The Astrophotography Manual: A Practical Approach to Deep Sky Imaging [3 ed.] 9781032613178, 9781032601236, 9781003463108

Citation preview

The Astrophotography Manual

Jellyfsh Nebula (false-color narrowband)

The Astrophotography Manual A Practical Approach to Deep-Sky Imaging Third edition

Chris Woodhouse

1

The Astrophotography Manual

Front cover: Chris Woodhouse Third edition published 2024 by Routledge 605 Third Avenue, New York, NY 10158 and by Routledge 4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

Routledge is an imprint of the Taylor & Francis Group, an informa business © 2024 Chris Woodhouse The right of Chris Woodhouse to be identifed as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers.

Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifcation and explanation without intent to infringe. First edition published by Focal Press 2015 Second edition published by Routledge 2017

Library of Congress Cataloging-in-Publication Data A catalog record for this title has been requested ISBN: 9781032613178 (hbk) ISBN: 9781032601236 (pbk) ISBN: 9781003463108 (ebk) DOI:10.4324/9781003463108 Typeset in Adobe Minion Pro and Myriad Pro by Chris Woodhouse Publisher’s Note This book has been prepared from camera-ready copy provided by the author. Visit https://www.digitalastrophotography.co.uk for additional book resources

Prelims

2

Contents

Preface About the Author Introduction The Curious World of Astrophotography

4 6 7 11

Pushing the Limits Down to Earth Optical Resolution On the Right Track

16 26 32

System Choices New Trends Support and Mount Options Telescope Options Camera Systems Capture Software and Electronics Options Portable Systems

36 43 54 61 71 83

Setting Up Catalogs, Maps, and Surveys Site and Image Planning Hardware Setup Software Setup Automation and Remote Control

90 92 101 110 117

Image Capture Exposure Planning Focusing Staying on Track Autoguiding and Tracking Tracking Models

127 137 144 149 165

Image Processing Image Processing Fundamentals Calibration, Registration, and Stacking Linear Processing Non-Linear Processing Narrowband Processing An Overview of StarTools

172 179 193 202 212 219

Assignments Practical Examples M92 (Globular Cluster) Iris Nebula (C4) Monkey Head Nebula (NGC 2174) Flaming Star Nebula Region (IC 405/410) Pelican Nebula (IC 5070) Bode’s Galaxy and Friends IC 342 / C5 The Hidden Galaxy Bubble Nebula Region (NGC 7635) Christmas Tree Nebula (LBN 912) Exoplanet TIC 468574941

227 230 235 240 246 251 257 264 270 276 282

Appendices Diagnostics Projects Bibliography, Resources, and Templates Glossary Index

290 297 298 302 305

3

The Astrophotography Manual

I was once asked by a 7-year-old “Why do you take pictures of space?” After a moment’s refection I replied, “Because it is difcult.” It still is... this book is for those who enjoy the challenge.

in memory of Jacques Stievenart, who was an inspiration

Prelims

4

Preface A third edition for the advanced amateur, using the latest developments, techniques, and thinking.

I

started astrophotography just over ten years ago, and as a hobby, it continues to fascinate, frustrate, and amaze me in equal measure. Over the years, I have discovered many new ideas to try out, projects to build, and enjoyed the companionship of like-minded folks worldwide. However, I realized there were no advanced guides on astrophotography and wrote Te Astrophotography Manual in 2015 to fll the void and, in doing so, become more profcient too. An expanded second edition followed in 2017, but in this environment, it is a never-ending task to keep up with the latest trends and ongoing obsolescence. Five years have passed since the publication of the second edition. In this time, much has changed, with no sign of slowing up. Te world reeled from the onslaught of a global pandemic in this period. One unexpected outcome was a surge in interest in astronomy and astrophotography. During that fateful summer, my introductory guide, Capturing the Universe was printed; a book aimed at those starting out and using conventional photographic equipment. Tis book was also an opportunity to re-focus on a new astrophotography guide for intermediate and advanced imagers. Constraining the book to a manageable size, however, has required me to resist the temptation to indulge in niche pursuits and to prioritize the more universally relevant content. With a vast subject such as this, the content of any book can never be deep, inclusive, or relevant enough for every reader. Tis book restricts itself to deep-sky astrophotography, i.e., imaging objects outside our solar system. Guided by the many glowing online reviews, I have a new set of practical case studies and processing fow diagrams. In response to several requests to write a PixInsight manual (a dull task to read or write), I expanded PixInsight content in previous books. Here, I use PixInsight as the core processing application and introduce some diversity with several forays using up-and-coming alternatives. Twenty years ago, it may have been possible to cover all the popular sofware and equipment in one book. However, the explosion in consumer choice makes that an impossible task today and requires careful pre-selection. I am not one who writes “reviews”

on the back of public-domain material. To thoroughly research alternative hardware and sofware solutions takes considerable time and expense, and there is a practical and fscal limit to what any individual can cover. In keeping with the book’s target audience, I have concentrated on those solutions that meet the needs of the more demanding practitioners. Astrophotography has been alluded to, disparagingly, as just “pretty pictures” and no big deal (though they would have amazed anyone thirty years ago). While making images with meager resources and knowledge is possible, these are mostly simple renditions of a few bright targets. Tose who persevere with modest equipment expand the repertoire and improve image quality, though some exhaust the accessible popular targets afer a few years. When this happens, the options are to revisit old favorites with better technique, seek out alternative targets or compositions, or image new targets remotely at a diferent latitude. Tis book considers these alternatives. Some practical ideas that have occurred to me over the years concerning adapters, mini-computer systems, and focus algorithms have occurred to others and become commercial products. However, not everything is commercialized and this hobby still ofers plenty of opportunities for making your own gizmos. I have moved the practical projects to an updated support website https://www.digitalastrophotography.co.uk with other resources. Appendices, glossary, bibliography, supporting resources, and an extensive index round of this publication.

general support website

practical projects

Fishtail Nebula, part of the Heart Nebula (false-color narrowband)

Prelims

6

About the Author The pandemic curtailed many activities but presented the perfect opportunity to start another book.

I

have always been fascinated by the natural sciences, making things, art, and photography. My frst exams used slide rules and log books at school, but two years later, scientifc calculators had wholly displaced them. At university, 8-bit personal computers were catching on. So, afer designing communication and optical gauging equipment and writing sofware in Forth, Occam, C++, and Assembler, I moved on to automotive development. My frst project evaluated the emergent navigation systems, and 30 years later, they almost work! My passion for photography and electronics also led to several darkroom designs (which are still sold today) and an associate distinction in the Royal Photographic Society. Nevertheless, I resisted the temptation of premature digital optimism and authored Way Beyond Monochrome, a book on traditional monochrome photography and, a few years later, followed with a second edition. Digital monochrome appeared to be the next logical venture until a chance camera club presentation awakened a dormant interest in astronomy. I quickly found astrophotography to be the perfect fusion of science, electronics, and digital photography. Like many before, my frst (and sometimes second, third, and fourth) attempts ended in frustration and disappointment. However, I quickly realized the technical challenges of astrophotography respond well to a methodical and scientifc method. Together with an artistic eye and decades of printing experience, these proved to be an excellent combination for producing beautiful and fascinating images from a seemingly featureless sky. Te outcome was Te Astrophotography Manual, acclaimed by many readers as the best book on the subject, followed by a groundbreaking second edition in 2017 and a more introductory guide to astrophotography, Capturing the Universe, in 2019. Acknowledgements Tis book and the intensive research it requires would not have been possible without the ongoing support of my wife Carol (who even helped dig the footings for my observatory) and the generosity of the wider online community.

Special thanks also go to Peter Carson, Dave Biddlecombe (for my portrait above), and others for loaning equipment and their valuable contributions to some specialist chapters. It is one of the pleasures of this hobby to share problems and solutions with other hobbyists, and this book builds upon the experience and wisdom of many. Tis hobby is a never-ending journey of refnement, knowledge, and development. It is a collaborative pursuit, and I welcome any feedback or suggestions for this book or the next. Tis global collaborative spirit is also the principal aspect of open-source development and fuels exciting developments in astrophotography. Long may it continue. Clear skies (if only)! Chris Woodhouse ARPS, FRAS

7

The Astrophotography Manual

Introduction The persistence and patience of the ancients is remarkable, all the more so when you consider their achievements were made without optical aids, accurate timepieces or computers.

T

he night sky has been a source of wonder for millennia. Today, we take modern technology for granted as we automatically slew and center on a distant and seemingly invisible quasar and expose, register, and combine multiple images over many hours. Yet, it was not always so and it is humbling to consider the great achievements of the ancients, who made their discoveries without access to today’s technology. Astronomy is such a fascinating and evolving subject that I like to think that astrophotography is more than just making pretty pictures. Researching the matter added to my sense of awe and, at the same time, made me appreciate the dedication of early astronomers and their patient achievements. A little history and science are not amiss in such a naturally technical hobby. Incredibly, the science is anything but static; in the intervening time since 2011, not only has amateur astrophotography improved signifcantly, but we have sent a probe 6.5 billion km to land on a comet traveling at 65,000 km/h, found phosphine in Venus’ atmosphere, frm evidence of water on Mars and the New Horizons space probe grazed past Pluto, just 12,000 km from its surface, afer a 9.5-year journey of 5 billion km. (It is incredible that its trajectory was calculated using Hooke’s/Newton’s law of universal gravitation, published in 1687.) Without today’s scientifc insights, mankind studied the night sky and placed mystical signifcance on eclipses, comets, and new appearances. With only primitive methods, they quickly realized that the position of the stars, the Moon, and the Sun could tell them when to plant crops, navigate, and keep the passage of time. But, driven by a need for astrology as well as science, their study of the heavens and the belief of an Earth-centric universe became interwoven with religious doctrine. It took the Herculean eforts of Copernicus, Galileo, and Tycho, not to mention Kepler, to overturn the dominance of the Catholic Church in Europe and determine the heliocentric solar system with elliptical orbits, anomalies, and detailed stellar mapping. We should not overlook the achievements in other regions: Astronomers in the Middle East and South America made careful observations and, without in-

fg.1a An abbreviated time-line of the advances in astronomy is shown above and is continued in fg.1b. The achievements of the early astronomers are wholly remarkable, especially when one considers not only their lack of precision optical equipment but also the most basic of requirements, an accurate timekeeper.

Prelims

8

struments, could determine the solar year with incredible accuracy. Te Mayans even developed a sophisticated calendar that did not require adjustment for leap years. Yet, centuries later, the Conquistadors all but obliterated these records at a time when, ironically, Western Europe was struggling to align their religious calendar with the seasons. Te invention of the telescope propelled scholarly learning, and with better and larger designs, astronomers were able to identify other celestial bodies other than stars, namely nebula and much later, galaxies. Tese discoveries completely changed our appreciation of our signifcance within the universe. Very few of us have looked at the heavens through a telescope and observed the faint fuzzy patches of a nebula, galaxy, or the serene beauty of a star cluster. However, to otherwise educated people it is a revelation when they observe the colorful glow of the Orion nebula appearing on a computer screen or the friedegg disk of the Andromeda Galaxy taken with a consumer digital camera and lens. Tis amazement is even more surprising when one considers the extraordinary information presented on television shows, in books, and on the Internet. When I have shared backyard images with work colleagues, their reaction highlights a view that astrophotography is the domain of large, isolated observatories inhabited by nocturnal Physics students. Tis sense of wonderment is one of the reasons why astrophotographers pursue their quarry. It reminds me of the anticipation as a black and white print emerges in a tray of developer. Te challenges we overcome to make an image only increase our satisfaction and the admiration of others, especially those in the know. When you write down the numbers on the page, the exposure times, the pointing accuracy, and the hours to capture and process an image, the outcome is all the more remarkable.

New Technology Te escalating amateur demand fuels the marketplace and supports an increasing number of astro-based companies. Seven years afer writing the frst book, innovation and engineering continue to improve afordable technology in mechanics, optics, computers, digital cameras and, in no small way, sofware. Te digital CCD sensor was chiefy responsible for revolutionizing astrophotography, but it is now at a crossroads. Dedicated imaging cameras piggyback of the sensors used in consumer devices, typically DSLR and, increasingly, mirrorless cameras. At one time CCDs and CMOS sensors

fg.1b Terrestrial and space-borne telescopes propelled astronomy, starting with visual observation and followed by examination of infrared, ultraviolet, X-rays and radio frequencies.

9

The Astrophotography Manual

were both used in abundance. Today, CMOS sensors dominate the photography marketplace and are the primary focus of sensor development, continually reducing noise and increasing size, efciency, and pixels. At frst, economics drove CMOS implementations. Tey were less expensive but had several performance issues that made them less suitable for astrophotography. Continual development has made most of those issues insignifcant, and in the space of a few years, CMOS designs now dominate the amateur astrophotography market too. Te demand for high-speed, high-resolution video photography, however, may make future CMOS designs less suitable for low-noise applications and quickly fll up computer storage. Tere are also defnite trends in the telescope marketplace. Bulky Newtonian refectors were once the goto instrument, as large aperture refractors were ofen expensive or of poor quality. Refractor price and performance have dramatically improved in recent years, with many doublet designs performing similarly to triplet designs of the prior decade. Better production techniques for large and especially non-spherical mirrors now make exotic optics like Ritchey-Chrétien and Dall-Kirkham afordable and, lately, fast-aperture Rowe-Ackermann Schmidt Astrograph (RASA) refectors. Te choice is bewildering. Computers have evolved at a similarly breakneck pace. Teir performance continues to improve, with lower power consumption and increasingly smaller footprints. Miniature PCs increasingly replace laptops for outdoor use. Initially Windows-based and increasingly using a Linux operating system, these small computing units are used without a display or keyboard with wireless remote control. Tese take the form of general-purpose computing boxes or integrated units with power, USB, focuser, and dew-heater control. Sofware trends are fascinating and fast-moving; innovation is the key to stay ahead. Te once popular cash cows are being put aside for new titles that ofer better value for money, responsive development, and compatibility with new systems. In addition it is generating considerable enthusiasm to feel a part of the development process. (Te same is true for mainstream applications; this, my 6th book, uses Afnity applications whose net cost is equivalent to two months of Adobe CC subscription and was immediately compatible with the latest Mac OS.) Back to astrophotography, ASCOM, which has been around for 20 years, is embracing cross-platform use with ASCOM Alpaca as it is jostling with INDI and INDIGO platforms for Linux-based systems in the market.

If that was not enough, some new mounts are bypassing conventional computing platforms and interface directly with tablets and smartphones for controlling mounts and in time, maybe taking pictures too.

About This Book I wrote two editions of Te Astrophotography Manual with the vision of being a fast track to intermediate astrophotography. Tese were ambitious tasks and quite a challenge. Many astrophotographers start with a conventional digital camera and conventional editing sofware. Tese make exciting images in the right conditions and for a few targets. Tis is the target audience for my book Capturing the Universe. Tis afords the opportunity for a new book, which assumes the reader already has some experience and needs less introductory matter. Tis is an opportunity to optimize the content to cover some topics in-depth and the latest developments. Signifcantly, this book is aimed at the enthusiast and, from the outset, assumes most will use a dedicated astronomical camera, rather than a consumer camera, to achieve the best quality results. Tis book focused on deep-sky imaging; my location is not ideal for high-magnifcation work and any references to planetary imaging are made in passing or by a lucky accident. Out of Scope Any book has limits; today’s consumer choice is overwhelming, as is the diverse nature of astrophotography. It would be presumptuous to recommend a shopping list, and it would be impossible and futile to try and evaluate everything. I am fortunate to have a healthy but not infnite budget. It forces me to carefully rationalize my purchases and evaluations for quick and reliable setups at home and abroad. My ultimate goal is to maximize the brief opportunities that the weather permits and extract the maximum quality from every exposure. My current setup is the result of a long journey of trial and error with a dozen telescopes and cameras and too many mounts. When explaining the principles of astrophotography, some things may be unique to one piece of equipment or another but the principles are common to many. We all get carried away occasionally; in my case, some of my gear was at the upper end of “portable” and the subsequent hernia prompted a permanent observatory. Tis book is divided into logical sections: Te frst looks at the subject matter, its diversity, and how to choose a deep-sky object. Te second discusses the es-

Prelims

sential truths of equipment in context to the environment. Te following section discusses the tools of the trade, including the latest developments in hardware, sofware, and automation. Te third section progresses with the practical aspects of setting up the hardware and sofware imaging system with permanent and portable systems in mind. In the following section, we do the same for image capture, looking at developments in process automation, guiding, focusing, and composition. Tis section evaluates the latest tech (at the time of writing). When we come to the processing section, although the emphasis on PixInsight in my earlier books was wellreceived by many, this application is not everyone’s favorite. In this book, the examples use various applications including StarTools, AstroPixelProcessor, SiriL, and ASTAP to the PixInsight content. Te assignments section lays out practical imaging projects that consider a particular target’s conception, exposure, and processing. Tese are selected to highlight unique capture and processing techniques. Over the years, I have deliberately used various equipment, techniques, and sofware to acquire and process images. Tat includes NINA, AstroPixelProcessor, StarTools, Afnity Photo, and Linux-based applications. Each shares considerable overlap with others but ofers unique approaches or tools. In these complex systems, we make mistakes, and worked examples are more valuable if they share and learn from mistakes. On the same theme, things fail, and in the appendices before the index and resources, I have included a chapter on diagnostics, with a list of common problems to help you troubleshoot. Fixing problems can be half the fun, but when they resist several reasoned attempts, a helping hand is most welcome. In my fulltime job, I used specialized tools for problem analysis and I share some simple processes to track down gremlins. Astrophotography and astronomy, in general, lend themselves to practical invention, and not everything is available of the shelf. Te practical projects are now placed online rather than in printed form. Tese include hardware and sofware projects. Te book rounds of with a bibliography and a comprehensive index. For some reason, bibliographies are a rarity in astrophotography books. As Sir Isaac Newton once wrote, “If I have seen further it is by standing on the shoulders of Giants.” (Tis is perhaps not the best example, as there is irrefutable evidence that our esteemed physicist borrowed more from Hooke and Halley than he would freely admit.)

10

fg.2 An updated time-line for astrophotography. The last ten years have been particularly busy periods for camera and sensor development; improving performance and lowering cost.

Te printed page is not necessarily the best medium for some of the resources, and the support website has downloadable versions of some tables, drawings, programs and source code, as well as any errata that escaped the various editors. Tey can be found at https://www.digitalastrophotography.co.uk as well as some unique resources for book owners, linked through a QR code on relevant pages. Share and enjoy.

11

The Astrophotography Manual

The Curious World of Astrophotography An introduction to this totally absorbing hobby, whose scope is only limited by location, budget, and patience.

A

strophotography is as diverse as the Universe is bizarre, and as a hobby, there are many paths to follow for pictorial, educational, or scientifc ends. In the main, however, the images adorning the many websites consist of star felds, special events, comets, planets, and deep-sky objects. Depending on viewing conditions, equipment, budget, and available time, amateur astrophotographers vary from occasional imagers using a temporary setup to those with a permanent installation, capable of remote control and operational at a moment’s notice. Similarly, astronomy is a vast and fascinating subject, and while it is not an essential prerequisite, it certainly helps to understand some basics; for instance, the coordinate systems used by catalogs and planetarium sofware, to fnd your way around. I’m ofen asked how far away a particular object is. I usually have no idea other than to know the fgures for distance and size are mind-numbingly large. It adds another dimension if I read up on the target I am imaging. I ofen wonder how folks know all this stuf. As amateur astronomers, we have an important scientifc role to play, even inadvertently. Amateurs have and continue to make signifcant contributions to scientifc research, including supernovae discovery, plotting meteor trails, identifying potential exoplanet candidates, and observing new nebulae or transitory solar system events. Tis is surprising until one considers there are far fewer professional observatories, each with narrow felds of view and that cannot cover the entire sky at any time. In comparison, the potential detection capability of thousands of amateurs worldwide is signifcant, especially for detecting transient events or maybe by classifying deep-sky images on zooniverse.org. Although I may happen upon something in my lifetime, I am content with making high-quality deep-sky images and inspiring others to do so too. If the conditions are right, I do stray on occasion; to image at the magnifcation extremes of Milky Way vistas or planets. A vacation during a new Moon at a dark site and the conjunction of Saturn and Jupiter in December 2020 (both equally rare events), were good excuses to try something new.

Specialties Time and money place a practical limit on what any individual can cover (as does their location), and it is ofen the case that astrophotographers specialize. Tese specialties ofen drive several distinct system types, driven by the extreme range of object intensity and size. For example, the intense, hazardous radiation of our Sun requires a specialized small-aperture and specially-fltered telescope ftted with a high-speed video camera and with no particular need for accurate tracking (or location aids!). Another uses long focal lengths and an imaging location with excellent atmospheric conditions to deliver highly-magnifed images of our planets rivaling those of the Voyager missions. Tese relatively bright subjects are ofen imaged with small, high-speed cameras over short intervals and are also forgiving of tracking errors. At the opposite extreme, those who monitor meteor events employ stationary, sensitive video cameras ftted with wide-angle lenses to cover most of the sky in their feld of view. Deep-sky astrophotography lies somewhere in between these extremes. It generally involves imaging objects of considerably less intensity and requires an entirely diferent strategy to deliver long exposures over extended imaging sessions, and highly-accurate tracking. Image processing is unique too. Subjects include galaxies, supernova remnants, clusters, planetary, emission, and refection nebulae, and even the dust of space itself. Tese subjects form a very diverse group and require specialized techniques and equipment to capture the surprisingly wide magnifcation range and subject intensity. Size Matters Subject size is intriguing, and while we intrinsically know how big the Moon or Sun are, deep-sky images rarely give any clue to their scale, and newcomers are ofen surprised at their apparent size in the sky. For example, fg.1 shows an approximate distribution of the apparent sizes of the deep-sky objects from the Messier, Bright Nebula, Caldwell, and Sharpless catalogs. Te Moon is shown for reference, as is the feld of view for two telescope and sensor combinations covering 40–400 arc minutes. Afer allowing for

Prelims

12

deep-sky astrophotography. One wrinkle is that the apparent magnitude of an object can be misleading, as it is not an accurate predictor of its intensity and hence, its exposure. For example, a large galaxy appears less bright than a star of the same apparent magnitude. Individual deep-sky exposures are uniquely long, typically in the range of 30 seconds to 20 minutes each. During this time, the camera and sensor’s electronics heat up and generate thermal noise. In general, this noise increases with exposure duration and sensor temperature and adds to the other sources of noise that make it into each exposure. Since some of the image details are so exceedingly faint, the deep-sky brigade spends considerable time and resources doing their best to reduce the impact of image noise. For the best results, in the case of camera technology, it is soon apparent that consumer cameras crucially do not have cooled or temperature-regulated sensors. In the war against noise, both these features are benefcial in equal measure, and this book assumes that digital SLR or mirrorless bodies are used for wide-feld vistas or bright subjects that require short exposures. While the individual exposures may be pretty similar between practitioners (since it is usually determined by the brightest objects in the feld of view), the length of the imaging session is ofen predicated by the light-pollution level. As we shall see later, the genCapturing Photons Te challenge that surprises many new astrophotogra- eral efect of light pollution can be easily subtracted phers is the amazing range of image intensity within from an image, but the random noise that accompanies it remains. In the case of imaging from an urban environment, the usual recourse is to lengthen the overall imaging session. Te efect is signifcant; I have seen equivalent images to my own that were exposed over a single night, for which I required many clear nights to achieve the same image clarity. In one case, my latitude was such that I only had a few hours per night when the target rose above my imaging horizon. As a result it took three months to capture enough data for a quality image. While exposure and gain settings may accommodate image intensity, it becomes more problematic when very dim and bright objects are in the same frame. Te image details in a typical print occupy a density range fg.1 The distribution of deep-sky objects has a surprisingly high number of objects of about 64:1 (in Ansel Adams’ Zone larger than the Moon, which can be imaged with modest equipment. System parlance, Zone II–VIII). Te image margins, this range covers about 500 objects. Tis is an approximation; many of the larger nebulae do not have convenient, well-defned boundaries, and it is not always possible to eliminate duplicated entries across catalogs. Fig.2 shows a pictorial comparison of similarly-scaled objects, including the Moon, for reference. Tese are just a handful of the better-known deep-sky objects. If you can fnd it, Jupiter is shown too and it illustrates why planetary and deep-sky astrophotography drive very singular needs. In general, however, a focal range of 350–2,000 mm with an APS-C sized camera gives plenty of scope, with an option to use a mosaic or a conventional camera lens to cover wider felds. In recent years, several manufacturers have also released high-quality instruments in the 250- to 350-mm focal length range, ideal for lightweight, portable systems. For deep-sky astrophotography, small and faint objects are particularly problematic. Tese require high magnifcation and long exposure times, over which the impact of atmospheric conditions and tracking errors decimate image resolution. We will examine these realities in more detail to understand how they relate to equipment choice. Unfortunately, very few of us image from a high-altitude observatory in the Atacama desert.

13

The Astrophotography Manual

fg.2 This pictorial depiction of several common astrophotography targets (at the same scale) shows a wide variety of shapes and sizes.

Prelims

14

narrowband shot in fg.3 has two circled areas. Tis image has been screen-stretched but, before manipulation, the intensities in each were 43 and 64,500 above the dark level, a ratio of ~1,500:1. Clearly, special techniques are required to show detail on-screen or in print in both areas without clipping. While some may choose to clip medium and bright stars (and become pure white dots), I prefer to expose and process my images carefully to preserve the intensity and color diferences between stars. Even so, some of my fltered exposures still exceed 15 minutes each, at which point the beneft of using a refrigerated sensor becomes even more apparent. The Soft Side It is very easy to be caught up in the glamor of “stuf ”. Less obvious and equally as important is the sofware side of things. Sofware is the glue that makes it all hang together. For the aspiring astrophotographer, an electronic cable release and Adobe Photoshop™ are simply insufcient. Deep-sky astrophotography uniquely combines many long-exposure images to capture sufcient photons from the object to make a meaningful manipulated image. To do this requires specialized sofware to precisely control the mount, cameras, and telescope hardware so that each separate exposure is as sharply focused and well-tracked as it can be. It is not uncommon in this challenging hobby to achieve focusing accuracy within 10 microns (0.4 thou) and tracking accuracy of 0.5 arc seconds (0.00014 degrees). For enthusiasts, this control extends to meridian fips, monitoring weather conditions, generating calibration exposures, accurate centering, and observatory control. Traditionally this may have required a desktop or laptop computer, but increasingly is within the capabilities of miniature computing bricks and low-cost computing modules running the Linux operating system. Just as the image capture process requires specialized sofware applications, each branch of astrophotography also has unique image processing requirements. For instance, the high-speed video-based streams of solar and planetary imaging use unique tools that perform a frame-by-frame analysis. Tese identify and align the best frames and discard the rest. Te result is ofen a magical transformation of a grainy orange blancmange into a smooth recognizable planet, ready for further enhancement. As you are probably aware, the image workfow for deep-sky images is complex, requiring signifcant efort to select, calibrate, align, and combine poten-

fg.3 A comparison of image values, before any manipulation shows an extreme range of pixel values, far beyond that of conventional photography. The intensities here have a 1,500:1 ratio and yet neither should be black or white.

tially hundreds of long exposures. Tese processes are not the forté of conventional image editors. From here, we start to manipulate the image beyond recognition. Te (usually) featureless black image is subjected to a barrage of processes, selectively reducing noise, sharpening, boosting color and faint details, and at the same time, avoiding highlight clipping. Tese manipulations are best achieved using high-bit-depth image fles (32or 64-bit). Even here, conventional image editors may struggle, and while I have seen reasonable outcomes, dedicated astronomical image editors do a better job, and I use a photo editor at the end, to make some small adjustments to the fnal image, size, color profle, and bit-depth and fle format for the intended use. Dedicated image processing applications are packaged in various forms; some are fully integrated with image capture features, and others are free/shareware utilities or paid stand-alone applications. Tese are niche products, and many are the work of a few individuals rather than large corporations. As such, the developers are ofen more responsive and nimble to suggestions. When I started in earnest in 2012, most of today’s capture and image processing applications did not exist. Many image editors do not necessarily have the same feature set; for instance, some assume using a separate calibration utility. Collectively, however, they ofer alternative solutions to the problem of manipulating high dynamic range images, in the presence of considerable noise, with an equivalent diversity of approaches in keeping with this hobby.

SH2-132 (The Lion Head Nebula)

Pushing the Limits

Pushing the Limits

16

Down to Earth A reminder of the signifcant challenges that astrophotographers face and how they might be met.

I

t is incredible to think that amateurs are routinely imaging deep-sky objects at a quality level to rival some professional observatories of twenty years ago. Even so, astrophotography is not a “walk in the park”, and for many of us, it is a constant crusade against forces mainly beyond our control. We operate at the very limit of our equipment, and yet making beautiful images is possible. Tese challenges form a signifcant part of the fnal reward and recognition. Your mission, should you choose to accept it, is to boldly go where no sane human has gone before. Tis chapter reminds us of the real-world challenges and the high-level techniques we use to overcome them. It touches upon many technical topics without too much detail, which would obscure the high-level message. Tese essential topics have a more rigorous treatment in later chapters.

Location Te location where we image from greatly impacts what we can achieve. As well as the limits imposed by our particular imaging horizon, the weather, light pollution and geography infuence what we image, how we image and how long it will take. It is important to recognize this limitation at an early stage, for though it is exciting to engage in a frenzy of retail therapy, it may not necessarily result in better images. In my case, while improved technique, equipment, and considerable patience have dramatically improved my image quality, my location is now the limiting factor and fghts further quality improvements. Imaging the Invisible Astrophotography is mostly about taking images of exceedingly dim subjects sprinkled with surprisingly bright points of light and usually in the presence of considerable light pollution. To give you an idea of the challenges, it is illuminating (pun intended) to compare the imaging conditions of a typical studio portrait with those of an emission nebula. Using a sensor’s gain, quantum efciency specifcations, and the resulting image fle’s pixel value allows one to estimate how many photons land on a pixel (technically, a photosite) in each scene.

In the case of the studio-fash portrait, about 25,000 photons hit a pixel during the brief fash. In the case of the nebula example, the image values of a faint glowing hydrogen cloud indicate just 40 photons (on average) emitted from the nebula land on a pixel during a 20-minute narrowband exposure. Tat makes the hydrogen cloud about 10 billion times darker, or, in photographic terms, over 33 stops. One would have to image for 1,000,000 seconds to get an equivalent pixel value. At the same time, the brightest stars saturate/clip sensor pixel photosites in a fraction of a second. Challenging times are ahead!

Object Brightness Te apparent magnitude of an object infers the object brightness. Tis measure is one of the various standard parameters in the stellar and deep-sky catalogs. However, this measure has to be used with caution; there are several terms that we use to imply brightness, namely luminosity, fux and magnitude. Luminosity relates to the total light energy output from a star, fux is a measure of energy over a unit area, and the intensity reduces with viewing distance, obeying the inverse square law. It is possible to measure fux with a telescope and a sensor, but it is not a convenient measure for comparing object brightness. Te apparent magnitude of an object provides a convenient unitless measure of its relative intensities from Earth. Magnitude is a back-to-front logarithmic unit; a one-unit increase is 2.5x less bright, and a 5 unit increase is 2.55x (100x) less. For instance, Sirius has a magnitude of -1.47, and the faintest object observable from the Hubble Space Telescope is about +31, or 2.4x1013x dimmer. Fig.1 shows a range of objects at diferent apparent magnitudes and what is typically required to detect them. Te note of caution referred to earlier relates to the apparent magnitude of an object. Tis is based on the amount of light emitted from the whole object and for large objects, such as some galaxies or nebulae, the “surface” appears fainter, compared to a smaller object of the same magnitude. As such, one cannot convert the apparent magnitude of an object into a simple exposure calculator.

17

The Astrophotography Manual

Dynamic Range We all appreciate that deep-sky objects are demanding to image and, on the whole, exceedingly dim and come in all shapes and sizes. We are also familiar with the general concept of reducing or increasing exposure in response to object intensity. In astrophotography, however, it is ofen the case that the various objects in an image will have very diferent intensities, i.e., the subject has a high dynamic range. With a fxed telescope aperture, exposure is controlled by sensor gain (or camera ISO) and exposure duration. Te best setting is elusive, and for scenes that include objects with a large intensity range, an exposure may have clipped highlights, lose faint details or both. Tis is not a new problem; in the days of flm, if one over-exposed transparency flm, the highlights washed out, whereas, with color and mono negative flm, underexposing shadows was of more concern. Transparency flm can distinguish tones over about a fve stop range (32:1), whereas modern sensors have a dynamic range of about 12 stops (4096:1) and are ~16x more efcient at capturing photons. Even so, capturing the faintest and brightest details of most deep-sky scenes in a single exposure is impossible. Sensor models have diferent dynamic range capabilities and may change with operational settings. Understanding these is useful when selecting and using a camera for deepsky imaging. Dynamic range is discussed with other sensor specifcations in a later chapter.

Sensors, Warts and All No sensor is perfect, and their imperfections become apparent when we operate them at their limit. For example, they typically waste 10–30% of the photons that fall onto the sensor surface and the electronics, which convert the electrical charge into a voltage and convert to a digital value, introduce their own errors. Tese include consistent conversion errors as well as random variations. Worse still, some sensors generate unwanted artifacts, the most obvious of which is amp glow. Amp Glow Amp glow is most noticeable in CMOS sensors and varies between models. Te unwanted signal is generated within the sensor chip, on account of the integrated conversion electronics and while successive sensor designs generally reduce the issue, it is still apparent in long exposures. For example, in fg.2 the amp glow intensity in this long exposure exceeds the narrowband image intensity. Removing the amp glow and the pixel-to-pixel inconsistencies requires precise ex-

fg.1 This table highlights the limits of perception for the aided and unaided eye over a range of conditions and indicates the number of objects within that range. The advantage of CCD/ CMOS imaging over an exposure of 5–50 minutes is overwhelming. For Earth-bound imaging, the general sky background and noise, indicated by the shading, will eventually obscure faint signals, from about magnitude 18 in suburban areas. The Hubble Space Telescope operates outside our atmosphere and air pollution and, at its limit, it detects magnitude 31 objects. Its sensitivity is approximately 150,000x better than an amateur setup. The James Webb Telescope’s mirror collects 6x more photons than the HST!

posure calibration, using perfectly matched dark frames with the same sensor temperature and exposure time (fg.3). In practice, the necessary precision favors temperature-regulated, refrigerated sensors and relegates the role of DSLR or mirrorless cameras to light duties. Once, the high price of CCD-based astro

Pushing the Limits

18

fg.2 Some modern CMOS sensors have amp glow, which becomes increasingly obvious and may exceed Hɑ emissions (as here) with long exposures. With care, however, calibration using temperature- and time-matched fles removes this glow, leaving behind a little random noise. Precise calibration is difcult to achieve with consumer digital cameras as it is impossible to regulate the sensor temperature without modifcation and even in a stable ambient, the sensor temperature increases with use.

fg.3 This is the same view as in fg.2. This image is a combination of 40 calibrated and registered exposures. The calibration process removes the amp glow and constant pixel-to-pixel conversion errors (and at the same time, things like vignetting and dust spots too). Combining 40 exposures reduces the random noise by about 6x, allowing one to stretch the image to reveal the faint Hɑ details. The amp glow area does, however, have a slightly higher noise level, on account of its accompanying shot noise.

cameras made this a signifcant investment, but today’s modern CMOS designs are more afordable and, once you have used one, there is no going back.

exposure. However, this only fxes the constant errors; things like hot pixels, pixel shifs, gain variations in the conversion electronics, and the mean level of thermally-generated electrons. Image calibration does not reduce random pixel variation. Tis random noise also reduces the ability of a sensor to record very bright and dim subjects simultaneously (I like to think of noise as fog – nothing in fog is high-contrast). Te sensor’s dynamic range is a computed specifcation from a photosite’s electron capacity (full-well depth) and its readnoise level. Each time the read noise doubles, the efective dynamic range halves. Random noise mechanisms include sensor read noise and the randomness of any thermally generated electrons during the exposure, which increases with temperature and duration. Te noise from these thermally generated electrons is most signifcant in DSLRs and mirrorless cameras as sensors heat up with operation and, without cooling, reach 30 °C or more with extended use. For best results, it is best to use a dedicated astro camera with a built-in cooler. I operate mine at -15 °C, achievable in most ambient conditions. By chance, my Fuji camera and astro camera use the same Sony sensor, and refrigeration reduces the thermally generated random noise by about 15x. Without cooling, the thermally generated noise from a 5minute exposure is obvious, at about 4x the sensor read noise. Te two equivalent images in fg.4 compare the thermal noise levels at two sensor temperatures.

