerations, independent from any particular technology used if possible. (Most) Communication technology specific details are addressed with the respective technologies in Chapters 7 and 8.

6.1 Supplying Power with the ­Communication

„ 6.1 Supplying Power with the ­Communication Supplying power via the communication cable does not only have the benefit of saving ­separate power cables. It also saves space and costs in the connectors, because neither ­extra pins nor a separate connector for the power supply are needed. However, “power-­over” is not always possible. The following subsections address this in detail, providing an overview on the relevant aspects to consider. First, not all communication technologies allow for the transmission of data and power on the same cable. Then, the power that can sensibly be supplied with the communication has limits that might be below what is needed. Finally, some additional circuitry is necessary on the PCBs to couple the power onto the communication wire in the unit that supplies the power, which is called “Power Supply/Sourcing Equipment (PSE)” in this context, and to separatse it again in the unit that receives the power, called “Powered Device (PD)”. This adds costs and requires space on the PCB, which is generally quite limited in small sensors. At the same time, sensors are often at the edge of a car and require long wires, which means power-over saves more cables. Whether to use power-over or not thus requires careful consideration for each individual case. This Section 6.1 provides the technical background and design considerations for such decisions. Section 6.1.1, first describes the nomenclature and relevant elements needed when transmitting power with data. Section 6.1.2 then discusses Power over Differential (PoD) cables based on the existing Power over DataLine (PoDL) standard IEEE 802.3bu for Automotive Ethernet. Section 6.1.3 discusses Power over Coaxial (PoC) cables as usable with the Automotive SerDes standards, with potentially single-ended Automotive Ethernet solutions in the future, and with many of the proprietary Automotive SerDes technologies today. This subsection shows, among other aspects, that the concrete power-over solution depends on the specific characteristics of the communication technology selected. Note that to transmit power with the data communication is not at all a new concept but has been known for a long time. One example is the so-called phantom power for microphones, where the DC is superimposed on the microphone’s audio cable [5]. Another example is a specific version of TV satellite antenna systems, where a Low Noise Block (LNB) converter at the antenna is controlled from the TV equipment. The LNB selects the polarization of the antenna depending on whether it receives 14 or 18 V via the coaxial antenna cable [6]. IEEE ratified its first specification that allowed to transmit power with Ethernet in 2003 [7]. It is called Power over Ethernet (PoE), a name intended to be used solely when power is transmitted with 2-pair Ethernet versions like 100BASE-TX. In addition to the single pair PoDL discussed in Section 6.1.2, IEEE also publishes specifications for power-over 4-pair Ethernet under the name of 4-Pair Power over Ethernet (4PPoE), for example, in IEEE 802.3bt [7] [8]. A last general remark: Power-over can always be used as a simple means to switch the powered units on or off (see also Section 6.2.2). Many rear-view cameras, for example, are only utilized when the reverse gear is selected. Activating the power of the camera only when data communication is needed is an obvious choice. However, this does not necessarily require power-over. It can also be done when the power of the satellite camera is controlled using separate wires. When power-over is available, it simply means that the possibility to switch the camera on and off by the power supply is built in.

165

166

6 Power

6.1.1 General Considerations for Power-over The key requirement for power-over is that it is possible to first mix and later separate power and data transmission, without the two impacting each other. For this to be possible, first, the frequency of the communication system has to be sufficiently high, and second, it needs to be possible to AC-(de)couple the data signal. This in return means that the communication mechanism cannot rely on DC content. The communication system Controller Area Network (CAN), for example, cannot realize power-over, because CAN asymmetrically relies on GND connectivity and is not free of DC contents, despite its differential transmission. In principle, the AC-coupling can be capacitive or magnetic via a transformer with two separated coils. Figure 6.1 shows the main elements of a power-over system. The unit that supplies the power is seen on the left. It is called Power Sourcing Equipment (PSE) in this context [9]. On the right side is the unit that receives the power. It is called Powered Device (PD). In a typical scenario of the use cases this book focusses on, the PSE is the ECU that processes the video or sensor data, and the PD is the sensor, but it does not have to be like this. The PD might be any (peripheral) ECU and it can also be a display, provided its power consumption is small enough. When the following text speaks of the PD as a sensor, it does not preclude that the PD might be another device with a different function. Next to a suitable communication system, power-over has two essential elements: first, the Bias-T, which is the component that mixes and separates power and data onto and from the communication wire, and second, a suitable voltage regulator, especially in the PSE that supplies the power. The following subsections address these in detail: Section 6.1.1.1 introduces the Bias-T, Section 6.1.1.2 discussed relevant considerations for the power supply, and Section 6.1.1.3 lists some general requirements designers of power-over systems need to consider. Specific details that depend on the type of cables selected, STP or coaxial, are discussed with Section 6.1.2 and Section 6.1.3 respectively. GND +Ubat PSE (ECU) Iin

Voltage regulator for ECU

Uin = Ubat

Voltage regulator for power-over

PD (ECU/sensor/display)

Iout=I

I

Uout = UPSE

Power

Power

Communi- Data cation chip P2

Bias-T

Cs

IRL URL

P3

P3 Application chip

UPD

Voltage regulator

Data & power P1

P1

Bias-T

Data Communication chip P2

Application chip

Figure 6.1 Main elements of a power-over realization

6.1.1.1 The Bias-T The Bias-T is the component on a PCB, which allows to join the power and (high-frequency) data stream onto a common channel at the unit that supplies the power. Figure 6.2 shows its principle setup. The Bias-T is a three-port component with the ports P1, P2, and P3 that

6.1 Supplying Power with the ­Communication

represents a technical nomenclature for a combination of inductive and capacitive e­ lements, L and C respectively. As shown in Figure 6.1 it is used in the PSE to mix power and data as well as in the PD to separate them. Power P3

Bias-T HF-data rejection

DC-power rejection Data

P2

L

C

Power path P1

Channel

Data path

Figure 6.2  The Bias-T as the basic element for coupling power onto a data line

The Bias-T makes use of the fact that inductances allow a DC current to pass with very small DC resistances, while blocking high-frequency (HF) currents with a very high impedance. At the same time capacitors do exactly the opposite. They block DC currents and let HF currents pass. Therefore, as is shown in Figure 6.2, HF data transmission is coupled onto and out of the transmission channel with the capacitor C, while C blocks the DC ­current. The DC current is coupled onto and out of the channel via the inductance L, which in return blocks the HF data. Figure 6.2 shows the basic physical principle behind transmitting power with data. The details of the exact circuitry, however, can vary significantly, depending on aspects like the cable, the Nyquist frequency, transmission mode(s), power regulator selected, etc. The specifics when using (S)TP or Coaxial cables are described in Section 6.1.2 and Section 6.1.3 respectively. The size and cost of inductors used in the Bias-Ts are a relevant design consideration. Section 6.1.3 provides some examples.

6.1.1.2 Voltage Regulators When realizing power-over, the voltage regulator in the PSE that supplies the power to the PD is essential. Naturally, also the PD has a voltage regulator. However, while there are some interdependencies with the voltage regulator in the PSE – especially with respect to the voltage level supplied, see also Section 6.1.3.4 and Section 6.1.3.5 – there is no prin­ ciple difference, whether the PD’s power is supplied from the PSE or whether it comes directly from the battery. The selection criteria for the voltage regulator in the PD are therefore the standard selection criteria in which the PD needs to be optimized for the PD’s application. This is different for the voltage regulator in the PSE that supplies the power for the PD. It needs to be optimized in accordance with a distant PD (that might have been developed by a different Tier 1 supplier). The following discussion is therefore targeted on the requirements of the voltage regulator in the PSE. It distinguishes between linear and DC-DC voltage regulators, and for the DC-DC voltage regulators between “step-down”, “step-up”, and “buck-boost”/Single-Ended Primary-Inductor Converters (SEPICs). Linear voltage regulators are an additional resistive load. They have a high-power dissipation, and therefore a comparably poor efficiency. When the input voltage Uin drops at the entrance of the linear voltage regulator – not uncommon in cars, especially in case of load

167

168

6 Power

changes – the output power, consisting of current Uout and Iout, (see Figure 6.1), drops as well. For a camera PD, a linear regulator might still be preferred, because the linearly regulated output has no DC-DC switching and does not interfere with the physics of the image sensors. For the PSE’s power-over supply, the loss and dynamically changing power dissipation of a l­inear voltage regulator is, however, not acceptable. DC-DC step-down converters change a higher input voltage Uin into a lower output voltage Uout. It is essential that Uin is always larger than Uout, which means that a minimum Uin_min needs to be guaranteed all the time. It is therefore not advisable to use a step-down DC-DC regulator in the PSE; again because of the potentially varying battery supply Ubat. Ubat might occasionally reach values below Uin_min. Inside a PD, this is less of a problem, and it might well use a step-down converter. After all, the PD gets a well-regulated power from the PSE (which is another advantage of power-over). DC-DC step-up converters change a lower input voltage Uin into a higher output voltage Uout. In this case Uout is always at least as large as Uin . A sudden drop in battery supply would thus not be an issue. If, however, for whatever reason the battery supply voltage increases beyond the intended Uout_nom, Uout increases beyond this value, too (see also Figure 6.16). DC-DC voltage regulator types that are better suited for the requirements in the PSE are “buck-boost”, or SEPIC. For these types, the input voltage can be lower or higher than the intended output voltage and they operate over a wide range of voltages. If used with the right parameters (see Section 6.1.3.4 for details), these converters keep the output voltage Uout constant at the nominal value, even at quite low or high input voltages. When selecting the DC-DC voltage regulator, the following impacts need to be taken into consideration: ƒ The size of the input current Iin: When the PD load stays constant, constant output voltage and power mean that also the current Iout stays constant, which in any case will be limited at the PSE using a fuse or active current limitation circuitry. So even when the input voltage Uin = Ubat changes, the power drawn will be the same. Therefore, the input current Iin increases when Uin = Ubat decreases. All components in the power path from the battery into the voltage regulator must be able to handle this increased current. This is a relevant design aspect in the PSE! Table 6.1 shows a simple example of the difference this can make for the current assuming an ideal step-up regulator with 100% efficiency. Between the highest and lowest voltage that might realistically occur for a nominal battery voltage of Ubat = 12 V, the current increases by more than factor four to the nominal value. Table 6.1 Example for possible input current values Iin . Uout = 10 V, Iout = 0.1 A, both are ­constant Ubat = Uin

16 V

12 V

10 V

8V

6V

3V

Iin

0.0625 A

0.083 A

0.1 A

0.125 A

0.167 A

0.33 A

ƒ Interrelations between operating frequencies: DC-DC converters work with switching frequencies in the range of 50 kHz up to 2 MHz. For operation, they apply a digital control loop working either with the switching frequency or a fraction of the switching frequency. It can happen, that the load of the PD is not constant but changes regularly

6.1 Supplying Power with the ­Communication

at a certain frequency, for example, because a camera imager does not need power during the blanking period (see also Section 2.1.2). In this case, it is important that the frequency of the control loop is selected such that it is noticeably different from the frequency of the load changes. Else, the output voltage of the DC-DC converter in the PSE might become instable. In such a scenario, the right frequency of the control loop is more important than the current. This also applies to the combination of voltage regulators, the one supplying the power-­ over in the PSE and the one in the PD. If both are DC-DC converters that readjust at an unfortunate combination of frequencies, they both might not provide the right output values. Using a linear voltage converter in the PD avoids this issue. ƒ Switching noise caused by the DC-DC conversion: Section 5.2.3.4 introduced the concept of power ripples that can cause interference in a communication system from the power supply. Such ripples may originate in the DC-DC converter used for the power-­ over supply in the PSE, in the voltage regulator in the PD (if here also a DC-DC converter is used), or in load changes of the PD itself (for example because of the already mentioned blanking periods in imagers). If the ripple originates in a DC-DC converter it is sometimes referenced as switching noise. The higher the power loss during the transmission on the channel and in the Bias-Ts of PSE and PD (see also Figure 6.11), the more prominent the effect of the ripple. The switching noise of DC-DC converters can be observed at harmonics of the switching frequency. They have an effect especially on the low-frequency control/status channel of Frequency Division Duplex (FDD)-based SerDes systems. The FDD-based SerDes high data rate channel, Time Division Duplex (TDD)-based SerDes, and HS Ethernet systems work at higher frequencies and therefore the impact is minor. For the proprietary (FDD-based) SerDes technologies, the tolerated maximum ripple values are part of the application notes for the specific SerDes product. A typical value for the maximum amplitude of the ripple is Vripple = 100 mV peak-to-peak. Note that the ripple noise potentially increases, when the DC-DC converter operates either at a very low load or at a very high load. At a very low load the pulses to charge the output circuitry are short. In this case the timing of the edges, the ratio dV/dt, creates noise. At a very high load, the DC resistance along the power-over channel intensifies the ripple. This form of ripple is the most common effect. ƒ High inrush currents at power on: Using a reasonably sized capacitor at the input of the voltage regulator of the PD limits the amount of voltage ripple. It is a necessary component (see the “Cs” in the PD part of Figure 6.1) that, however, also has a downside: The capacitor needed for a stable operation, can easily reach the value from a few nF up to several µF, depending on the voltage regulator selected. During power-on, the capacitor is discharged and charges when the voltage rises. Even a small capacitance value can cause a large inrush current during the power-on. The maximum value of the current pulse depends on the DC resistance on the power-over channel and the Equivalent Series Resistance (ESR) of the capacitor. Together with the ESR, the capacitance value of the capacitor determines the duration of the current pulse. This effect must be considered in the design and selection of the components, mainly the inductors in the Bias-T. The effect of the inrush current must be handled in an appropriate form in the system design too, for example by implementing a timeout for failure detection at power-on.