Sensor Noise It is easy to become fxated on sensor noise and specifcations. Unfortunately, these specifcations lead to misleading model comparisons, particularly when their pixel sizes are diferent. Te term “noise” is ofen used as a general term to refer to any unwanted (error) signal. Under normal circumstances, it would be invisible, but in astrophotography, it requires extreme image stretching to reveal faint details that increase local contrast 100-fold or more and reveal hidden issues. In a nutshell, there are several error mechanisms: • • • •

pixel gain diferences (constant) pixel ofsets (constant) pixel variation (random) pixel variation (temporary patterns)

Image calibration is the key weapon in our arsenal to combat sensor pixel to pixel inconsistency. Tis is something that conventional photographers seldom do knowingly. Te image calibration process evaluates your sensor, operating under the same conditions as the image exposure. It uses exposures taken in the dark (dark frames) and of an even illuminated target (fat frames) to correct the error in each pixel in each image

19

The Astrophotography Manual

dient, leaving behind a uniformly neutral dark background with faint details of the deep-sky object. Te real demon is still there, however, lurking in the shadows (literally). Tis is where we come face to face with the immovable force of physics, and this crucial and inescapable property requires a little explanation. Shot Noise Photons behave as discrete particles and are emitted randomly because of the laws of physics (they are also waves, but we are not going into that just now). We normally do not perceive that randomness but know the average emission rate sets the subject brightness. Over a suitable duration, a photosensor converts enough photons into electrons and converts them into a digital value at the end of the exposure. It is useful to think of photons like raindrops; we understand the diference between a shower and a downpour, but in each case, the raindrops land randomly over the back yard, and you cannot predict where the next one will land, or when. Photons behave similarly, and even in an image of an evenly-illuminated uniform target, each sensor pixel captures a slightly diferent number of photons (just as the number of raindrops hitting two precisely equivalent targets is diferent). Te unwanted randomness between pixel values is noise (or shot noise, to be precise). Tese combine with other sources of noise caused by interference and sensor electronics. Te shot noise level increases with the number of captured photons (or raindrops). Tere is an important distinction here; we are not talking about light level (brightness) but the total exposure. If 10,000 photons land on a sensor pixel over 0.001 or 100 seconds, it makes no diference to the shot noise. Even more interesting, it is the same deal if 10,000 photons are captured over 100 x 1-second exposures and then added together. Te diference is noticeable and the example in fg.5 compares two short but slightly diferent exposures of the same bright scene. (Tese images were carefully processed to isolate the random noise from the mean exposure.) Tankfully, the shot noise does not increase at the same rate as the exposure; if the captured photons in-

fg.4 The magnifed image on the left is a 10-minute exposure with the sensor at 5 °C. On the right, another at -15 °C. In both cases, the mean level was subtracted from the image and a similar stretch applied to each, to reveal the dark noise. Some DSLR and mirrorless cameras report their sensor temperature over the USB connection and in practice, can rise above 30 °C. This is thermal noise and increases with exposure and doubles with every ~5 °C rise in sensor temperature. Most cooled astro cameras are capable of reducing the sensor temperature 35–40 °C below ambient, which lowers the dark noise level by 128–256x.

Without diverting ourselves into too many technical details at this time, the key takeouts are to use a cooled camera to reduce the thermally generated current (and hence noise) and calibrate every image exposure with matched dark and fat frame exposures. But, unfortunately, the elephant is still in the room.

The Problem with Light Pollution Light pollution is a growing issue for many of us. Urban living conspires with personal safety requirements and encourages households and local authorities to illuminate our surroundings at night. Even those in uninhabited deserts sufer each month as the Moon illuminates the sky. Each of its forms afects our imaging diferently, as do the coping strategies. Even with the same amount of illumination, light pollution may vary from night to night, depending on the amount of dust, water vapor and aerosols in the atmosphere. Tese scatter and refect the incident light in all directions. A few weather forecast applications predict atmospheric transparency and infer light pollution to some extent. I have found some of the best imaging periods occur afer it has been raining, as the air is usually cleansed and has less backscatter. Many believe the objection to light pollution is due to the general obscuring glow in an image. Tis glow is typically brighter near the horizon, making it even more objectionable and, practically, may limit an exposure duration. Tis, however, is not the real issue since, with care, image processing removes the glow and gra-

Pushing the Limits

20

fg.5 Shot noise increases with exposure on account of the nature of light. These magnifed images have had their mean levels adjusted and equally stretched to reveal the noise. The left image is a 0.01-second fat exposure and the right is 0.04 seconds. Increasing exposure increases shot noise, which can dominate the other noise sources in each exposure.

crease by 4x, the noise level only increases by 2x. Conversely, quartering the exposure halves the shot noise. In fact, a simple equation predicts the shot noise for a given exposure level. Te same equation holds true for other signals; for example, the photons from the object and the thermally generated electrons in the sensor.

average glow is removed during image processing, but the random shot noise is lef behind. Here, the faint signal also competes with increased shot noise considerably more than the combined object shot noise and sensor noise. Te relative noise level is indicated with a red error bar on the graphs and almost halves with the reduction in light pollution.

shot noise level = √(signal level)

Competing with Light Pollution So how does this relate to light pollution? Just this; if we consider two identical exposures of a dim object, taken from an area with low and high light pollution, we can measure the damage caused by light pollution. Figs 6–8 illustrate this in image and graphical form. At the dark site, the faint signal competes with sensor noise and its shot noise. For the light-polluted site, the general

Tere are two strategies to lessen the efect of light pollution: reduce the amount of light pollution hitting the sensor and increase the exposure. As you have guessed already, increasing the exposure is an all-round good thing to do and we will consider that last, in a broader context. Let us look at the various forms of light pollution and the potential specifc workarounds. Tere are

fgs.6,7 These closely-cropped stretched images of this galaxy indicate the sample areas for sky background and the spiral arm, used in the analysis of the image in fg.8. The image on the left (fg.6) is a single, calibrated 300-second exposure through a red flter. The image on the right (fg.7) is a combination of 53 calibrated and aligned images. This one image demonstrates the beneft of multiple exposures for improving dynamic range and noise levels beyond any doubt.

21

The Astrophotography Manual fg.8 Using the values from the images in fgs.6/7 together with the calibration fle values, it is possible to break a single exposure’s pixel values into their constituent components. On the left-hand side are the image pixel values before calibration, in the presence of high and low light-pollution conditions. The light pollution dominates the faint galaxy signal in both cases, with increased noise levels in the more light-polluted image. The magnifed view on the right shows the results after calibration and background adjustment. It can be seen that the exposure with high light pollution has almost double the noise level of the one taken at a dark site. The only way to reduce the infuence of this random noise is to average multiple exposures, resulting in the image in fg.7.

several approaches to reduce the impact of light pollution on our astrophotography: • • • • •

light-pollution flters separate RGB flters on a monochrome sensor narrowband flters (on color or mono sensors) image from a darker site (capture more photons)

Tese all improve the signal-to-noise ratio but in diferent ways. In the frst four, we reduce the amount of light pollution hitting the sensor and, more signifcantly, reduce the substantial shot noise that comes with it. Filtering Strategies If your local street lighting uses traditional sodium lamps, a color image of what appears to be a dark sky reveals a muddy orange color as the light bounces back to Earth from all the airborne contaminants. Tese emission lamps emit distinct colors, and although there are several diferent types, their primary outputs are at 590 and 600 nm. Te less popular mercury vapor lamps emit over a broader spectrum but with strong components at specifc wavelengths in the violet, blue, green, and yellow-orange colors. Te coping strategies in these conditions take advantage of the specifcity of the light-pollution colors. We use unique thin-flm

(dichroic) optical flters to block the unwanted artifcial colors and pass the others. In this instance, light-pollution flters precisely exclude the main street lamp colors. Tere are many on the market; most exclude the two sodium and four mercury lamp emissions yet pass the common nebula emission colors with remarkable efciency. Common sizes are 1.25- and 2-inch sizes and are increasingly made in smaller tailored packages, to insert inside a consumer digital camera. Many telescopes have an internal 48-mm thread in the focus tube or feld fattener to accept a 2-inch flter. Te established manufacturers are Hutech (IDAS), Baader, and Astronomik, with new entries from SkyTech and STC. If you hold one of these up to a street lamp, it all but disappears, and while these flters signifcantly reduce specifc wavelengths they do not eliminate light pollution as a whole. At the same time, they ofen improve the color balance (and sky gradient) of an image taken on a color camera. As a result, they are most commonly used with conventional (color) cameras, but not exclusively. Monochrome sensors present more fexible fltering options. Tese sensors do not have Bayer flter arrays and have a broader and higher overall sensitivity. However, they are rarely used without fltration since they are sensitive to UV and IR light, and it is typical to use them with discrete flters, the choice of which afects the transmitted light pollution. A natural color

Pushing the Limits

image requires one to image through Red, Green, and Blue flters (RGB) or perhaps Sloan (g’,r’,i’) flters, for a slightly diferent efect. Some sets, as well as the flter mosaic in front of a color sensor, have overlapping flter responses (similar to human vision). In contrast, others intentionally omit the yellow light wavelengths associated with low-pressure sodium street lighting and yellow light from the deep-sky object. Te big guns in the monochrome arsenal are narrowband flters. Teir incredibly selective transmission passbands exclude all light other than a specifc emission nebula color. Te principal nebulae emissions are fortunately in the blue-green and deep red regions and do not correspond to those from urban lighting. Tis removes most of the light pollution and the associated shot noise. As a result, many inner-city astrophotographers almost exclusively do narrowband imaging of emission nebula. New Lighting Developments Te efciency benefts of LED lighting are hard to ignore. Many countries are steadily replacing older vapor lamp systems with white LEDs. Tis is a worrying problem for astrophotography as, for all its benefts, the backscatter from the ground appears worse. Even more signifcant is the spectrum of the light itself. An LED is monochromatic, with the wavelength set by the manufacturing doping process. A white LED is a combination of separate red, green, and blue LEDs, and consequently, its output covers a broad spectrum and is more difcult to flter out. LED color is defned by its wafer doping; there are no color standards and, unlike a sodium vapor lamp, whose color is determined by quantum physics, there is considerable variation between manufacturers. Our traditional coping strategies are less successful with white LED urban lighting; while some recent light-pollution flters, for example, the IDAS D2 LPS flter, have been designed to remove the intense blue spectrum associated with LEDs, they still pass the lower intensity green and red wavelengths. At the same time, they are reducing useful light from the deep-sky object and on color sensors, may afect color balance. Unfortunately, this intense blue light pollution passes through the blue flter of an RGB flter set, but again, the discriminating nature of narrowband flters will continue to work well. Time and Place Sometimes the only recourse is to image from a darker site, with less light pollution, or when condi-

22

tions are at their best. For instance, the Moon is a regular nuisance and, for about 10 days each month, the refected light from its surface foodlights the sky and degrades image backgrounds (especially when the atmospheric transparency is poor). In addition, moonlight has a broad spectrum, peaking in the orange-red wavelengths. Te obvious coping strategy is to avoid imaging during this period, or when the Moon is below the horizon. One may also consider imaging with narrowband flters when the Moon is up or choose targets away from the Moon in the opposite direction or the two northern quadrants.

Exposure, Exposure, Exposure Finally, our universal solution to reduce random noise (from light and sensor) also improves the efective dynamic range and allows us to capture faint fuzzies. Tis golden rule is to capture more photons. Each time you quadruple the captured photon count, the shot noise level only doubles. At the same time, the sensor read noise in any exposure stays the same. Any combination of the following will help: • • •

use a bigger aperture (diameter, not f/ratio) take longer exposures take and combine multiple dithered exposures

I chose my words carefully; these three are about capturing more light by increasing light intensity or efective exposure duration. It is not the same as increasing the gain (camera ISO). Tis does not afect the amount of captured light and potentially reduces the photon count. Te temptation is to shorten the exposure duration at high gain (ISO) to avoid over-exposure. Sensor gain may afect image noise for other reasons; in some models, especially those with 12- or 14-bit ADCs, increasing internal sensor gain reduces the quantization error, which adds to the read noise. Tis is a marginal improvement, compared to the overwhelming shot noise from typical light pollution. Aperture Each time you double the aperture of an instrument, its photon capture capability quadruples. In astrophotography, the aperture of a telescope is the diameter of the primary mirror or glass lens. Tis is initially confusing to photographers who ofen incorrectly use the same term to refer to f/ratio. A telescope is ultimately a light-gathering device. If you compare the pupil in your eye with a telescope’s aperture, it is easy to appreciate the beneft. Big is better. Over a given period, a

23

The Astrophotography Manual

2,000-mm focal length f/10 with a 200-mm aperture captures 8x more photons than a 200-mm focal length f/2.8 lens with a 71-mm aperture. (Incidentally, an iPhone has an aperture of about 2.5 mm, capturing 6,400x less light.) Telescope prices increase rapidly with aperture size, and there is an economic and physical limit to how big you can go. Exposure Te second option is to take a longer exposure. Doubling the exposure time doubles the photon count and flls up the sensor photosites. At the same time, this makes full use of the sensor’s dynamic range capability and registers faint details out of the noise. Exposure duration and aperture size are two sides of the same coin, each trying to use the full range of the sensor. In one case, it does so by increasing light intensity and, in the other, recording for longer. However, both have the same practical limit, which is the onset of unwanted highlight clipping. Exposure is, therefore, almost always a compromise between the extremes of registering faint details and highlight clipping, and the optimum setting depends on an individual’s artistic goals. For instance, I like my images to have plenty of color variation and, for that reason, my exposure durations use the full range of the sensor and just clip on a few pixels corresponding to the brightest star cores. However, as a colored object brightens, it loses its apparent color saturation and when bright image pixels are stretched, the result is usually an unsightly white blob. Terefore, I double-check that my exposures are not over-exposing a bright galaxy core or nebulae before committing an exposure sequence. Multiple Exposures In the quest to increase the number of captured photons through intensity and duration, the sensor design sets an upper limit, typically in the range of 15,000– 80,000 electrons. Unfortunately, this is insufcient for photographing faint fuzzies in the presence of light pollution. Te solution is to combine multiple exposures. Conceptually, we can either add successive exposures, creating a “super sensor” with unlimited electron capacity or, more usefully, average multiple exposures, usually in a 32-or 64-bit fle format. Tere are subtle diferences between the strategies; quadrupling the intensity allows one to capture the same number of photons in a quarter of the time and halves the random dark noise in the exposure. Doubling the exposure count and doubling the exposure duration also have slightly diferent results; in the for-

mer, we have multiple read-noise contributions and in the latter, just one. For that reason, the optimum exposure strategy is to use the largest telescope aperture that you can aford/lif, with an exposure duration that makes full use of the sensor capacity but keeps clipped pixels to a minimum and then I combine lots of them. Combining multiple exposures increases dynamic range, allowing us to capture faint fuzzies and bright stars in the same fnal image. We have only touched upon this earlier, but as sensor noise levels increase, it reduces our ability to distinguish diferent tones. If the sensor uncertainty is 2 electrons, it efectively halves the ability to discern discrete electron levels. Now the fun part; if we consider four identical deep-sky exposures, they are not truly identical, because of random noise. Tis is due to the photons and electrons occurring randomly in time. Averaging these four exposures also averages (smooths) out the randomness, reducing it by 2x. If our sensor uncertainty was 2 electrons, in the averaged version, it is now 1. Te simple act of averaging four exposures doubles the dynamic range (increased by 1 stop). If we average 64 exposures, we improve dynamic range by 8x or three stops (bits). With sufcient exposures, it is possible to have an efective dynamic range that exceeds the number of levels in a 16-bit fle, hence the recommendation to combine exposures to 32- or 64-bit fles.

Optical Resolution and the Environment Advertising and consumer pressure tempt us to overindulge in telescope purchases for astrophotography. Many optical and physical properties distinguish a good telescope from a bad one. Just like with any other pursuit, knowing what is important is the key to making the correct purchasing decision. In the case of resolution, the certainty of optical performance, backed up by physical equations, is a beguiling one for an engineer. I have to frequently remind myself that these are only reached under perfect atmospheric conditions (which I have yet to encounter). Te fnal image quality in all forms of photography has many factors, and the overall performance is a consequence of all the degradations in the imaging chain. It is easy to misinterpret the image and blame the optics for any defect. Te simple objective truth is that many amateur telescopes resolve fner detail than an astro camera can record and the atmosphere allows. Assuming well-corrected optics, the resolution of a telescope improves linearly with the size of the aperture diameter. Unlike conventional photography, astronomers are more interested in angular resolution,

Pushing the Limits

24

and we conveniently express telescope resolution in arc seconds. As apertures increase, the shimmering of our atmosphere, what we call astronomical seeing, sets a ceiling on resolution performance. Astronomical Seeing Astronomical seeing is an empirical measure of the optical stability of our atmosphere and, conveniently, is also measured in arc seconds as it afects angular resolution. Air refracts light like glass and its index is afected by pressure, temperature, and humidity. Turbulence causes rapid localized changes in air density and parallel beams deviate fg.9 This shows the star size and size variance for exposure times from 0.1 to 60 through refraction, causing stars to seconds. In this case, astronomical seeing is a short-term abrupt anomaly in the shimmer or blur when viewed through atmosphere that occurs every second or so, rather than more frequently. a telescope. At any one time, the light beams pass through adjacent small air pockets with fg.10 shows the theoretical limits of visible light resodiferent refractive indices. Astronomers look through lution for several popular amateur telescope sizes comabout 20 miles of the atmosphere (looking straight up) pared to typical seeing conditions. It is sobering to reand double that, closer to the horizon. Turbulence is alize the limitation imposed by typical seeing condimost signifcant in the denser air near the ground or tions through the atmosphere is equivalent to a telefrom tiny convection currents within the telescope tube. scope with an aperture of ~3 inches (~75 mm). A series of brief exposures or videos show how a star Te overall imaging system resolution is a combijumps about, with some badly blurred and others re- nation of the separate resolutions for the optics, senmarkably sharp. During a long exposure, the photons sor and atmosphere (and tracking errors during the from this twinkling image accumulate onto the sensor, exposure). As a result, it is always worse than the creating a smeared star image. How long is long? I ran weakest link in the chain. In a typical system, these an experiment and measured the star sizes for thou- might be 1, 3, and 2 arc seconds, respectively, making sands of exposures taken at diferent exposure durations the system resolution of 3.7 arc seconds almost 4x (fg.9). With 0.1-second exposures, the stars were con- worse than the telescope’s. sistently small, increasing rapidly and in variation with 0.5- and 1-second exposures. Te trend changed with Coping with Seeing longer exposures, with less variation and diminishing Our options are limited. Seeing conditions are sensisize increases. Tis suggests that, in this experiment at tive to inconsistencies in the dense atmosphere closleast, image shifs occurred around a sub-1-second est to Earth and generally improve with altitude and time-frame, rather than fractional or double-digit expo- proximity to large expanses of water (due to the moderating efect on thermal generation). Te sure values. Astronomical seeing changes with location, season, mountain observatories in Hawaii and the Canary Istime, and weather conditions and can be forecast to lands are good examples of prime locations. Seeing some extent. Some of the weather apps targeted at as- conditions also change with the season and the tronomers include seeing forecasts. Tese apps are amount of daytime heating. Te local site also has an constantly changing and evolving, and a search will es- immediate bearing; it is better to image in a cool tablish the current ones that apply to your location. For open feld than over an expanse of concrete that has a prime site, a seeing condition of 0.5 arc seconds is received a day’s sunshine. Likewise, telescopes are possible, but values in the range of 1.5–3.5 are more best acclimatized to their surroundings for several typical. More ofen than not, the prevailing seeing con- hours before use to reduce convection currents and ditions will limit any telescope’s resolution. Te table in stabilize the lens and mirror dimensions. As-

25

The Astrophotography Manual

tronomers choose remote sites not to be anti-social; they need to fnd high altitudes, clear skies, low light pollution, and low air turbulence. Some also resort to “lucky imaging”, a technique borrowed from planetary and solar imagers. Tis takes thousands of very short exposures, ruthlessly rejects the blurred images, and combines the rest. Te result improves the system resolution closer to the optical limit. However, it goes against some of our golden rules for reducing noise and for it to work efectively, requires a bright target, small pixels (for resolution), large apertures (for resolution and photon capture), sensors with very low read noise, and a capacious disk drive. Tracking, Focus, and Resolution It is worth remembering that seeing conditions are not the only resolution-robbing efect. Tracking errors from mechanical fexure, alignment, and manufacturing tolerances cause a star to drif or shif during the exposure, blurring the result. Autoguiding and complex multi-parameter mechanical and atmospheric refraction modeling can help, if done well, or make matters worse. Tese are complicated subjects and they have their own chapters. As far as the fundamentals are concerned, best practice includes carefully polar-aligning a telescope mount, making sure everything is as rigid as possible, and using autoguiding to correct any tracking errors that the mount cannot correct for itself. In practice, I consistently achieve 0.5 arc second accuracy with autoguiding, better than typical prevailing seeing conditions. Not surprisingly, poor focus also ruins image resolution. Unlike conventional photography, it is not a one-time thing; the mechanicals are less stable; tube length and optics change with temperature, and heavy mirrors move around. Te focus position also changes with the flter in the optical path. Accurate focus, repeated checks, and adjustments are necessary; you guessed it, focusing has its own chapter.

Tere are no shortcuts; the signal-to-noise ratio (and hence quality) of an image is improved by: • • • • • •

using a bigger aperture (captures more photons) taking longer exposures (without clipping) taking and combining more exposures cooling the sensor imaging from a darker site using flters to exclude light pollution

In my semi-rural location, a typical image has 20– 50 hours of exposure with a cooled astro camera over several weeks or months on account of the weather. A year’s efort may result in half a dozen images. Te prevailing weather conditions dictate the terms of this hobby and are an unavoidable consideration. Keen amateurs are relocating their system to a remote dark site in increasing numbers, in search of more frequent clear nights and the chance to operate at an alternative latitude. I have started using dual-camera systems to make the most of each rare opportunity. One of the exciting developments in acquisition sofware overcomes a limitation of existing interface standards and allows dual-imaging systems to operate more efciently in tandem on a single mount. Tis enables doubling-up exposures or, with careful alignment, side-by-side images for a wide-feld mosaic image.

Summary Image noise is the major hurdle in the path of achieving quality results. Te mind-bogglingly low light levels from the object are competing with larger forces. In many ways, image noise has more impact than focus and tracking errors, which are more process and mechanical in nature. Unfortunately, image noise is ofen not entirely understood, and many practitioners concentrate on noise sources within the sensor and forget about the random nature of light itself.

fg.10 The chart above indicates the difraction-limited resolution for visible light, in arc seconds, for any given aperture in relation to the practical limits imposed by typical seeing conditions ( 0.5–3 arc seconds).

Pushing the Limits

26

Optical Resolution Physics has a habit of dictating what we can and cannot see in amateur astrophotography.

F

ollowing on from the previous chapter, it is worthwhile delving a little deeper into optical system resolution. We already appreciate that telescopes, sensors, and our atmosphere have a big part to play in what we can see and that the overall result is worse than the weakest link in the imaging chain. Te frst step to understanding this in greater depth is to clarify what we mean when discussing resolution. Resolution and Sharpness Many amateurs and not a few professionals confuse resolution and sharpness. Tey are not entirely unrelated but, in an image, they convey diferent visual attributes. In simple terms, resolution is the ability to discern two close objects as separate entities. It has a physical measure. Resolution is measured in several ways in traditional photography. A popular amateur method is to use a test chart, like the venerable USAF1951 or more modern ISO12233 targets, and measure the resolution at the scale where the image contrast drops to 5%. A more scientifc method is the Modulation Transfer Function (MTF). MTF graphs typically record the contrast of a small, medium, and coarse-scale pattern from the image center to the edge. Te better the optics, the higher the contrast. Te clever thing about MTF characteristics is that the overall MTF of an imaging system is the multiplication of the individual component MTFs and ideally should also consider the result of multiple, registered, and interpolated images, well beyond the scope of this book. Image sharpness has no agreed measure but is our perception of contrast between adjacent light and dark areas, especially in the transition area. We can also use test charts and MTF to infer sharpness, as they measure contrast. One proposal is to use the contrast measure of a coarse scale to compare sharpness between optical systems. Tere are a couple of other distinctions; post-exposure image manipulation cannot restore lost image resolution, but it can increase image sharpness. Te traditional unsharp mask tool (a digital remake of an analog flm process of the same name) can potentially create a target contrast that exceeds 100%. It is not pretty. Te other distinction is that it is possible to have a highly resolved image with generally

low sharpness (think old Leica lenses). In my mind, sharpness is overrated. Before we leave sharpness behind, however, it is also worth mentioning that mild sharpening ofen improves our perceived resolution of coarser image details but, at the same time, ofen bludgeons delicate detail. It also accentuates image noise. Tis was especially true of early sharpening tools, which were generally awful, giving early digital images lousy press. Tey have since become increasingly sophisticated. One such specialist tool used in astrophotography is deconvolution. Some other tools are increasingly using artifcial intelligence to guess missing details. System Resolution It is essential to realize the fnal image quality in all forms of photography has many components, and the overall performance combines all the imperfections in the imaging chain. In general photography, the common lp/mm (flm) or lines per image height (digital) are common measures of optical resolution. Tese do not relate well to celestial object separations that are defned by angles. For that reason, astronomers quote angular resolution in arc seconds (”), arc minutes (’), or radians. An arc minute is 1/60°, and similarly, an arc second is 1/60 arc minute (1/3,600°). Tere are 2π radians in a circle, making a radian about 57.3°. If we had MTF characteristics for each element, we could multiply them together to predict system resolution. In astrophotography, we make a simpler equivalent computation using the resolution of each element in arc seconds. Suppose x, y, and z are the arc second resolution for say atmosphere, telescope, and sensor. Te system resolution is then given by the following equation, defned by the quadratic sum of the various components: system resolution = √(x2 + y2 + z2) Tis is a helpful equation and highlights how easy it is to make the wrong judgment and blame the optics for any defect. For instance, long before digital cameras were popular, the premium 35-mm camera optics had more resolution than could be recorded on flm. In the case of a lens and a fne-grain flm inde-

27

The Astrophotography Manual

pendently resolving 200 lp/mm, the measured system resolution on flm is close to 140 lp/mm. It is easy to forget the combined efect; it was not uncommon to fnd self-proclaimed experts conducting a lens test using a digital camera with a sensor resolution equivalent to 50 lp/mm, half that of a medium-speed monochrome flm!

fg.1 Difraction at a circular aperture causes a point light source to form a difuse blob. The FWHM is defned by the separation of the 50% intensity points. The HFD measurement is similar but has a diferent threshold, which encloses 50% of a star’s fux. At the same time, this is the point spreading function (PSF) of a perfect telescope with a fnite aperture. The resolution of a system is very similar to its FWHM value and for practical purposes, the same.

Optics and Resolution Many telescope designs are general-purpose, and are used for visual and imaging purposes. Te needs of the visual observer and the astrophotographer are diferent, leading to commercial and design choices. For instance, the human eye has a higher resolution and less sensitivity than a conventional sensor but can dynamically adjust focus as it roves over the feld of view. Visual users value apochromatic refractors that focus all wavelengths of light at the same point, a quality also appreciated by those imaging with a color camera or a monochrome camera with a luminance flter. It has less signifcance if separately-focused exposures are taken through narrowband or individual red, green, or blue flters and combined during image processing. Te vision through the eyepiece on a color-corrected doublet or triplet refractor is an illusion, as this image, projected onto a fat sensor, will not be in focus across the entire frame. Unlike camera lenses, refractor optics focus light onto a curved surface and require additional negative lens elements near the sensor to focus to a fat plane. Tese optical cells are either a separate optical module called a fattener or reducer/fattener or, in the case of an astrograph, integrated into the telescope design. It is usually the case that the fattener or reducer has an optimum spacing to the sensor. Some designs are optimized for a particular telescope but most are general-purpose, suiting a range of focal lengths and apertures. To get the best corner-to-corner resolution requires a unique spacing, found by trial and error and which changes with focal length and aperture ratio. Refector telescope designs have their particular optical issues depending on their design. Spherical, elliptical, and hyperbolic mirrors, also require optical correction with glass elements, either at the entry or exit of the optical system of imaging purposes. Resolution, Difraction Limits, and FWHM Although a star is a massive object, it is so distant that it should focus to a single pixel. Even perfect optics cause it to appear as a difuse blob, due to difraction. Te brightest part is at its center, and one measure of its blobbiness is the diameter at which its intensity is half its peak value (fg.1). Tis defnes the Full Width Half Maximum, or FWHM for short. Another similar measure is the Half Flux Diameter (HFD), the diameter within which half the photons fall. Either are ofen displayed and used by image capture and focusing applications to measure goodness. For instance, most autofocus algorithms assume the optimum focus is the position at which this measure is at a minimum. Te unit changes with the application; it is reported in sensor pixels but, if the imaging scale is known, arc seconds. Scientifc applications prefer radians. Assuming perfect optics, the size of our stellar blob is purely dependent upon the wavelength and the aperture diameter. Te minimum FWHM (in radians) of a point image for a wavelength λ and aperture diameter D (in the same

Pushing the Limits

units) is given by the equation below and pictorially in fg.1: FWHM = 1.03 ∙ λ / D Difraction limits our ability to distinguish neighboring stars and is similarly dependent upon the aperture and wavelength. Te theoretical resolution for resolving two close objects is shown below and pictorially in fg.2. It is known as the Rayleigh Criterion.

28

robs the optical resolution during long exposures, as do the tracking and focus errors. Tere are other factors at work, in astrophotography, that are associated with optical resolution. For example, one of the less desirable visual outcomes of image stretching is the bloating of bright stars. While optical difraction has a part to play, star bloat is mainly a result of light scatter and seeing through the atmosphere and from the optical surfaces. All glass surfaces should be multi-coated to reduce internal refections and kept clean and condensation free, with dew shields and heated bands around the exposed optics for best results.

Aperture has several meanings, especially if one has a background in traditional photography. Astronomers refer to the aperture of a telescope as the diameter of the main glass element or mirror. In photography, “aperture” is ofen used incorrectly to imply the aperture ratio, (f/stop) as well as the physical adjustment mechanism.

resolution (radians) = 1.22 ∙ λ / D Te equations are very similar, difering only by about 20% and, for practical purposes, can be treated as the same and more conveniently in diferent units: resolution (arc secs) = 0.251 ∙ λ(nm) / D(mm) Te interesting feature of these equations is the resolution improves with aperture diameter and is signifcantly unafected by focal length. Tese equations work for a simple aperture (i.e., a refractor). Refector telescopes typically have a central obstruction and a more complex equation because of the additional difraction from the circular obstruction, which reduces resolution. Practically, I cannot detect any meaningful resolution improvement moving between my 6-inch refractor and 10-inch Ritchey-Chrétien refector telescopes. Visual astronomers particularly value optical resolution for splitting double stars but how much resolution do we need for astrophotography? I would propose it is not a priority where the stars are the supporting act and appear randomly sprinkled throughout an image with plenty of space between them. We do not necessarily require a high resolution to see them, only contrast and, to a large extent, image processing to enhance their appearance. Tose nebulae and galaxies with indistinct object boundaries or subjects spanning a wide feld do not necessarily require a high resolution either. On the other hand, a high angular resolution is useful to distinguish the individual stars in a globular cluster. Modern manufacturing techniques polish and assemble optics to a very high standard and more efciently. Most amateur telescopes have more resolution than the digital sensors we use. Even then, atmospheric turbulence (seeing) along the optical path

Sensor Resolution An imaging sensor has a grid of photosites of fxed pitch, typically 2–9 microns. Sensors are either based on CCD or, more recently, CMOS technology. While the conversion efciency varies a little with color, the photosites are monochromatic. A natural color “pixel” requires a combination of exposures taken through red, green, and blue flters. Tis is achieved either from separate exposures, taken through a flter over the entire sensor, or a single exposure through a color flter array (also called CFA or Bayer array) bonded to the sensor. Astrophotographers use both approaches, each with benefts and drawbacks and are discussed later on. In terms of resolution, however, combining adjacent fltered pixels in a Bayer array to form a color pixel has a slightly detrimental efect on resolution. Te pitch or spacing of the photosite grid directly afects the potential image resolution. Up to now, the discussion has been about angular resolution and, to consider the physical relationship between angular and linear measures on an imaging sensor, we need to take into account the focal length fL of the optics. Tis starts with the pixel scale, the angle subtended by 1 pixel (arc seconds per pixel). Tis usually requires trigonometry but can be simplifed as the tangent of a small angle is the same as the angle expressed in radians. In the equation below, fL is in mm and the pixel pitch is in microns: pixel scale (arc sec/pixel) = 206 ∙ (pixel pitch / fL) To understand how this afects image resolution requires us to consider how the Rayleigh Criterion ap-

29

The Astrophotography Manual

plies to sensors. It requires at least three adjacent pixels to resolve a close pair of stars. Most experts agree on 3.3 pixels to guarantee the resolution of two points (fg.2). (Stars do not always align themselves conveniently with the sensor axis, and one must consider all angles.) Te angular resolution of a sensor is therefore about 3.3x its pixel scale in arc seconds/pixel. To match the optical resolution: 680 (pixel pitch / fL) = 0.251 ∙ λ(nm) / D(mm) or… pixel pitch = 0.18 ∙ F (F is the aperture ratio) Let us consider the case of an f/7 refractor, with a measured focal length of 924 mm, an aperture of 132 mm, and a sensor with a pixel pitch of 4.6 microns. Te telescope has a difraction-limited resolution of approximately 1 arc second, but the sensor’s resolution is 3.4 arc seconds. For the sensor to match the difraction-limited performance of the optics, it would require a much smaller pitch of 1.26 microns. At frst, this looks pretty damning, but another consideration is the efect of astronomical seeing. A sensor resolution of 3.4 arc seconds is only marginally worse than typical seeing conditions of, say, 3.0 arc seconds. One has to consider the entire system resolution and its sensitivity to one or more contributions:

better resolution (remember, each time you halve the pixel pitch, the pixel count quadruples for any given area) but this advantage is ofset by a host of other performance issues associated with vast numbers of small photosites.

Many optical equations use radians for angular measure. Tey have the property that for very small angles, the sin or tan of that angle is the same as the angle expressed in radians. Tis provides a handy method to simplify formulae for practical use. Tere are 2π radians in 360 degrees and as most of us are more familiar using degrees, nanometers and millimeters, rather than radians and meters, you will encounter, in more convenient equations, the numbers 206 in varying powers of 10 to convert angular resolution into arc seconds.

image resolution = √(0.92 + 3.42 + 32) In this example, the overall resolution is 4.6 arc seconds. If we halve the sensor pitch to 2.3 microns, it would only improve the system resolution to ~3.6 arc seconds. Tis is still a best case, as the resolution is degraded further by atmospheric refraction and tracking errors. In this typical setup, the telescope’s optical difraction has little infuence on the fnal resolution, as the efects of the sensor and atmosphere dominate. Te seeing and sensor resolution are similar though, and while sensors with a fner pitch can be used (albeit with other issues), in astrophotography the most difcult thing to change is one’s environment. All these factors are evaluated in the balance between resolution, image requirements, feld of view, signal strength, cost, and portability. Te conventional wisdom is to choose a sensor whose efective pixel scale is about 1/3rd of the limiting condition, either the seeing condition (as in this case) or the difraction limit of telescopes with small apertures. Smaller pixel pitches have theoretically

fg.2 Optical and sensor resolution both afect the hardware performance; here, up to 3.3 pixels are needed to discriminate two close stars and is broadly equivalent to the FWHM value.

Pushing the Limits

Efect of Poor Focus We are accustomed to smooth helicoid focusers on flm cameras and invisible autofocus mechanisms that can predict subject motion and likely focus position. Whereas yesteryear's large, medium, and full-frame formats had ground glass screens, split-image, and micro-prism focus aids, modern cameras use their small viewfnders for composition. Similarly, the concept of depth of feld or hyper-focal distance are largely forgotten. Instead, focus and exposure have become something we take for granted through technology. For many starting astrophotography, the efect of poor focus and the challenges are a surprise. In my early days of astrophotography, I ruined many hours of exposure with poor focus, blurring small details, and bloating star sizes, at the same time. Poor focus also makes optical aberrations more apparent towards the edge of the frame. In this discussion, we are interested in the efect of poor focus on image resolution rather than methods to focus accurately. Te illustration in fg.3 shows the concept of a focused beam of parallel light (refractor or refector). At the best focus, it does not quite become a singular point, due to difraction and the other issues we have previously mentioned. Autofocus systems (sofware and hardware) have tolerances and during an exposure, the focus point can shif too. Tere is a zone, however, on either side of the best focus position, which gives an acceptable result and for which there is general agreement. Tis is exceedingly shallow and is called the Critical Focus Zone or CFZ. Te following equation is used to calculate this zone for a telescope with an aperture ratio F and wavelength λ: CFZ = ±2.44 ∙ λ ∙ F2 Te thing to note here is that it only changes with f/ratio (the focal length divided by the aperture diameter). In practice, the William Optics FLT132 refractor used in our earlier example is an f/7 optic and has a CFZ for green light of about ±0.07 mm. Te difraction-limited resolution is 0.9”, and a focused star has a FWHM of about 4μ, similar to the size of a single pixel in this case. If we shif the focus position by 0.1 mm, this grows about 4x, covering 16 pixels. In comparison, if I imaged this with prevailing seeing conditions of 2” it would only double in size. Comparing this CFZ value with the efect of thermal expansion is useful. A 1-meter long aluminum tube shortens by 0.07 mm for a 4 °C drop in temperature. Te optics deform too, changing the focal length,

30

fg.3 A conceptual parallel beam of light focuses down to a small fnite blob, defned by the aperture and optical prowess. Either side of this point, the cone of light rapidly grows. Fast apertures (e.g., f/2) are more sensitive to focus inaccuracies than slow apertures (e.g., f/8).

sometimes reversing the expected focus shif. In practice, I acclimatize any telescope for an hour or two before imaging. I also monitor the ambient temperature during imaging, re-focusing each time the temperature changes 0.5–1 °C. Te focusers on my telescopes have gear-reduced stepper motors. A single step is about 4μ, making the CFZ for this f/7 optic about ±17 steps. Typical aperture ratios used in astrophotography range from f/2 (RASA) to f/10 (SCT), a 25x range of sensitivity to focus error. In everyday photography, the subject matter ofen disguises many optical issues; I have what are considered quality primes for an APS-C mirrorless camera and in conventional use, they bear out this judgment. However, if I photograph the night’s sky with them, the images show some coma, chromatic aberration, and curved focus. Te problem is that we inherently know that a star should resemble a tiny round dot, making it easy to spot the result of poor focus and optical aberrations. Knowing that stars are supposed to be round dots does have its advantages though, in the form of the mysterious deconvolution technique. Efect of Poor Tracking Poor tracking over the exposure duration is the equivalent of camera shake in photography and has a dramatic efect on resolution. Telescope mount tracking errors will elongate a star along the RA axis and polar misalignment, fexure, and the uncorrected efect of atmospheric refraction will cause drif in either RA or DEC axes, smearing the image. External stimuli do too, e.g., wind, cable drag, and so on. We use a host of mechanical, sofware, optical, and assembly best practices to minimize these error mechanisms. Even so, nothing should be taken for granted and the outcome of a dozen exposures always yields a range of efective

31

The Astrophotography Manual

resolutions, and for the best resolution, it is essential to sort through the images as the frst step during preprocessing. Some applications (both acquisition and processing apps) analyze each image in turn and provide a sorted range according to statistical data. In these, I discard those exposures whose stars have a higher-than-average FWHM or eccentricity (elongation) metric. I also reject any exposures compromised by poor transparency, usually detected by an unusually low star count or a high mean level. Point Spreading Function and Deconvolution Deconvolution is a process that slightly improves spatial resolution and defnition. It relies upon a Point Spreading Function (PSF), which describes the efect of the optical system on a point light source. In astrophotography, this conveniently describes the appearance of a star in an image fle, the result of the entire optical path from the star to the sensor. Te smudgy appearance in the image is said to be convolved. Not surprisingly, the PSF characteristic is very similar to that of a difraction-limited star (fg.1). Te argument goes if you know what the star should look like and what it does looks like, it should be possible to create an opposite transformation to re-create the perfect star image. Tis is the essence of deconvolution, a post-processing technique used extensively in astrophotography and microscopy. Furthermore, the argument goes, what is good for star correction is good for everything else. Well, not quite. Deconvolution makes stars smaller and less fuzzy and can make other details clearer too (fgs.4,5). Te complex iterative algorithms use an artifcial or measured PSF from sampled stars and also have the potential to correct elliptical star shapes too (providing they are consistent across the frame). However, a deep-sky image is not a single dot but a complex mixture of pictorial elements and no simple transformation can untangle the blurred photons. As such, it is not a perfect process, emphasizing noise and creating dark rings around bright objects (afectionately called “Panda eyes”). In terms of resolution, deconvolution is a valuable tool, but it cannot work miracles. It also works best with over-sampled images, in which a star’s FWHM diameter is at least 3x the sensor’s pixel pitch. (In the case of under-sampled images, where the pixels are too big, dither and drizzle techniques during capture and processing, respectively, may improve resolution closer to the optical difraction limit and increase image noise slightly.)

fg.4 This crop from IC 1848 (Soul Nebula) is the result of 41 x 5minute calibrated, registered, and combined luminance exposures. No other processing apart from an image stretch.

fg.5 The efect of deconvolution is subtle. Small stars are more obvious, larger stars smaller and some double stars are now apparent. The nebulosity has more defnition too, though it might be difcult to see in half-tone reproduction.

fg.6 The full, false-color narrowband image (HαSIIOIII). The star color is corrected using a few hours of RGB-exposure.

Pushing the Limits

32

On the Right Track Last, but not least, accurate tracking is an essential requirement for deep-sky astrophotography.

T

he last chapter on essentials considers the remaining linchpin for efective image capture. Finally, we have an idea of how well an imaging system resolves and exposes extremely faint objects under a shimmering, light-polluted sky. What remains is the “simple” task of keeping the telescope on track during each exposure. When planning the system, it is the mount, not the telescope or camera, that has the lead role and is the most important consideration. Before we even consider rotating a telescope around the celestial pole, we need to realize that every mechanical interface in the support, mount, and imaging system afects the general stability, and, in practice, it cannot be assumed. Te same goes for every component’s stifness, as wood and metal bend under load, and when we measure tracking errors in fractions of arc seconds, a tiny mechanical anomaly has a signifcant impact. Even with a perfect mechanical system, our environment has a few curve balls to throw at us. Understanding and respecting these are essential for accurate tracking over long exposures. When is Good, Good Enough? Good tracking aims to keep the feld of view unchanged during each exposure. Tis is a combination of two factors: center location and rotation. Te frst is afected by polar alignment, atmospheric refraction, fexure, and the mount’s tracking error. Te second is usually a secondary efect, caused by poor polar alignment, and made worse by the choice of guide star, which cause the image to rotate slightly during the exposure about its center point or the guide star. Autoguiders measure tracking errors in guide camera pixels. If the guide system’s pixel size and focal length are known, the imaging scale is additionally reported in arc seconds. Errors can be positive or negative, and the reported measurement updates a root mean square (RMS) value as each error is calculated. Te RMS value is not the peak error but statistical average magnitude. I aim to achieve 0.3-0.6 arc seconds. If it increases above 0.8, stars noticeably grow or elongate and resolution sufers. To prevent feld rotation from becoming a potential concern, I ensure my polar alignment is within 5 arc minutes, which is easily achievable.

In a Perfect World Except for simple trackers, telescope mounts have a system with two motors, set orthogonally, to center the imaging system on the target and continuously track it. Tey come in several alternative architectures, but all have one axis orientated to the celestial pole that rotates with a constant and accurate tracking speed. With few exceptions, all mounts track a target’s movement around the sky with a single motor: Te object’s declination is fxed and the mount has to “simply” rotate around the right ascension (RA) axis. If only! Perfection is an RA axis pointing directly at the celestial pole, with no fexure in the entire system, a perfect drive mechanism, and no atmosphere. Even if the mount was perfect, it is useful to consider what is outside our direct control and adjust our perspective. Drift from Atmospheric Refraction Ours is not a perfect world, and there is no such thing as perfect polar alignment. It is a compromise between diferent error states, such as drif and rotation. Te real celestial pole is not where we think it is, and stars do not “simply” orbit in a perfect circle either, on account of atmospheric refraction. A perfectly smooth motor and mount will lose track as it traces a star across the night sky. (Te two exceptions are if you image from the North or South Pole.) So, what does this mean in practice? At sunset, when the Sun appears to be just touching the horizon, it is already fully set, about 0.5 degrees lower. Te precise fgure is afected by pressure, humidity, temperature, and wavelength. At 10 degrees above the horizon, the error has dropped to about 0.1 degrees, and for a London latitude, the actual celestial pole is about 1 arc minute lower than its apparent position. As a star sets, it appears to slow down. From an imaging standpoint, in an otherwise “perfect” system, if one is imaging at 30 degrees altitude, a star drifs of track by about 1 arc second per minute because of atmospheric refraction. Tat drif rate halves at 45 degrees. Tis drif does not necessarily confne itself to the RA axis, and any long-duration narrowband exposure requires a “Plan B”. Te obvious solution is to correct for atmospheric refraction. Tis is difcult to do precisely;

33

The Astrophotography Manual

our atmosphere is a complex non-uniform medium and its refraction continually changes. We know, in general, that its refraction properties are afected by altitude, ground-level pressure, temperature, and humidity. Several studies have arrived at equations that model the refraction but are frst-order approximations. Refraction increases by approximately 1% for every 0.9 kPa increase in pressure and every 3 °C decrease in temperature. Planetariums and high-end mounts typically use these models. Even then, long unguided exposures are difcult to achieve, even with a high-quality mount, sky model, and accurate polar alignment. Drift from Polar Misalignment For some, accurate polar alignment is a crusade to achieve unguided operation, with a growing armory of clever tools and techniques. Good polar alignment is considered something around 1–2 arc minutes and excellent alignment 0.5’ or better. It should be possible to achieve alignment within 2–5’, with minimal efort, using a traditional polar scope (that has been centered). A 5’ alignment error on an otherwise perfect imaging system will have image drif of 6.5 arc seconds over a 5-minute exposure (at low declination). In a typical system, this will cause noticeable streaking of 5–10 pixels. Tere are more potential issues; if on a tripod assembled system, that has been carefully polar aligned, one foot sinks slightly during an exposure, the polar alignment and tracking error will sufer. Te imaging system is extremely sensitive to micro-movement; in the case of a tripod, whose feet are 40” apart on sof ground, if one sinks by just 0.1 inches, it causes the tripod to tilt by 0.16°. Tis does not sound signifcant, until one converts 0.16° into arc minutes (~10’) and evaluates in the context of polar alignment error… and a worse-case drif of 13 arc seconds over a 5-minute exposure. In summary, a simple thing like sof ground can ruin careful polar alignment and tracking, and, more worryingly, do so without your knowledge. The Imperfect Mount Mounts are not perfect and their drive systems introduce angular tracking errors into this reality. Te best known is the cyclical error that corresponds to the mechanical tolerances of a worm or drive as it rotates. Tis is known as Periodic Error or PE, a consequence of the RA axis alternatively rotating a little fast and slow for diferent gear angles. It is not the only tracking error; every part of a rotating system has cyclical toler-

ances. Tese combine to form a complex cycle of tracking errors, though usually with less impact than PE. During a long exposure, PE causes a star’s image to elongate along the RA axis to an oval, or worse. Although linear tolerances cause PE, the recorded error is angular, typically in arc seconds. Te declared peak-to-peak tracking error is only part of the story. Some specifcations are worse-case; others are typical values. Quoted specifcations vary widely, between 1 and 60 arc seconds before correction. Price and brand do not always guarantee good PE performance; many forums indicate signifcant diferences between identical models, an unavoidable outcome of manufacturing tolerances. Several manufacturers side-step the issue entirely and do not specify a value, declare a value afer PE correction, or their drive system has zero PE. At best, PE specifcations are an indicator of tracking error, and at worse, they are misleading. It is instructive to compare the typical magnitudes of tracking errors over a 5-minute exposure. Assuming a 5’ polar alignment error, and a ±15” peak-to-peak periodic error, the tracking errors are: • •

polar error 6.5 arc seconds (RA or DEC) periodic error 15–30 arc seconds (RA)

In this example, the drif from periodic error is the dominant error over the exposure and needs correction, for example, using Periodic Error Correction, which typically removes about 90% of the error. Tis improves things, leaving a signifcant tracking error over the exposure. An obvious solution is to use autoguiding, which typically looks to measure and correct tracking errors every second or so. Autoguiding Perspectives An autoguider cannot instantly correct a tracking error; the cycle of exposure, calculation, command, and movement takes time. It works retrospectively and the cycle time limits how fast an autoguider system can correct a tracking error. It also afects how quickly it can react to a changing tracking error. In autoguiding systems, the rate of change of the tracking error signifcantly infuences the efectiveness of any autoguider system. Our altered perspective must consider the efect of polar misalignment and periodic error on the rate-ofchange of tracking error. We now need to re-examine the respective tracking errors over a typical guider cycle of, say, one second, during which the image will also temporarily shif due to atmospheric seeing.

Pushing the Limits

34

In the case of the drif from polar misalignment, this is a simple division and using the above example is a negligible 0.02”/second. In the case of periodic error, if we assume the error is a smooth sinusoidal, for a worm with a cycle of t seconds, the approximate peak rate of change is related to the peak period error value PE, as the tracking error changes polarity: arc secs/second = 6 ∙ PE / t In our example, the tracking error changes by 0.3”/ second, ~10x more than the efect of polar misalignment, but still within the capabilities of an autoguider. Tese values are typical of popular mounts. Better models will have lower PE or have PEC, but, at the same time, the polar alignment will likely be better too. Tese tracking errors, however, pale into insignifcance compared to the efect of atmospheric seeing, which rapidly changes 1–3 arc seconds from one second to the next and is about 10x larger still in magnitude. Unless one uses adaptive optics, the response time of an autoguider and a mount cannot keep up with the transitory tracking errors caused by atmospheric seeing. Worse still, if the autoguider settings are too aggressive, it starts to react to seeing conditions, and it is more likely to introduce tracking errors of its own making. It is a common misperception that any autoguider tracking graph represents the mount’s tracking error during the course of an exposure. It certainly is an indicator but to use it with any precision requires the guider outputs to be disabled and to use a long exposure time (4–10 seconds) so that the random efect of atmospheric seeing does not obscure the assessment. Fig.1 shows how the apparent tracking error, as reported by PHD2, using a long guider exposure is better than a short one when they are really similar. Summary Te essential takeaway from this chapter is an appreciation that the sample-to-sample tracking error, as measured by an autoguider, is primarily made up of atmospheric seeing (random noise) and usually dwarfs the underlying real tracking error caused by any periodic error in the mount, fexure, polar misalignment, and atmospheric refraction. Te trick with autoguiding is to fnd the balance between reacting to real tracking errors and not overreacting to seeing conditions. Less is ofen more.

fg.1 These two screen captures, from the PHD2 autoguiding application, show the RA (blue) and DEC (red) tracking error for the same mount. These were taken immediately in succession with the same system. In both cases, the autoguider output is disabled, so that it simply records the star position error. In the top image, the exposure time is just 1 second and most of the apparent tracking error is simply the efect of random seeing conditions on the apparent tracking error. The bottom trace uses a 4-second exposure, which averages out much of the atmospheric seeing and produces a smoother graph and is more indicative of the underlying tracking error of the mount. The slight overall slope in both cases is an indication of the underlying drift and periodic error.

Heart and Soul Nebulae

System Choices

System Choices

36

New Trends A surge in amateur interest, and the unrelenting development of innovative hardware and software, continue to advance astrophotography and make it more afordable and accessible.

T

he astrophotography industry is constantly re-inventing itself, due to new technologies, advances in manufacturing, sofware, and novel applications of existing technology. Some others are market-driven, especially for lightweight quality systems. “New” is perhaps a misnomer; it may simply be a point where an item or idea is more than an interesting curiosity. For example, in previous books, I classifed CMOS astro cameras as a new trend. At the time, the available models were too unreliable to recommend although it was obvious that, once remedied, they would shake up the established industry that had largely stagnated due to the slow-down of new CCD sensor development.

Optics Te fundamentals of optical refraction and refection have not changed. What has happened, however, is the introduction of durable, low-dispersion glasses and more efcient manufacturing processes. Tere are now many inexpensive doublets, with near APO performance, using these glasses and large refector telescopes, taking advantage of more afordable hyperbolic and parabolic mirrors. Tese feed the surge of new interest in astrophotography and, for more established users, new optical designs that make the most of the latest full-frame (or larger) sensors. In recent years, several well-regarded refractors have been discontinued because of their small imaging circle and expensive glass. You can see this trend in their camera couplings; when I started, the 42-mm Tthread was the standard, whereas today, it is increasingly a 48- or 54-mm thread on the back of a telescope or fattener. With so many new small and medium telescopes (although a closer look suggests many share the same optical cells), it is interesting to pick out a few to tempt the more seasoned astrophotographer. RASA Te original SCT (Schmidt-Cassegrain Telescope), made popular by Meade and Celestron, was an all-purpose telescope using a mixture of mirrors and correctors. Long, slow focal lengths are not ideal for a lot of deepsky imaging. Celestron ofered the HyperStar™ modifcation that replaced the secondary mirror with a multi-element lens coupled with a slim in-line camera mounted out the front. Tis shortened the focal length and dramatically improved the aperture ratio. Later, a large purpose-designed imaging version was developed to improve the image quality, the RASA (RoweAckermann Schmidt Astrograph), followed by a smaller and less expensive model. Tese designs use one spherical mirror and multiple glass elements to correct and fatten a wide imaging feld. Te aperture ratios are about f/2 with 8- or 11-inch mirrors. Compared to a typical f/5–f/7 refractor, the design is more sensitive to focus, and collimation errors but individual exposures and sessions become much shorter as a result.

fg.1 As imaging sensors increase in size, the telescope couplings are also increasing to ensure there is no vignetting. This illustration has scaled the coupling diameters as they would appear on an f/4 telescope, at the typical 56-mm spacing to the sensor. The 54-mm is just big enough for a full-frame sensor and the once common 42-mm coupling is being displaced by 48-mm to ensure adequate coverage of APS-C and FF.

fg.2 The AstroTrac 360 is an innovative, portable, premium mount with extremely low periodic error. It has two main confgurations, GEM (as seen here) or as an accurate RA tracker (without time limitation). In the GEM confguration, small loads are balanced by sliding the DEC motor position and may not require any additional counterweights.

37

The Astrophotography Manual

fg.3 This Rainbow RST-135 mount is ridiculously small and is usually dwarfed by its payload. It uses strainwave gearing, commonly found in robots, aerospace, and lunar rovers. It uniquely does not require balancing or counterweights. Unlike the AstroTrac and Fornax units, however, it does have cyclical tracking errors that require guiding out when imaging. Several manufacturers have announced their own compact mounts based on the same technology.

Tese conditions make them ideal for creating deep, real-color images using a color CMOS camera; over an equivalent imaging period, these fast scopes produce amazingly deep images, revealing faint nebulosity. However, the centrally-mounted camera is an obvious light obstruction and while it is possible to use conventional cameras, the small cylindrical astro designs, with their short fange distances, are better suited. Fast aperture ratios in general, however, work less well with standard astronomical dichroic flters. Te coating design of these flters assumes normal incident light, and their passband changes with the angle of incidence. Several flter manufacturers have developed new flter coatings specifcally for use with fast aperture ratios. Newtonian Astrographs Traditionally, Newtonian telescopes used small secondary mirrors optimized for visual use. Unfortunately, these have a short back focus, which makes imaging a mechanical challenge. Te latest astrograph designs reposition the secondary mirror to extend the back focus. As a consequence it is larger but ofen incorporates a corrector lens in the focus tube to fatten the focus plane. A few years ago, these would have been f/4 devices, but the latest are afordable f/2.8 instruments. Unlike RASA telescopes, these suit visual use too (and with bulky flter wheels, without worrying about obscuration), but, in common with RASA designs, these fast-aperture instruments are sensitive to focus and collimation and beneft from a sturdy construction to maintain alignment.

Mounts Worm-driven mounts have dominated the amateur and enthusiast market for many years. Te trend of replacing spur-gear couplings with toothed belts is almost complete, and even rotary encoders, once the preserve of high-end mounts, now feature in amateur models. More recently, a few models have launched using diferent drive mechanisms. Astrophotography places an incredible demand on smooth rotation, and introducing a novel drive is risky. Two such models to satisfy the highend portable market trend are the AstroTrac 360, using an advanced friction direct drive, and the Rainbow RST-135, using a harmonic drive, joined by several competing models.

fg.4 The StarAid Revolution is an integrated autoguider/polar alignment/plate-solving widget housed in a guide-camera body. Using a WiFi-coupled smartphone, its apparent simplicity and ease of use hide the sophistication of its intelligent control and measurement algorithms. This unit does not require a PC and additionally has a trigger output for a DSLR/mirrorless camera and will dither between exposures through its ST4 interface.

Hybrid Tracking Mounts Like the simple Fornax LighTrack tracker, the AstroTrac 360 drive rotates a polished metal wheel pressed against a circular track. In both, the simplicity of the machining enables close tolerances, and both deliver tracking with little periodic error. Te novelty of the AstroTrac design is it breaks the 2-hour constraint and allows 360-degree rotation. It also has the unique facility to bolt two tracker units together to form a dualaxis mount. Tese deliver GEM-like operation and ensures accurate pointing and tracking using in-built rotary encoders. Tis stylish, premium design adapts to several ultra-portable scenarios and, as with a conventional GEM, requires balancing; small payloads are balanced by sliding the DEC tracker position, but heavier loads require a traditional counterweight system. Te single tracker confguration weighs about 2.5 kg, which doubles for two-axis operation. Te counterweight system in-

System Choices

38

creases this further to a respectable 8 kg and a payload of 10 kg. Tis mount is controlled through a wireless browser interface and for imaging, with a custom X2 plugin for TeSkyX and an INDI driver. Harmonic Drives For 30% more, the Rainbow RST-135 mount, one of the smallest harmonic drives, accomplishes two-axis imaging (Alt/Az and GEM confgurations) in an even more compact and lightweight package. Harmonic drives (a.k.a. harmonic gearing or strain wave) deliver high torque density and are commonly used in industrial robots. Tey uniquely do not require balancing and reportedly track more accurately when unbalanced. Although they have an optional balance shaf, this is mainly to ensure the stability of a portable setup. Tis particular mount is fully ASCOM-compliant, weighs just 3.3 kg (without a saddle plate) and has a 13.5-kg payload, at 4x its weight. Tis is a remarkable achievement; a conventional GEM’s payload is usually similar to the combined weight of the mount and counterweight system. In addition, it has a fully-featured handset and has USB and WiFi connectivity for remote operation with image capture applications, such as EKOS, SGP, or NINA. ZWO introduced a competing model, which weighs 5.5 kg, has a 13-kg payload without counterweights, and a further 7 kg with them. It is ideal for portable use and closely integrates with ZWO’s ASIAIR imaging computers. Tis model uses harmonic drives on both axes, though it is arguable that the DEC axis does not beneft from the high torque, as it is easy to fore-af balance a telescope in the dovetail saddle. Te technology is catching on, and iOptron and Pegasus have joined the fray. Te iOptron unit uses a conventional worm drive on the DEC axis. Te worm drive is less expensive and keeps costs down. In common with the Rainbow unit, it has a model with an RA encoder to reduce tracking errors and permit unguided imaging with short focal lengths. Te Pegasus unit, released in January 2023, is a little heavier, at 6.4 kg, but supports a 20 kg payload without counterweights, sufcient to support a fully-laden 11-inch SCT. It does not have a handset but has a free mobile application that facilitates polar alignment and goto functions. It has USB and wireless connectivity, and its in-built environment sensors help level the tripod and adjust for atmospheric refraction. Both the Pegasus and Rainbow units have signifcant periodic error (~±20”) over a period of about 430 seconds, and employ rapid guiding for imaging purposes.

Cameras New CMOS-based camera designs continually grab the headlines, with ever-larger pixel counts and formats. Tis is not a surprising trend, given that it follows the consumer digital camera evolution. What is slightly less obvious is the trend to use higher exposure counts using shorter exposure durations. Tis plays to CMOS’s strengths and weaknesses; their read noise is typically low enough to minimize the noise penalty of multiple short exposures and the fast download speeds lessen their time-wasting efect over a given session. At the same time, this avoids issues that become progressively worse with long-duration exposures; namely amp glow and in some models, practical inconsistencies with efective fat feld calibration, probably caused by non-linearities at low signal levels.

fg.5 A diminutive Raspberry Pi 4 in an aluminum enclosure, sitting on an Intel NUC (which is not exactly big) fully loaded with StellarMate Linux distribution, including K-Stars, Ekos, ASTAP and PHD2 applications. The Linux equivalent to ASCOM is the INDI platform, which is increasingly supported by camera and telescope manufacturers and, when coupled with an integrated power control, USB hub, focuser, and dew-heater controller (fg.6), form an extremely compact imaging control system.

fg.6 At the same size as the Intel NUC PC above, the UPBV2 is designed to be mounted with the NUC (or RPi) on the telescope itself, keeping focus, power, dew and USB leads to a minimum. It gives full power control, an industrial USB 3.0/2.0 hub, focuser output, and adaptable dew-heater control. It has a fully-featured application and has ASCOM support too, as well as INDI drivers that work on the RPi. In many installations, the only trailing wire is a single power cable to 12 volts.

39

The Astrophotography Manual

Not all trends are positive; the rush to market new camera designs seems to inevitably lead to customer frustration as the sofware (and sometimes hardware) were not robustly design-verifed at launch. ADC Bit-Depth Te other signifcant trend in camera design is the improvement in apparent read noise and dynamic range from using higher bit-depth on-chip ADCs. Sensors that once commonly used 12-bit ADCs are displaced with new designs with 14-, 16-, or 18-bit conversion. At the same time, with all other things being equal, these sensors have lower read noise at low gains on account of less quantization noise.

Gizmos Tere is always room for a new widget. Two favorite candidates are those to speed up polar alignment and stand-alone guiding. Over the last few years, they have been several; the QHY PoleMaster and the Lacerta MGen3 stand-alone guider come to mind. More recently, an entirely new product, StarAid, does both. However, what looks deceptively like a guide camera is much more. Tis little unit is a standalone guider, a novel polar alignment tool, and does astrometry and telescope alignment too. While none of these are revo-

lutionary, the smartphone interface's integrated nature and simplicity is benefcial for portable users. It uses a combination of astrometry and drif measurement to make drif alignment and guiding as efortless as possible. (Te latest frmware defaults to polar region polar alignment, similarly to PoleMaster and SharpCap Pro, which makes it less susceptible to periodic error.) StarAid also uniquely purports multi-star guiding with a wide-feld lens for optimum results. Another established trend is to replace separate modules, like dew-heater and focus controllers with an integrated module, with full ASCOM/INDI interfaces. Tese typically combine multiple dew-heater outputs with focus motor operation, a resettable USB hub, environment sensing, and intelligent power distribution. Tese compact units complement the mini-computer modules to make a fully portable control system that can ride on the telescope. Te Pegasus Astro unit in fg.6 is a good example, and works equally well with PCs running Linux or Windows operating systems.

Computing and Software Advances in computing and sofware never stop. Ten years ago, an image capture system would likely use a Windows desktop or laptop, running an established and ofen expensive application. Te sofware market

fg.7 A relative newcomer, NINA is a highly customizable image acquisition application for Windows. Unlike some other applications, that have built-in logic and constraints, NINA invites more experienced users to form complex operational structures, with endless permutation. This application is under rapid development and many new features are being introduced through plug-ins.

System Choices

has since undergone some fundamental changes, driven by the concept of low-cost applications on mobile platforms and subscription/hire rather than ownership. Popular utilities are no diferent; many now use free plate-solving applications. ASTAP has replaced PinPoint in my system. ASTAP is fast, reliable, works with 64-bit applications, and on multiple platforms too. It also includes stacking and image analysis tools that largely replace the venerable CCDTools utility. Te great news is that my experience with these lowcost sofware titles is extremely positive. In addition, the developers are responsive to customer suggestions and input, which is a positive trend in itself. Some free and low-cost entry solutions have existed for many years (e.g., PHD and Nebulosity), but for more advanced capture sequences, Sequence Generator Pro was the frst to break the mold with intelligent capture automation and afordable pricing. SGP has been joined by others, including NINA (donate-ware), Prism, and Voyager, covering a range of price points and sophistication. Tese ofer more diverse levels of user customization, and equally, demands on customer input and knowledge. Tis trend of evermore sophisticated and specialist image capture is a perpetual dilemma for developers; it is easy to over-burden the average user with unnecessary features and encumber the user interface. To overcome this, these new applications use alternative approaches. In the case of Voyager, and to some extent, SharpCap Pro, the logic and utility is extended with customer-designed scripts. NINA’s approach is perhaps less intimidating. It has an architecture and accessible user interface that perform the standard capture functions and augments this with an alternative advanced interface, and further, with optional specialist plugins. Tese plugins are written by advanced users and NINA’s developers. At frst, they added such things as hardware-specifc utilities, incremental sequencer controls, orbital target management, and multiple camera support. (Te latter has proved invaluable and permits synchronization between multiple imaging systems on the same mount to make the most of a clear night.) NINA’s trend of evermore sophisticated and interesting plugins now extends to imaging exoplanets (thanks to Nick Hardy) and lucky imaging. With the increasing use of home and remote observatories, complex multiple-target planning and sequencing is now a reality in SharpCap Pro, Voyager, and NINA (thanks to Tom Palmer). Tese dynamically prioritize, schedule, and acquire multiple targets according to a set of user-defned criteria.

40

Computers Computing platforms have gone through equally radical changes. In the past, we commonly used old desktop and laptop computers. Older computers typically have slow USB2 interfaces, but USB3 interfaces are now the standard for CMOS cameras. Te speed diference is most noticeable during autofocus and lucky imaging. Combined with the increased readout speed of CMOS cameras, my 24 MP CMOS camera downloads an image over USB3 in less than a second. My 8 MP CCD camera took 15–20 seconds over USB2. I would recommend using a modern computer; they are typically smaller, use less power, and have more robust WiFi, Bluetooth, and USB interfaces. Today we are increasingly switching to remotely-controlled diminutive computing bricks, running Windows OS or Linux, mounted on or close by the telescope. It is not essential to have a high-powered model for image capture. Some of the best units are re-purposed media PCs (e.g., Intel NUC series). Tese can run both Windows and Linux operating systems. More commercial solutions integrate interfaces for focuser, mount, power monitoring, WiFi, GPS, dew control, and with extensive USB connections. For the Windows operating system, the PrimaLuceLab Eagle series is a good example. Tese fully integrated models are an appealing single box solution and permit remote operation via WiFi from a tablet or laptop. My preference is to separate the computing and control units (e.g., NUC/Pegasus combination). I fnd this easier to maintain, has less radical upgrade paths, and allows me to change the operating system to Linux by simply exchanging the chewing gum-sized solidstate drive in the brick PC (or exchanging the NUC for a Raspberry Pi unit). Conventional PCs are expensive and require a Windows license at about $100. Tis operating system is the target for constant malware and cyber attacks, and Windows increasingly decides when it will upgrade itself and reboot, regardless of whether you are in the middle of an imaging sequence. (It is difcult to entirely defeat these update mechanisms or downloads and it is now an ever-present consideration before starting an imaging run.) An increasing number of astrophotographers are switching to Linux for reasons of economy and self-determination. Raspberry Pi/Linux It is hardly surprising that the trend to use inexpensive Linux-based Raspberry Pi boards or derivatives for image capture is gathering pace. For $60, one can

41

The Astrophotography Manual

have a complete computing unit with the free AstroBerry astro operating system, or another $50, the similar StellaMate Linux distribution, which ofers extensive remote control options, to reduce the local computing burden. Similar to the PrimaLuceLab imaging computers, there are several highly-integrated commercial products based on Linux, including TeSky Fusion, ZWO ASIAIR Pro, and ATIKbase. TeSky Fusion operates a Linux version of TeSky Imaging Edition. and the popular ZWO ASIAIR integrated imaging computer is small, and optimized for a ZWO imaging ecosystem. ATIKbase and StellaMate use free, open-source applications like PHD2, EKOS, Kstars, CCDciel and CDC. Changing the operating system on these is as simple as changing the micro-SD card. (Interestingly, it only takes a few minutes to swap out the solid-state drive in my more powerful NUC PC and try out the same Linux imaging applications.) Open Systems Open systems refer to a diversifcation of the control of astronomy equipment. A (typically small) Windows or Linux computing unit is physically connected to the imaging devices and runs without a monitor or keyboard (i.e., headless), from a remote client, typically via WiFi. Tese make for neat installations, with minimum cable clutter. Te user interface may be a PC, Mac, or tablet but in this case, not necessarily running a remote desktop application (e.g., Microsof Remote Desktop or TeamViewer). Standardizing the protocols between applications and devices makes it simpler to adapt to diferent operating systems. If this sounds familiar to ASCOM, it is, with one exception, that ASCOM works entirely within a Windows .Net environment. Suppose you break the system apart between the device drivers and the application sofware. In that case, the communication can be internal, within the same computer, or between computers, permitting greater freedom of choice. Tis recent concept has seen rapid development for both Windowsand Linux-based hosts. For example, the ASCOM Alpaca (JSON) protocols allow any computer to interface with Windows-based device drivers. Similarly, the INDI (XML) protocols do the same thing for the Linux operating system. Tese are “simple” text-based protocols, and for the sofware savvy out there, there is no reason why one cannot use Alpaca with a Linux host… you “just” have to write drivers that recognize and adhere to the protocol. To add to the melee, there is also an INDIGO

framework, currently for MacOSX and Linux platforms, which purports to be a “modern” replacement for INDI, with its unique suite of (paid) applications and Linux distributions. Of course, in sofware, “modern” is a relative term, when one considers that many astro devices use RS232, developed in 1960! Too many standards, however, confuse users and developers alike, as well as increasing development costs. I hope that this can be improved by more collaboration between the well-meaning developers. Many of us know our way around Windows foibles and do not think twice about nested menus to fnd a setting or other. Over time we become familiar with its “logic” and stumble with something diferent. Linux is very diferent and although its graphical user interfaces (GUIs) are improving, there are many times when you need to remember and type what appear to be obscure commands into a terminal window. It is not for everybody, though I believe it will become more user-friendly since there is infnite potential to customize the interface with a new Linux distribution (distro) for an intended purpose. However, the trick is to keep things simple, as the low overheads of the Linux system are the reason it runs complex tasks on lesser hardware. Image Processing Applications Image processing application development has been equally exciting. PixInsight established itself favorably with many enthusiasts. It is a powerful tool with an underlying philosophical approach to being “truthful” to the data. However, it remains an enigma to some amateurs, because of its ease of use and incomplete documentation. Some new entrants, for example, StarTools and Astro Pixel Processor, take a very diferent approach and are gaining favor. I have tried both, and I found documentation patchy or missing altogether. I should add ASTAP to this list, which also performs image calibration. Tese applications run on Windows, Linux, and Mac OSX. StarTools takes a unique approach to image processing that can iteratively and selectively adjust process settings to suit the subject, given sufcient computing power and patience. Star removal in image processing has been around for several years, using several tools and processes. Te results have, however, been hit-and-miss. StarNet, which I use as a plugin in PixInsight, went some way to remove stars automatically, and works best on non-linear images. More recently, Russell Croman released StarXterminator, a remarkable PixInsight and Photoshop plugin that automatically removes stars and per-

System Choices

42

mits easier image processing, both before and afer image stretching. Tis interesting utility is joined by two others using artifcial intelligence to content-aware fll pixels. Tese are BlurXTerminator (a deconvolution tool) and NoiseXTerminator which efectively removes noise without losing defnition. Smart Telescopes Electronically assisted fnders have been around for some time. Te idea has evolved into fully integrated imaging systems, using an Alt-Az mount architecture and a fully-enclosed telescope/camera system. Ordinarily, seasoned astrophotographers would pass over these models, but they are becoming extremely sophisticated (and expensive) and track, focus, align, image, de-rotate, calibrate, and live-stack images. Tese are primarily designed for portable use and fold up into a self-contained tube when not in use. Tey ofen use a dedicated smartphone application to control all aspects of operation. Vanois and Unistellar are leading companies, with models ranging from $2,000 to $45,000. Te premium Hyperia model is a 150-mm f/7 APO refractor with a built-in 61 MP fullframe camera and direct drive on each axis. Te premium cost refects the convenience of its design but exceeds a conventional equatorial mount-based system by some margin. Te Hyperia is a substantial investment and, fortunately, its camera sensor is upgradeable as new models are released. Te specifcation is impressive and in time it will be interesting to see how the value, practical convenience, and image quality measure up to more traditional methods. Summary Te only constant in the world is change and in these exciting times, I have been using and exploring many of these on-trend items. On the hardware side, these include StarAid, Rainbow RST-135, Pegasus NYX-101, Raspberry Pi 4/StellaMate and several new CMOS astro cameras. On the sofware side, these include NINA and CCDciel to capture images, and ASTAP, StarTools, PixInsight, and Astro Pixel Processor to process them. Consumer choice is overwhelming and invention shows no sign of slowing down. I just need a gizmo to blow away the clouds.

fg.8 The Pegasus NYX-101 harmonic drive joins Rainbow, iOptron, and ZWO. Although a little heavier at 6.4 kg, it has a load capacity of 20 kg (without counterweighting). Unlike the Rainbow mount, there is no handset and it relies upon either a computer application or its smartphone application for alignment, control, and pointing to a catalog object. Here it is mounted on its lightweight carbon tripod, with a Pegasus PowerBox and Intel NUC computer for either Windows and Linux operating systems. The counterweights are a precaution with this heavy refractor to improve stability and additionally reduce lateral stresses/fexure on the tripod, pier, and mount.