169

170

6 Power

ƒ Ringing effects: So far, only the resistive effect of the channel during power-on was discussed. As the channel itself is also an inductor, together with the capacitive load, an uncontrolled power-on might cause a certain ringing. One method for handling the ringing effects of the inrush current is to implement an additional current limitation circuitry in the PSE and ramp up the voltage and current during power-on in a defined way, staying always in the defined limits of the circuitry. A short channel is more critical than a long channel for the current peak, as the DC resistance of the channel is a main factor for the inrush current. A long channel represents a bigger inductivity and is more critical to ringing effects. This Section 6.1.1.2 listed some general effects that have to be taken into consideration with the voltage regulators in a power-over system, especially in the PSE. Section 6.1.3 provides more concrete examples for the consequences in a PoC system.

6.1.1.3 Failure Detection and Protection When realizing power-over, the system must provide several mechanisms that detect whether the system is functioning correctly and prevent certain events. The power system can do so based on DC-parameters (only). An error-free state of the power system does therefore not allow to draw conclusions on the quality of the data communication. However, when the power supply is seriously impaired, it is possible to deduct that the data communication is likely not functioning correctly either. Designers of ECUs with power-over should set several limit lines and parameters for the power supply elements in the PSE and PD. As a basis, the nominal voltage, current, and power range needs to be defined. Limits for maximum temperature, maximum current, and maximum voltage, which may not be exceeded, are needed in addition. Ideally the implementation provides protection mechanisms that prevent (longer) over-temperature, over-current, and over-voltage events and set respective error flags. Typically, also a minimum voltage is needed at the input of the voltage regulators. If the actual voltage falls below in an under-voltage event, proper operation of either the complete power-over system or the PD cannot be guaranteed (see Section 6.1.3.4 for examples). In the latter case, the PSE needs to be notified not only to ensure stable power again but also to restore the configuration of the PD once the voltage is back up. However, without proper voltage a PD might not be in the position to store and notify the PSE about the failure event. Designers of the system need to consider this. A restart of the PD following an under-voltage event as well as a voltage drop that was noticed but too short to cause a restart, should also be recorded. Opens and shorts normally happen in the external cabling. Open-load and short-load are registered in the PSE only. An open-load detection needs to be deactivated, when the PD is in a low power mode. A short-load detection is very similar to detecting that an over-current and the system should have a similar mechanism protecting it from damage from shorts. In most cases, failure detection mechanisms are implemented with timeouts and controlled by timers in order to support a stable power-on or failure recovery procedures. Typically, timers are in place for the detection of over-temperature, over-current, over-voltage, and PSE restart events. Because the PD is not directly connected to the power supply system of the car, it does not have to fulfill most of the requirements resulting from the battery connection itself. For

171

6.1 Supplying Power with the ­Communication

e­ xample, the PD does not need Reverse Battery Protection (RBP) against reverse polarity in the power system. The PSE needs to support this. PD and the PSE together have to ensure they use the correct pins in their connection.

6.1.2 Power over Differential (PoD) Cables Figure 6.3 shows two possible setups for power over a differential communication line using an STP cable to connect the PSE, generally an ECU, with the PD, which might also be an ECU, but is more likely a sensor. Note that power transmission over UTP instead of STP ­cables uses the same principal setup. Because we expect the use of shielded cables for the high-speed use cases discussed in this book, this section focusses on STP cables. We would like to point out though that most aspects are equally applicable for UTP cables. Figure 6.3 also shows the key element for the data transmission: Because differential transmission uses two wires, two Bias-T circuits are needed per unit. GND +Ubat 3

PSE (ECU) Voltage regulator

2

1 Communication chip (PHY)

Voltage regulator UPD

UPSE

Application chip

PD (ECU/sensor)

I

I

I

Bias-T

Bias-T

I

CMC

CMC

I

Communication chip (PHY)

Application chip

I

GND +Ubat

3

PSE (ECU)

CMC

Voltage regulator

1 Communication chip (PHY)

Voltage regulator UPD

UPSE

Application chip

PD (ECU/sensor)

I

I

2

I Bias-T

Bias-T CMC

I

I

Figure 6.3 Different PoD options for differential transmission

CMC

I

Communication chip (PHY)

Application chip

172

6 Power

An element in addition to the Bias-T to take into consideration is the Common Mode Choke (CMC), which is used for reducing the impact of common mode interference on differential signals, for example, for 100BASE-T1 Automotive Ethernet communication. The main advantage of the setup shown in the top graphic of Figure 6.3 is that the current transferred from the PSE to the PD does not have to pass through the CMC and the CMC can be the same, independent of whether PoD is used or not. In the bottom part of Figure 6.3 the current of the power supply passes through the CMC. A potential advantage of this approach is that the CMC helps to improve the symmetry behavior of the power injection. As the current flows through the CMC, the CMC needs thicker wires and is therefore generally significantly larger than a CMC without PoD. This means that for such a setup generally a different CMC is used than without PoD. When transmitting power with the communication also the GND connection of the PD is provided by the PSE. This is, in principle, easier to realize with PoD and STP cables than with coaxial cables, because for STP the shield is not part of the communication path, as is the case for coaxial cabling (see also Section 4.1.4.1). Figure 6.3 shows two different approaches. These are independent of the position of the CMC and are usable with either setup. In the setup shown in the upper part of Figure 6.3 the shield of the STP cable is connected directly to the cases of the units (see “1” and “2”). In the setup shown in the lower part a capacitive coupling connects the shield to the PCB GND of the PD (see “2”). This can be done in order to avoid unintended effects of GND shifts. Such coupling is not possible for PoC systems. In case of PoC, the PCB ground cannot be separated from the shield of the cable, because if that was done the communication backchannel would be ­interrupted. In the setup in the upper part of Figure 6.3, the system ground is connected directly to the case (see “3”). This is a common approach to attenuate radiated emissions from ICs on PCB level. The setup in the lower part shows an approach mainly used for UTP applications. In area “3” a power CMC is used, which raises the common mode impedance of the ECU serving as PSE and thus the complete system. Because UTP systems cannot rely on the interference reducing effect of a shield, it is important to have a high common mode impedance in the system in order to reduce the effective power or RF noise. Both the power CMC and the CMC in the signal line raise the common mode impedance to mitigate the effects from unintended noise on the signal wire, meaning that PoD can further optimize the EME and EMI. The standard that exists for PoD is the IEEE 802.3bu Power over DataLine (PoDL) specification [10]. It was completed in 2016 and defines the power-over realization for the single pair (Automotive) Ethernet solutions 100BASE-T1 and 1000BASE-T1. The different technologies use the same PoDL setup (as shown in Figure 6.3, especially the setup shown in the lower part, see [9]) but with the electronic components selected such that they are optimized for the respective frequency of the speed grade. In the specification, these are identified with Type A for 100BASE-T1 and Type B for 1000BASE-T1. Type A+B works for both [10]. The defined power classes, see Table 6.2, are, like many other aspects, independent of the speed grade and thus useful also when considering other, higher data rates [11]. The IEEE 802.3bu specification defines ten different power classes, from 0.5 W to 50 W. In the authors’ experience, the most relevant power class for cameras using proprietary SerDes technologies with coaxial cabling was, at the time of writing, the Class 3, which supplies up to 5 W.