43

The Astrophotography Manual

Support and Mount Options Getting these right is the frst, second, and third most important thing in astrophotography.

T

his chapter, and the following ones in this section, consider the crucial considerations when planning an imaging system. Tese include permanent, luggable, and portable setups. Tere are many things to discuss and several ways to approach the subject of priorities, best practices, and things to consider. Astrophotography is a fast-moving target right now on account of its many branches and the industry response. Seemingly logical choices go wrong for one reason or another, and over the years, some of mine have been, too, on account of the wrong compromise or changing needs. Bargains are great, but they incur a higher overall cost if they lead to an early upgrade. Systematic consideration of the essential requirements and pertinent specifcations is one way to prioritize, but you must arrive at your own purchasing decisions to meet your particular needs. While some personal preferences emerge in this and later chapters, it is arrogant and futile for me to make frm recommendations. Te principles, however, will endure longer than any particular model. One logic is to work from the ground up, but before we set to, I must bring an essential and usually overlooked subject to your attention.

Safety Astrophotography includes activities that move heavy items, late at night, in darkness, ofen alone in cold, remote locations, and without cell coverage. What could possibly go wrong? Closer to home, safety considerations are still necessary. Health and safety planning is part of modern life, and before we consider the more exciting aspects of astrophotography, one should consider the potential hazards. Tese include personal safety, clothing, handling heavy equipment, lighting, communications, and a fair degree of common sense. Personal liabilities to third parties are another consideration in today’s litigious world. For instance, when I started to do outreach events in school playgrounds, I quickly realized the trailing cables to the laptop were a trip hazard. I initially obstructed the critical areas and, aferward, eliminated the issue by transitioning to wireless control. In the case of portable setups for deep-sky pho-

tography, the equipment is still substantial and, at the same time, delicate and easily damaged. One needs to apply the best practices for manual handling, reminding ourselves it will likely be darker, colder, and damper when we pack up. Tese include an unobstructed passage, sensible footwear, and knowing one’s limits. I have had a few near-misses; a dewy 12kg counterweight slipped and missed my sandaled foot by a few inches, and I also walked into a black iron garden chair, lef in the middle of the lawn, while carrying a heavy refractor. We sufer for our art; while I am writing this, I am still sore from the unaccustomed efort of carrying my equipment to the end of the road, to image the 2020 Jupiter/Saturn conjunction. Safety is rarely mentioned in any books on astrophotography, and I’m not going to call out every potential risk, with one potentially lethal exception, electrical safety. Electrical Safety Astrophotography equipment, in general, is usually powered with 12-volt DC. However, there are some exceptions; a few large mounts use 24- or 48-volts DC and an ever-increasing number of small devices are also being powered through USB cables, operating at 5 volts. Low voltages are rarely lethal; the current is dangerous, and a higher voltage carries a greater risk. Te risk of electrocution mainly concerns 110–240 volt AC mains and power supplies. Most power supplies are not safety-approved for outdoor use; including small plug-in-the-wall and larger stand-alone switched-mode power supplies. Te less efcient transformer/rectiferbased models are not exempt, even if they appear enclosed. Tose power supplies certifed for indoor use are a potential risk of failure and electrocution, as is any exposed mains connector when exposed to dampness, dew, and rain. An earth-leakage circuit breaker is mandatory, but its use is not an excuse to take liberties. Te problem is widespread; none of the power supplies bundled with my equipment are certifed for outdoor use. Tere are a couple of workarounds; battery power is an obvious choice, whether lead-acid or much lighter (and expensive) lithium polymer models. Another option is to run 12-volt DC out from the

System Choices

44

fg.1 The four phases of astrophotography, with some ideas of budgets and equipment. In many cases, the choice of mount defnes the system. There are fewer choices in the middle-ground between budget and premium mounts in the ~$1,500–$4,000 range. One of the wonderful things in this hobby is the extent and quality of free or inexpensive software (which is just as well!).

house using low-impedance cabling. If this is impractical, it will require mains electricity at your installation and additional protection of all power distribution and DC power supplies using sealed enclosures (with sealed entry and exit ports). I use all three strategies for remote, backyard, and observatory setups. I have also used a plastic box designed for gardeners who wish to use electrical appliances outdoors temporarily. (An Internet search for IP55 weatherproof boxes will fnd likely candidates). If you are in any doubt, please ask a qualifed electrician. I want you to image the heavens, not be a resident member.

Equipment Overload, a Public Health Warning I have a knack for choosing expensive hobbies (confrmed by my enduring wife), and I have realized over the years that high-quality items ofen depreciate less, due to the infuence of brand and quality. I am fortunate that my hobbies pay for themselves and I have been able to indulge myself, without being extravagant. Te rapid product development and subsequent obsolescence in digital technologies is a diferent matter. Tere are two forms of obsolescence: psychological and physical. In the former, older equipment or sofware is prematurely cast aside as it is no longer consid-

45

The Astrophotography Manual

ered as leading-edge. For example, that is happening with the side-lining of older CCD cameras in favor of CMOS. As a result, there are some excellent used bargains on the Internet. Regarding material obsolescence, electronic interfaces are usually the worse culprit. Manufacturers abandon interfaces to make room for the latest or switch entirely to wireless protocols. Physical obsolescence is also accidentally triggered by sofware or hardware upgrades that drive incompatibilities; for instance, moving to a 64-bit operating system or application that requires 64-bit drivers, only to fnd that your device is no longer actively supported. For example, I write my books on an Apple Mac. It, or rather OSX, is not a practical proposition for image capture, and only a few image processing applications have 64-bit OSXcompatible apps. When I started astrophotography, I was guided by an out-of-date book and made the wrong purchasing decisions. I bought an entire used SCT outft from eBay that included a variety of telescopes, cameras, and eyepieces. Over several months I realized it was not entirely suitable for my needs but fortunately, I was able to refurbish it and sell it for a small proft. It was a helpful experience that highlighted what was important. I needed a computer-controlled equatorial mount that tracked well, a refractor, a cooled camera with a flter-wheel, and a guiding system. Since then, I have changed all my systems several times over, mostly to support the research for my various books and to grow with an increasing budget. Upgrading can be addictive, however, and I occasionally remind myself of that with my forum tagline: “too much stuf, too few clear skies to use it”. Systems vary enormously, and in an attempt to propose various scenarios, fg.1 has examples of alternative systems. Planning Considering whether your imaging system will be at a remote location, a dark-feld site, or near a residential property, there are a couple of obvious considerations for placement. Te frst is the horizon, including its limitation on alignment and how it afects access to the parts of the sky you wish to image. A portable system has the potential to operate nearer the horizon than, say, a roll-of roof observatory with fxed walls. In practice, however, this only afects a few low-declination objects since routinely imaging below 35° altitude is not ideal, because of the increase in light pollution and atmospheric seeing. Access to low declinations expands the potential repertoire. Automated drif methods and modern sofware sky model alternatives di-

minish the necessity to see the celestial pole for polar alignment. Increasingly, a contemporary installation requires Internet and home-network access, ofen using WiFi (with a distance limit) or a buried Cat 5/6 network cable, or strategically placed WiFi access points and repeaters. Te best optical and imaging systems are compromised if they cannot accurately track the subject during the entire exposure. Tracking correction techniques, such as autoguiding and tracking models, are of little use if the whole system is not stable, rigid, and robust to environmental disturbances. A cursory assessment will only identify extreme issues as, during a typical exposure, it is normal to track with an angular error of less than 0.001 degrees. Just as the foundations of a house are essential and dull, the ground upon which we place our equipment is too. It must be stable, resistant to local disturbances (like you), and preferably fat and level for the best results. An isolated concrete block is ideal when it is decoupled from its surroundings, and is the usual base for an observatory pier. A large concrete pad is a good second best or a few concrete pavers placed on sof ground. In previous books, I shared my tripod lawnspike design, which when hammered into the earth, provided a discreet, stable, and precise tripod (re)location. Tese are relatively stable and isolate the tripod from pedestrian activity. Compliant couplings, such as rubber feet or sof ground, are not ideal. (An example of the sensitivity to small localized subsidence to polar alignment and drif is discussed in the prior chapter.) With a typical pixel scale of 0.5–2 arc seconds, the slightest vibration will cause image blur. Before I used lawn spikes, I rested three paving slabs on the grass. One evening, my autoguider suddenly went berserk, and looking outside, I saw my kitten sitting on one, staring intently at the blinking LEDs. Ground vibration from heavy lorries or nearby trains potentially imposes a performance ceiling and more exotic forces; I remember an amusing forum thread in which a bunch of us were trying to help a fellow imager with inexplicable tracking errors. Afer several increasingly imaginative posts, another member realized the user lived in California and the strange anomalies corresponded to recent San Andreas seismic activity. (Note to self – update the diagnostics chapter!) Some of my colleagues image from the roof of apartment blocks or the deck in their back yard. Both are challenging environments due to setting up and fnding a stable footing. A wooden deck or a fat roof with wooden beams fexes with load, and your presence will

System Choices

46

likely cause tracking errors. Tis installation is not ideal, but a useful workaround is to operate remotely from a safe distance.

Tripods and Piers Tese are the silent unglamorous edifces upon which we lavish more exciting goodies. At one time, they were quite distinct; tripods for portable use and piers for permanent installations. However, the splayed legs of a tripod are a cause for concern when imaging close to the zenith, as the back end of the imaging system may collide with them. Tese “leg clashes” are at best inconvenient but may also damage delicate mechanisms. Te initial response to these concerns was for the mount manufacturers to ofer “pier extensions”. Tese typically pipe-like extensions attach to the top of the tripod and lif the mount by 6–12 inches and, in many cases, are sufcient to avoid calamity. A natural progression followed, in the form of the portable pier (fg.5). Tis hybrid design has three, low-profle legs and an extended center column. Tese are semi-portable, either resembling a lunar lander or cleverly stowing the legs inside a hollow column and using wire braces to stifen the entire assembly. Some have bolt-mounted feet with adjustments for accurate leveling. In theory, leveling feet are a great idea, provided they have a locking feature that stops the threaded bolt from wobbling around. During visual astronomy, it is usual to extend the tripod legs to make the eyepiece more accessible. Height is not a priority for imaging and, to reduce fexure, it is best practice to only partially extend legs sufciently to stabilize the platform and provide wriggle-room for leveling. Te tripod designs to support equatorial mounts and telescopes are deliberately more substantial than photographic models for load stability. Tripod materials vary including wood, steel, aluminum, and carbon fber. In much the same way as bicycle frame design, the material choice is only one consideration; the form and clamping methods are equally important. For instance, ash hardwood has remarkable shock-absorbing properties, but if the extended legs are too thin, or insufciently braced, a model may be too fexible for the demands of astrophotography. Te resistance to a twisting motion on the top plate is ofen an indicator of stifness. Four Legs Good, Two Legs Better A three-point clamp in a system (eyepiece, focus tube, or tripod) is uniquely stable. Te best tripod designs have a brace system to reduce leg fexure and have stif pivots (or locking ones) near the crown. Tose mounts supplied with a inexpensive tripod, made with tubular chromed steel legs, usually have a single lock on each leg. I have had several and the clamp mechanism is not as secure as other designs, and they were quickly replaced with an upgrade. Premium mounts usually have more substantial (and expensive) tripods or portable pier options, in a variety of materials and designs, with a dedicated top plate to suit the mount. Independent manufacturers do the same, with multiple fxing holes or available with alternative top plate options to suit the popular models. Te better ones have a spiked-foot option and double-leg clamps, to improve rigidity. I have used Berlebach Planet (wooden) and Avalon TPod 110 (aluminum) models. Both are stable platforms, capable of carrying 100 kg. Te TPod is the more compact and conveniently has a top plate with multiple fxing

fg.2 A stable footing starts with the ground and works up. After hammering three 1-foot metal rods in the lawn, this is all that shows; a large washer (to enhance visibility) and an M8 bolt screwed into the top. The bolt facilitates removal and the tripod spikes locate into the bolt head. An even simpler alternative is to use three long coach bolts, either with a cap head, or drill an indent, to accept the tripod spikes.

fg.3 A sturdy tripod is essential to reduce movement and suppress vibrations caused by external infuences. This Avalon TPod is a versatile aluminum design, with a variety of tapped fxings holes to suit many mounts. It has the same 100-kg capacity as my traditional Berlebach Planet ash tripod. Both have double leg-clamps and bracing to improve stability.

47

The Astrophotography Manual

fg.4 The Paramount mount range are dependable worm-driven GEMs, with through-mount cabling and convenient USB, guider, and power connectors on the saddle. This MyT model is semi-portable but the two larger models are better suited for permanent installations.

fg.5 The Rainbow RST-135 is a premium compact mount. Its slim dimensions make leg clashes more likely and it requires a long pier to avoid leg clashes with long refractors. Even so, a large flter wheel or an of-axis guider may still hit the column if the mount ofset is small.

options. When I buy a new mount, I have a local machine shop fabricate an adapter plate and pier extension to suit my telescope mounts (fgs.4,5). Mechanical interfaces are also important, and I fnd the bolted assembly of the Rainbow and Paramount models to the tripod is more secure than the single, hand-tightened central fxing of my SkyWatcher system. In ultra-light systems, using simple trackers (or the more sophisticated harmonic drives) a carbon fber tripod keeps the system weight to a minimum. Carbon fber is a remarkable material, but it also has limits. Te ultra-light compact models are less suitable because of their multiple and thin leg-sections, lack of leg bracing, and rubber feet. Te central column of any conventional tripod is its weakest link and one should avoid extending the central column for imaging. My Gitzo carbon fber tripod is a substantial 3-section leg design, with secure clamps and a detachable center column. For astrophotography, I swap this for a simple plate, ftted with a 3/8-inch central bolt. Some of our requirements are common to those of professional videographers. You will notice that many quality video tripods have complex braced leg designs to increase stability in all directions and dampen vibrations too. Video tripods (without the head) are worth considering for a truly portable system with light-weight equipment. Pier Pressure Having a pier tells others you mean business! Tese ugly, no-frill obelisks are usually either welded-steel constructions or concrete castings. Tey are heavy, uncompromising, and their installation usually requires one to plan the ground and mount fxings before fabrication (it is not easy to drill four precision holes in quarter-inch steel using a hand tool). Some installations have a generic three- or four-point fxing on the top, requiring a custom adapter plate for the mount. Other installations have the top plate mounted on substantial bolt stilts (to permit leveling). Tis is not a necessity for a permanent German Equatorial Mount and a potential source of fexure. Te steel pier in my observatory has missile-like fns running down its length, allegedly to reduce vibration. I cannot confrm either way; it is a heavy and uncompromising beast. Filling the cavity with dry sand has the same efect, (audiophiles fll hollow loudspeaker stands with an inert material to stop ringing). Some piers also have the option to use the hollow cavity as an optional cable conduit. Piers are hand-built to order or constructed on-site. I chose a pier height that allowed my observatory roof to just clear a parked telescope and to give a low imaging horizon (fg.6). To keep cable lengths to a minimum I made a cradle to hold my electronics housing close to the pier. Everything in this environment is sealed or from materials that will not corrode in damp conditions. To reduce humidity, I placed a paper bag of rice in each enclosure. I also make a habit to remove and re-insert all cables every month to clean their contacts and spray diluted lavender oil around the observatory crevices to dissuade spiders. I’m told peppermint oil also works as a deterrent.

Telescope Mounts Te choice of mount has a signifcant impact on the performance of an imaging system and one’s wallet. Tere is a vast range with an increasing emphasis on ultra-portable and lighter models to tempt newcomers.

System Choices

48

(Portable tracking mounts are a special case and are discussed separately.) Tey all have commonly-listed specifcations and exhibit other characteristics that are difcult to measure, but equally important. Marketing hype and practical realities are not always aligned and, whatever the model, a mount is primarily classifed with the following attributes: • • • • • •

mechanical integrity payload capacity tracking capability guiding capability computer control additional features

Evaluating these before a purchase is not always possible, but user forums usually ofer insights into strengths and weaknesses and, importantly, the manufacturer’s response (or lack thereof). Mechanical Integrity Mounts are mechanical devices and every housing, bearing, gear, pivot, and adjustment is an opportunity for unwanted movement. Furthermore, everything is made to a budget (cost and weight) and the efectiveness of mundane things difers greatly between models: • • • •

secure mounting to pier/tripod non-slipping clutch design secure mounting clamps for dovetail stable, micro-adjustable polar alignment

Sometimes, the simplest things are a source of continual frustration and potential calamity. In the past, I have had issues with all these aspects on one mount or another. Something as simple as knob design has a practical impact on functionality; my frst premium mount had an attractive machined clutch, baseplate, and dovetail clamp-knobs that were too small and smooth. In practice, I and others found them challenging to tighten by hand, and if it were not for a safety tether, my telescope would have fallen out on one cold night. Function should follow form and a few holes for a tommy-bar would have sufced and not sullied the appearance. Similarly, polar adjustment mechanisms are equally challenging. Various designs involve worm and threaded push bolts and even friction rollers. Tese are required to articulate the considerable mass of the mount and imaging system yet ofer smooth micro-movements and clamp securely without

fg.6 A Paramount MX, mounted on a substantial steel pier, provides a solid support for larger telescopes. In this roll-of roof observatory, it must be parked horizontally before the roof can open or close safely. This requires sensors to detect the roof and mount position and an intelligent roof controller that works, irrespective of any high-level imaging applications or device drivers.

shifing the position. Te altitude adjuster is especially difcult as the center of gravity of a fully-loaded mount is rarely directly above the pivot point causing unequal raise and lower forces. (In these designs, to avoid backlash, it is best to approach polar alignment against gravity.) In the simpler mounts, basic pushbolts engage a cast lug close to a pre-loaded pivot. Te stresses are considerable; enough to deform bolts, and strip threads in sof aluminum castings, and are usually difcult to micro-adjust. Te better designs have threaded adjusters further away from the pivot, with separate locknuts, which reduce the adjustment force and permit micro-movement when undone. When locking the polar altitude alignment, gradual opposing tightening of the knobs on each side lessens the chance of ruining the alignment.

49

The Astrophotography Manual

Azimuth adjusters have a diferent set of criteria. Most designs use opposing bolts to rotate the mount within itself or on the tripod/pier. Te design must permit rotation (i.e., slackened fxings) without allowing the mount to tilt and yet be easy to lock down securely. Te designs with the least potential for inadvertent movement have two or more fxings or a secure clamp. Payload Capacity A key mount specifcation is its maximum payload. Tis usually refers to the telescope and camera system, but some manufacturers include the counterweights in the payload tally. For best results, the consensus is not to exceed 2/3rds of the mount’s stated maximum imaging capacity with the weight of the telescope, camera, and hardware. Te capacity must exceed your current imaging system (and potential upgrades). Te right mount balances portability requirements with a capacity to cope with future telescope upgrades. Te payload calculation includes tube rings, dovetail plates, guide scopes, cameras, and fnders (it soon adds up!).

fg.7 A selection of intermediate and premium mounts, with gleaned specifcations from the OEM websites and typical pricing in 2021. Periodic error assessments are tricky, on account of unit-to-unit variation and, in practice, the fgures above are more on the conservative side of typical.

Tracking Capability Except for simple trackers, telescope mounts have a system with two motors, set orthogonally, to move the imaging system. When one motor is oriented to the celestial pole it becomes an equatorial mount, of which the German Equatorial Mount (GEM) is a familiar variation, found in small and medium systems. Several manufacturers ofer hybrid solutions, in which the mount body can be orientated in either equatorial or altitude/ azimuth (Alt/Az) orientations. Others have realized that the tilting (altitude) adjustment is a potential weak point in the design and for permanent installations, ofer a substantial metal wedge to an approximate specifcation, with micro-adjusters for precise polar alignment. However, the heavy mass of professional imaging systems ofen demands stronger architectures, including open fork-mount, cross-axis, and horseshoe designs. Some of these have been down-sized for amateur use, for example, in the form of Center Equatorial Mounts (CEMs).

fg.8 It may sound boring, but welldesigned control surfaces make it a joy to quickly polar align and lock tight (and similarly applies to dovetail clamps). These azimuth adjusters have an extremely fnepitch for precise control and has a locking clamp at the front.

System Choices

Tere are several mount drive systems with individual strengths and weaknesses. Tey all, however, have a common goal to provide a consistent and accurate tracking speed, with minimum hysteresis on both axes and, for most users, be responsive to guiding commands. At their heart are usually stepper or synchronous motors. A motor’s angular resolution is too coarse without a reduction drive, using a combination of belts, spur gears, friction drives, strain-wave gears, or worm drives. Some of these work “open-loop” without any positional feedback systems, while others have precision optical position encoders on the motor, axis, or both. In theory, all equatorial mounts track a star’s movement around the sky with a single motor but as we have already shown, atmospheric refraction, fexure, and poor polar alignment require both axes to be active. Up to the last decade, the dominant architecture used spur gears with a worm drive, made popular by SkyWatcher, Celestron, Orion, and Meade. Te spur gears were the weak point of this design and are now mostly replaced by toothed belts, reducing backlash and improving tracking accuracy. Te dominant tracking error is usually associated with each worm cycle, which we commonly refer to as periodic error (PE). Periodic Error PE is the cyclical error caused by machining tolerances of a worm (think fusilli pasta) and repeats for each rotation, typically over a 3- to 8-minute period. Tiny machining errors translate into the axis, alternatively moving slightly faster and then slower as the worm rotates. Although linear tolerances cause PE, it is the peak-topeak angular error that is recorded (in arc seconds). If you use the same worm with worm gears of diferent diameters, the angular error decreases as the diameter increases. Consequently, smaller mounts place higher demands on worm accuracy.

50

Periodic error is an important specifcation that requires qualifcation. At best, PE specifcations are an indicator of tracking error. At worse, they are misleading. One has to read the specifcations carefully, as quoted values may be a worse case, typical, or afer sofware correction. Quoted specifcations vary widely, between 0.3–60 arc seconds before correction. Price and brand do not always guarantee good PE performance; many forums indicate signifcant performance diferences between identical models, an unavoidable outcome of manufacturing tolerances. Several manufacturers side-step the issue entirely and do not specify a value or declare their drive system has zero PE, due to it having no worm drive (but have signifcant tracking errors over a cyclical term, nonetheless). Traditionally, PE is the most prominent and identifable tracking error, due to its short period. Lef unchecked, the peak-peak PE value has a direct efect on long-duration exposures. If the mount supports periodic error correction (PEC), the improvement may be sufcient for unguided exposures, especially if one uses a short focal length. However, I have noticed a trend for low-cost units to market built-in PEC capability rather than control machining tolerances. Tis is, at best, a band-aid on mounts with poor uncorrected PE, as PEC is not 100% efective and I have had large PE corrections interfere with guiding commands. In general, it is difcult to set an absolute upper PE limit (the acceptable tracking error changes with focal length and declination), but as a guideline, ±2, 7, and 15 arc seconds would be excellent, good, and adequate. Te disclosed peak-to-peak periodic error value is not the whole story. While an autoguider system works “afer the event”, in practice, it will fully correct a slowly changing tracking error. If the tracking error is changing rapidly, the autoguider is continually playing

fg.9 While a large peak-to-peak periodic error is not a good thing, it is not necessarily the whole story. Here, the unguided PE of a mount with a harmonic drive looks scary, but due to its fast, low backlash, and high-torque mechanism, PE is efectively eliminated by a combination of short guider exposures (0.2–0.5 seconds), low aggression and a high guide rate. (fg.10)

51

The Astrophotography Manual

catchup, ofen with less favorable results. For that reason, it is ofen the rate of change of the tracking error that is critical and, unfortunately, is not usually specifed. How an uncorrected tracking error changes with time is complex, and some characteristics are features of the design. For instance, the drive-wheel in a directdrive mount is a simple disc, easy to manufacture and polish to a high standard. It typically has excellent periodic error but the design is susceptible to contamination on the mating surfaces, causing abrupt changes. (Premium models enclose the drive mechanism and use rotary encoders to correct tracking errors.) Worms are less easy to manufacture with the necessary tolerances, especially when they engage with small-diameter worm gears. As a result, they have higher PE but, on the whole, the error changes smoothly. Plastic spur gears and toothed drive belts are challenging to mold with high accuracy and, made of plastic, fexible. Spur gears, in general, introduce specifc periodic errors and usually introduce additional backlash. In the case of drive belts, they typically engage over multiple teeth, which dilutes the impact of any single-dimensional issue and backlash is minimal, associated with the slight elasticity of the belt when it changes direction. Similarly, the strain-wave gear (a.k.a. harmonic drive) has a complex mechanical construction that defes a consistent PE characteristic but generally has little backlash. In all these cases, the drive system has complex unspecifed interactions, which will vary from model to model and unit to unit. Te only way to truly understand how a mount tracks, is to monitor the unguided and guided tracking errors over an hour or so. Guiding Capability One will likely require an additional form of tracking correction. For many, this will be an autoguider system, while others, with selected premium mounts, may turn to modeling and automatic compensation for refraction

and mechanical error mechanisms. Tese do not necessarily have to be mutually exclusive; I have used both systems and fnd on mounts employing a tracking model (without rotary encoder feedback), autoguiding improves tracking errors. Tere are two main guider interfaces; the venerable ST4 hardware interface and sofware, in the form of pulse guide and its variations. Most guide cameras and mounts support both. Some premium mounts employ precision encoders to reduce periodic error and combine that with sophisticated sky modeling to run without a guider. In my experience, these mounts perform best afer careful calibration, in a permanent setup, and when not autoguided. (From an engineering perspective and in practice, designing a stable system with nested feedback loops is challenging.) While unguided operation is a goal for some, I prefer to keep my options open, and all my mounts have ST4 and sofware-based guiding interfaces through ASCOM. Sofware-based guiding is as good as, if not better than, an ST4 hardware input, as it has the potential to combine with PEC in real-time and minimize conficting mount movement commands. How a mount responds to guiding is not easy to describe, nor is it entirely consistent. A mount’s dynamics change with its motor and transmission design, load, balance, and environmental conditions. However, what might be called out, is a mount’s backlash behavior, without necessarily putting a fgure on it. To make life even more entertaining, guiding parameters that work one night may require tuning, on a follow-up session, according to the seeing conditions. Backlash is a mechanical property that afects guiding performance (particularly on the DEC axis) and slewing accuracy. Mechanical play, also known as backlash or hysteresis, is a consequence of every little gap or fexure between each gear, belt, and bearing. Backlash becomes an issue when a system changes direction (or when the balance point fips over and the

fg.10 This is the same mount used to generate the tracking in fg.9, with PHD2’s predictive PEC in operation. A remarkable improvement.

System Choices

engagement forces reverse). It is measured by noting the reverse input required to change the system direction. It is present in both axes, but while tracking, the RA axis is constantly moving in one direction, and so backlash usually creates a dead zone for the DEC axis, in which initial reverse guider moves have no (or diminished) efect. Some mount manufacturers claim zero backlash for designs that do not use spur gears. While technically improbable, they usually have low enough values to be efectively ignored over other errors. A simple thing like incorrect worm-gear engagement, using too little force, or lateral movement on the worm’s bearings, is enough to introduce enough backlash to afect guiding.

Computer Control It is possible to do astrophotography without a computer; it keeps things simple and is usually confned to short-duration wide-feld images taken with a DSLR or mirrorless camera. For deep-sky imaging, however, a computer is considered an essential element for intelligent control of the mount, camera, focuser, autoguider, and flter wheel over an extended period. Computer control of a mount has several advantages over the smartest of handsets. Tese include; automatic location, date, and time setting, as well as alignment, PEC, accurate slew/centering, sofware-based guiding, managing meridian fips, multiple targets, and facilitating remote operation. Phew! Unlike cameras, the data fow to and from a mount is small and well within the capabilities of RS232 serial. Te computer connection may be on the mount body or linked through a smart handset. Some older mounts confusingly use the common RS232 DB9 connector on the mount body for the handset connection. Tis is a serial interface, but uses TTL signal levels (0–5V) rather than RS232 (±12V). Te signals are incompatible and if you try, the TTL interface will fry. A 5V TTL-level USB–serial adapter (FTDi) will allow a safe computer connection for those afected mounts. While RS232 is slow, this venerable system is reliable and works over long cable lengths (unlike USB2/3/C). A few mounts use Ethernet-based communications, which are efective at long distances. RS232 is not natively supported on modern computers and mandates a USB adapter and a virtual communication port driver. Mounts with USB connections need localized computer placement or gizmos to break the USB transmission bottleneck. In my case, I use a mini brick PC on or under the mount and control it through a wireless connection.

52

Having established the need for a computer interface, mounts do not have the same level of external control. Again, this is not obvious from an advertisement but interpreted from downloaded instructions and forum discussions. For instance, not all of the available commands in the ASCOM telescope interface are mandatory, and the exclusion of some may afect automatic meridian fips, parking, synching, setting time, and homing. In practice, a robust computer interface is essential. Although the text-based protocols allow any computer system to operate a mount, most drivers and applications are Windows-based, a few support Apple OSX and, increasingly, Linux. Additional Features Tere are plenty of add-on features to lure a prospective buyer. However, a word of caution is required and one has to be wary of glamorous features on less expensive mounts. Tese make a sale but do not necessarily compensate for poor mechanical design/tolerances. Some features are more valuable than others, depending on whether the system is used in a temporary or permanent setup. For instance, integrated GPS sensors, polar scopes, and camera triggers are convenient for portable use. A GPS receiver provides time and location information, but the commonly-used slow, serial interface introduces a small delay; enough to introduce RA slew errors. I use an Internet-based time server as it is a more accurate timebase. Some other features are universally helpful, such as programmable slew limits/safety stops, through-mount cabling, and USB/power outlets on the dovetail clamp. Depending on the friction of the bearings, electronic balancing can aid assembly, and permanent PEC and rotary encoders potentially improve tracking. However, the added value of bundled sophisticated handset/tablet catalogs and planetariums is diminished by the duplication of equivalent computer systems and free applications. Ultimately, added features are largely irrelevant if the basics are not right. Buying with Confdence It is presumptuous to recommend any particular model for imaging. I suggest prioritizing good mechanical design and manufacture over gimmicks and buying the best you can aford. Encoders on lightweight mounts make sense for unguided imaging with wide-feld lenses but less so in demanding situations. Reliability is critical for unattended imaging; electrical connectors need to be robust and sofware (frmware and driver) quality is essential. All too ofen the sof-

53

The Astrophotography Manual

ware is the Achilles’ heel. With a mount purchase, it is ofen the case that customer support is required to iron out initial setup issues. A supportive dealer is a must but regrettably, as in every other trade, some retailers are only interested in selling. It is useful to check the Internet forums to see if a model has persistent customer issues (mechanical and operational) that remain unresolved. Te specifc forum for the product is usually the most relevant. For instance, before purchasing the Rainbow RST-135 mount there was a long thread on one generic forum about its assumed performance. Several posts were critical, but crucially, not by actual users. So, I joined the user forum and asked other users to share their views on specifc issues and their PHD2 guide-logs. In particular, a guide-log is very instructive and it tracked well, especially with short guider exposures. Evaluation and Tuning It is sensible to evaluate a mount as soon as possible and it is pretty likely that it will not be perfect out of the box. Tracking performance is an excellent place to start and a simple method that does not involve sofware is in the setting up chapters. (It is an unfortunate truth that, with such a complex hobby, many customer issues are of their own making, which makes it all the more difcult to convince a dealer you have a real concern. In such circumstances, a foolproof method is useful, as it cannot be easily dismissed.) I have also noticed that the tracking performance on two of my mounts improved slightly with use, probably caused by the polishing of “hot spots” on the gears. When I purchase a new mount, I check the mechanism does not bind at the fastest slew rate and make multiple meridian fips at diferent slew speeds. Tis additionally exercises the mount, distributes lubrication, beds-in the gears, and confrms the worm tension. Methods to evaluate guide performance are more difcult to pin down; a good place to start is to use default guider settings and see how the mount responds to corrections. Under- and over-damped systems, backlash, and stiction can be detected from the autoguider graphics, or by running a specifc utility in the popular autoguider applications. Some examples are shown in the chapter Autoguiding and Tracking. Te default guiding parameters are just a starting point. Te high-torque Rainbow RST-135 works well with 2–5 guiding commands per second, the Pegasus every 0.5–1.5 seconds, and the larger Paramounts work better with a high confdence correction, every 3–8 seconds.