6.1 Supplying Power with the ­Communication

Table 6.2 Power-over classes defined in IEEE 802.3bu [10] 12 V unregu­ 12 V regulated PSE lated PSE

24 V unregu­ 24 V regulated PSE lated PSE

48 V regulated PSE

Class

0

4

8

Voltage [V]

5.5–18

1

2

3

14–18

5

12–36

6

7

26–36

9

48–60

Current [A]

0.1

0.22

0.25

0.47

0.10

0.34

0.21

0.46

0.73

1.3

PD power [W]

0.5

1

3

5

1

3

5

10

30

50

When switched on, IEEE 802.3bu defines that the PSE first tests the link. If the link is as expected, the PSE can either directly inject the power (provided it knows upfront which PD with what power requirement is connected) in a so-called “fast start-up”, or it can use the optional Serial Communication Classification Protocol (SCCP) to negotiate the power classes between PSE and PD, which is useful in a plug&play scenario. The SCCP is a low speed (333 bps) self-powered, bidirectional protocol based on the Maxim 1-wire serial protocol [12]. The SCCP also supports some simple control functions and error reporting. In the predefined automotive scenarios, it is unlikely that the power class will be negotiated. Especially for rear-view cameras, stringent start-up requirements exists: It is expected that a camera image is displayed within two seconds after power-on [13]. In consequence, time consuming negotiations are counterproductive. Furthermore, an in-vehicle sensor system often relies on other control formats like I2C or SPI anyway (see also Section 9.5) and it is likely that these will also be used for controlling the power-over.

6.1.3 Power over Coaxial Cables It was stated at the beginning of this Section 6.1 that power-over has limitations, for example in the power it can supply. Many nodes in the in-vehicle network will need more power than can be supplied via the data communication. Also, they might be located so close to the power supply anyway, that the saving in cable does not compensate for the added cost and weight in components. That power-over is used in the core in-vehicle network is not so likely. Power-over is particularly beneficial for (camera) sensors that are at the edge of the car and require a) limited power and b) long wires. At the time of writing, some of these cameras used compression and 100 Mbps Ethernet as the communication technology. The PoDL, as discussed in the previous Section 6.1.2, was developed with this application in mind. Cameras without compression used, at the time of writing, one of the many proprietary Automotive SerDes technologies. For these, the PoC capability is an essential feature. Without PoC, a SerDes camera had to either use STQ connectors, or STP or coaxial connectors plus an additional connector for the power supply. It is obvious that two connectors or a hybrid connector require more space (on the space limited sensor) and costs than one simpler connector. In case of a single connector, coaxial is typically less costly than STP or STQ. The savings realized with PoC made a difference for SerDes cameras in the competition with 100 Mbps Ethernet c­ ameras using compression. Because PoC is thus such an important element of the SerDes communi­ cation/camera sensor connectivity – potentially also for future, single-ended ­Ethernet ver-

173

174

6 Power

sions – its use is discussed in more detail in the following. Many aspects described in the following are, however, also applicable in case of PoD. Section 6.1.3.1, first, looks at the inductive elements, which are the basis of the PoC system and responsible for its properties as well as many of its limitations. These limits might be  the power limit, which is discussed in Section 6.1.3.3, or the stability of the system, which is discussed in Section 6.1.3.4. Section 6.1.3.2 provides a concrete Bias-T example and Section 6.1.3.5 a field manual on how to decide on the PoC components.

6.1.3.1 Required Inductive Elements Figure 6.4 shows the principle setup of a (SerDes) communication system using PoC. Here only one Bias-T is needed per unit. In difference to the differential system shown in Figure 6.3 the depiction in Figure 6.4 also includes a potential control path for the power-over voltage supply via I2C or SPI (or any other I/O). The processing chips for the data, which both, PSE and PD, will have in one form or other, are not included in the depiction. However, the power these processing chips require are part of the overall load of the system(s), in addition to the power the communication chips use. PSE (ECU)

PD (ECU/Sensor)

I2C, SPI I/O

POC PSE SW IF

On/Off

Voltage regulator

Comm. I2C, SPI DeserialiI/O status zer bridge control or PHY

I

I

UPSE

UPD

C

L

Bias-T I

I

L

C

Bias-T I

Voltage regulator

I2C, SPI I/O

Serializer I2C, SPI Comm. I/O bridge or status PHY control

POC PD SW IF

Figure 6.4 Principle PoC setup in a single-ended (SerDes) communication system Figure 6.4 shows the basic Bias-T element consisting of one capacitor C and one inductance L. For real SerDes implementations the Bias-T might be significantly more complex, depending on the actual features of the specific transmission technology selected (see also Section 6.1.3.2 for an example). The reason is the following: The inductance in the Bias-T has the purpose of blocking the data signal from passing into the power path. For it to do so, it needs an inductive AC resistance/impedance above 1 kΩ minimum (the higher the better) over the complete bandwidth of the data signal. Most proprietary SerDes technologies and the MIPI A-PHY SerDes standard make use of the asymmetric data rate requirement for SerDes by using FDD to separate the data streams being transmitted. The video data is transmitted at a high frequency in the GHz range in one direction and the control channel uses a significantly lower frequency in the reverse direction at the same time, so that the different frequencies of the data streams do not interfere with each other. The actual frequency of the reverse channel varies, depending on the technology, between a few 100 kHz and about 100 MHz (see also Chapter 7). Most inductive elements are not capable to provide the needed inductive resistance over the thus very large frequency range but show a comparably narrow band frequency dependent

6.1 Supplying Power with the ­Communication

0.1

Impedance [kΩ] 100 1 10 1

10 100 Frequency [MHz]

1000

0.1

0.1

Impedance [kΩ] 1 10 100

optimum. The left diagram in Figure 6.5 shows an example of the principal characteristic of a single inductive element. To be able to block data signals over the complete frequency range, different inductive elements are therefore connected in series in order to achieve a continuous, broad range AC-resistance. This is shown in the right chart in Figure 6.5. The effective inductive resistance is the sum of the shown three characteristics.

0.1

1

10 100 Frequency [MHz]

1000

Figure 6.5 Example of frequency dependent impedance of inductors in power path Figure 6.6 shows two examples of the respective Bias-T. On the left side a Bias-T with a single inductance is shown. On the right side, a Bias-T is shown that covers a broader frequency range by putting three inductances in a row. Note that the lower the frequency the inductance must block, the larger the component. Implementations with a control/status channel in the range below 10 MHz, for example using only I/O and I2C, need a Bias-T that provides enough impedance at these frequencies to separate the control/status stream from the DC power. Implementation with a control/status channel above 10 MHz, for example using 100 Mbps Ethernet data for that purpose, can shift the requirements for impedance to higher frequencies; with the effect that a smaller component with a smaller impedance value is sufficient for the Bias-T. Full-duplex Ethernet systems or TDD SerDes technologies like the ASA Motion Link have the advantage that they only use higher frequencies in the system and therefore might be designed such that the use of smaller inductors with respectively smaller impedance values is sufficient. Power

Power

L1 L

Data

L2 Power path

C

L3 PoC channel

Data path

Data

Power path

C

PoC channel Data path

Figure 6.6 Bias-T options in case of PoC suppressing a smaller (left side) or larger (right side) frequency range of HF data

175

176

6 Power

Figure 6.7 and Table 6.3 provides some examples of different inductive elements and shows their sizes as well as electrical properties. The ferrite beads are defined by their nominal impedance Znom, their parasitic resistance RDC, and maximum current Imax. The inductors are identified by their inductance L, their parasitic resistance RDC, and saturation Isat. The elements shown are exemplary components and not necessarily suitable for a PoC Bias-T. Inductors

Ferrite beads

0402

0603

0805

1206

1 mm Size „A“ 7.2 x 7.2 x 4 mm3

Size „B“ 12 x 12 x 6 mm3

Figure 6.7 Sample sizes of ferrite beads and inductors (Photo: Michael Kaindl) Table 6.3 Examples of ferrite beads and inductor properties; for size categories see Figure 6.7 Ferrite beads Size 0402

Size 0603

Size 0805

Inductors Size 1206

Znom = 220 Ω

L = 1 µH

RDC = 0.05 Ω

RDC = 0.01 Ω

Imax = 1500 mA Znom = 1000 Ω

Znom = 1000 Ω

Size “A”

Znom = 1000 Ω

Size “B”

Isat = 5.31 A Znom = 1000 Ω

L = 10 µH

L = 10 µH

RDC = 1.0 Ω

RDC = 0.5 Ω

RDC = 0.35 Ω

RDC = 0.3 Ω

RDC = 0.045 Ω

RDC = 0.018 Ω

Imax = 200 mA

Imax = 200 mA

Imax = 300 mA

Imax = 1000 mA

Isat = 2.6 A

Isat = 5 A

Znom = 1500 Ω

L = 56µH

RDC = 0.35 Ω

RDC = 0.228 Ω

Imax = 700 mA

Isat = 0.93 A

Znom = 2200 Ω RDC = 0.45 Ω Imax = 200 mA

Table 6.3 shows the variability of parameters of ferrite beads and inductors. When one parameter changes, all parameters tend to be different. There is no linear relation, but just some basic tendencies. Using a higher inductance comes with an increased DC resistance and a decreased saturation current. If more current is needed, a larger size inductor has to be selected. This size increases over-proportionally. When looking at the 10 μH examples in Table 6.3, a current increase by factor 5/2.6 ≈ 1.9 leads to a DC resistance decrease of factor

6.1 Supplying Power with the ­Communication

0.045/0.018 = 2.5 and a size increase of (12.2 × 12.2 × 6) / (7.2 × 7.2 × 4) ≈ 4.3. For about twice the current, the DC resistance about halves, but the size increases by more than the square (1.92 = 3.6). As Table 6.3 shows, ferrite beads are significantly more compact, so that their size increase is less critical than for inductances. Because they are lossy components that absorb RF ­energy in the core material, they are ideal to improve the EMC behavior of electric circuits [14]. They are also efficient for power-over circuits, albeit in an overall smaller frequency range than inductances. For PoC circuits they are typically used for a frequency range between 20 and 300 MHz [15]. For lower and higher frequencies inductances must be used instead. Note one other difference between the ferrite beads and inductances: ferrite beads are defined by the maximum current their wires support Imax. Inductances are defined by a saturation current Isat. Their frequency behavior changes significantly if the actual current is higher than 50% of their saturation current, which is why for inductances Imax ≈ 0.5 Isat. Figure 6.8 shows some examples for the typical frequency behavior of ferrite beads in comparison with inductances. Example inductance Frequency characteristic (Z)

Example Ferrite bead Frequency characteristic (Z) 300x

3x

Impedance [Ω]

Impedance [Ω]

4x

2x x

0

1

10

100

200x

100x

1000

1

10

100

1000

Frequency [MHz]

Frequency [MHz]

Frequency Characteristic (L)

Inductance [µH]

6x 4x 2x

1

10

100

Frequency [MHz]

Figure 6.8 Schematic example impedance characteristics of ferrite beads and inductances

177

178

6 Power

6.1.3.2 Bias-T Example Figure 6.9 shows an example of a current implementation of a Bias-T for an FDD-based ­SerDes technology. This setup was created to investigate the impact of the Bias-T on the electrical performance of the data path. It allows to separate the SerDes communication and power communication for test purposes. As can be seen in the schematic, the implementation of the Bias-T consists of two Ferrite Beads (FBs), two inductors and three capacitors. The data bus connector also carrying the power (at P1) is an automotive connector at the right side of the PCB. The power path (at P3) runs through the bottom left SubMiniature version A (SMA) and the data path is at the bottom right (at P2). Power P3

FB2

C3 L2

C2

L1 FB1 Data

P2

C1 P1

Figure 6.9 Example Bias-T implementation for an FDD-based Automotive SerDes solution (Photo: Michael Kaindl, source: FTZ) Figure 6.10 shows different measurements performed with the implementation introduced in Figure 6.9. They visualize the impact of the power transmitted with the data on the data path and power path Insertion Loss (IL) values. As can be seen in the left two graphics, there are, first, variations that simply depend on the DUT. For the data path, depicted in the upper two graphics, a low IL is the design goal. In the top right graphic, the IL varies depending on the current transmitted, especially for very low frequencies and in the range of 700 MHz to 2 GHz. The assumption is, that some components react with larger parasitic losses than others. The inductor active in the 1 GHz range seems to be particularly affected. The graphic at the bottom left shows the IL of the power path. For decoupling the power from the data path, a high insertion loss is required for the used bandwidth. The selected circuit works particularly well in the range between about 10 and 100 MHz.

6.1 Supplying Power with the ­Communication

One DUT with different currents

IL data path

Four different DUTs

IL power path

No load

Figure 6.10 Example measurements for different DUTs and with different loads (Source: FTZ)

6.1.3.3 Power Limit One of the most important aspects around power-over is how much power can be transmitted with the data. In principle, the elements that need to be taken into consideration are the same for a PoD or a PoC system but vary in some details. This section uses a PoC system as an example, simply because it is more common for the use cases. How much power can be transmitted primarily depends on the following three factors: a) the maximum current the connector supports b) the maximum current at which the Bias-T inductors do not saturate c) the overall power loss on the power supply path between PSE and PD Ad a) The maximum current the connectors support This depends on the exact connector used and varies with the respective model. Typical automotive connectors available at the time of writing supported 1–3 A. This means that they do not necessarily represent the limiting factor but must be selected accordingly. Ad b) The maximum current at which the Bias-T inductors do not saturate As a rule of thumb, the current during the operation should be ≤ 50% of the defined saturation current Isat of the inductor(s) used in the Bias-T, meaning Imax = 0.5 Isat. The actual value of Isat is given with the product description of the inductive elements used and varies with the inductance and the diameter of the wires used in the inductors. The larger the diameter of the wires, the higher Isat, but the larger the component. In case multiple inductors are needed (see right graphic of Figure 6.6) the smallest saturation current is decisive. Ad c) The overall power loss of the power supply path between PSE and PD The overall power loss of the power supply path depends on its DC resistance. In an ideal system, none of the elements in the power path – the cable, the connectors, the induc-

179

180

6 Power

tances in the Bias-Ts, the PCB traces – would have a noticeable DC resistance; after all they are intended for conductivity. However, in the real world this is different, and the DC resistance is one of the key parameters when designing robust PoC application. The ­major share of the DC resistance comes from the cable, RDC_cable, and the inductors of the Bias-Ts in the PSE and PD, RDC_LPSE and RDC_LPD. It is normally indeed acceptable to ­neglect the DC resistances of PCB and connectors. Figure 6.11 shows the respective model.