Mount Tuning Some mount designs are sealed and dissuade customer tuning, while others suggest one periodically cleans and lubricates gears or adjusts the worm block. Budget designs are ofen upgraded, for example, by replacing the Alt/Az adjustment bolts or a better dovetail plate clamp. On those mounts that use a lubricated gear system, the lubricant slowly degrades and traps potential debris, increasing friction and irregularities. Every few years, I clean out and replace the lithium grease in the Paramounts, making an immediate improvement to guiding performance. Other mounts that use a bolted assembly rather than a complex casting ofen beneft from careful adjustment. Tis ensures the worm axis is aligned with the bearings and engaged with the worm gear with the right force and is perfectly tangential. An Internet search on a user forum will identify what to do, and more ofen than not, there will be a YouTube video too. Finally, there are more extreme tuning examples; with upgrade kits to replace worms and to replace spur gears with a toothed belt. Tese projects are not for everyone, though, but are an valuable way to improve upon an inexpensive used bargain. Caring for Your Investment It is worth remembering that mounts are substantial and fragile at the same time. For instance, disengage the drive mechanism (if possible) to avoid unnecessary loads on the precision interfaces while assembling the system, physically moving, or during transport. In the case of the Paramounts, the worm block is disengaged and an internal peg secures the mount to stop it fopping about. Te Rainbow mount is permanently engaged but the unpowered gear mechanism is not damaged by load forces. Most mounts have integrated electronics and rapid temperature changes may cause condensation on cold surfaces and cause electronic failure. Tis most likely occurs when it is not powered or afer bringing a cold mount into a warm house. It is worth noting that some mount retailers will not fulfll a warranty claim if the outdoor installation does not have a dehumidifer installed in the observatory. If there is accidental water ingress into a mount (or any electronics for that matter), remove the power immediately and dry thoroughly before turning it on again. You might be lucky.

System Choices

54

Telescope Options Amateur telescope models have moved on from the humble Newtonian and Achromat.

T

he ever-expanding range of telescope mounts is bewildering at frst sight, but it is possible to quickly narrow the search based on budget and biceps. On the other hand, telescope optics come in an even greater array of types, sizes, and costs and defy simplifcation. Tere is no one all-purpose model, and the enthusiast is likely to own two or three to cover a range of object sizes. A choice is a good thing; at the end of the day, all telescopes use glass lenses, mirrors, or a combination of both. Te trick is to flter down to a few alternative designs and then choose one of the many broadly-equivalent models. Recommendations are a two-edged sword; useful to a newcomer but potentially alienating to those with an opinion. Some models have beguiling compound names to honor their creators. Still, we should remember that all optical designs obey simple physical laws, and the outcome is predictable to a large extent (tolerances permitting). In years gone by, one either bought a 3- to 4-inch achromatic refractor or a Newtonian refector with a 6- to 10-inch aperture for about the same money. Since then, afordable refractors have evolved into APO triplets and purpose-designed 4- or 5-element astrographs, while refector designs have had a more radical makeover with ultra-fast designs, astrographs, models using exotic mirrors, and aberration-correcting glass optical elements. Before computer-aided design, optical development was a slow, laborious process and typically limited to simple formulae. Computers can quickly model and optimize complex designs without expensive prototypes. Choosing between models is difcult, and a good place to start is to distinguish between telescope types according to the laws of optics in terms of contrast and aberrations. Like today’s modern digital cameras, there are few “bad” ones and most will perform well enough; the mechanical attributes ofen set them apart, on account of the overall construction, optical element stability, and especially the focus mechanism quality. Te following few pages generalize on the diferences between optical confgurations. Tese comparisons set an upper limit on likely performance, but imaging conditions and the physical execution of a particular optical design ofen limit the actual real-world experience.

Nowhere to Hide One of the challenges in astrophotography are the extreme demands placed upon any optics. Unlike conventional photographic subjects, which are surprisingly forgiving of optical defects, small intense light points show up every faw. Tese demands multiply with the aperture size; as the glass or mirror area doubles, quality control becomes increasingly difcult. Telescopes operate at full aperture, unlike photographic lenses, which are ofen stopped down to improve image quality (until difraction outstrips the beneft). Difraction is also the culprit for the characteristic 4-point spike around bright stars on those refector telescopes that use a secondary mirror support (spider) in the optical path. Te physics of difraction causes, in general, the simplest optical designs to have the highest image contrast. Since each optical boundary introduces additional difraction, a refractor has a higher image contrast than a Newtonian, which is higher than the more complex folded refector designs. Similarly, each optical surface fails to refect or transmit 100% of the incident light, with slightly dimmer and lower contrast images. Glass versus Mirror Glass and mirror systems have unique characteristics; when light refects of a mirrored surface, all colors refect along the same path, and the resultant image has no colored fringes (chromatic aberration) around the stars. Still, there are transmission losses and difraction from boundaries or obstructions. Glass optics bend (refract) light, with some transmission losses, difraction, and dispersion, that cause diferent colors of light to bend by diferent amounts, producing chromatic aberration. For the same aperture diameter, a lens-based telescope has less difraction and potentially higher resolution than a refector design (due to the additional difraction from its central obstruction). Both architectures have optical defects (aberrations) with various strategies to overcome them. For instance, in the case of a refractor, a lens cell will have two or three elements of diferent strengths and use glasses with diferent dispersion characteristics to minimize chromatic aberration. Te venerable Newtonian parabolic refector has a unique optical de-

55

The Astrophotography Manual

fg.1 For lightweight wide-feld imaging, this small 51-mm f/4.9 refractor has built-in fatteners and is robust to handling and transport. It covers a full-frame sensor and the focus mechanism is similar to a normal lens, but with an additional lock ring to ensure the optical assembly is rigid.

fg.2 At the other extreme, this 250-mm f/8 RCT requires careful handling and occasional collimation checks. Once aligned, the mirrors are locked in place and focusing is achieved with the rear mechanism. It represents the practical upper limit for imaging in my domestic location.

fect, coma, which is increasingly apparent away from the image center. Te two other mirror surfaces are spherical and hyperboloid, each with plus and minus points. Te image plane of a telescope is an imaginary surface upon which the stars come to focus. Ideally, this should be fat and coincident with the imaging sensor. Unfortunately, most simple systems have curved image planes and require optical correction to make them suitable for high-quality imaging onto a large fat sensor. To reduce these issues requires an additional lens or mirror element, or a combination of both, into the optical system.

multi-coated, multiple cemented, or air-spaced elements replace the simple single-element primary and eyepiece lenses of the 1600s. Combining positive and negative elements with diferent dispersion characteristics reduces chromatic aberration in the doublet design and better still in the triplet designs, ofen tagged with the term APO. Aspherical elements, now common in photographic lenses, also appear in some astronomy products.

Imaging Circle Te imaging circle is the area over which a telescope projects an acceptable image and relates to the diagonal of the imaging sensor. Many have an imaging circle of over 30 mm, enough to cover an APS-C-sized sensor with minimum darkening (vignetting) in the corners. Te premium and latest models exceed 44 mm, enough for full-frame formats or larger. If one uses a reducer to fatten the feld, the image circle may shrink and not entirely cover a big sensor.

Refractor Designs Te frst telescopes were refractors made from a positive primary lens and a negative eyepiece lens (ofen using available spectacle lenses) and produced an upright image. Today, we use later (Keplerian) designs that use a positive eyepiece lens and creates an inverted virtual image. Tese designs have a wider feld of view, higher magnifcations, and longer eye relief (the distance from the eyepiece to the eye). Today,

fg.3 Doublet and triplet refractors, and some refectors, require additional feld fatteners to ensure focused stars across the frame. A fattener/reducer will additionally widen the feld of view and increase the light intensity hitting the sensor. The two large reducers are for refractors and have adjustable spacing, the smaller one is for the RCT and is spaced with extension tubes. Most modern units have a screw-coupling option to telescope and camera, which is preferred for structural rigidity and orthogonal alignment.

System Choices

fg.4 These telescopes essentially are as long as the focal length of the design. At the top is a simple refractor doublet design. Beneath that is a triplet design, using positive and negative lens elements with diferent dispersion characteristics to minimize chromatic aberration. A feld fattener lens is inserted before the camera to ensure a fat focus plane across the whole sensor area. An alternative is the astrograph that integrates corrective optics within the telescope body to deliver consistent results (at a cost).

Field Flatteners When used for imaging, refractors require additional compound elements just in front of the sensor, to fatten the focus plane and modify the efective focal length. Referred to as feld fatteners or reducers, they are a required accessory for astrophotography. Teir design is relatively generic; a design will work well enough on telescopes of similar f/ratio and focal length. Depending on the telescope, they have an optimum spacing to the sensor (usually 55–85 mm).With increasing fattener spacing, the concave plane of focus at frst fattens and then becomes convex when the spacing is past the optimum position. Te best spacing has round stars in the image corners. Te better fatteners and reducers are screw-coupled for rigidity and incorporate an adjustable spacing mechanism to tune the distance to the sensor, which usefully permits their use on diferent telescope models without resorting to spacers, shims, and extension tubes. In the case of astrographs (telescope designs intended for photography), these instruments have 4 or 5 elements to ensure a fat feld and low aberrations. Once limited to premium brands, these are now increasingly common in small and medium enthusiast instruments at increasingly alluring economics.

56

Low Dispersion Glass For centuries, crown and fint glass have been combined in optical designs to improve chromatic aberration. In the 19th century, lens elements made from small naturally-formed calcium fuoride crystals were used in microscope optics to improve image quality, and later, larger crystals were synthesized, to make elements for camera optics. Te latest new glasses approach the performance of calcium fuoride, but without its manufacturing drawbacks. By introducing trace elements, glass can change its optical characteristics, and you will see refractors using FPL-53, FPL-51, and Lanthanum glasses. A note of caution, however; just as using carbon fber does not guarantee a great bicycle, the optical design and its execution are equally as meaningful as the use of exotic glasses alone. When choosing a model, look out for a published Strehl ratio (a measure of optical performance) greater than 0.95 for a refractor and about 0.9 or above for folded refector designs (on account of the extra difraction).

fg.5 The three Newtonian designs show increasing sophistication: The classic simply uses a parabolic mirror but the bottom two are optimized for astrophotography and make use of a more economical spherical mirror and a glass corrector. These are designed for imaging and their focus point extends further outside the telescope body, which gives room to insert the motorized focuser, flter wheel, and the camera system.

57

The Astrophotography Manual

Refector Designs Newtonian Designs A standard Newtonian is optimized for visual use, with a short back focus position that is insufcient for imaging. Te open body of a Newtonian refector cools down quickly, but unprotected mirrors tarnish more quickly and accumulate dirt. Imaging Newtonians use a larger fat mirror and closer to the main mirror to extend the back focus. Te Schmidt Newtonian design addresses some of these limitations. It has a sealed tube with a secondary mirror mounted to the rear of a glass corrector element at the front. Te primary mirror is spherical, making it more economical, and the absence of a spider means there are no difraction spikes. Te corrector plate reduces aberrations and produces a fatter feld. Tese designs are ideally suited for a narrow feld of view, and as with the Newtonian, it is also necessary to check that a particular model has enough back focus for imaging purposes. Another sealed-tube Newtonian variant is the Maksutov Newtonian, with lower aberrations and considered the best Newtonian designs. Te meniscus corrector plate at the front of the telescope is thicker than that in the Schmidt Newtonian and, potentially, takes longer to acclimatize to temperature changes, during which time there is potential for focus shif and “seeing”, from tube air turbulence. In all of these designs, the primary mirror is not moved for focusing but has three opposing screw adjustments to tilt it to precisely align (collimate) the optics and produce round stars. Tese usually need periodic correction over large temperature swings, handling, or afer transporting between sites. Some collimation adjustments are possible visually, using a defocused star, while others require a collimation tool to align the refecting surfaces. Te adjustment process is ofen unique for each telescope design.

Rowe-Ackermann Schmidt Astrograph (RASA) An 8-inch RASA costs about the same as a 4-inch APO refractor. In comparison, it operates at f/2 rather than at f/6, capturing 4x more photons and 2.2x more sky area. For deep, wide-feld imaging, it is a remarkable instrument. Te RASA uses a spherical mirror with a four-element lens group to correct feld fatness and aberrations, supported in the middle of a glass Schmidt corrector plate. Accurate focus, camera spacing, and collimation are critical at this f/ratio. Celestron makes three sizes, 14-, 11-, and 8-inch. Te 14and 11-inch require a substantial telescope mount. Small-bodied cameras are best for minimum obscura-

System Choices

58

tion. Color or mono cameras usually require fltering, accomplished by screw-in flters or a compact manual flter drawer.

Folded Refector Designs Advanced folded telescope designs were once expensive until Meade and Celestron led the way with more afordable designs and made them increasingly popular. Te folded light path makes them lighter and more compact. A 2,000-mm, 200-mm aperture telescope is about 0.5 m long and 6.5 kg. By comparison, a Newtonian with the same aperture and half the focal length is about 1 m long and 9 kg. Unlike the Newtonian designs, which use a fat secondary mirror, the folded Cassegrain designs use a convex secondary to extend the focal plane through a hole in the spherical primary mirror and out the back. A feature is their long black bafe through the middle of the primary mirror to prevent stray light from reaching the sensor. Tese designs incur additional difraction, slightly lowering the image contrast over simpler optical designs. With their large aperture and long focal lengths, these excel at planetary imaging. Te Schmidt and Maksutov derivatives, as with their Newtonian cousins, ofer increasing levels of refnement and lower aberrations. High-quality imaging over a large sensor requires a feld fattener. In these designs, many combine image reduction and feld fattening, with the outcome that a 2,000-mm f/10 design transforms into a 1,260-mm f/6.3 or 660-mm f/3.3. Te shorter focal lengths are more useful for imaging nebulas and large galaxies, with the added advantage of plenty of light-gathering power. Some scopes with longer focal lengths achieve focus using two systems; a small external standard focus mechanism with limited travel and moving the primary mirror back and forth along the optical path. Tis moving mirror can be an Achilles’ heel since it is difcult to engineer a sliding mirror mechanism without introducing lateral play. In practice, the primary mirror sets the coarse focus afer which it is locked in place. Even so, the slightest mirror movement causes the image to shif during long exposures. If the guiding system uses independent optics, an image shif in the main mirror may cause elongated stars. One solution is to use an of-axis guider, which monitors and corrects image shifs through the imaging system, adjusting for shifs from fexure and focusing. Te Ritchey-Chrétien (RCT or simply RC) is a specialized version of the folded refector design that uses two relatively expensive hyperbolic mirrors to elimi-

fg.6 The schematics above and the table on the next page compare several folded telescope design concepts. The top two both use a spherical main mirror and are similar to their Newtonian namesakes. In these designs, the principal diference is the secondary mirror is convex, which allows the folded geometry to work and the focus position to be behind the main mirror. The Ritchey-Chrétien design uses two hyperbolic mirrors. These are more expensive to make but the large image circle has signifcantly less coma than the others. There is no front corrector plate and the secondary mirror is supported on a spider. The RASA design puts a multi-element corrector and sensor where the secondary mirror would be, to create an ultra fast, medium focal length instrument at about f/2. This is an update on the earlier Hyperstar™ modifcation to an SCT, pioneered by Celestron.

59

The Astrophotography Manual

nate coma rather than use a glass compensator. Since there is no front glass element, large aperture designs are feasible and suitable for professional observatories. Te optical design is not entirely free from feld curvature and requires a feld fattener for imaging onto a large sensor. Te Hubble Space Telescope is an RCT with a 2,400-mm mirror and a focal length of 57.6 m! A typical amateur 200-mm aperture telescope, with a focal length of 1,625 mm, is only 450 mm long and 7.5 kg. Tese systems have a fxed primary mirror and use a focuser tube in front of the camera or move the secondary mirror to achieve focus. Two other, more esoteric designs are the modifed DallKirkham and the Riccardi-Honders. Te “modifed” in the DallKirkham uses a lens group in front of the sensor to correct the of-axis coma that would otherwise occur with its elliptical primary and spherical secondary mirror combination. Te seductive RiccardiHonders telescopes are short focal length refectors with fast aperture ratios and command premium prices. Tey use a weak positive lens at the front that supports a central secondary mirror and multiple integrated lens elements to correct aberrations. Te primary mirror is a Mangin mirror, a negative meniscus lens with a refective rear surface. Tese optical designs require precise alignment and stability for the best results. Te reward is a very large well-corrected fat feld. In practice, fast-aperture optics may cause circular halos around bright stars due to refections from the sensor surface and optical system. Big is Beautiful Amateur astronomers now have a fantastic choice of instruments to quench their aspirations and drain their wallets. Te thing to remember with all these remarkable instruments is that although larger apertures enable visual observation of dimmer objects and capture more photons (lowering SNR), they do not necessarily provide better resolution in typical seeing conditions. Seeing conditions are not confned

to atmospheric conditions; they also occur between air cells a few centimeters across, when convection occurs. For that reason, larger apertures are more susceptible to thermals and their massive glass components take longer to reach ambient temperature. In addition, the efect of turbulence in the telescope’s optical path typically increases with focal length. Although folded refector designs may have less air in the tube, the light beam passes several times through the medium. Long focal lengths and high magnifcation generally amplify any mechanical or optical issues. A larger aperture does not directly overcome the efect of light pollution, though capturing more photons (in any given period) improves the image’s signal-to-noise ratio. Several UK authors suggest that imaging resolution rolls of above a 250-mm aperture. (A 250-mm aperture has a difraction-limited resolution of about 0.5 arc seconds, equivalent to fantastic seeing conditions.) It sounds plausible, but I do not have the budget to fnd out.

System Choices

60

Focusers No evaluation of telescope alternatives should exclude a discussion on focusers. With so many models on the market and many sharing similar optics, it is ofen the focuser that sets them apart. A focuser plays a critical role in image quality; any fexure causes image shifing and tilting, to the detriment of focus quality across the frame. Diferential fexure also undermines tracking accuracy, causing elongated stars. Focus mechanisms are expensive to manufacture and set up, compounded by the individual needs of visual astronomy and imaging. Consequently, some top brands ofer a heavy-duty focuser option for imaging or alternatively supply their optics ftted with a third-party focus mechanism, such as a Feather Touch™ or Moonlite®. A good focuser will account for a signifcant proportion of the telescope cost, and its important attributes are smooth operation, low fexure, no slippage, and preferably low backlash. Depending on the optical confguration, a focuser may be a conventional draw-tube or move the primary mirror (or occasionally the secondary mirror) position. Tose that move mirrors are an inherent part of the telescope design since they must do so without introducing tilt. Some of the latest designs combine focus and rotator mechanisms in one unit. Te familiar draw-tube has several architectures, Crayford and rack and pinion are the most common, and, when set up correctly, work equally as well in practice. However, they have diferent strengths and weaknesses; while a rack and pinion mechanism is more likely to have some backlash, a Crayford mechanism is more likely to slip. In practice, however, a slipping tube is the larger issue. Te reasoning is as follows: astrophotographers usually motorize their focus mechanisms. Te electronic control eliminates all backlash by making its fnal focus-position move in one direction only (usually against gravity) using sufcient steps to comfortably exceed the backlash amount. Tis is similar to the process used by some mount controllers when slewing; in one direction only they deliberately overshoot and double back, ensuring they arrive at the destination from the same direction. Te best systems couple a reduction gearbox and stepper motor assembly to the main focuser control shaf. In practice, motor/gearbox systems work better than simple motors coupled to a typical “fne” control reduction drive, designed for eyepiece focusing. Te stepper motor and gearbox form an efective brake and lock the focus position. A Crayford friction drive tension must be adjusted, with the telescope pointing up and with the heaviest imaging system, just beyond the point of no slippage during an inward movement. Most motorized focuser mechanisms have a movement resolution of about 3–6 microns. Tis is sufcient for most use cases, but in some ultrasensitive and fast-aperture ratio systems, micro-stepping may be required, to reduce the step size, or a motor gearbox with a larger reduction ratio. On my Lakeside focuser system, I have one module and a motor/ gearbox on each scope (fg.8). A temperature sensor is sampled by the imaging application’s autofocus trigger. All of my scopes have a rotation feature on the focuser, some sturdier than others. If the tube has three sof nylon grub screws, I suggest replacing them with nylon-tipped stainless steel to improve the clamping force, but without damaging the metal fange. Little things matter!

fg.7 The underside of a rack and pinion focus mechanism, showing the slanted teeth, that improve backlash and stability. The many fxings hold bearings and bushes in place and set mechanism tension too.

fg.8 While this may not win any beauty competition, this highly functional motor and gearbox assembly works well with this large Feather Touch focus draw-tube mechanism. It uses a stepper motor and is driven by its own control module or any standard unipolar motor drive. Alternative gearboxes alter the step size, to suit more demanding telescopes.

fg.9 For the hobbyist, a small Arduino Nano and a darlington driver is all you need to make your own focusmotor controller. Many such projects are on the Internet.

61

The Astrophotography Manual

Camera Systems Some essential perspectives to help choose a sensor and camera system.

I

n earlier chapters, we touched upon some of the challenges we face each night, one of which is operating sensors on the edge of their performance envelope. Exposure, and particularly sensors, are emotive subjects, probably because of the constant comparison between consumer camera models. Less-experienced astrophotographers equally put too much emphasis on sensors, not helped by a new model launch every few months. When writing in 2017, the emerging CMOS astro cameras at the time were beset with reliability and technical issues, and I excluded them for that reason. In the scheme of things, I rank reliability as the #1 consideration for long-duration and unattended operation, followed by choice of mount, focuser, telescope, and fnally, sensors. Five years on, and thanks to early adopters, the OEMs have sorted most of the design issues. A complete discussion on sensors involves three main overlapping topics: understanding specifcations, how to use them to their full potential, and choosing between models. To make it more accessible, I favor an approach that divides the discussion into three increasingly detailed exposé, starting with the essentials. It is easy to obsess over new technology. I have 3 CMOS imaging cameras of 12, 24, and 26 megapixels yet my favorite images were taken with a 16-year-old, 8-megapixel 4/3-inch CCD sensor that took forever to download a frame but, crucially, its hardware and sofware were utterly reliable. In the fast-moving world of astrophotography, all books are out of date, in some way or another, by the time they reach print. However, it is important to recognize the remaining 95% of fundamental concepts, and best practices, endure. A few more home truths are in order; conventional digital photography has reached a point where diferences between sensors are increasingly irrelevant. Equivalent 16x20-inch prints from each are indistinguishable at a normal viewing distance. What remains are camera comparisons based on autofocus speed, ergonomics, and increasingly, video capability. If you come from a conventional digital photography background, it is almost impossible to escape the continual rivalry between sensors and cameras. Astrophotography is an order of magnitude more demanding than

conventional photography. Any evaluation is unavoidably technical, as understanding the inner workings is required to correctly interpret published specifcations or ill-informed amateur “reviews”. Even from the same manufacturer, the growing number of competing sensors is bewildering to novice and seasoned astrophotographers alike. At the same time, any evaluation must indicate the practical insignifcance of various sensor attributes in practical conditions. Finally, even though astrophotography is more demanding on sensor performance, one should keep in mind the quality ceiling imposed by the ever-present environmental conditions highlighted in previous chapters. Several considerations infuence our choice of a camera for astrophotography. Of course, there is no perfect camera, and to help, I have ranked several attributes in order of importance; sensor size, cooling, color type, pixel size, technology, and package.

Sensor Size Four popular sensor sizes used in imaging for sensors are 1-inch, 4/3-inch, APS-C, and full-frame. Tere are smaller and larger ones, typically for planetary and professional purposes. Teir relative sizes are shown in fg.1. Te size alters the feld of view with a given telescope and there will be a natural craving to go big. Big is not always beautiful, however, as many telescopes, even with feld fatteners or reducers, will not cover the entire sensor area with their feld of view, or the quality fall-of at the image corners will require extensive image cropping. Less obviously, large sensors take more energy to cool down, and camera models typically achieve 5–10 °C less cooling capability than those with small sensors. Big sensors, especially with a fast aperture, require big flters. A set of 2” flters is almost double the cost of 1.25”, and a high-quality full RGB and narrowband flter set is a signifcant investment, ofen with a bulky flter wheel too. A set of mid-priced 36-mm flters (LRGBHSO) cost more than $1,000 in 2021. Small objects (for example, the Crab nebula) are best imaged with a long focal length, and a large sensor is simply a waste of money, weight, download time, drive space, and image processing muscle. Big sensors are at their

System Choices

62

best creating incredible wide-feld vistas under dark skies and using premium optics. A slight tilt error that is almost invisible on a 1-inch sensor will be evident at the edges of an image taken with an APS-C or fullframe sensor. Te better cameras ofer some tilt adjustment on their faceplate. Lastly, the quoted imaging circle of a telescope is not a sharp cut-of but a subjective quality threshold. Tat subjective assessment may difer from your own; the ofen specifed 44-mm diameter grazes full-frame sensor corners. Several of my mid-range refractors quote a 44-mm imaging circle, but I notice some aberrations at the margins of a smaller APS-C camera.

Cooling and Temperature Regulation Long, sustained exposure runs beneft from temperature-regulated cooling. Without cooling, long exposures of 5 minutes or more become unacceptably noisy and, without accurate temperature control, impossible to calibrate sufciently well to withstand the rigors of extreme image manipulation. Cooling a sensor reduces its thermal current and hence its thermally-generated random noise. Each 5–6 °C reduction approximately halves the dark current (and reduces the dark noise by 40%). Setting an exact temperature allows one to create calibration and image frames with the same dark current properties, essential for efective calibration. Astro cameras typically employ two stacked Peltier coolers that provide 30–40 °C of sensor cooling below ambient conditions. Tis cooling reduces the dark current by over ~200x (since the sensor would otherwise be 10 °C or more above ambient conditions) and its random noise by ~15x. But, again, this is not noticeable until one stretches an image to reveal faint nebulosity and the shot noise from light pollution is modest. In comparison, cooled DSLR enclosures reduce the ambient temperature, but they have no direct means of precisely regulating the sensor temperature, causing imprecise image calibration.

Color or Monochrome? Color and monochrome sensors difer by the addition of a color flter array (CFA) between the sensor’s micro-lenses and photo-sensitive layer. Tere are several designs, the most common being a Bayer array, which resembles a stained-glass window (fg.2) and is usually accompanied (in conventional cameras) by a UV/IR blocking flter. Te sensor circuitry is otherwise unchanged. Choosing between color and monochrome is a strategic decision and a popular subject for online

fg.1 A range of common sensor sizes, reproduced at full scale compared to a 44-mm image circle. The quality may deteriorate towards the projected image margins and it is worth remembering that a reducer/fattener may shrink the image circle.

debate, usually involving poor science. In summary, monochrome cameras are more fexible, with higher quantum efciency, extended spectral sensitivity, and marginally better resolution. On the other hand, they are less convenient and, as a system, they are more expensive, as they usually require more extensive fltering and the extra bulk and complexity of a flter wheel. Color cameras are convenient for natural-color images and an efective choice in low light-pollution locations, possibly with a flter to improve nebulosity or reduce light pollution. In uncertain weather conditions, they capture all colors with every exposure and do not leave you with an incomplete exposure set. I use both systems, using the more compact color camera system for vacations and wide-feld shots in favorable conditions. I also double-up, using the monochrome camera for narrowband images on one system and the color camera to capture natural star color on a second telescope. Te CFA flters are less efcient than most separate RGB flters, and the red flter in the CFA typically reduces the sensor relative efciency by about 30% for the common Hα and SII narrowband nebulosity wavelengths and more if it has an integrated UV/IR blocking flter. Unlike consumer cameras, the sensors in most astro cameras do not have a UV/IR blocking flter and require an external flter to fulfll that role. In the case of CFA models, there is ofen a facility for a 2-inch screw-in flter on the camera or the telescope’s fattener, reducing the need for a flter wheel. Not all sensor models are available in color and monochrome versions. Many CMOS sensors are only available in color, taking advantage of the lower cost of

63

The Astrophotography Manual

fg.2 A sensor is a layered device, with micro-lenses on top, followed by (in the case of a color sensor) a Bayer flter array, photosites, and readout circuitry. CMOS sensors additionally have conversion electronics integrated in-between or behind the photosites.

nation of flter and photosite also has a lower quantum efciency. (Human vision is extremely sensitive to green light, and the doubling-up of green pixels in a CFA is to reduce visible noise in conventional images. During astro image processing, distracting noisy green pixels are routinely removed and replaced with a neutral color. Fuji camera owners will fnd their X-Trans sensors have an even higher proportion of green-fltered pixels in their sensors.) In summary, a mono camera generally captures more deep-sky photons over any given time, especially when one is imaging red nebulosity. Tere is one marginal exception; if one is imaging a natural-color narrowband image containing OIII and Hα emissions, a special dual-narrowband flter on a color camera exposes the red-, green-, and blue-fltered pixels simultaneously with a mild efciency advantage over separate narrowband exposures.

mass production. However, the situation is improving and the monochrome choice now includes all of the Pixel Size popular frame sizes. Astronomical cameras with CCD We saw in earlier chapters that the common pixel sizes sensors are mostly monochrome, as the generally in a sensor cannot match the difraction-limited resosmaller production batches do not penalize specialized lution of a typical amateur telescope. So, the obvious options (for instance, with or without micro-lenses, solution is to choose a sensor with smaller pixels, UV/IR flter, and cover glass). which increases the number of pixels for a given area? One last aspect of the color vs. mono debate is the Te reality is a bit diferent. I have a range of quality question of imaging efciency. By efciency, we refer telescopes, including a 51-mm f/5, 85-mm f/5.5, 102to the number of photons captured over a given pe- mm f/7, 132-mm f/7, and 250-mm f/8. In practice, with riod, the driving force behind all we do. Te CFA of a sub 0.5” RMS guiding (i.e., excellent tracking), an 8color camera has less efcient flters than the special- megapixel KAF8300 CCD, with its 8 million 5.4μ pixels, ized dichroic ones used with mono cameras. Con- has sufcient spatial resolution in practice. My CMOS versely, for a normal photograph, one would need to sensors have 2.4–3.9μ pixels and are already small and expose through three flters in a given time if you plentiful enough. Te one situation where smaller pixels were using a monochrome camera. In one case, se- are welcome is with short, fast telephoto lenses, as the lected pixels are exposed to either red, green, or blue low image magnifcation causes the overall resolution to wavelengths for the entire duration and in the other, all pixels are Aperture (mm) 50 60 70 80 90 100 110 130 150 200 250 300 exposed, for a third of the time, through red, green, and blue flF/ratio ters. For natural-color images there 2 0.5 0.6 0.6 0.7 0.7 0.8 0.8 0.9 1.1 1.4 1.7 2.0 is an argument that color cameras are more efcient; the most ef2.8 0.7 0.8 0.9 0.9 1.0 1.1 1.2 1.3 1.5 1.9 2.4 2.8 cient option entirely depends on 4 1.1 1.1 1.2 1.3 1.4 1.6 1.7 1.9 2.1 2.8 3.4 4.0 the colors you are imaging. For ex5.6 1.5 1.6 1.7 1.9 2.0 2.2 2.3 2.7 3.0 3.9 4.7 5.6 ample, in the color sensor in fg.2, half the flters in a Bayer array are 8 2.1 2.3 2.5 2.7 2.9 3.1 3.3 3.8 4.3 5.5 6.8 8.1 green. Tis is fne for landscapes, 11 2.9 3.1 3.4 3.7 4.0 4.3 4.6 5.2 5.9 7.6 9.3 11.1 but in astrophotography, green objects are rare, For the principal red fg.3 Maximum pixel size recommendations are tricky. This table suggests the pixel size for sky color, only a quarter of the pixa 50% FWHM deterioration, assuming 2 arc second seeing, using the equation els are red-fltered, and the combiopposite. It does not account for multiple image processing later on.

System Choices

be dominated by difraction and pixel scale, rather than by astronomical seeing. An ofen suggested guide is to choose a sensor whose imaging scale is a third of the typical seeing conditions or difraction-limited resolution, whichever is worse, however, it is not an exact science and the table in fg.3 suggests pixel sizes using a more scientifc approach. Tis works out the pixel size, to cause a 50% increase in FWHM in the presence of 2 arc second seeing for diferent optics. Ignoring the efects of post-processing and tracking errors, the equations in the chapter Optical Resolution can be re-arranged. Te limiting pixel size p (microns) to increase the FWHM by Q%, in seeing conditions of S (arc seconds) and for a telescope of focal ratio F and diameter D (mm), is defned by the equation: p = (F∙D / 680) √((((Q / 100) + 1)2 -1) (S2 + (125 / D)2)) Tese results are a guideline, and, in practice, imaging conditions and tracking errors may cause more resolution loss than expected. (A spreadsheet version of fg.3 is on the support website.) If the sensor pixels are smaller than required, it is always possible to bin exposures on the sensor (additionally benefcial on a CCD) or during image processing.