RDC_loss

I

PSEDC

UPSE

RDC_LPSE RDC_cable

RDC_LPD

UDC_LPSE

UDC_LPD

UDC_cable

PDDC

UPD

IL

UDC_loss

RL

URL

Figure 6.11 Main contributors to the DC resistance in the power path The cable and inductors of the Bias-Ts cause a power loss on the channel and that the voltage at the PD is reduced accordingly (see also Equation 6.1). A developer of a power-­ over system must be aware of the reduced voltage and needs to define the maximum loss on the path. What is obvious: the larger the current I needed by the load, the larger the power loss. (6.1) Table 6.4 shows two examples for DC resistance values of coaxial cables. It distinguishes between the resistance of the inner core Ri and the outer shield Ro. These values are used for the example calculation below. It is worth noticing that the cable that allows for frequent bending (identified as RG 174 in Figure 5.13) has a significantly higher DC resistance value than the coaxial cable used for static routes. Apart from its typically higher costs, the significantly higher resistance value is an important reason for using the flexible cable as little as possible in a PoC system. Table 6.4 Example values for the DC resistance of typical automotive coaxial cables [16] Type

Core Ri [mΩ/m] @20 °C

Shield Ro [mΩ/m] @20 °C

DACAR 302, 3.3 mm, low loss

~50

~20

DACAR 462/RG174, 2.8 mm, flexible with copper cladded steel core

~320

~35

The following shows the calculation for two examples. Both example cables consist of two segments connected by an inline connector. “cable1” uses different cable types for each segment. One segment, “cable11”, uses 9.5 m of the low-loss cable and the other s­ egment, “cable12”, 3 m of the flexible cable. In the other example “cable2” both channel segments

6.1 Supplying Power with the ­Communication

use the low loss cable. The DC resistances for the cables are calculated according to Equation 6.2. With this formula, RDC_cable1 = RDC_cable11 + RDC_cable12 = 0.665 Ω + 1.065 Ω = 1.73 Ω and RDC_cable2 = 0.875 Ω. (6.2) Table 6.5 shows the resulting power dissipation PDC_cable and voltage drop UDC_cable on the cable for the two example channels for current values up to the 3 A limit seen with current data connectors. 1 A current results in moderate voltage drop and power loss values. When considering providing 3 A, at least for the channel with the flexible segment, the voltage drop and the power dissipation on the channel are significant and would require a careful system design to support this. For the channel that uses the non-flexible cable only, the loss values are reduced by about 50% and thus significantly more tolerable. Note that the values were calculated for room temperature. With increasing temperature, the DC r­ esistance of the cables increases as well, again depending on the material used. For wires made of copper of copper alloys like used in the static cable (CuMg; CuSn; CuAg) RDC_cable@80 °C ≅ 1.2 × RDC_cable@20 °C and for wires made of copper cladded steel or steel as such like used in the flexible cable, RDC_cable@80 °C ≅ 1.3 × RDC_cable@20 °C. Table 6.5 Power dissipation and voltage drop values for the example channels I [A]

RDC_cable1/2 [Ω]

UDC_cable1/2 [V]

PDC_cable1/2 [W]

0.25

1.73/0.875

0.43/0.22

 0.11/0.05

0.5

1.73/0.875

0.87/0.44

 0.43/0.22

0.75

1.73/0.875

1.30/0.66

 0.97/0.49

1

1.73/0.875

1.73/0.875

 1.73/0.875

1.5

1.73/0.875

2.60/1.31

 3.89/1.97

2

1.73/0.875

3.46/1.75

 6.92/3.5

3

1.73/0.875

5.19/2.63

15.57/7.88

The most typical PD is a camera. In a camera, the image sensor is sensitive against heat, which is particularly tricky, as the size of the camera is typically very small with little room for cooling measures. Assuming a limit for power dissipation of 1 W for the complete camera, the inductive elements for the PoC supply should be selected with a maximum power dissipation of 200 mW as design goal. Table 6.6 shows respective DC resistance, RDC_LPD_max, and voltage drop values, UDC_LPD_max. It is noticeable, first of all, that the voltage drop of the cable UDC_cable (see Table 6.5) is larger than the voltage drop of the ­inductance of the PD UDC_LPD, at least from a current of 0.5 A on. The current influences the relation to the power of two and thus has significant impact.

181

182

6 Power

Table 6.6 Voltage drop and DC resistance values for the inductance in the example camera PD I [A]

PDC_LPD_max [W]

RDC_LPD_max [Ω]

UDC_LPD [V]

0.25

0.2

3.2

0.8

0.5

0.2

0.8

0.4

0.75

0.2

0.356

0.27

1

0.2

0.2

0.2

1.5

0.2

0.09

0.13

2

0.2

0.05

0.1

3

0.2

0.022

0.07

The most important part is therefore to select an inductance for the PD that meets the targeted current and power dissipation values. The thicker the wire for the inductance, the lower its DC resistance but the larger and more expensive it gets and the more space it needs in the space limited camera (see Table 6.3 for examples). Table 6.6 shows that for larger currents, these low RDC_LPD values are needed, especially when considering that several inductances might have to be used in a row. Note that the PSE does not have the same constraints as a PD camera. The thermal conditions are generally not as stringent in the PSE as for the PD. The PSE typically allows for a somewhat larger loss PDC_PSE. Still, the BIAS-T in the PSE also adds to the overall constraints in the system.

6.1.3.4 PoC Stability Any power-over system needs to observe some stability aspects, which again depend also on the DC resistance of the channel and Bias-T components [17]; in the following summarized as RDC_loss (see also Figure 6.11 and Equation 6.3). The key stability criterion is the voltage at the load UPD in relation to the supplied voltage UPSE. The load of the PD has to be significantly smaller than the maximum load PPD  UPSE_min 16 14 UPSE_min [V]

12 10 8 6 4 2 0

0

PPD

1

{

2

3

4

RDC_loss [] 1W

2W

3W

4W

5W

6W

7W

8W

9W

10 W

5

Figure 6.12  UPSE needs to be ­significantly larger than UPSE_min in order to ensure a stable power over system.

The description up to here explained the theory of the interrelation between the voltage regulators in the PSE and the PD. In the following, practical examples are given for different types of voltage regulators, just looking at the PD side. The test setup is depicted in Figure 6.13. A low frequency triangle ramp is applied as input voltage to the different voltage ­regulator types used in the PD. For simplicity reasons, the loss resistances of any Bias-T are neglected in this setup.

2N3055 Generator UPSE = UPSE_out Triangle voltage ramp U

RDC_cable UPD = UPD_in

Vreg PD

URL = UPD_out

t Figure 6.13 PoC test setup, note that the DC loss resistance of the Bias-T inductances are neglected so that UPSE = UPSE_out and UPD = UPD_in.

RL

183

184

6 Power

For the tests, the loss resistance varies from 0–5 Ω and the load resistance is 10, 20, or 30 Ω. The voltage regulators tested, and their basic description is provided in Table 6.7. Naturally, this is only a small subset of all the available solutions. Also, only a meaningful subset of the possible measurement results is shown. Table 6.7 Overview of tested voltage regulators Name

Type

Company

Comment

LT3089

Linear regulator

Linear Technology (now Analog Devices)

For rough industrial environment; supplier boards

LT8640

DC/DC step-down

Linear Technology (now Analog Devices)

For EMC critical applications; supplier boards

LT8617

DC/DC step-down

Linear Technology (now Analog Devices)

For EMC critical applications; supplier boards

LM2596

DC/DC step-down

National Semiconductors Mounted on low-cost (now TI) ­consumer boards

XL6009

DC/DC step-up

Maxim Integrated (now Analog Devices)

Mounted on low-cost ­consumer boards

EL910.26

DC/DC buck-boost, SEPIC

Elmos

Supplier boards

Figure 6.14 shows different measurement results for four of the five voltage regulators listed, here with a loss resistance of 1 Ω and a load resistance of 10 Ω.

Figure 6.14 Input and output voltages at 1 Ω loss and 10 Ω load for different regulators in the PD At factor ten between the loss resistance and the load resistance, the figures show almost no difference between the linear regulator LT3089 and the two step-down converters LT8617 and LM2596, except that the two step-down regulators need to have passed a cer-

6.1 Supplying Power with the ­Communication

tain threshold before they linearly ramp to the maximum voltage UPD_out. The LM8617 shows some instability for very low voltages. The EL910.26 SEPIC/buck-boost regulator, in contrast, regulates UPD_out already at very low UPSE_out values. Figure 6.15 shows the same setup at 5 Ω loss.

Figure 6.15 Input and output voltages at 5 Ω loss and 10 Ω load for different regulators in the PD With a ratio of just two to one between load and loss resistances, no voltage regulator reaches a stable point of operation for UPD_out. The exact behavior varies depending on the regulator selected. The figures recorded for the LT8617 and the EL910.26 are particularly impressive with respect to the impact when the stability criterion is not met. Further tests not depicted here showed that with the selected voltage regulators a loss resistance of maximum 3 Ω in case of a 10 Ω load seemed reasonable. To complete the examples, Figure 6.16 compares the behavior of the step-up DC/DC converter with the buck-boost behavior with a different load resistance.

Figure 6.16 Comparison of a step-up converter with the buck-boost behavior

185

186

6 Power

The XL6009 step-up regulator converts, within its operational range, a low input voltage to a stable output voltage, if the input voltage is lower than the output voltage. When the input voltage is higher than the target output voltage, the XL6009 has no regulatory effect, and the output voltage is the same as the input voltage. This is different for a buck-boost converter. This type converts any input voltage to a stable output voltage within its specified operational range. The input voltage can be lower or higher than the output voltage. Another important observation for the buck-boost and step-up converters visible in Figure 6.16 is that the input voltage UPD_in is almost at a constant level, up to a certain point. For the buck-boost regulator this is when the target output voltage UPD_out is reached, which coincides with UPD_out being almost larger than UPD_in. For the step-up regulator this is, when the target output level is reached. In both cases, the output of the power supply, UPSE_out, shows some discontinuities in the ramp. This effect is caused by the sudden change of the current consumption, despite the very low internal resistance of the power supply. When using a step-up or buck-boost converter, this behavior should be considered, because at low voltages the current can be several times higher at the input of the voltage regulator than when the voltage regulator operates close to its nominal voltage. A simple example is an ideal regulator operating at UPD_in = 12 V, which supplies the output at a defined voltage and current. At UPD_in = 3 V instead of 12 V, the input current is four times larger, provided the output voltage is already regulated to its target value. The input current thus cannot necessarily be neglected. Furthermore, the efficiency of the voltage regulator can be very poor at low voltages, which might additionally impact the current drawn. For each regulator, its very individual characteristic needs to be taken into consideration. This section only showed a very small selection of different voltage regulator types with limited variations in the setups, while there is an almost infinite number of applications and implementations possible. The recommendation is, to use the theory of operation points and limits as a starting point but add detailed simulations and measurements with the planned circuitry. Not all effects can be seen immediately; this also applies to the load. All the descriptions so far assumed a resistor with a linear behavior as a load, without any timing effects or alike. Real loads, for example a microprocessor, an image sensor, or other IC’s, often have a non-linear relation between applied voltage and current consumption. And this relation might change with operating states and other, environmental conditions. The setup shown in Figure 6.13 is a fast and efficient starting point. Note though, that the internal resistances of any used test tools and the power supplies may also impact the measurements.

6.1.3.5 PoC Compendium After having discussed all relevant elements and their interrelations in a PoC system in this Section 6.1.3 so far, this subsection concludes the topic with a proposal on how to design a suitable Bias-T. Simple approaches to design the Bias-T are: 1. Use the example from the application note of the chip supplier. 2. Use the proposed solution of a supplier of inductive elements. 3. Use the proposed solution of a supplier for voltage regulators. These three methods will most probably deliver a functioning technical solution but not necessarily the best match for the application specific requirements. Such a match needs to

187

6.1 Supplying Power with the ­Communication

include the parameters and requirements for the wiring harness, meaning the minimum and maximum cable length, cable type, and allowed DC resistance(s) in the channel between the devices. The following design proposal assumes a camera as a PD with a step-down DC/DC ­converter. For the PSE, different types of voltage regulators are considered. Figure 6.17 summarizes the nomenclature of the elements and parameters used in the following. PSE (ECU) Ibat = Iin

I

I

UDC_loss

PSE supply

Uin = Ubat

PD (ECU/sensor/display) PD supply

UPD

UPSE

IRL URL RL

RDC_PSE

RDC_PD

UDC_cable

PHY UPSE_out

RDC_cable

PHY UPD_in

Figure 6.17 Reference nomenclature used for field manual Step 1: Determine the power requirements of the PD. First, the properties of the load need to be known. This includes the required voltage, URL, and power consumption, PRL, and, consequently, current, IRL. As for a DC/DC step-down converter UPD ≥ URL, then check how much larger UPD needs to be than URL in order for the DC/DC converter to provide a stable output voltage. Ensure, with help of the data sheet of the voltage regulator that the voltage regulator is not only stable but also functions efficiently at this value UPD. Else, somewhat increase UPD accordingly. With help of the conversion efficiency coefficient PRL = n PPD of the selected regulator, the current I can then also be calculated. With the current and the PoC requirements of the system, a first selection for the Bias-T inductor elements in the PD can be made, which then determines their resistance RDC_PD and power loss PDC_PD and with that PPD_in and UPD_in. Step 2: Determine the minimum supply voltage in the PSE, UPSE_min. The two voltage regulator types recommended for the PSE are buck-boost and step-up (see also Section 6.1.1.2). The selection needs to be made carefully, to ensure, based on Ubat and Iin, that PPSE > PPD_in and UPSE > UPD_in , so that the voltage regulator is operating at a stable working point. Estimate UPSE_min and UPSE_max from this. Again, with the known current I, a first assessment of the required inductor elements and their losses PDC_PSE can be made. Thus, the nominal, the min, and the max values for UPSE_out can be determined. Step 3: Determine the cable requirements and loss values. RDC_cable can be calculated based on the selected cable type and maximum cable length. Add the temperature coefficient to the cable resistance to make sure the system can work at the worst-case temperature. The cable resistance must fit the available loss resistance range that the selected voltage regulators and Bias-Ts allow for.