Camera Technology Tis is always an contested and fast-moving topic. Tere are two principal contenders; CMOS and CCD sensor architectures, and the understandable desire to use an existing consumer digital camera (which uses CMOS sensor technology). DSLR, Mirrorless, or Astro? DSLRs and mirrorless cameras are an excellent way to get started in astrophotography, and it is possible to use them for deep-sky imaging. In my book Capturing the Universe, I use them to capture a variety of bright subject matter and, recently used the video mode on a Fuji X-T4 to capture the Saturn/Jupiter conjunction. Many popular telescopes include Canon or Nikon bayonet adapters. Consumer digital cameras do not have cooled, temperature-controlled or monochrome sensors. Tat immediately puts them at a slight disadvantage. Tey are not ideal in other ways; an extended imaging session quickly depletes a battery (especially in cold conditions) and they are best used with external power. Teir small buttons and dials are also fddly in the dark, particularly with gloves, While tethered image capture is an option, it relies on the manufac-

64

turer sharing their application interface or API to allow third-party USB control. Tat is not always the case but Canon, and now Nikon, do. Not all digital cameras are weatherproof, and their bayonet couplings sufer fexure. Capturing with RAW fles is the only real deep-sky option, but this is not necessarily unadulterated sensor data, as some camera models manipulate the sensor data before RAW output, in the worse case interpreting stars as hot pixels and, more frequently, manipulating the image’s background level and ruining traditional calibration techniques. Over the same imaging period, a dedicated cooled astro camera using the same sensor should always produce a higher-quality image than a consumer camera. Consumer models work well when coupled to a fast f/ ratio camera lens or telescope, for natural-color widefeld shots, in good imaging conditions and on brighter targets. In addition, many DSLR and mirrorless models include an interval timer feature in their menus, for the simplest of setups. As conditions become less favorable and with fainter targets, more exposure is required to improve the signal-to-noise ratio. Tis is when sensor cooling and the additional control of a dedicated model are increasingly benefcial. I believe the keen pricing of the latest comparable astro cameras makes them, on balance, the best choice for deep-sky imaging, and they are used almost exclusively throughout this book. CCD and CMOS At one time, most cameras (photographic and astronomical) used CCDs. CCD sensors were to be found in professional observatories and space telescopes and still are the preferred choice for photometry. At that time, CMOS sensor performance was poor in comparison, but the equivalent systems were less expensive. Tis encouraged their accelerated development, and, at present, I am not aware of a consumer camera that uses a CCD. Sensor design has generally improved over the years, with better efciency, larger sizes, higher pixel counts, lower noise, lower power consumption, and faster operation. Tere are now many more new CMOS sensors for the astro camera manufacturer than CCD models. CCD development continues, ofen in smaller sizes and for specialist applications such as low-light surveillance and some video applications. CCD and CMOS sensors have unique architectures; while both convert incident photons into an electrical charge, the way each photosite converts charge into a digital value is diferent. A CCD chip has a single voltage output pin. Each photosite’s charge is

65

The Astrophotography Manual

amplifed into a voltage and passed, in turn, to the output pin. An external analog-to-digital converter (ADC) converts the voltage into a digital value, and additional external electronics control the sequential read of each photosite charge. Tis sequential sampling is fundamentally consistent but is slow and image downloads are similarly tardy. Te performance of a CCD-based camera is afected by the external circuitry; camera models using the same sensor ofen have diferent readout speed and noise levels. CMOS sensors are more self-contained; the integrated circuit includes the conversion electronics to output a digital value directly. In CMOS architectures, duplicate conversion circuits simultaneously convert multiple photosite voltages, resulting in fast image downloads and potentially introduce more complex readnoise characteristics. MOS doping defects occasionally cause some pixels’ mean value to fip high and low (i.e., three distributions). Tis is called Random Telegraph Noise (RTN), or salt and pepper noise. For imaging purposes, calibration and stacking remove this variation, but it can degrade photometric accuracy when images are calibrated but not stacked. High readout speed is most appreciated during activities like autofocus but less impactful on imaging sessions with individual exposures of 300+ seconds. One outcome is that diferent cameras that share the same sensor model share a similar overall performance. Camera model diferences mainly concern cooling, anti-fog features, anti-refective coatings, and operating strategies to reduce digital artifacts. Te diferences between CCD and CMOS architectures have interesting consequences: ADC Bit-Depth Analog to digital conversion electronics are complicated, and the duplicated circuits in a CMOS sensor potentially add cost. Early CMOS sensors integrated lowbit-depth ADCs to reduce cost. For example, the venerable KAF8300 CCD has a single external 16-bit ADC but my current CMOS sensors have multiple integrated 12- and 14-bit ADCs. Te smaller bit depth afects the sensor’s ability to accurately read a photosite (with just 4,096 and 16,384 states, respectively). Te latest CMOS sensors use 14- or 16-bit ADCs; their initial high prices are already falling with competition and scale. Amp Glow Te integrated electronics on a typical CMOS sensor generate heat and increase its thermal noise, more so than a CCD equivalent. In the case of the Peltier coolers, they remove thermal energy and, as a result,

achieve a better cooling performance with smaller sensors and equivalent-sized CCDs over CMOS. Sensor heating is particularly noticeable during movie capture, but even in the slow-paced world of astrophotography, the readout circuits cause localized heating efects, or amp glow, which become increasingly obvious with exposure duration and potentially brighter than faint nebulosity. Amp glow varies considerably between sensor models. Modern CCDs efectively avoid it by only energizing the readout circuits when needed. Tis is more difcult on CMOS sensors, but removing these artifacts is possible if one has precisely matched fat and dark-fat calibration exposures. Afer calibration, though, the localized random component of the thermal noise remains. Te chapter Down to Earth showed an example of CMOS amp glow and removal. Small amounts of amp glow are not a practical concern for deep-sky imaging, providing the lights are correctly calibrated. Dark Compensation In a sense, CCDs are primitive compared with the highly integrated nature of CMOS models. Afer years of working with CCDs, the cleverness of CMOS has its downside, an example of which is the efect on the calibration process. CCDs and CMOS architectures have dark current, a small, random stream of thermally-generated electrons. Over time, these slowly accumulate in the sensor photosites, increasing the pixel values. Photographers do not want gray shadows, and the CMOS sensor manufacturers employ a technique to reduce the pictorial efect. In addition to the exposed photosites on the sensor, they include a masked strip of photosites (overscan area). In some sensors, the image pixel values are automatically reduced by a constant value during the readout of long exposures relating to the mean level of the shaded pixels. It is not well documented, and the efect varies between sensor models. On some, the mean level of a 900-second dark exposure is signifcantly less than one of 0.001 seconds. Tis behavior catches the unwary, as the well-established calibration processes include bias subtraction and dark scaling. However, these processes only work if there is no internal sensor compensation for dark current. In this case, subtracting a bias frame from a dark frame to form a “master dark”, a common calibration step, creates clipped black pixels with zero value, which then introduces noise during the calibration process. Calibration is a critical topic in its own right and is fully explored in a later chapter. For

System Choices

66

now, the golden rule for calibrating CMOS exposures is to use precisely matched (temperature, gain, ofset, and duration) dark frames for both fat frames and image (light) frames. Te enhanced external electronics of a CCD have the potential for better linearity, which is important in scientifc felds, but less so for imaging purposes (that is, the pixel value goes up consistently with increasing exposure). Non-linearities are usually associated with the brightest levels near a pixel’s electron capacity. Some CMOS sensors, however, also have small non-linearities at low exposure values and, during calibration, may require fat frames of similar mean exposure to the image frames (rather than a value in the middle of the range) for best results.

fg.4 These sensor characteristics are from real data for a 4/3-inch CMOS sensor with a 14-bit ADC. As the gain increases, the system gain steadily reduces, as does the full-well capacity. In this particular model, an internal mode changes at a midgain setting and the readout noise suddenly drops with a small increase in gain. In this case, it might be sensible to have two gain scenarios; 0, for high dynamic range objects and about 1,600 for nebulosity. Other CMOS sensors will behave diferently and it is good practice to understand the particular characteristics for your camera’s model.

Binning Binning is a process that combines several adjacent photosites on-chip to form a larger virtual photosite. Te process has been used on CCDs for years, commonly combining 2x2 or 3x3 pixels. A binning setting of 2x2 reduces a 16 MP sensor to 4 MP, with a corresponding halving of the spatial resolution. At the same time, the readout speed is four times faster. If an imaging target does not require high resolution and has a high dynamic range, binning a sensor potentially improves well-depth and lowers noise. Te reason, however, requires a little understanding of the conversion process within each architecture. In the case of a CCD, all the photosites share the same external ADC, and on-chip binning combines the photosite charges before conversion into a digital value. In the case of 2x2 binning, the signal level is the combination of 4 pixels, but there is only one conversion event. As a result, the conversion into a digital value introduces read noise, measured in electrons, it is an important quality measure of any sensor, and a key specifcation. With a CCD, 2x2 binning increases the signal-to-(read) noise ratio by 4x. Te CMOS architecture is diferent and (usually) converts all photosite charges into digital values before binning. In efect, binning is simply the combination of the individual pixel digital values and is identical to combining 2x2 pixels to halve image size during image processing. As each photosite is converted separately, the outcome is similar to taking 4x more exposure, which increases the signal-to-noise ratio by 2x. One cannot simply add random noise, as it is combined in quadrature. Te binned read noise is therefore:

67

The Astrophotography Manual

2x2 binned read noise = √(4 ∙ (read noise2)) Consequently, binning a CCD 2x2 compared to a standard mono CMOS sensor/image has a 2x advantage in signal-to-noise ratio (read noise). Unfortunately, one cannot bin image exposures from any color (CFA) sensor; to do so would combine photosites sitting under diferent color flters in the Bayer array, ruining the color information. Binning CFA sensor exposures is good enough, to speed up the process of framing, plate-solving, and focusing, as these processes do not rely on color or perfect defnition. Specifcation Gotchas CMOS sensors grab the headlines for several legitimate and misleading reasons. Compared to more venerable CCD models, they have remarkable size, megapixels, and noise properties. However, the prospective buyer should additionally understand the pitfalls of taking these specifcations at face value. One way to do this is to consider what has changed. A key parameter is the Quantum Efciency or QE. QE is at the core of everything and is the proportion of incident photons that are converted into electrons. No matter what technology is used, two sensors with the same QE capture the same amount of light. Two technologies have improved QE: micro-lens design and, in the case of CMOS sensors, maximizing the photosensitive surface by moving the electronics to the back side of the chip (BSI). Te photo-sensitive material itself has seen modest improvement too. Te KAF 8300 CCD has a peak QE of 60%, not dissimilar to more modern sensors with QEs in the 40–80% range. Te very best sensors peak at 90%, and, considering the losses at each air/glass surface, further improvement is unlikely. QE specifcations are not all the same. Some OEMs quote the relative QE for Hα. Tis value is relative to the peak QE, which is never 100%, and, in absolute terms, the QE for Hα wavelengths may be as low as 40%. From experience, a small efciency gain from the latest sensor is lost by external flters with low transmission. Te peak transmission of a dichroic flter is, at best, about 98%, but it can be much lower. Te flter manufacturing process is highly variable and manufacturers have diferent rejection criteria. I have sent back flter sets whose passbands had less than 40% transmittance. Te remaining sensor processes convert the electron charge into a voltage and sample it, introducing

as little error and noise as possible and associated with a plethora of confusing specifcations. Pixel Games Sensor specifcations are reported at the pixel level, and it is tempting to directly compare models using these values without considering the pixel size. Smaller pixel sizes generally have smaller dark current and noise values. Tis does not necessarily make them better; at the end of the day, the image counts, not the pixels. For example, consider two sensors, A and B, with the following characteristics: sensor pixel size read noise dark current (x10-3)

A 5.4μ 7 e3

B 2.7μ 4 e1

At frst glance, sensor B looks best; it has smaller pixels and less noise. Interestingly, if you consider the pixel size, sensor A has a lower read noise and dark current per unit area and a better image SNR for any given light intensity. A more helpful way to compare sensor specifcations is to normalize them and calculate the full-well capacity, dynamic range, noise, and dark current per square micron (or for a particular angular measure on a specifc telescope). Tis is more meaningful to the fnal image than the published pixel fgures and is associated with the concepts of information theory afer Stan Moore. Fig.5 is a table of popular sensor specifcations using the following equations for normalizing. Te specifcation is normalized (N) per square micron in each of the following. rn is read noise, wc is well capacity, dr is dynamic range, dc is dark current and p is pixel size in microns. for normalized read noise: Nrn = √(rn2 / p2) for normalized well capacity: Nwc = wc / p2 for normalized dynamic range: Ndr = Nwc / Nrn and for dark current:

System Choices

Ndc = dc / p2 and its noise contribution over x secs is √(x ∙ Ndc) Afer normalizing key specifcations, it is not unusual for sensors with apparently very diferent values to end up with remarkably similar performance, as shown in fg.5. Gain Games Te second pitfall relates to specifcations that change with a sensor’s gain setting. Te system gain is defned as the number of electrons required to change the ADC digital output by 1 unit. Tis is not to be confused with “gain”; an arbitrary number that refers to the voltage amplifcation before the ADC. A high gain requires fewer electrons to change the output by one unit and lowers the system gain. CCDs ofen use two gain settings; high for 1x1 and low for 2x2 binning. CCDs made life simple (well, almost). Most CMOS sensors have a variable gain setting that the image capture sofware can change. A perusal of typical CMOS sensor specifcations, like those in fg.4, indicates that the sensor read noise and full-well depth capacity decrease with an increased gain setting. Te readout noise for this sensor varies from 6.2 e- to 1.24 e-, with a corresponding full-well capacity of

Sensor

gain pix. size e/ADU μ

68

61,000–1,700 e-. One can guess which values are used for advertising. Unfortunately, however, these occur at entirely diferent gain settings. Te other hidden issue is dark current, measured in electrons/pixel/second. Over a 5-minute exposure, the gain amplifes this unwanted signal and its unwanted shot noise. Tere is no such thing as a free lunch. Bit-Depth Games CMOS sensor specifcations have another source of confusion, as the system gain is usually quoted in e-/ ADU. ADU refers to the digital unit value coming out of the ADC. When the sensor employs a 16-bit ADC, as is with most CCDs, the pixel value also changes by the same amount. However, those CMOS sensors using 12- or 14-bit ADCs multiply the ADC value by 16x and 4x, respectively, to create a full-range 16-bit output value. In these cases, the system gain in e-/ADU must be multiplied by 16x or 4x to arrive at a fnal image value. To distinguish between these confusing gain terms, some applications refer to the image value as the data number or DN. With all this potential variability, it is sensible to settle on a few gain settings for a CMOS sensor and then use these same settings for both the calibration and image exposures.

rn e-

dark curr. (-10 °C)

FWC

QE (Ha)

ADC

dr (steps)

FWC /μ

rn /μ

dr (steps) /μ

dark noise /μ (600s)

IMX455 FF

0.77

3.76

3.6

0.0046

51,000

75

16

14,166

3,607

1.0

3,767

0.44

KAF16200

0.7

6

8

0.06

40,000

50

16

5,000

1,111

1.3

833

1.0

IMX571 APSC

1.0

3.76

3.5

0.0014

63,000

60

16

18,000

4,456

0.9

4,287

0.24

KAF8300 4/3

0.5

5.4

8

0.02

26,000

50

16

3,250

891

1.5

601

0.64

IMX294 4/3

3.8

4.63

6.3

0.0056

61,500

75

14

9,761

2,868

1.4

2,108

0.39

IMX294 4/3

0.85

4.63

1.56

0.0056

14,000

75

14

8,974

653

0.3

1,938

0.39

MN34230 4/3

1

3.8

1.6

0.009

4,000

40

12

2,500

277

0.4

657

0.61

IMX183 1”

3.75

2.4

2.6

0.004

15,500

65

12

5,692

2,690

1.1

2,483

0.64

IMX183 1”

1

2.4

1.75

0.004

5,000

65

12

2,857

868

0.7

1,190

0.64

ICX695 1”

0.27

4.54

3.1

0.002

19,000

72

16

3,600

873

1.1

793

0.24

fg.5 There are many sensors on the market today. This is just a selection of popular CCD (blue) and CMOS (salmon) models in full-frame, APS, 4/3- and 1-inch sizes, with their published specifcations and computed (normalized) specifcations per square micron. Some of the CMOS models have multiple entries, to show the efect of gain on their various specifcations. Each has its own strengths and the best normalized values are randomly distributed, and change with gain too. The “best” sensor depends on the use-case. The last column has the computed dark noise per square micron for a 600-second exposure and may be more or less than read noise. If these sensors were operating without cooling, dark noise would likely exceed read noise in all cases after 60 seconds.

69

The Astrophotography Manual

fg.6 These QHY CMOS cameras are designed to work in a number of confgurations. The mono camera on the left is directly screwed to a large flter wheel, with spacers for an of-axis guider tube. To achieve the right spacing to a feld fattener, the slim-bodied color APSC model on the right will also work well on a RASA telescope. The silver rod in each case is a tube of silica gel desiccant.

Exposure and Gain Te act of digitization, converting a signal into a digital value, is imperfect. Simply put, the digital output hops between discrete stepped values that approximate to the signal voltage. Te error is referred to as the digitizing (quantization) error and is a form of noise that is associated with the sensor read noise. Te practical efect of quantization noise increases with the number of electron charges required to change the digital outcome. Terefore, in those sensors with a variable gain setting, increasing the value decreases the electrons/ ADU and reduces the quantization error. Tis is the mechanism behind the read-noise improvements of the sensor in fg.4. Tis has a 14-bit ADC, and an even more noticeable improvement occurs with sensors that have lower 10- or 12-bit resolution ADCs. Tis noise improvement has not gone unnoticed and is probably the rationale behind some misplaced recommendations for DSLR users to use short exposures at a high ISO (high gain) setting. I hope by now that you recognize the faw in this argument; the way to improve quality is to capture more photons. In addition, longer exposures make full use of the available pixel well depth, and images have proportionately less read noise. Fewer, longer exposures have better overall SNR as they have fewer read-noise events. Tat is not to say all exposures should use the minimum gain setting. Extended exposures increase dark noise, and it is sometimes benefcial to use higher gains with very faint, low-contrast subjects (like nebulosity), and espe-

cially when light pollution is low. Tere is ofen a “sweet spot”, about 1/4–1/3rd along the gain setting scale, that is a good value for a variety of situations. Te characteristic curves in fg.4 also hide an interesting outcome concerning the change of dynamic range with gain. A pixel’s dynamic range is defned by the following equation, using the published specifcations for well capacity and read noise (in electrons): dr = wc / rn At high gain settings, however, the maximum voltage at the ADC occurs before the photosite reaches capacity, and the efective well capacity reduces. At the same time, the read noise reduces with gain because of less quantization noise. Tese changes are shown by the characteristic downward curves in fg.4. Te sudden reduction in read noise at a mid-gain setting causes the dynamic range to increase temporarily. Tis is a peculiarity of this sensor as it switches modes at this particular gain setting. Other CMOS models exhibit a diferent trend of decreasing noise or dynamic range with gain. Several specialist large-format CMOS sensors additionally have several internal operational modes that bin pixels with varying gain settings to increase full-well capacity and dynamic range. Choosing between the modes and gain settings is a complex trade-of and is highly dependent on the dynamic range of the subject and its intensity and belongs in a discussion on exposure.

System Choices

Camera Package Astro cameras come in a variety of confgurations. At their simplest, they typically have a cylindrical-based aluminum housing, suitable for mounting on either end of a telescope, including front-mounting to a RASA or HyperStar system (fg.6). Additional hardware is required for flter management and autoguiding. Other (monochrome) models integrate the camera with a flter-wheel housing and an of-axis guider tube. Tis larger package makes a lot of sense for backend mounting. Te reasons are twofold: Te tight integration of the flter wheel and camera reduces the distance between the flter and the sensor and reduces vignetting or permits a smaller flter size. Tis reduces the diameter of the flter wheel/housing and improves the clearance with tripod legs or piers. Tight integration also permits a closer coupling of the of-axis guide camera and removes the requirement (and obstruction) of additional USB/control/power leads. My medium-sized QHY camera, flter wheel, and of-axis guider assembly is wider than the prior QSI683wsg, and it grazes the pier in some orientations. I miss the compact simplicity of the QSI camera, and while separate camera/flter wheel/of-axis guider components ofer endless customization, its assembly reminds me of a Meccano kit of the 1980s. Non-Specifcations Not everything is neatly specifed on a sensor’s data sheet; amp glow, image banding, condensation control, orthogonality, linearity, and other noise sources are usually qualitative, at best. Download speed varies considerably too; there are clear diferences between CCD and CMOS readout speeds and between USB 2.0 and USB 3.0 interfaces. (Not all USB 3.0 cameras work well through USB 2.0 hubs.) How well a model performs over an extended period is hard to defne. Of my many cameras, some were more reliable and robust than others. It is hard to be more defnitive since the complex interactions with the acquisition system always throw a little doubt on probable root causes and expectations. Te rush to compete during this period of rapid product evolution is good for consumer choice but not so good for robust engineering. Rushed product developments, in my experience, are less likely to be fully validated, and require 6 months or more for customer feedback to percolate back into improved hardware, frmware, and sofware. Again, a perusal of user forums is an excellent place to pick up on quality and design concerns and their remedy. Based on my

70

own experience and that of others, I believe some otherwise excellent imaging sensors are not suitable for astrophotography. USB and Bufers A CCD’s slow, sequential read nature meant that it was best to read quickly in one go, at a speed that was not too fast to afect read noise. If the USB transmission was interrupted, it was possible for severe banding to appear across the image. For that reason, notebook computers are preferred to slower, resource-constrained netbooks. Today, powerful computers encourage the take-up of modern CMOS cameras’ high-speed video and burgeoning pixel counts. Tese high demands on the camera require large bufer memories (e.g., 0.25–1 GB) to permit high-speed imaging with an uninterrupted sensor read and managed download. However, this is not the whole story. In practice, the default driver settings cause regular image corruption when multiple cameras download simultaneously through a common high-speed hub, and the bufer memories do not appear to help. Tis is confrmed by the presence of a USB Trafc (throttle) control in the camera driver that alters the frequency of the sensor readout. To fnd a value that works, I set up each camera on a 1-second exposure loop and change the USB trafc (throttle) values until the downloaded images are good. Most downloaded images from CMOS and CCD cameras have faint random horizontal banding, which is commonly removed by combining multiple frames. In some cameras, the USB trafc value makes it better or worse, and experimentation with diferent values is advisable. Summary Te introduction of CMOS technology has dramatically reduced the cost of dedicated astro cameras, making high-quality astrophotography accessible to those with smaller budgets, and opens doors to alternative imaging techniques, such as lucky imaging. CCD and CMOS sensors have diferent strengths, and weaknesses and as part of a complete imaging system, operating in real conditions, it is difcult to tell them apart. In time, CMOS improvements will likely dominate the market. Unfortunately, trends can be good and bad and the current trend for highspeed, high-resolution video favors speed above all else. Tis is not a priority for the astrophotographer; worse, it may compromise read noise, full-well depth, amp glow, even smaller pixels, and long-exposure stability, afecting some models.

71

The Astrophotography Manual

Capture Software and Electronics Options The glue that holds everything together is more than likely to make a sticky mess.

T

here is an indulgent satisfaction from the feel of winding on a Leica M6 and fring the shutter. It is pure and simple mechanical bliss, honed by decades of cautious product evolution. Te same cannot be said for modern electronic life, which is fraught with an endless cycle of upgrades, incompatibilities, confusion, obsolescence, security risks, and change. Electronics has revolutionized astrophotography, not only in the form of sensors but also with the easily-overlooked numerous applications, databases, apps, advances in wired and wireless communication, battery technology, miniaturization, and operating systems. Just as with traditional photography, taking the image is the sexy part, and the subsequent “darkroom” processes do not have the same cachet. In both arenas, the two processes are distinct and for clarity, image-processing applications are found in later chapters. When I started, computing and sofware options were limited, and the common wisdom was to give a new lease of life to an old desktop or laptop computer running Windows with one of a handful of applications. Today, astrophotographers are easily disorientated by the growing and evolving PC market. Key decisions consider the choice of operating system, hardware platform, imaging applications, imaging hardware, and remote operation abilities. Tere is no simple decision path since many choices afect others. Here, we consider the following and highlight the decisions that impact others. • • • • • •

operating systems and computers remote operation applications and utilities USB electronic modules power

Operating Systems and Computers Considering the whole, the heart of the acquisition system is a computing unit, that interfaces with various electronic devices and applications. Depending on your needs, budget, and environment, this may be a desktop, laptop, or miniature PC. It does not operate in isolation and requires connectivity to external de-

vices through USB, serial, and wired/wireless network interfaces. Tere are a few devices (mostly older video cameras) that use Firewire (IEEE1394), which is an uncommon interface for modern PCs but most of these have been replaced with USB 3.0/C and Gigabit Ethernet models. Computer Choices Te computer also afects the choice of possible operating system; there are exceptions to every rule, but Linux is generally more universal than Windows. You can run Linux on many hardware platforms, including the tiny Raspberry Pi models. Selection depends on several factors; for instance, while the computational requirements of image capture are much less than gaming or video editing, the amount of image data has steadily increased with camera speeds and pixel count. To operate as designed, these ideally require high-speed USB 3.0 connections and short cable lengths. Older desktop machines increasingly fall short of these requirements, and some choose laptops as an alternative. Modern laptops have their unique good and bad points. Tey are suitably portable, and many ofer high-speed USB interfaces and wireless connectivity. However, they need to operate close by the mount and require a solution for extending battery life. Laptops are fragile and need protection from the elements and accidental damage. In a group setting, etiquette demands any monitor is shielded from visual astronomers. Afer some experience with all three, my preferred solution for portable and permanent systems is a brick PC computer, for example, an Intel NUC. Tese are small enough to mount on a telescope or tripod/pier. Tey are low-power units and usefully run of 12–19 volts DC and have multiple USB 3.0/C ports, solidstate disks, and built-in WiFi, Ethernet, and Bluetooth. Used as a desktop, they plug into the usual monitor, keyboard, and mouse. As a portable PC, they operate “headless” and simply communicate with a remote client computer. Intel NUCs are also available as “barebone” units, without an operating system, perfect for using an old license or trying out an alternative OS. I have three NUCs, two with Windows 10/11Pro and a

System Choices

72

third for experimenting with Linux. One is permanently housed by the observatory pier in a plastic sandwich box, and the other sits on the telescope or under the tripod of my portable system. Operating Systems Choosing a host operating system requires some thought; not all operating systems support a particular piece of hardware (in the form of low-level drivers or application availability). For instance, Windows is almost universally supported, but Linux (and OSX) derivatives are not. Te situation is improving, but even if your current hardware is supported, that may not be the case with a potential upgrade. Tere are three main operating systems amateur astrophotographers use: Windows, OSX, and Linux. Of these three, OSX is increasingly sidelined; due to Apple’s policies that dissuade small developers and the challenge of physically interfacing with its anorexic laptops. I use the best tool for the job; all my image processing, photography, and publishing have been Apple-based for the last 20 years, but their computers and tablets work best as client-facing systems for image capture. David and Goliath Tat leaves Windows and Linux. For many, the Windows operating system is the “devil we know”. We are painfully aware of its limitations, but it still has plenty to commend it, including extensive backward compatibility (unlike OSX) and a hardware interfacing fexibility that makes it useful as an image capture tool. On the other hand, it attracts ongoing cyber-threats that cause bumps in the road, in the form of mandatory system updates/restarts and the occasional OS “enhancement” that cause operational issues for low-level communications, common in acquisition processes. Te rich user interface and do-it-all capability have caused Windows to require ever-increasing computing resources to function well. A capable mini PC, loaded with the Windows 10 Pro operating system, is about $400 at the time of writing. Linux is an open-source operating system that has been around since the mid-90s. Some users love it simply because it is not Windows, and others discount it as a curiosity. It is surprisingly more prolifc than Windows due to its simplicity, reliability, and security. It works on many hardware platforms, and powers Android and most of the world’s Internet, supercomputers, home appliances, and stock exchanges. More recently, its public profle has boomed due to its use in the Raspberry Pi miniature PCs and in dedicated as-

fg.1 A Raspberry Pi4 next to a Core i5 NUC, showing the very diferent resources. (The i5 processor is on the other side too!)

trophotography computers, including Sofware Bisque’s Fusion and ZWO’s ASIAIR range. Linux’s efciency enables it to operate with fewer resources. Te OS is free and a Raspberry Pi (RPi) module costs less than $80. What’s not to like? Tere are several reasons why the world is not using Linux for astrophotography just yet. First, we need to distinguish between the operating system and hardware; some devices are not supported with Linux device drivers, though this is improving steadily. Similarly, not all control and acquisition applications have a Linux-compatible version and, depending on their sofware design, not likely to as it would require a complete re-write. Finally, there are a few dedicated Linux applications, some of which are being ported to OSX and Windows, for a reason I will explain later on. Another reason for the slow uptake is the user experience. Since the 90s, PC operating systems have increasingly used mouse and graphical interfaces to replace typed commands and now, fngers/stylus on smartphones or tablets. Users are very familiar with this way of working. Te Linux OS has many variants, which can be tailored and pre-loaded with applications to make a distribution, or “distro”. For example, the pared-back Linux astrophotography distros require typed commands for changing settings, updating sofware, and other activities that are a few button presses in other OSs. At the time of writing, it is necessary to open a terminal window and type in an unfamiliar (and forgettable) command, word-perfect, to perform basic operations. Tis is inconvenient (especially when remotely working with a tablet) and adds more complexity to an already challenging hobby. Furthermore, anything that requires a Google search to work out a basic function is less likely to gain universal acceptance. For wider ap-

73

The Astrophotography Manual

peal, astrophotography distros, in general, will require self-explanatory graphical interfaces. Tese interfaces, however, will increase demands on PC hardware and erode some of the OS’s efciencies. Linux hardware is an interesting dilemma. Te commercial appeal of the diminutive Raspberry Pi (RPi) series is hard to ignore, and it is possible to run a full open-source astrophotography distro, such as AstroBerry (free), headless and controlled remotely, via WiFi, using virtually any browser running on the client machine. However, the latest 8 GB RPi4, the most powerful model at the moment, struggles with some activities and, at times, has a frustratingly slow user response. Te similar StellarMate product uses the same core applications and version of Linux but, for a small fee, includes additional versions of the main applications (KStars/EKOS) that run in Windows or OSX. Tis relieves the RPi hardware from these processorintensive activities and relegates it to less demanding device-interface duties. Tese applications and the RPi communicate through a wired or wireless network using a standardized “INDI” protocol. In this confguration, image fles pass to the client computer over the network for plate-solving, focusing, or storing. A slow WiFi link will afect some activities, and for home operations, a wired network is best. Te KStars/EKOS planetarium/data acquisition combination is freely licensed. It has many features, though not yet with the refnement of the Linux versions of TeSkyX, or the fully integrated and more user-friendly behavior of the ZWO ASIAIR computer. Tere are a few other applications on the Linux platform that work well on low-powered PCs including CCDciel and the established Cartes du Ciel, both by Patrick Chevalley. Like StellarMate, CCDciel works on multiple operating systems, including OSX, and interfaces to a PC connected to the devices. Te combination of a tablet with an RPi-based host is a great combination for ultra-portable use, especially when combined with the integrated hub/dew/focus/power control of a module, such as the Pegasus Powerbox, which supports ASCOM, INDI, and INDIGO protocols. I like the idea of an RPi but found the initial excitement of using this low-powered computing module diminished with use. It has its use cases where size, cost, and power are more important than speed and sophistication. Te user experience and slow performance of some apps are not as consumer-friendly as we are accustomed to on Windows, OSX, and smartphone operating systems. Tere are also a few interface issues; setting up the device serial COM ports so that they are

recalled on power-up, regardless of the physical connection, is confused by USB hubs. It will likely improve with time. At the same time, it is important to distinguish between the limitations of the operating system and that of the RPi. In time, and with suitably higherpowered computing modules, there is no reason why Linux-based applications cannot overtake Windows in quantity and quality and avoid Windows’ unnecessary overheads. Current mainstream applications written in portable languages have made the transition, including TeSkyX, PHD2, ASTAP, PixInsight, and CdC. Entirely new applications are more likely to use languages that allow cross-platform publishing with minimal efort. It is possible to install Linux on an Intel PC and I installed Linux Mint on a NUC using a spare M2 SSD. One can predict that more powerful miniature computers will allow Linux distros to develop their interface more fully, which in turn will encourage further development. For the moment, I will continue to use Windows on an Intel NUC and, from time to time, check out the latest Linux developments. It is fun to try things out on an inexpensive RPi/Linux platform and at some time they will likely converge in power and usability, providing a real choice without compromise.

Remote Operation Tese headless confgurations are a form of remote operation and accomplish two things: it physically permits you to operate the system from afar and it allows the device-facing PC and the user interface to be diferent systems. WiFi connections are convenient over medium separations and safer than trailing cables. On a clear night, I can run two separate imaging systems, one on the lawn and the other in the observatory. Each has a NUC PC that connects to my home network and is monitored and controlled from an Apple Mac. In practice, a big screen is easier on the eye and permits a full view of several application windows. If I need to set something up on the rig, for instance, to focus the of-axis guider, I just temporarily connect to the NUC from a tablet. In this scenario, the NUCs are doing all the work, and the home computer/tablet is just a user interface. Te integrated astro-imaging computers expand on this idea and add a focuser, dewheater, and power control functionality into a single compact unit. Tese are convenient and reduce wiring between separate modules. Remote operation works at diferent scales too; one can equally control these same systems from anywhere in the world with an Internet link, a key enabler for purpose-built remote dark sites. Setting up remote operation is discussed in a later

System Choices

74

chapter; for now, however, a small brick PC permits the most fexible arrangement of hardware and sofware platforms. Driven to Despair My acquisition computers interface with a dozen hardware peripherals, each requiring a specialist driver(s). Many devices are interfaced with USB directly or with a USB-serial converter in the cable or the module. Each USB interface ofen has a low-level driver and, in the case of serial devices, a unique protocol too. Each driver has a specifc set of functions and data elements (methods/properties in sofware speak) that ofen evolve to accommodate new features within its product family. Te potential driver count reaches four fgures when this complexity is multiplied by the alternative product choices. It is an impossible task, regardless of the operating system, for any application developer to incorporate, update, and recompile their application to keep up with every driver release. Fortunately, the community recognized this problem early on, and they collaborated on a solution to ease this burden.

Keeping Up Standards

fg.2 This has been the overriding architecture in astrophotography for many years, where applications reside on the same (Windows) PC that connects to the physical devices.

Although working directly through a low-level device driver is efcient and accesses all the device’s features, there is little standardization between models or vendors. A few applications interface directly with popular cameras, but this requires constant vigilance and frequent application updates. Te solution is to introduce another driver layer in between the device driver and the application. Tis “hides” the device’s uniqueness and allows any application to work through one standard interface. For Microsof Windows device drivers, this is the familiar ASCOM platform. For Linux drivers, this is INDI or INDIGO. Te upside is that an application has a standardized interface to a device class, for example, a camera, focuser, or telescope mount. Te downside is that this interface may not give access to all of a device’s features, and conversely, with a complex class, such as a telescope mount, the device driver may not support all the possible commands and has to reject some. To accommodate this, the ASCOM standard mandates some commands are implemented and allows others to be optional. INDI and INDIGO device support is less extensive than Windows, but growing fast. To add to the confusion, PCs and operating systems are presently a mixture of 32- and 64-bit, and while 32-bit applications will work in a 64bit Windows environment, a 64-bit application requires 64-bit drivers. Similarly, Linux also has 32- and 64-bit versions. Upgrading from 32- to 64-bit may also cause expensive obsolescence; for example, the last few versions of OSX do not run 32-bit applications, obsoleting expensive paidfor sofware that is otherwise perfectly good. Multi-Platform Operation Te infuence of networks and the Internet are never far away. Te trend of remote operation is popular, either on a local or over a wide-area network. I use an Intel NUC, located by the mount, with Microsof Remote Desktop running on an indoor PC, iPad, or Mac. In this confguration, the functionality resides in the NUC and connects to my home network (which, with a fxed IP address, allows me to operate it from a local drink-

fg.3 With the addition of a remote desktop protocol, this increasingly popular approach allows the user to operate their Windows or Linux PC from an alternative computer (or tablet), via a wireless/wired connection or over the Internet.

75

The Astrophotography Manual

ing establishment!). In common with the earlier comments about using smaller resources to interface to the hardware, the trend for remote operation and distributing the processing burden continues. While a host Windows- or Linux-driven computer physically interfaces to (mainly) USB devices, the latest developments allow the user or client computer to be any platform. Tis is a logical expansion of the ASCOM and INDI concept of a unifed interface between device drivers and applications. At one level, this can be within the same computer or be an interface specifcation between computers using human-readable text like XML or JSON. Tis is hidden from the user and is one of the emerging ASCOM Alpaca API, INDIGO, and INDI server protocols. Over time, I hope the developers behind these alternative standards will agree on a single approach. Tey all have a common goal but for the beneft of all, a single standard focuses meager developer resources to improve implementation, reduce R&D costs, and consequentially, price.

Applications and Utilities Tere will likely always be more choice of acquisition applications and utilities for the Windows OS. However, a functioning system can be installed on either platform, unless a device manufacturer or small utility has not released a Linux driver or app. Te table in fg.5 indicates the broad capabilities of the more popular applications and operating system compatibility. Tis does not include the ever-changing list of utilities for weather sensing, weather forecasting, USB and power control, dew-heater control, polar alignment, time syncing, collimation aids, calibration aids, and observatory control. Most utilities have Windows support, a few are Linux-only, and a growing number are universal. fg.4 With greater choice of computing platforms, operating systems, and resources, the latest system architectures are more fexible still and do not dictate the host or user computer system. This approach also permits the device-interfacing PC (the host) to be physically smaller, using less resources and processing power (for example, a Raspberry Pi or Arduino). There are subtle diferences between the three main protocols but this simplistic overview represents their intent. INDI Server, INDIGO, and ASCOM Alpaca all have merit – it would just be better if they got together and decided on a single standard.

Imaging (Capture) Application Choices Te scope of imaging applications varies widely; data acquisition includes planning, exposure, flter selection, mount control, plate solving, guiding, focusing, calibration activities, and observatory control. Some apps integrate all or some of these, others concentrate on the core activities of focusing and exposure and make use of public-licensed utilities to fulfll guiding and plate-solving functions. Some apps are fully suited for deep-sky imaging, while others work as an imaging front-end to an external scripting application. Price is not an indicator of performance. Some of the most innovative and able apps are open-source, with multiple contributors. Te developers are invariably responsive to problems and suggestions. Tey drive the industry forward, and it is important to support these endeavors with constructive feedback and a fnancial contribution. Recommending one app over another is a fool’s errand, as it assumes we share the same requirements and logic. Reliability issues with one or another are misplaced as they are ofen caused by things outside the developer’s control, namely USB, driver, and device robustness. Genuine bugs are usually minor or diferent viewpoints of what is “logical” to the minds of the user and the developer. I have previously used heavyweight apps for imaging; I bought Maxim DL5 when I started, but I realized that a combination of dedicated imaging and editing applications better suited my needs. TeSkyX controls the Paramount but its imaging capability does not permit advance sequences without an external script. I

System Choices feature\app Nebulosity Windows

Y

APT

BYE/BYN

SGP

NINA

TheSkyX

Maxim

Voyager

Prism

EKOS/ KStars

CCDciel/ CdC

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Linux OSX

Y

DSLR support

Y

Y

astro cam

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

video cam multi cam

Y

Y

Y

planetarium planner

Y

Y

centring

Y

Y

Y

autofocus

Y

Y

Y

autoguiding Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

multiple targets

Y

complex sequences

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

shutdown protocol

Y

Y

Y

Y

Y

Y

resumption protocol

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

simple sequencer

Y

Y Y

Y

meridian fip

76

Y

weather monitoring

Y

Y

Y

observatory control polar alignment

Y

calibration frames

Y

image calibration

Y

image processing

Y

scripting

Y

Y

Y

Y Y

Y Y

Y Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

fg.5 A selection of imaging applications for Windows, Mac OSX, and Linux with a snapshot of their features. Applications are developing all the time and one should check the latest updates. These are the integrated features and many, for instance, interface to other utilities or apps for autoguiding, plate-solving, or planetarium support. Scripting is often used to add further, more complex functionality, usually associated with complex targets and control protocols for observatories.