188

6 Power

Step 4: Check for plausibility and adapt where necessary. All values must match. The overall voltage values UPSE = UDC_LPSE + UDC_cable + UDC_LPD + UPD, the power values, the resistances and the currents supported. Where necessary, adaptations and iterations need to be made. If, for example, the power dissipation in the inductances needs to be reduced, inductors with the same inductance but lower DC resistance may be used. This increases the size of the inductor, which might be counter-productive when considering the limited space available in cameras and other small sensors. Designers have also to take this into consideration. Table 6.8 provides some examples based on the setup shown in Figure 6.13. Table 6.8 System values based on the setup shown in Figure 6.13 Type

RL [Ω]

PRL [W]

n

PPD(_in) [W] I [mA]

RDC_loss [Ω] UPSE [V]

LT8617

30

0.833

1.3

1.078

154

1

7

LM2596

30

0.833

1.96

1.638

182

3

9

Be aware that the real behavior of an inductive component is often not completely described in its data sheet. It is therefore recommended to put all selected components and values in a real hardware design and simulate as well as perform tests, to ensure a stable operation of the voltage regulators in the PSE and PD.

„ 6.2 Power (Saving) Modes Electronic systems inside the car are not just on or off but need to support several different modes with different purposes. While the instinct is to focus on the intended mode of operation, it is important that the other modes are also correctly supported in the design. This has to include the different requirements of each mode on the power supply, also and especially when transitioning between modes. Transitions pose a particular risk to the stability of a system and stability is of utmost importance. Figure 6.18 shows a simplified diagram of the different dedicated modes (white circles) and states (grey boxes). Similar versions were used to develop the SerDes standards but are not part of the documentation for the proprietary technologies. The sequence shown in Figure 6.18 starts with “Power On”. Before power on, in the “Off” state, no operation is possible. Because of discharged capacitors in the power paths of the system, the state during power-on is called “Inrush” (see also Section 6.1.1.2 and Section 6.2.1). Following the power on, the system starts in “Operation” state with the “Init” mode. In this mode, the system is configured and stabilized. If there is a power-over system in place, the initialization first needs to take place in the PSE and then in the PD. It is essential that the initialization is done in a controlled, system-dependent order. Also, without power-­ over, there generally is a detailed start-up and initialization sequence that needs to be followed, before a stable communication and system operation is possible.

6.2 Power (Saving) Modes

Off

Low Power

Test

Diagnostic Power Save

Inrush

Init

Power On

Normal Error

Operation

Error

Figure 6.18 Typical functional modes and power states In the “Normal” mode, the unit functions in the framework of its purpose. The development process of an ECU generally focusses on this mode as it represents its target deployment. Normal mode allows that all elements are active and that the hardware (HW) and software (SW) can access all resources in the system. Most system parameters, like communication requirements, processing power, power supply system, and more, are defined based on the Normal mode operation. Note though, that Normal mode operation might comprise other modes not depicted in Figure 6.18, and that system and use case dependent methods may be used to systematically save power (see Section 6.2.2.2 for an example of a power saving method during Normal mode operation). The test state contains a “Diagnostic” mode, important for the development and maintenance of any ECU. One function of the Diagnostic mode is, for example, the execution of a Built-In Self-Test (BIST) or an explicit cable fault detection routine, which an external tester would initiate and evaluate. After the completion of the test mode, the unit would go into Init mode, before going back to Normal mode. While Diagnostic is an intended system mode, implementations might also foresee an “Error” mode for unintended system behavior. Not all systems necessarily have this as a dedicated mode and state. But, because Error mode is always a possibility, implementations of Automotive SerDes Systems, for example, foresee a dedicated set of commands in order to stabilize the system and to process according failure symptoms (see also Section 6.1.1.3 for respective parameters in a power-over systems). In many cases, an Error mode would reset a system into “Init”. Finally, a unit generally supports a “Low Power” state and “Power Save” mode. Considering the exigent need to limit power consumption in cars, such mode is an important aspect in vehicular electronics. There are various, structurally different possibilities to realize power saving in systems. Not all of them result in changing into the low-power state. The difference between possibilities relevant in the context of this book are described in Section 6.2.2. Before that Section 6.2.1 focusses on the impact the modes have on the power requirements and important aspects to consider for the transition between modes. In Figure 6.18, the off state is only re-entered from Init mode. This represents the orderly fashion of a system to shut down. However, in case of a power loss, the system will go into off state, no matter what mode and state it was in before. This is not explicitly depicted in Figure 6.18, but needs to be considered in every system design.

189

190

6 Power

6.2.1 Transitioning between (Power) Modes The three important active power states depicted in Figure 6.18 are: Inrush, Operation, and Low Power. Inrush deals with the fact that in off state all capacitors are discharged. When the power is switched on again, these capacitors represent, at first, a situation comparable to a short circuit, which gradually “heals” with the charging of the capacitors. The strength of the inrush – meaning the size of the current that flows during the inrush – depends on the capacitive load and the DC resistance in the path. In case of a power-over system as described in Section 6.1, the DC resistance is the resistance especially of the cable and impedances in the power path. Possible strategies for handling the inrush current are: using a suitable power switch (meaning it sustains the needed power, it can switch off when current, voltage, or temperature are too large, and alike), have a current-controlled source, design the system such that the capacitive load does not create an excessive inrush current, accept a temporarily high load, and suppress any error checking diagnostic features until the power is stable. In the Operation state the system is first in Init mode, which configures the system and further stabilizes the power if need be. The amount of power needed in the following Normal mode depends on the implemented functions and the environment conditions, like temperature or lighting conditions. In most cases, Normal mode is assumed to require a continuous power supply. In the design phase of a unit, it is generally the Operation state, which is used as a basis for dimensioning the components in the power path. The Operation state should also include regular checks for potential error conditions of the power system as part of the normal operation (meaning: without changing into Test state). Normal mode implies that all elements and resources are available and usable by the HW and SW. As is detailed in the next Section 6.2.2, this is not necessarily so. Also, in Normal mode some partial functions or components might be deactivated in order to save power. Depending on how much of a unit is deactivated, HW and SW intentionally have only restricted access to the resources and the residual power needed varies accordingly. Key is that in Normal mode parts of the application are still running in an intended way. This is different in the separate Low Power state of Figure 6.18. In the Low Power state, the ­complete application is temporarily not needed and almost all of the unit is deactivated. Depending on the exact realization, see Section 6.2.2.1, the current drawn varies from a few 10 s of μA to