77

The Astrophotography Manual

fg.6 CCDciel reminds me of a modern, cross-platform version of Nebulosity, providing good sequencing features and a deceptively simple, uncluttered interface.

fg.7 While a full version of TheSkyX can run on the Raspberry PI4, the Lite version (that installs with TheSkyX), is simpler, clearer, and faster and ideal for simple imaging tasks. TheSkyX is an unusual application (on all platforms) as it additionally has its own device protocol called X2, but it does additionally allow interfacing using ASCOM/INDI protocols.

only use its imaging capabilities for building PEC curves, pointing, and tracking models. My preferred capture application has changed from SGP to NINA, which permits greater control and supports dual cameras. Both use PHD2 for guiding and ASTAP for plate solving. Tese apps accomplish similar goals, but conceptually, SGP has an underlying logic, with options to deviate to accommodate system variations. In contrast NINA’s sequencer is a blank canvas and ofers greater fexibility but, at the same time, requires experience and care to get it working precisely in the way you want. Some of my friends have completely diferent preferences, and only use Maxim DL for scientifc study; familiarity is an important attribute. With such diverse and complex systems and individual needs, some combinations simply work better than others.

I have tried several other imaging applications along the way; two extremes are Nebulosity (written by the author of the original PHD guiding application) which pares things back to the bare essentials, and APT, aptly self-described as a “Swiss Army Knife” of numerous features. Early on, many imaging apps did not support DSLR users, and BackYardEOS (BYE) flled the gap, followed by BackYardNikon (BYN). Tese have found innovative means to control the focus position on autofocus camera lenses, which is an novel development. However, the need for such apps has lessened somewhat as, with the release of camera sofware developer kits (SDKs), deep-sky imaging applications now interface to mainstream photographic cameras in addition to astro models. I bought and tried each of the above during my research, and while they did the job and had some novel features, SGP’s clean, uncluttered interface was the bedrock of hundreds of hours of unattended imaging. NINA has largely replaced SGP in my observatory; an evolving and more powerful capture application that uses templates to store complex sequence instructions. Less is ofen more, however, and the clarity of CCDciel or Nebulosity is welcome on a small screen when the brain is slow, and the fngers are numb. Many paid applications have a free, time-limited trial version. It is useful to compare them in turn and, afer thrifing out those missing key features (for example, observatory control), go with one or two that you are most comfortable with. SGP, NINA, and Voyager’s approaches to the same problem have diferent perspectives. SGP may not be as fexible as the other two, but its logic is ideal for a range of users. NINA’s advanced sequencer entertains more complex logic and sequences, with fewer constraints, to appeal to more experienced users. Voyager goes further; it majors on being reliable and has many setup options (not necessarily pertinent to your setup) to overcome known device shortcomings. It relies on a built-in scripting utility to customize its operation, which is ideal for complex tasks and research. However, as systems become more complex and require more programming, the user is increasingly likely to be the weak link in the imaging system. Tere are too many capture applications to evaluate them all. Te most signifcant diference between them is ofen the look, feel, and “logic”. Depending on your computer experience, some will feel more familiar than others. Te Linux applications are in a similar state of fux, and my recommendation is to periodically review what is available and try them out, especially since most are free and several work on all platforms.

System Choices

78

fg.8 NINA has an intuitive imageplanning section, which is more informative than the equivalent in SGP and many other applications. The source can be an on-line database, an image (which it platesolves) or manually typed in. It manages mosaics and outputs the framing as a target fle, to be used later, create a “simple” imaging sequence, or a more complex sequence.

fg.9 Here, the M31 target data has been sent to a simple sequence. Even so, it has many options including autofocus trigger options, centering and end-of-sequence instructions. The exposure sequence is a series of exposure/flter combinations. This screen grab gives a favor of what is on ofer, including the visibility of the object (which shows I’m imaging M31 in the wrong season!).

fg.10 The more advanced sequencer allows full customization of the imaging plan, including start up and shut down sequences, multiple nested loops and triggers (e.g., meridian fips, safety conditions and autofocus events). It cleverly allows the user to drag instructions, triggers and targets from the right-hand side panel to the main screen. This is very powerful and fexible but it only does what you ask, so it is best used by more experienced imagers.

Tere is a cautionary warning from other disciplines; many years ago, photographers would endlessly swap and change flm developers based on mystical claims of grainless, ultra-sharp negatives. Te reality was that there was more variability in the developing process than with the chemistry. Te same applies to

applications; I recommend using an app for a while before switching; each time you change is an opportunity to make fresh mistakes and waste clear nights before being accustomed to its foibles. Tis home-truth is revealed in the forums; you will notice an imager expects immediate success and blames the application for

79

The Astrophotography Manual

setup issues before switching to another, only to be similarly confused and disenchanted and switch again before fnally returning. Astrophotography is not a turn-key hobby; it would be half the fun if it were. Utility Applications Tere are many utility apps (many of which are free, or donate-ware) that ease the burden of imaging applications. Tese include PHD2 and MetaGuide for autoguiding, ASTAP and PlateSolve2 for astrometry, and several planetariums, including CdC, C2a, and Stellarium. My favorite iOS/OSX planetarium is still SkySafari, and its graphics are excellent for general awareness of what is rising, object sizing, and image framing. I then use the image-planning capabilities within the imaging apps to determine the precise framing. FocusMax, once the go-to utility, is now a commercial app, and most imaging apps now integrate advanced protocols using curve fts. ASTAP, known for its excellent plate-solving capabilities, has replaced PinPoint in my system and also analyzes and stacks images. It also diagnoses image issues such as focus drif, collimation errors, distortion, and estimate feld curvature, making it a viable alternative to the aging CCDInspector. It runs on Windows, Linux, and OSX. Other minor apps include D4 time synchronization, SharpCap Pro (which has an excellent polar alignment utility), and focus mask sofware.

USB, Serial, and Hubs In practice, USB communications cause more issues than anything else due to poor power delivery, propagation delays, interference, and chipset quality. Te need for good-quality cables increases with data speed, and the maximum cable length reduces. USB 2.0, 3.0, and 3.1 have maximum cable lengths of 5, 2, and 1 m, respectively. To transmit further requires a hub (preferably powered) or a USB extender (which has a wide range of efectiveness). Many inexpensive cables save costs using ultra-thin copper conductors, or, worse copper-clad aluminum and poor-quality shielding. If the cable delivers power, a high resistance causes a critical voltage drop. Some devices, like flter wheels and guide cameras, are entirely powered through their USB connection, and any signifcant voltage drop causes a malfunction. Another less obvious issue is poor RF screening. Screening reduces radiated interference and susceptibility from other sources. In particular, USB 3.0 radio frequency interference from modules or cables can interfere with WiFi networks. Te better cables are double-

fg.11 The combination of a small computer sat on the Pegasus Ultimate Powerbox™ is a convenient and compact combination, with integrated power, focuser, dew, and USB-3.0 hub control. It makes setups quick, efcient, and keeps cable lengths to a minimum.

shielded and, if they are going to be kept in place for a while, have gold-plated terminals. Hubs of Evil USB hubs are another source of issues that ofen are mistakenly attributed to sofware. It is tempting to use one of the hundreds of inexpensive hubs that are designed for a domestic environment. Unfortunately, many stop working in cool conditions and, in the case of USB 3.0, have error-prone (or poorly counterfeited) chipsets. Te astro community has converged over the years, and, like many others, I use industrial USB hubs that have metal enclosures, ESD protection, and work in freezing conditions. Tey are expensive, but they have proved reliable in all conditions. I ofen lend my spare to friends with suspected hub issues, with a 100% success rate. Hubs also exist inside some modules and cameras; a great convenience, but only if implemented well. It is best practice to power all USB hubs; for outdoor operation, this may require a small sealed 12V– 5V DC converter module for some, though most industrial hubs accept a 12-volt DC supply. Moving the hub close to the equipment permits short lead lengths, which improves USB transmission reliability. Similarly, placing the PC by or on the telescope is a good option. Alternatively, using a USB extender allows a distant PC placement though some are better than others. Before linking my backyard units with WiFi, I operated with a USB 2.0 over Cat 5/LAN extender, comprising a hub and a module, linked by an ethernet cable. Tese work up to 300 feet and the latest versions also work over a home network. However, they are expensive and exceed the cost of a NUC PC, which is arguably more versatile.

System Choices

USB-Serial Adapters Like USB hubs, several chipsets interface USB to RS232 serial devices; over the years some have been more reliable than others. Tose using FTDi chipsets are consistently good, but it is not always possible to choose, if they are integrated into a device module (which unfortunately limits the connection cable length). FTDi cables, with the conversion electronics built into the USB connector, are also available with 5- or 3.3-volt logic levels (ideal for Arduino module connections) and RS232 serial for long distance communication. Te aging Keyspan converter (now Tripplite) also fnds favor, although this chunky unit assumes the RS232 serial connector is a 9-way D Type. USB Control Hubs Before leaving the subject of USB, it is a common enough nuisance that USB devices need resetting occasionally, either because they do not “connect” or for some other failure. Tis ofen implies a cable pull and power cycle, to reset the device. An increasing number of USB hubs ofer full reset capability, which, when coupled with power control, allows one to reset a device without pulling connectors. Tis is essential for remote observatories and convenience in a domestic environment. It also enables a power-on sequence (and shutdown) for all the connected devices.

80

fg.12 A high-quality USB hub is essential for outdoor use. This USB 3.0 model usefully accepts 12V DC power input, works at low temperatures, and its inputs are protected against electrostatic discharge.

Electronic Modules Tere are many alternative gizmos for focusing, environment sensing, dew-heater control, voltage conversion, communication adapters, and power control. Te simpler ones in my setup are usually homemade or repurposed modules (like USB-controlled relays) with my ASCOM driver. I purchased a Pegasus Ultimate Powerbox (fg.13) for my portable system to replace and upgrade my previous assemblage of standard modules. It is small enough to mount on a telescope’s tube rings, with the NUC mounted on top. Te assembly has short power and USB cables, and only requires a single power cable to ground level. I leave the Pegasus/NUC assembly, with all its connections, attached to the telescope, and, to set up each night, I clamp the telescope to the mount and connect up a single power and USB cable. In common with others, Pegasus has a growing range of smaller modules for focusing, dew control, USB control, and environmental sensing, with ASCOM, INDI, and INDIGO driver support. Tis is useful and keeps OS options open for the future. As an electronics engineer, I fnd designing modules gives a lot of satisfaction and sometimes saves a little money. Te support website has a growing list of projects for those of you who are interested. Te most recent is a simple Arduino-based environment sensor, and, like the others, includes an ASCOM driver installer. Te others include an ASCOMcontrolled 4-way relay for power control and a similar one for operating a single relay to trigger a camera shutter, with an ASCOM driver that behaves like a camera device. With this book, I have added a few projects for the adventurous. One is a general environmental module that measures temperature, humidity, pressure, sky quality, cloud, and rain. I use this for all-night imaging; my imaging applications continually scan the readings and shut down the observatory or set of an alarm if they exceed safety thresholds. It does not cost anything to develop non-com-

fg.13 An alternative to the unit in fg.12 is an intelligent USB controller that allows remote control and reset over its USB 3.0 connections. This is particularly useful for cameras that do not reliably connect on powerup at remote sites, where otherwise the USB connections and power would have to be manually removed and replaced.

fg.14 There are several chipsets that work in connector housings to convert USB to RS232 serial (or lower-voltage TTL serial). Some are more reliable than others. FTDi is favored by many, followed by Keyspan and Prolifc.

81

The Astrophotography Manual

mercial sofware; the developer tools are free and there are plenty of online examples. I did have to re-learn the C language afer a 25-year hiatus and I found the driver prototypes on the ascom-standards.org website made it possible to create device drivers afer a few weeks of playing. When the INDI documentation and developer resources are similarly user-friendly, I will attempt to write an INDI driver.

fg.15 The DewBuster has multiple dewheater outputs with fxed and variable power. The variable power outputs are regulated to maintain a constant ofset between ambient and telescope housing temperatures with the help of two thermopiles. This quality unit also carefully modulates the power output to avoid electrical interference.

fg.16 Many focus mechanisms use unipolar stepper motors. These motors have 6 or 5 wires and are compatible with many diferent motor controllers (with the right cable). The Lakeside controller above additionally has a temperature input, used by its optional temperature compensation feature and also reported through to the capture application, permitting temperature-triggered autofocus events. Simpler, USB-only controllers are possible using Arduino-based boards with a power transistor output stage. Many such fun projects are documented on the Internet.

Dew Heaters Several years ago, when commercial dew heaters were $100 or more, a favorite frst-time project was making your own from discrete components or re-purposing a motor/LED control module. Some designs are more suitable than others, and a few examples were described in previous books. Several others failed on account of radio frequency interference. However, things have moved on and there are several available inexpensive commercial units. Te more advanced models use two temperature probes or monitor external temperature and humidity to regulate heater power automatically. Interestingly, the low cost of miniature Arduino microcontroller units makes it possible to move up-market too and design your USB-controlled dew-heater controller, with a slow-switching output to 12 volts, for the price of a round of drinks. Focuser Modules Many manufacturers ofer electronic focusing systems with button and computer control. Te better designs use a stepper motor. Te motors themselves are usually unipolar designs and have two center-tapped windings. It is usually possible to mix and match stepper motors and controllers with an appropriate cable, providing the motor current/voltages are comparable. Te Pegasus unit mentioned earlier has a focuser output compatible with a dozen motor systems. All focus drives have backlash to some degree and require deliberate overdrive and reverse to eliminate it. A note of caution, I found some control modules do not have a backlash feature, as they assume the imaging app performs the function. Still, equally, some of my imaging apps do not have a focus backlash adjustment, as they assume the focus module handles it!

Power Some CCD and CMOS cameras are sensitive to power supply variation, especially electrical noise on their power lines. Electrical modules use DC power in diferent ways, and those devices, like pulsed dew heaters, mounts, and focus motors, are more likely to create electrical interference on the power lines. Tis interference potentially afects camera image noise and, if close-coupled to USB cables, may cause USB dropouts too. Electrical interference generated by poorly-designed electroluminescent fat panel power supplies can also jam WiFi signals. Terefore, I use four high-quality switched-mode power supplies in my observatory, sealed in a large plastic box with a desiccant bag, sealed cable conduits, and loudspeaker terminals for the DC outputs. Tese are highly efcient and do not get hot; one powers the camera and USB hub, a second does the focuser/mount/dew heater, a third powers the NUC, and the last one supplies the roof control system.

System Choices

My portable rigs work of one or two supplies, either from batteries or a bench power supply in the garage. Tese have a typical power consumption of 25– 40W. Using two supplies balances the current draw and reduces voltage drops across the power cables. It is not a good idea to overly discharge batteries; not only does it reduce their charge capacity but some mounts and the NUC computers do not work when the supply is less than 12 volts. To calculate the useful operation time of a battery, one needs to know the average power consumption. My desktop power supply has ammeters, and in warm weather, my mobile setup uses ~3 amps on average. In the case of a small 24,000 mAh lead-acid cell, the useful capacity is about 50% of the specifcation, giving a run time of just 4 hours. Lithium cells usefully deliver nearer 85% of their stated capacity, and a 500 Wh mobile power station will power the same setup for 11–12 hours. In both permanent and portable setups, the power supplies to the imaging system have their 0-volt wires connected to a single spur in the distribution/fuse box. Tis is what audiophiles refer to as a “star-point earth”. Tis reduces ground ofsets, which may cause low-voltage communications to fail, and the chance of a ground loop that can similarly pick up interference. Te power feeds (and USB cables) have ferrite RF chokes clamped on, and I use heavy-duty locking XLR connectors for DC power. Battery technology has changed dramatically over the last ten years. Lithium technologies are steadily replacing lead-acid and when I send mine to a recycling center, I will replace them with a lithium derivative. Although they are more expensive, they have higher power density and unlike lead-acid designs, they keep a stable output voltage over their useful (and deeper) discharge cycle. Tere are several types of lithium batteries, but all are made up of multiple cells to provide the designated output voltage. Be careful, some designs rated as 12-volt “nominal” mostly operate at less, and it is necessary to carefully check its discharge characteristics before purchase. One way to guarantee a minimum 12-volt DC supply is to purchase a 24-volt model and use a high-quality DC-DC converter to generate 12 volts. All lithium batteries require careful handling, charging, and discharging, and to operate safely, always follow their instructions and use the recommended charger. Like lead-acid cells, lithium batteries require careful disposal.

82

fg.17 The front panel of a typical lithium battery-based portable power-bank, with USB, AC, and DC outputs. The DC outputs are nominally 12 volts, but not all models regulate these and near depletion, the voltage may fall below 12 volts, which will cause issues with some equipment.

Some mounts require 24- or 48-volt DC, and the RPi PC is 5 volts. For these, DC-DC conversion modules are a convenient way of providing power. However, it needs to have sufcient current capability, thermal and short-circuit safeguards, low ripple current, and high efciency. Many are on the market, with diferent quality levels. As a precaution, I suggest soaktesting a new module for a few days before use, using an appropriate dummy load (for example, low-voltage light bulbs or a large power resistor). If the entire imaging system uses a single, shared power source, the current surge during power-up can cause problems with some modules. For instance, if I turn on my mount afer the NUC, the brief current demand causes a transitory voltage drop, sufcient to reset the NUC. For a common power source, I turn on the main 12-volt supply (actually 13.5 volts), power up the mount and, afer a few seconds, press the power button on the NUC. Managed power control is now possible through USB-controlled modules, and automated sequencing of power on (and of) with the modules control panel or switch control in one of the better image capture apps.

83

The Astrophotography Manual

Portable Systems A fast-growing sector with a broad appeal to newcomer and expert alike.

W

e collect more “stuf ” in many hobbies as we move on. For example, photographers purchase more lenses and bodies than they need for those “just in case” moments that rarely occur and cannot be comfortably carried around all day. Astrophotography is no diferent, and our equipment and setups slowly become larger, heavier, and more elaborate. Te additional complexity is an opportunity for more things to go wrong and making it less likely to be transported to a dark site or taken on vacation. On the other hand, keeping things simple is ofen a liberating experience and brings us closer to the roots of the hobby and, if it uses smaller and lighter equipment, makes it additionally ideal for portable use. In a pursuit that thrives on precision and rigidity, light portable setups are at a disadvantage. Chosen and used with care, however, they can produce remarkable results, especially when they unlock the potential of a better imaging site. Why Portable? Although Moore’s law persists for electronics miniaturization, the physics of photon capture and image noise does not change. For those who live in or around densely populated areas and do not have the luxury of a second remote site, a portable system is an opportunity to travel conveniently to darker skies or image at a diferent latitude. Portable systems are diverse; initially conceived to appeal to the increasing numbers of those wanting to “have a go”, many compact components are a compromise of weight, performance, and, initially, economy. Te evolving market now includes premium equipment aimed at experienced astrophotographers who require portability without compromise. In efect, the market is polarized. Te entry-level systems make use of existing photographic equipment where possible. Tey are at their best taking short exposures of the Milky Way and deep-sky objects, with optics up to ~350-mm focal length. Compact, premium mounts and optics, tightlyintegrated miniature imaging computers, and advanced automation aids work together as a competent system. A system can also be a hybrid of existing photographic equipment and specialist items. Portable systems are engaging, and for those with existing static set-

ups, put some fun back into the hobby. In the following few pages, we consider the choices within both “camps” starting with the heart of any system, the mount. Ultra-Portable Mounts Tere is always the possibility to use a static tripod and a camera system for astrophotography. Tis setup works best with short, wide-aperture lenses, in dark conditions, and stacking several brief exposures for a pleasing vista. Longer exposures or focal lengths, however, soon cause noticeable star trailing. “Noticeable” is a subjective assessment but if one is using an APS-C camera, a guideline for the maximum exposure time is given by the following equation (fL is in mm): exposure (seconds)=300 / fL For example, an exposure exceeding 8 seconds will likely have elongated stars through a standard lens and anything over one second with a 300-mm telephoto lens. To keep it simple and light, ultra-portable mounts have a single motorized RA axis and are ideal for frsttime users. Even so, using one unlocks the door to longer exposures, extended imaging sessions, and longer focal lengths. With accurate polar alignment, premium models from Fornax and AstroTrac exhibit low periodic error and work well unguided. For example, 5-minute unguided exposures taken through a William Optics RedCat51 telescope (250-mm focal length) on a Fornax Lightrack2 had beautifully small circular stars. Te same imaging setup on an inexpensive single-axis tracker would likely require the additional complexity of an autoguiding system to match the image quality. Choosing between the ever-growing number of models is tricky; conceptually, single-axis units fall into several camps: miniature motorized rotators, single-axis mounts, and swing-arm devices. However, they all have one thing in common; since only the right ascension (RA) axis is motorized, these designs cannot correct for any declination tracking errors arising from polar alignment error and their lightweight construction is more likely to fex and limit their efective payload.

System Choices

84

Rotators (Camera Tracking Mounts) Tese small, inexpensive units are simple tracking devices designed for light consumer cameras. Tey are simple to operate and are usually battery-powered. Tey employ a traditional worm-drive mechanism, which in practice, has a signifcant periodic error due to the small worm-gear diameter and worm tolerances. Nevertheless, these units are fun and ideal for teaching purposes. Teir tiny payload and tracking performance, however, limit their wider use. Single-Axis Mounts Moving up, these devices resemble small equatorial telescope mounts with one motor. Tey are more robust than rotators and mostly use worm-gear mechanisms. Like the simple trackers, their periodic error is challenged by economics and the tolerance demands of small worm gears. Some have an RA guider input that can mitigate that to some extent, but at the expense of system complexity and weight. Most have a selection of alternative tracking rates and an integrated polar scope for landscape astrophotography and convenience, respectively. Tese units require an additional attachment for DEC adjustment, usually a small tripod head or an accessory plate that additionally facilitates counterbalancing. Some of these have a built-in wedge adjustment, and others use an accessory to make small adjustments easier to accomplish. With their more signifcant periodic error and used without autoguiding, as before, they are best used with short exposures and focal lengths. Unlike swing-arm units, these rotate continuously and do not need resetting every few hours. If your goal is to run unguided, the swing-arm designs are an interesting alternative and ofer a unique combination of weight and precision. Swing-Arm Designs Swing-arm designs are a logical evolution of the homemade barn-door trackers of yesteryear. Tese units comprise a caliper-like arrangement with a drive unit at one end that changes the arm’s angular separation. Te clever aspect of this architecture is that the angular tracking error of a mount is proportional to the movement error divided by its distance from the pivot. By creating an extended arm, these mounts exhibit extremely low periodic error, rivaling that of more expensive and heavier mounts. Te downside is that they have a limited rotation period, typically 1–2 hours, afer which they must be reset and re-aligned to the target before continuing the imaging sequence. Two examples are the AstroTrac TT320, and Fornax LightTrack mounts, both of which feature an almost redundant RA guider input. Te Fornax unit in fg.1 and fg.2 uses a deceptively simple friction drive system (also used in Mesu and Gemini equatorial mounts), whereas the AstroTrac TT320 uses a long lead screw. Tese units consume less than 200 mA and will work from eight AA cells for many hours. Tese have no high-level control or handset; they simply rotate the pivot at the sidereal rate until it reaches the end-stop. I used a guider to evaluate Fornax’s low periodic error claim. I found my unit performed better than the specifed worse-case periodic error of 1 arc second peak-to-peak, a remarkable result for a tiny unit.

fg.1 The simplest tracking confguration makes use of existing photographic equipment and with care, requires no balancing. This mount runs of 8 AA cells and an interval timer fres the camera shutter (some cameras have an intervalometer built-in but limit exposure length to 30 seconds.).

fg.2 With a heavier payload, a substantial tripod and balancing is a necessity, diminishing the weight advantage of the lightweight tracking mount. The camera is a modifed Fuji X-M1, to improve Hα/SII sensitivity.

85

The Astrophotography Manual

AstroTrac has now replaced its swing-arm design with a modular system that employs a friction drive and rotary encoder. Tis can be used as a single-axis tracker or by assembling two units orthogonally, a dual-axis mount. Each has a guide port, and its philosophy places a premium on portability, and price-wise, competes with fully-featured but more substantial equatorial mounts. It uses wireless communication to a smartphone in place of a wired handset. Tese portable mounts have an unguided tracking performance in the range of 1–5 arc seconds peak-topeak, and, in practice, tracking quality is mainly afected by polar alignment. Unlike its predecessor or the Fornax unit, its rotation period has no time limit. It does require careful balancing; small loads from altering the assembly geometry, but larger loads, such as a short refractor, require counterweight ofsets.

fg.3 Good balance on small mounts is important. This assembly is simply resting on the ball and socket base and, as it is balanced, does not fall over in any direction. This angle is set to the required declination and then the assembly is screwed onto the tracker. The panoramic base of the ball and socket head allows repositioning in right ascension when a swing-arm tracker is reset.

fg.4 Another balanced system, this time with counterweights, hanging from a cord attached to the 3/8” fxing.

Wedges, Counterweights, and Balancing As mentioned, single-axis systems cannot correct for DEC and require accurate polar alignment to minimize drif. Physically, alignment involves some form of an accessory wedge, adding further weight and cost. Something as simple as the original tripod pan/tilt head or ball head may sufce for less demanding situations, but a fne adjustment is almost impossible without a geared mechanism. A middle-ground maybe something like the Manfrotto 410 geared head or more specifcally, a dedicated astro wedge from iOptron, SkyWatcher, Fornax, or AstroTrac. A good wedge is an investment; it supports the combined mount and imaging system and permits fne, controlled adjustments to achieve accurate polar alignment. Light-weight trackers have limitations; their payload is limited, and their mechanisms are delicate and will likely slip or stall if the payload is unbalanced. Te balance does not have to be perfect, most will cope with an APS-C camera ftted with a short lens, but needs care with larger fullframe DSLRs and longer lenses/telescopes. It may be possible to fnd a mounting position that balances on a simple ball and socket head, but more likely, it will require a counterweight system. For the swing-arm systems, the counterweight and wedge accessories signifcantly increase the mount’s cost and weight and may also require you to mount the imaging system on a dovetail plate. It is likely that the system will grow. Balancing any mount places the rotating mass's center of gravity onto the rotation axis. It is usually done by releasing clutches and seeing how the free mass moves. Tis is not always possible on a portable mount, and alternative evaluations are required. For example, the assembly in fg.3 is balanced and shows a camera and lens sitting on an unsupported tripod head. It is balanced because the center of gravity is directly above the center of the head, i.e., on the axis of rotation. If the center of gravity were not, this assembly would topple over. It is useful to visualize that tilting the camera over sideways (to go from landscape to portrait mode) will always create an imbalance and is best avoided. Another way of assessing the balance of-mount is to turn the problem on its head, in reverse, by seeing how the assembly hangs from its central fxing bolt (fg.4). One frst does the check illustrated in fg.4 and then rotates the telescope 90° and slides it along the dovetail to achieve fore-af balance.

System Choices

Premium Portable Mounts Ironically, the small dimensions of light-weight trackers place a greater demand on worm-gear tolerances than their larger cousins. Friction drives, using rotating discs, are easier to machine to fner tolerances and, if coupled with a rotary encoder, achieve a remarkable tracking/pointing accuracy of about 1 arc second (as in the case of the AstroTrack360). A highresolution rotary encoder is expensive and nudges a mount into premium territory. As mentioned in New Trends, however, a recent and novel design architecture using industrial robot technology is causing considerable excitement in the premium compact mount market. Tese use strain-wave motors (a.k.a. harmonic drives), which have incredible torque density, making these mounts smaller and lighter than conventional designs. Hoybm Observatory, ZWO, Pegasus, and Rainbow Robotics produce a range of small mounts with high payloads. For instance, the smaller Rainbow model weighs just 3.3 kg and has a remarkable payload of 13 kg. Tese mounts can work in Alt/Az or GEM confgurations and have a built-in polar adjustment mechanism. Most signifcantly, they do not require balancing and the counterweight system is an accessory. Counterweighting is primarily for tripod stability and increases the maximum payload by several kilograms. Unlike trackers, these are fully-featured telescope mounts with a handset, PC connection, and ASCOM, INDI, and INDIGO drivers. Teir premium pricing refects the precision machining and componentry used in their construction. Tese mounts challenge many commonly-held preconceptions. Teir mechanisms have an inherently high peak-to-peak tracking error, at about ±20 arc seconds but have virtually no backlash. Tis signifcant tracking error on the non-encoder version restricts unguided imaging to short exposures using short focal lengths. Te harmonic drive has a cyclical tracking error that has obvious periodic error components that change slightly with load, balance, and orientation. It is difcult to fully correct these with traditional PEC methods. However, some of the latest guiding algorithms have some intelligence and “learn” the cyclical errors afer a few cycles and update themselves. Generally, sub-second autoguiding corrections are essential, which is ofen seen as a portent of doom on its own. Fortunately, the torque and speed of these mechanisms are such that the tracking correction is almost instantaneous and works well at high guide rates. For example, the rapid guiding commands (~4 Hz) from the StarAid Revolution, achieved a tracking accuracy that approached the guided tracking error of my Paramount MX and MyT. In addition, the latest Rainbow model has an optical encoder on the RA axis, which lowers the tracking error by 8x to about 5 arc seconds, permitting unguided imaging in many portable setups and makes it easier to guide more challenging imaging systems. Tese premium mounts are extremely compact and work well with a conventional photo tripod base or portable pier, but their diminutive size and the small ofset to the column make leg collisions more likely with bulky cameras and flter wheels. It is usual to add a pier extension on a tripod or mount on a portable pier, to improve clearances (fg.6). Unlike some other mounts, these motors have high torque, and the transmission has no clutch; a collision between the support and a delicate mechanism will end in a crunch.

86

fg.5 In this tripod-mounted lightweight confguration (albeit without a few cables), the mount is supporting a short refractor ftted with a mirrorless camera. The camera uses its internal intervalometer and a stand-alone guider does the tracking.

fg.6 At the heavy end of “portable” this pier mounted system is capable of extended deep-sky imaging, autofocus and sequencing. The NUC is remote- controlled from an iPad. The bulky flter-wheel can potentially hit the pier and will be replaced by a slimmer flter-drawer system. The small counterweight improves the stability of the pier on soft ground.

87

The Astrophotography Manual

Portable Imaging Systems Choosing the mount is just the start; the rest of the system is ideally equally light and compact, including: • • • • • • • •

batteries mount support telescope camera flter systems guider systems computer system electronic modules

We have touched upon most of these before, including lithium batteries, carbon fber tripods, miniature computing/control modules, and stand-alone guiding systems. It is worth elaborating on the imaging components, including the telescope, camera, and flter system. Telescope Choices Compact telescopes will likely be small refractors or robust, folded refectors with fxed mirrors, such as Maksutov Newtonian designs. Imaging on vacation or at a dark-feld site benefts from short, fast optics to ensure sufcient exposure depth in a short duration. Most manufacturers ofer several refractors with apertures of 50–90 mm using high-quality doublets or triplets with matching feld fatteners. At the short end, the RedCat51, by William Optics, is a 250-mm focal length telephoto lens and is ofen doubled-up on a side-by-side saddle plate to increase capture hours and create a two-panel mosaic. It accommodates 2inch screw-in flters and has a secure guide-scope mount, which permits the use of DSLR/mirrorless cameras and astro cameras. Te RedCat has a conventional helicoid focus mechanism with a lock ring to secure the optical assembly. Te focusing can be motorized, but, in practice, it focuses well enough with its Bahtinov mask and keeps its focus as it cools. Askar is another brand with a range of high-value short-focal length astrographs. Tere are dozens of high-quality “grab and go” telescopes, with little to choose between. For example, William Optics and others have many refractors in the 60–80 focal length range, including an increasing number of 4- and 5-element designs that do not require additional feld fatteners. If you look carefully, it is clear that many share the same optical cell and difer only in aesthetics and focus mechanism. At the top end of this market, the Takahashi Baby Q FSQ85 is a fast wide-feld refractor. Once ftted with an imaging

adapter and reducer this “baby” weighs over 5 kg and is more “luggable” than “portable”. Camera Choices If you already have a DSLR or mirrorless camera with you, they can do double duty for astrophotography. Teir large sensors (typically 4/3, APS-C, or 35-mm) are ideally suited for wide-feld imaging and vistas. Tey potentially side-step the computer-controlled focus, cooling, download, and exposure control paraphernalia. Tey permit the most straightforward deepsky imaging setup comprising a mount/support, handset, telescope, and camera. Imaging without a computer is liberating and works best with short focal lengths that disguise tracking errors or using an independent tracking device, such as the StarAid. However, the extra weight and complexity of computer control has many ofsetting benefts; in particular, accurate focus, framing, dithering, and exposure control. While USB control of Nikon and Canon models is generally supported by imaging apps, support for other conventional cameras is poor or non-existent. Tese cameras, (or any camera ftted with a Bayer array) are sensitive to dominant light-pollution colors and ofen beneft from a light-pollution flter or narrowband flter to reduce shot noise and isolate nebulosity. As shown in earlier chapters, photographic cameras have higher dark current and hence dark noise than their cooled astro counterparts and are at their best taking short exposures under dark skies. Although cooled astro cameras take higher-quality exposures, on account of lower dark current, less internal image processing, and calibrate more efectively, they require an acquisition system to run them. Tis implies a computer, more cabling, external power, a focusing, guiding, and flter system. If narrowband imaging is your thing, a monochrome camera has better sensor efciency (especially for deep red wavelengths) and allows more control over the imaging process. On the whole, astro cameras are generally more compact, robust to the operating environment, and have no fddly dials, buttons, or menus to fnd in the dark. Filter System Choices Both monochrome and color cameras beneft from selective fltering to exclude some wavelengths and pass others. Astro cameras are not fltered and require a UV/IR blocking flter. Most astronomical flters automatically block UV and IR wavelengths. Photographic cameras typically have a UV/IR blocking flter bonded

System Choices

to the sensor. Additional fltering is ofen needed, depending on the camera, conditions, and target. Traditionally, astro flters are mounted in a manual or mechanized flter wheel as a separate unit or integrated with the astro camera. Tings have moved on, and in addition to 1.25- or 2inch screw-in flters, there are 31-, and 36-mm circular flters and complex shapes, to ft within the lens throat of a photographic camera. Filter wheels are convenient but bulky. On a small mount, such as the Rainbow, they almost guarantee collisions with a tripod or pier, especially if there is an of-axis guider and a smaller solution is benefcial for a portable system, even if it requires some manual intervention. Tere are a couple of alternatives; fxing a flter in the optical path or a manual flter drawer. In the frst case, several telescopes now ofer a convenient 48-mm flter thread for a 2-inch flter in the focus tube or feld fattener. My QHY cameras have a Meccano-like adapter kit that holds a 2inch flter between two screw-in adapters. Tese systems, and those that snap a flter into the lens throat of a digital camera, are workable for extended imaging runs with a single flter but require disassembly to change over. Tis is not ideal during an imaging session and may cause accidental misalignment, dust, and fngerprints. In addition, I am wary of screw-in adapters; they have a habit of seizing, even afer lubrication, and require rubber gloves or surgery with a drill and a lens wrench to unscrew. Manual flter drawers are potentially the solution; designs from Baader Planetarium, Astrodon, and others are suitable for compact premium systems (and RASA users). Here, a slim but rigid frame is assembled in front of the camera and accepts a flter frame with a single flter (fg.7). A couple of flter frames permit quick and easy flter changes, with less risk of misalignment or contaminating the flter surfaces. Application Choices Imaging at home and “in the feld” are very diferent afairs. Perusing the multiple docked modules of an advanced application on a 27-inch monitor is a very diferent experience from squinting at an iPad or 12-inch laptop screen with a sleep-numbed brain and cold hands. Not all capture applications work well on a small screen; some have a minimum screen resolution, or their ergonomics are unsuitable for small screens. To some extent, equipment confguration and sequence planning beforehand make life easier. Still, even when running a sequence, those applications with small controls, fne graphics, and confusing iconography are more challenging on a small screen. In addition to display clarity, it should also be possible to select controls or enter data easily. Without a separate keyboard, the tablet interface becomes cumbersome, and a stylus is ofen useful for precise selections. I have also noticed that some touchscreens do not work well with cold fngers. Te good news is, as portable systems become more popular, applications are considering these alternative environments, and the simple layout of Nebulosity and CCDciel, for example, adapt well to a small screen. One thing to note about the allure of a portable system used for feld trips and short-stay vacations is as you become more profcient and go deeper, the probability of acquiring sufcient exposure is devastatingly annihilated by poor weather or conficting demands on your time. Been there, have the T-shirt, and subsequently sold the gear.

88

fg.7 This flter drawer system from Baader Planetarium has an extensive set of adapters that suit most of the current camera and telescope interfaces. The drawers can accept various flter sizes, including 48-, 50.4- and 50x50mm flters and adapters for 1.25”, 31and 36-mm. One side of the adapter has a camera rotation feature, held securely with 6 grub-screws.

Iris Nebula (natural color)

Setting Up

Setting Up

90

Catalogs, Maps, and Surveys There are more cells in the human body than there are stars in the Universe. Discuss.

T

he examples in the earlier chapters hint at the amazing variety that awaits. Te vast number of possibilities is, however, bewildering. Over the centuries, as objects were discovered, they were variously recorded and passed on to the next generation and into astronomical catalogs. Tese are an invaluable resource to the astrophotographer, mainly to aid object selection and facilitate accurate framing and tracking. Today, gigabytes of astronomical data are held in Internet databases and integrated into PC-based and mobile planetariums, which we largely take for granted. However, it was not always that way and we owe a debt of gratitude to generations of dedicated and unbelievably patient astronomers. Early Catalogs Te frst astronomers were not as fortunate; naked-eye astronomy only permitted the brightest stars to be depicted in ornamental images, ofen intertwined with other elements of astrological signifcance. Te Ancient Greek Antikythera mechanism and, later, the Islamic astrolabes of the 13th century indicated the potential for celestial map making, but if there were any, none have survived. Translating a 3D solid model into a 2D map requires some form of spherical projection. Te earliest known printed star chart originates much later in Europe c.1440 and was still based mainly on the ancient work of Ptolemy. Some early catalogs have survived. Hipparchus cataloged 850 stars in detail, and Ptolemy revised it and included the fuzzy outlines of a few bright nebulae. In the 16th century, Galileo, Tycho, Kepler, and Copernicus revolutionized mankind's understanding of the universe and propelled the science of astronomy, and still largely without the aid of a telescope. Once telescopes were employed methodically, the number of observable objects increased exponentially, creating the need for systematic catalogs by type, position, and brightness. Some of the earliest catalogs and classifcations still exist today; Johann Bayer started the convention of prefxing the constellation name with a letter from the Greek alphabet in the order of the star brightness, and John Flamsteed, in his star atlas of 1725, listed stars using numbers combined with

the constellation in the order of their right ascension. (John Flamsteed was the frst Astronomer Royal at the Greenwich Observatory, London. He built the observatory on the meridian, and his telescopes only pivoted in altitude, making it convenient to label stars in the order they crossed the line of sight.) By the early 20th century and with the aid of photographic plates, the Henry Draper Catalog listed more than a quarter of a million stars. Later still, using digital and subsequently satellite imagery, catalogs expanded further. Lost in Space Amateurs mostly use these stellar catalogs for celestial navigation. Visual astronomers use the brightest recognizable stars as signposts for fainter observing objects, and most early telescope mounts used these for calibrating their orientation, usually by manual alignment. Tis method is still in use today and assisted by automatic means; by comparing catalog data with a camera image using plate-solving sofware. A popular catalog for plate-solving was the Guide Star Catalog (GSC), originally compiled to support the Hubble Space Telescope. With a suitable catalog, the plate-solving process matches the relative positions of the brighter stars in an image with a catalog database. In a matter of seconds, it derives the actual image scale, position, and rotation with incredible accuracy. Most modern image acquisition applications use this technique to precisely align a telescope mount (afer a slew) to the target coordinates rather than rely upon the telescope mount to know exactly where it is pointing. In addition to plate-solving, the amateurs use massive stellar databases to detect supernovae. Tese days, it should be almost impossible to be lost in space. Catalogs for Astrophotographers In addition to the detailed stellar databases are the catalogs of interesting objects. We mostly use these when choosing and planning an imaging session. One of the most famous is the Messier Catalog. In 1781 the French astronomer Charles Messier published Nebulae and Star Clusters. Tis was not a catalog of stars but of 110 deep-sky objects that were not to be con-

91

The Astrophotography Manual

fused with comets. He used a simple index, prefxed with “M” to identify these objects; for example, M31 is the Andromeda Galaxy. Since observations with a telescope at that time only showed the most obvious deep-sky objects, it follows that these objects, in turn, are prime subjects for amateur astrophotography. Te Messier Catalog is very convenient and, ironically, became the backbone of amateur astrophotography in the northern hemisphere. Te late Sir Patrick Moore generated a supplementary hit list of 109 objects for his Caldwell Catalog. He noticed that Messier had excluded objects that were only visible from the southern hemisphere and had missed several interesting bright deep-sky objects too. Since Messier had already taken the “M” prefx, Moore used his middle initial “C”. His catalog is listed in numerical order of degrees away from Polaris (declination). In addition to these two, a group of astronomers selected 400 deep-sky objects from the 5,000 listed in John Herschel’s Catalog of 1864, which are observable from mid-northern latitudes and with a modest telescope. It is called the Herschel 400. About 60 objects in the Herschel 400 also occur in the Messier or Caldwell catalogs. One hundred years afer Messier, the New General Catalog (NGC), compiled by J. Dreyer, listed about 8,000 objects, stars, and deep-sky objects and remains a useful comprehensive catalog, still in use today. It is astonishing that the early catalogs were compiled by hand, without the help of computers or photographic records, but by patient observation. Tere are many catalogs, and it is common practice to specialize for a particular object type. Many are for specialist star types, but, some are for deep-sky objects, e.g., Sharpless for hydrogen emission nebulae, Abell for galaxy clusters, and Barnard for dark nebulae. A selection of useful catalogs is shown in fg.1. It is no longer necessary to pore over large books of numbers. One can now use planetarium programs on computers, smartphones, or tablets to select objects for viewing or imaging and display their location, size, and brightness. Today the main catalogs are available in digital formats and are freely available; for example, from U.S. and ESA websites. Clearly, as new objects were identifed in the early days, the catalogs expanded and overlapped previous editions. Subsequently, as measurement techniques improved, those with more accurate astrometry, brightness, and color replaced earlier surveys. Even so, stars and galaxies are on the move relative to the Earth and each other, so any catalog’s accuracy will change in time. Tis has less signifcance for the amateur, but renewed surveys are re-

quired to update their databases for scientifc use. In addition to data catalogs, the Digitized Sky Survey (DSS) is one of several photographic atlases of the night sky available from both planetarium and image acquisition applications. In summary, the astrophotographer has more objects to photograph than a lifetime of clear nights. Te choice is bewildering, and while planetarium apps ofer recommendations for the more popular targets, other references (like Charles Bracken’s Te Astrophotography Sky Atlas) list more unusual targets. Putting the principle target coordinates in the middle of the frame is not always the most pleasing composition, and I ofen use its DSS image to plan the composition of the frame or mosaic. Increasingly, image acquisition apps and some mobile apps use these survey images and combine them with computed altitude, Moon phase, and weather conditions to further assist the planning process. Increasingly, these are linked directly to intelligent telescope mounts for instant alignment.

fg.1 A selection of catalogs that include deep-sky objects such as galaxies, nebulae, and clusters.

Setting Up

92

Site and Image Planning A little planning goes a long way to make astrophotography more productive.

I

once documented all the equipment and sofware setup settings, calculations, and steps to start an imaging sequence. Te result was scary and highlighted how complicated astrophotography can be and how easily the unwary can fall. At the same time, I realized it was essential to simplify the process and make it as robust as possible to human error. To improve the chances that a night’s imaging completes without incident, one can do several practical things (without going mad). On any given night, there are many variables, not all of which are under your control (like the weather). Te trick is to care for as many of the ones as possible. It is rarely the case that an imaging session goes fawlessly by chance. A little preparation and planning beforehand help, starting with deciding what you want to image, from where, and when.

Systems and Siting We have already seen from a few examples just how amazingly diferent deep-sky objects can be. Tese, in turn, place very diferent demands on an imaging system. A key question at this stage is, “What do you wish to image?”; the answer is usually a selection. Te most demanding targets are generally the smallest and dimmest, infuencing the system choice and where you are most likely to image from. For instance, these objects beneft from long focal lengths with large apertures that, in turn, require heavy-duty mounts, which are less convenient for portable or temporary setups. On the other hand, imaging expansive objects is less demanding because of the more forgiving image scale and falls within the capability of smaller and lighter systems. In the middle, a growing number of amateur systems handle a range of assignments at home or further afeld. Many of us start with a temporary setup at home, some progressing to a more permanent solution, for convenience and longer imaging sessions. Others may drive with a luggable system to a dark site or travel to an astro-tourism destination with an ultra-compact system. Ultimately, a location’s light pollution and the typical atmospheric conditions set the upper-performance limit for astrophotography, and an increasing number of backyard users, discontent with their local

environment, ship and install their equipment at a managed remote dark site, in a diferent county, country, or even continent and operate it remotely through the Internet. At best, a temporary system takes about 30 minutes to set up or pack away, with travel time on top. Weather conditions and forecasts for some countries are a hit-and-miss afair, and inevitably, there will be several false starts each year. Early enthusiasm can wane, and although we might start with good intentions, the practicalities and efort of imaging away from home may be worthwhile for just a few guaranteed imaging runs in fantastic conditions, an astro camp, or a social context. On the other hand, a backyard system is convenient, especially for all-night and extended imaging sessions, though ofen with more light pollution, an interrupted horizon, and a neighbor’s insecurity lights. In my case, I have a worthy dark site about 20 miles East near the coast, with low light pollution and a convenient grassy car park surrounded by preserved salt marsh. Mosquitos come free of charge. Maybe another day. Temporary (a.k.a. Portable) Setups A portable setup needs to be just that, and it is surprising how quickly the repeated assembly of a large, articulated mass in cold, dark, and damp conditions reduces its appeal. For example, my frst purchase was a used 8-inch Meade LX200 SCT. Te previous owner had bought it for his retirement but quickly changed to a lighter telescope, and I soon did the same. It is not just a case of lifing a heavy weight; equipment requires transport, carrying, and assembly, ofen in the dark, without trips, damage, or injury. Tese larger systems require a more permanent setting where they can remain fully assembled. Many of us would love the convenience of a turn-key observatory, but with a bit of preparation, however, it is possible to deploy a temporary setup and be imaging at short notice. If a rain shower takes you by surprise, a large waterproof cover is thrown over in seconds, potentially permitting the setup to remain for a few days. For a portable system, weight is prioritized in the tradeof between weight and rigidity. Temporary setups, by their nature, also require

93

The Astrophotography Manual

repeated assembly, alignment, and handling. Tese cause additional wear and tear on the mechanical and electronic components. For instance, the optical alignment of some refector telescopes is sensitive to handling and may require collimation before each imaging session. For that reason, in the case of temporary and portable setups, I almost exclusively use refractors. Permanent Setups Permanent setups usually involve some structure to protect the assembled system while not being used. Being permanent, a structure's siting must consider the imaging horizon, access, planning regulations, and electrical installation. In addition, its orientation needs to be compatible with domestic requirements and the geometry of a pole-aligned, articulating telescope. Amateur-purchased solutions include roll-of roof and dome-based observatories. Others make their own, and there are some wonderfully inventive designs, with articulating pitched roofs, sheds on rails, and even one in the shape of Doctor Who’s Tardis. As for the equipment inside, it can be as simple as the original portable solution on a frm footing, tripod-based, or mounted on a central pier. A word of caution; while my wooden roll-of roof observatory was the cost of a medium triplet refractor, the subsequent furry of mount and telescope upgrades far exceeded it! Although the equipment is permanently assembled in these setups, it is exposed to a harsher environment. While aluminum and stainless steel are robust to outdoor life, electronics, connectors, and optical surfaces need protection from corrosion, humidity, dust, and condensation. Sometimes, the mount warranty re-

fg.1 My original master box has survived the test of time. I have one permanently in the observatory and another for backyard use. Today, there are more compact integrated solutions, which are small enough to ride on the telescope and coupled to a miniature PC.

quires an observatory to use a dehumidifer. A permanent solution on your doorstep also invites unattended all-night use. To avoid an unexpected rain shower ruining things requires an automated shutdown strategy that predicts or detects poor weather and takes appropriate precautions before closing the roof. In both my systems, I house all the hubs, modules, and low-voltage electronics in an enclosure to speed up the assembly, minimize exposure to moisture, and for connection consistency (fg.1). Tis practical project is in earlier books. Te concept has been improved upon by several manufacturers who, by integrating the electronics and connectors onto a single circuit board, have shrunk their units sufciently to suit on-telescope mounting and supply them with their own Windows and Linux drivers for focuser, dew, USB, and power control. Leave It to Chance? Having an idea of what you want to image beforehand makes a lot of sense. For many of us, the weather is unpredictable, and when unexpected good conditions do occur, we scramble to image… something, but what? Sensible planning and preparation, such as the ones below, avoid relying on good luck: • • • • • •

target availability imaging system planning capture preparation exposure and sequence planning nightly timetable weather planning

Target Availability Deep-sky imaging is not a quick fx, and it is frustrating to start a promising imaging sequence only to be obliged to break of before you have acquired enough exposures to make a quality image. Dim galaxies and nebulae beneft from at least 10 hours of data, or more, if you use narrowband flters. Depending on the target coordinates, month, and conditions, it will likely require several nights’ worth of imaging. Deep-sky objects pass by slowly over the year, and their season spans from when they rise before dawn until they have not quite set at dusk. Te optimum period is when the target crosses the meridian, around midnight. A target’s declination and right ascension set the visibility over the year and for each night. For instance, if I compare the Pleiades with the Orion Nebula, at my latitude, the Pleiades has an imaging season from August to March, and at its peak, rises over my imaging horizon for 8 hours. Te Orion Nebula is visible for

Setting Up

94

many months over the winter, but its low declination shrinks the imaging time to an hour per night, just beneath my nominal imaging horizon. Tat situation improves with a lower latitude. It helps to use a simple diary of what and when to image. Tere are several ways of doing this; for instance, Te 100 Best Astrophotography Targets by Kier and now Te Visible Universe by Bracken and Whitby are month-by-month practical target suggestions. One can also use a planetarium application that shows deep-sky object outlines and scroll forward in time to evaluate a target’s visibility over your imaging horizon. Te process may be time-consuming but, at the same time, displays other potentials in the vicinity that may be worth following up on or combining in a wide-feld composition. Various applications that combine catalog data, geographic location, and imaging preferences readily identify potential imaging targets. On mobile platforms, the iOS app Observer Pro (fg.2) is an excellent tool for understanding an object’s imaging opportunities over the year using an innovative graphical interface. For computer users, AstroPlanner, DeepSky, and SkyTools4 are several more traditional applications, as is the image planning tab in NINA, that select and sort catalog entries by visibility, size, magnitude, and so on. Finally, if you prefer a more leisurely perusal in paper form, Charles Bracken’s Te Astrophotography Sky Atlas has a comprehensive set of maps of varied deepsky objects for all seasons and latitudes.

Image Aesthetics With all the glamor of equipment, computers, and sofware, it is easy to lose sight of the purpose of astrophotography. It is convenient to slew to a popular target, center it, and 20 hours later, voila! It is a record, yes, but is it art? Many of us, myself included, have done just that and it is reasonably efective when the scale of the object fts conveniently within the feld of view. It is possible, however, that a diferent composition will improve it. Artists have the natural skill of making these choices without apparently thinking, while the rest of us sometimes need a gentle aide-mémoire. Tese choices are half the story, as, similar to a photographic negative, the magic of image processing creates something that is a world apart from the faint smudges on the screen. What Makes a Good Image Better? Te saying, “Art is in the eye of the beholder” holds true and although astrophotography is essentially record-taking, a quick Google search reveals an amazing variety of interpretations, that allows one to turn multiple subexposures into photographic art. Tese include both technical and aesthetic choices. While many successful images conform to common guidelines this is not an exclusive club and if the image is strong, deliberately breaking composition rules may cause a stronger emotional response. A good part of photography is knowing what you want to achieve before you press the shutter button. Te sophistication, immediacy, and automation of digital photography encourage bad habits. Tirty years of experience using flm in all formats is not lost; I am still frugal with each of my subjects and consider the details before the shutter is pressed.

fg.2 The iOS app Observer Pro gives useful object data for long-term planning, including the best months and visibility by month, day, and hour. The orange line represents object altitude, green is computed visibility and the dashed lines for the Sun and the Moon. Another screen shows a one-line summary for sorted multiple objects, facilitating a quick comparison of likely subjects.

95

The Astrophotography Manual

fg.3 Both imaging applications and planetariums can assist with precise object framing. Here NINA and SkySafari are using downloaded deep-sky images to show the extent of the nebulosity. On the right, the C2A planetarium shows a flled outline, that approximates the brightest parts of the nebulae. As you can see from the monochrome image, the nebulosity fades of, with no clear boundary.

Technical Considerations Te technical aspects of a good deep-sky image have less room for interpretation and bridge image capture and processing. Stars should be tightly focused and round, all the way into the corners of the image. Ideally, they should retain some of their natural colors in the fnal image. Stars are ruthless indicators of poor capture technique, including focusing, tracking, optics, and exposure errors. While it is possible to mitigate some issues during image processing, it is better to avoid the issue during image capture. Te same is true for the appearance of the sky background and noise levels. It is better to consider the minimum altitude, exposure plan, and suitable fltering upfront, rather than rely on fxing problems during image processing. Aesthetic Considerations With a fxed subject that cannot be re-arranged, our choices are limited to orientation, scale, framing, the position of the center of interest, balance, and directing or confning the view. Deep-sky images are also two-dimensional; everything is in focus, and there is no foreground or background. Before exposure, we must consider orientation, scale, and framing. It is all the more difcult as a single test frame usually reveals little detail, e.g., the galaxy core and bright stars. For that reason, I ofen plan my images in two passes; a planetarium to confrm the target availability and general Field of View (FOV) and the system choice, followed by the detailed framing assistants in the capture application, which overlay the FOV on a downloaded sky-survey image (fg.3). On more than one occasion I have been caught out by an inaccurate object outline shown in a planetarium.

Deep-sky objects rarely have welldefned limits, and their displayed outlines are somewhat arbitrary. However, the sky-survey images show fainter nebulosity and permit precise centering and rotation. On more than one occasion, I have changed to a wider FOV to capture the fainter margins of a nebula or galaxy. Another trick is to look up others’ images on the Internet and use these fully-processed images to plan your own framing. Tose of nebulae ofen indicate the narrowband content too. Rotating an image can make a surprising diference; for instance, vertical and horizontal presentations change the dynamic mood. Image dynamics are an interesting subject, with some orientations and placements “feeling” more natural than others. Landscape orientation is ofen considered passive compared to portrait orientation. Tere are few portrait-orientated astrophotographs and those that exist convey power (e.g., the Hubble Space Telescope’s image “Te Pillars of Creation”). An object’s angle also generates diferent emotions. If one considers two similar portraits, the eyes are level in one and the other, tilted, provoking a diferent emotion. Te angle and direction of the tilt also have a surprising efect. If an object has an axis, in so much that it is not an amorphous blob or perfectly symmetrical, tilting that feature increases the image’s dynamic (fg.4). For example, try reversing the angle; it has a surprising efect. In the West, it is said our brains “read” an image from lef to right, like a book. In addition, a swooping diagonal from the bottom lef corner to the top right feels more natural than the opposite. As these are images rather than scientifc records, one can also consider if the image is more compelling when refected on a vertical or horizontal axis. Other

Setting Up

96

fg.4 It is interesting to compare the diferent emotional efect of an image’s orientation. There is no “up” or “down”, yet the one on the right looks upside down. The middle image looks more engaging than the one on the left to my mind.

conventions from general photography include the rule of thirds, avoiding distractions near the image periphery, and image dynamics. Te rule of thirds is a common guideline for rectangular images. For reasons that are obscure, placing an object of interest on the intersection of thirds has a pleasing efect, especially if there is some balance in the image to ofset the main attraction. However, this does not always work for an image with a single object (a cluster or galaxy) and sometimes a square image with a centered object is more powerful. Te brain seeks out bright and high-contrast areas, and a distracting object near the edge of an image draws the eye away from the center of attention; for example, a bright star or its difraction spikes close to an edge is annoying. With a familiar object, the brain is easily distracted to miss these things; one trick is to look at the image upside down. Similarly, it is usually necessary to slightly crop images (or change their aspect ratio). Tis is an opportunity to exclude an exceptionally bright star on the periphery and ensure the border does not bisect a small galaxy or bright star. In regular darkroom work, one would burn in a distracting highlight or clone out in the case of a digital image. To some, this is painting, not photography, and is considered heretical, yet it is still art. Regarding aspect ratios, common sensors have 1:1, 4:3, and 3:2 ratios, with some video-orientated models at 16:9. Tere is no rule that your fnal image has to retain this shape. A composition does not have to be of a solitary deep-sky object or place it dead-center. Some of the best images combine several in juxtaposition. Others combine transitory neighbors; a comet passing a galaxy or a nebula adds a diferent dimension to an image. Including several targets also allows one to bal-

ance the image, for example, an open cluster ofsetting a bright nebula. At the extremes, a composition might be a small part of a deep-sky object; a fascinating patch of nebulosity, dust pillars, and globules are favorites. Alternatively, it might be wide-feld, taking a well-known target (e.g., Pleiades) and showing it in the context of the clouds of gas and dust that surround it. Te universe, so to speak, is your oyster.

Imaging System Planning Te apparent size of deep-sky objects varies enormously, and composition is usually a compromise from the available telescope, reducer, and sensor combinations. Te nearest combination may be anything from a generous frame to a close crop. If the available FOV is too small, a mosaic is another (if time-consuming) option. Te Sky X planetarium application calculates the FOV combination for one’s telescopes, reducers, and cameras. Tis is compared with a catalog object’s dimensions, (remembering difuse boundaries are vague at best), and the objects are usually random shapes. Other planetarium applications include StarryNight, CdC, C2A, HNSKY, and Stellarium for PC users. SkySafari is my choice for Apple devices, and on Linux, KStars, TeSkyX, and HNSKY ofer planning capabilities, with varying integration with capture apps. Te better image capture applications (e.g., SGP and NINA) also display deep-sky survey images in relation to your equipment’s FOV and allow precise framing and pass orientation data to the exposure sequencer. Te graphical representations vary enormously, and fg.3 shows a few equivalent charts and planning view examples for the same object. Most image capture applications also have composition planning utilities that link to the target coordinates and

97

The Astrophotography Manual

camera rotation controls. Tese include object altitude maps, frame superposition over catalog or imagebased sky maps, and controls for planning overlapping mosaic images. An object’s center point is not always the best choice, and these permit fne-tuning and updating the target in an imaging sequence. In addition to the applications above (several of which are free), there are many free online planetariums. For planning purposes, the ones I have sampled are not as efective as a dedicated application and are usually accompanied by distracting adverts. Nevertheless, they are ever-evolving, and it is an area worth checking out occasionally.

tion and its impact on focuser settings, dither, imaging scale, and feld of view (assuming the other devices are common). Te capture applications I use permit profles or confgurations to be modifed. My initial approach is to create a basic equipment profle and use it for an imaging session. Ten, I closely monitor the session, and if some things do not work perfectly, I make manual changes during the imaging session and update the profle aferward (this is usually associated with matching various settings to image and guider scale). In this way, the improvements are easily carried over to any session using the same equipment. When I introduce new imaging equipment, the underlying sysCapture Preparation tem is the same, and I modify a profle/confguration Te better capture (acquisition) applications have the for focal length, pixel size, and so on before saving it facility to remember the key operational characteris- as a new profle. tics for multiple equipment confgurations. Variously I also create distinct profles designed to capture fat known as equipment profles or confgurations, they or dark calibration frames without linking to mount, have a signifcant role in improving the repeatability focuser, or guider devices. Te autoguider applications and reliability of your imaging system and are a prime also have equipment profles that store the device inexample of pragmatic planning in action. Tese per- formation, calibration data, and link to bad pixel maps manently store and recall important information or dark frames associated with a particular guide camabout the connected physical devices, including fo- era. PHD2, arguably the most popular guider sofware, cusers, mounts, cameras, rotators, observatories, and auto-updates the current profle with operational setenvironmental monitoring. (Some may include addi- tings without an explicit update. tional sofware control over power and USB connecTe popular image capture applications also store tions, dew heater, and fat-frame accessories.) Tese exposure, target, and operational commands in fles, confgurations also store important operational infor- for instant recall or modifcation. Te precise organimation about the guider system, plate solving, autofo- zation depends on the application. For example, SGP cus, and startup/shutdown processes. It is possible to stores the target and exposure information in a semanually enter this data for an imaging session, but it quence fle along with the current control panel. Te takes considerable efort and increases the chance of control panel is, in efect, a local copy of an equipment human error. Furthermore, the diference between profle kept with the sequence. It can be loaded/saved profles is ofen related to camera and telescope selec- to a profle, modifed in use, and keeps equipment choices and many settings in one place. Another app, NINA, stores similar information but structures it diferently, using targets, templates, and sequences. Te emphasis in both applications is on reuse. A modern imaging application requires a vast selection of user input to function. As a result, there is a lot of information to recall, and keeping a record in a spreadsheet is helpful. I have mine in an fg.5 SGP equipment profles repay the efort to create them several times over, storing the online fle, an example of which is many settings required to operate a complex system. NINA does similarly, with on the book’s support website. diferent screens for equipment selection and operational settings.

Setting Up

98

Tis interactive sheet calculates the imaging scale, feld of view, focal lengths, guide settings and records autofocus characteristics, camera gain/ofset settings, and fattener spacer notes. I refer to these when assembling an imaging system or setting up a new equipment profle. Tese sheets are also an opportunity to record notes about a particular setup likely to be forgotten.

Exposure and Sequence Planning An exposure plan varies from a simple instruction to trigger a camera release a dozen times to a complex sequence with multiple targets, exposure settings, conditional actions, and additional activities. Similar to equipment profles, saving a tried and tested plan and re-using or modifying it for a new sequence is helpful. Tese exposure plans usually link to an equipment and location profle to form a cohesive group of settings. For extended imaging of deep-sky images, a plan allows one to continue an imaging session on another night with a minimum of user intervention. For instance, Sequence Generator Pro takes three mouse clicks from the PC boot to continue an entire imaging sequence from one night to the next. If the sequence ran smoothly last time, it will likely do so again, and by using stored sequences and profles, the user is implicitly planning an imaging session. Te specifc content of exposure plans varies with application, as do the boundaries between equipment and exposure settings. Typically, if a setting is considered generic to a group of exposures or targets, it is more advantageous to be stored in the equipment profle. Examples include autofocus trigger settings and safety monitoring. Determining the optimum exposure duration is considered in the chapter Exposure Planning, but when it comes to estimating the likely total imaging time, in the absence of prior knowledge, the Internet is a handy resource. It ofen helps to look up the exposure details for online images and, afer considering the equipment used and conditions, estimate your requirements. For instance, there are some remarkable wide-feld images with amazing depth and, amazingly, taken in a single night. I was initially disbelieving until I realized they were imaged with a RASA telescope (a hybrid refector telescope) operating at f/2.2. Compared with my f/5 refractor of the same focal length, equivalent image depth was achieved, in a ffh of the time! Te key takeout is that equipment and exposure plans require a little upfront work and experimentation, but if they are kept up to date with your latest optimized settings, they make life easier and more reli-

fg.6 When imaging multiple targets on a given night, the individual capture periods are optimized for the target’s altitude and the available astronomical darkness.

able later on. Tat is not to say that things never go wrong, far from it, but when trying to debug a problem, the general consistency helps isolate the cause. At the heart of an exposure plan (a.k.a. sequence) lie the details for the exposure including: • • • • • •

camera gain and ofset setting flter setting camera rotation exposure duration exposure binning exposure count

In several capture applications, this group is called an exposure event, and there may be several events for a target, with individual settings for each flter in the flter wheel. In many applications, a set of events is repeated, to ensure that each flter receives the same overall exposure. NINA’s balancing plugin stores progress from one night to another and prioritizes exposures accordingly.

99

The Astrophotography Manual

Te target itself will also have unique settings: • • • • •

image center coordinates camera rotation angle slew and centering behavior start and stop times or altitudes options between exposures (scripting, pauses, etc.)

In the case of targets with brief imaging opportunities, it makes sense to image multiple targets to make the most of a clear night. Most capture applications have the additional capability to manage multiple targets in a sequence and automatically move from one target to another when the prior one has concluded, timed out, or reached its altitude limit. To manage the optimum changeover times, modern acquisition applications like NINA and Sequence Generator Pro (SGP) have a facility to show the altitude profle and imaging window for a given target. Sequences for creating mosaics are an example of a multi-target exposure plan in which the exposures overlap with a safety margin. NINA and Voyager also go one step further and automatically sequence multiple targets for any given night, depending upon a number of user-set criteria. If I consider a single target, an ideal imaging window is set up at dusk when the target is low on my Eastern horizon. By the time the system has cooled down, centered, and focused, the target is above the nominal altitude threshold and ready to go. Earlier in the season, I would have to wait for it to rise before imaging, and conversely, later on in the season, I start imaging as soon as it is dark. For sequences with multiple targets, the frst up is the one that sets frst, and the target’s end-time is set to coincide as its altitude falls lower than a threshold (nominally 35–45° in my case). If the second target in the sequence is viable when the frst sets, it does not need a start time, but it usually has a target end time (or sequence end time) to stop imaging before it sets or dawn. As each exposure plan for each target times out, the sequencer moves to the next to optimize the time under clear skies. It is easier to see this in graphical form, and fg.6 shows how one would do this in practice. Each night has a slightly diferent optimum plan based on altitude and darkness. Nightly Timetable As the night progresses, a sequence with one or more targets benefts from fne-tuning to adjust to the changing dark hours and the best changeover time

from one target to another. It only varies by a few minutes each night and is only signifcant if one is continuing a sequence for a week or two since the last imaging session. In the case of NINA, it can automatically adjust for darkness, an object’s altitude, and lunar light pollution from session to session. To conclude our discussion on exposure planning, we fnish of with a top layer of sequence-level settings and conditions, including: • • • • • • •

connection behavior sequencing behavior safety monitoring image and fle header management camera cooling and warming meridian fips shutdown and recovery behavior

We can do a few other things to keep exposure plans simple. As discussed in greater detail, CMOS cameras have temperature, binning, gain, and ofset controls that allow countless exposure combinations. Tese multiply with potentially multitudinous exposure durations to create a crazy situation where no two events use the same settings. Tis obfuscates best practices and requires matching calibration exposures for each combination. Binning options add further complexity. Binning an image within a CCD sensor has a distinct image noise advantage over sofware binning. However, on my KAF8300-based CCD camera, 2x2 binning causes bright stars to bloom into irregular shapes and causes other unwanted artifacts. As we have previously discussed, CMOS sensor architecture is diferent, and binning on-chip is usually identical to binning during image processing (except on a few models). For simplicity, I set all exposures to 1x1 binning and exclude the overscan area. What is lef is to decide upon a suitable gain and ofset for the various exposures. In general, I use two and sometimes three gain settings; low (minimum), medium, and high, with a sufcient ofset in each case to ensure that every pixel is registering noise signal (I check the minimum pixel value changes on repeated dark frames and is in the range 0–150). I deploy low gain settings for maximum dynamic range when exposing through LRGB flters, typically for stellar objects. I reserve medium gain settings for narrowband deep-sky targets to trade of exposure length, peak intensity, and read noise. (Extended exposure durations have considerably more amp glow.) Finally, I reserve high gain set-

Setting Up

tings for those occasions I have a particularly dim nebula. I combine with other exposures at diferent settings to improve the overall dynamic range with HDR image processing. Night to Night Light pollution and atmospheric seeing afect exposure plans from one night to another. I typically image with a monochrome camera ftted with a flter wheel. To avoid continual autofocusing and settling afer dither, I use pre-determined flter ofsets and dither every 2 flter cycles. I separate narrowband and LRGB exposures into two imaging loops within NINA, making them conditional on Moon illumination and target angle with diferent sensitivities to lunar light pollution. In that way, the same sequence can run over several weeks without any intervention. When the frst target expires, I delete it from the beginning of the sequence and add a new one at the end. Te sequence stages (startup, target acquisition, and shutdown) are saved as instruction templates, making it quick to assemble an entirely new imaging sequence process that just needs its target information populated.

100

Detection I use multiple forecasts to reach a medium-term view and DarkSky (now Apple’s Weather app) to provide short-term notifcations. Both my observatory and portable systems have environmental monitoring, connected to the image capture applications for recording observing conditions and crisis management. If you have access to the Internet, openweathermap.org is available as an observing conditions ASCOM device in real-time for capture applications. Reaction I prefer to predict the onset of rain rather than react to it. Terefore, I included safety triggers for humidity, cloud cover, and precipitation when planning the control strategy. As conditions deteriorate (i.e., damp and cloudy), the image capture pauses, the mount parks, the roof closes, and it waits for conditions to improve before reopening and resuming. Te shutdown safety plan has PC-independent redundancy (in case the PC fails) that shuts the roof if it detects rain or sounds an alarm if it cannot close the roof without colliding with the mount.

Keep Planning Simple Weather Planning Tis may seem an oxymoron; in the UK, meteorology is an inexact science, and while weather systems are predictable to some extent, localized variations are common, so how can one possibly plan? In a planning context, it breaks down into three considerations; prediction, detection, and reaction. Prediction Weather applications come and go. Most are alternative graphical front-ends based on a dozen global data sources. Several astronomy-focused ones add Moon phase and seeing condition predictions. Overall, the regional forecasts are the worst case, and there has been more unpredicted good weather than bad. A little knowledge of cloud sequences is helpful; for instance, small cumulus clouds in the evening generated by thermals are more likely to disperse than high cirrus clouds that signal a warm front. Te online weather reports and those with dedicated sky forecasts are useful for anticipating clear skies in the next 48 hours, although the exact timing is less specifc. During the day, the local cloud conditions help to re-time the forecast, especially if corroborated with information from other sources, say a weather station. As far as imaging goes, these systems are only useful to determine the approximate probability of a clear night.

Astrophotography is complex enough without making it worse. For example, I restrict the combinations to allow me to standardize exposure plans and re-use dark frames. Tis strategy has little downside; there is no such thing as perfect exposure (all exposures are a compromise between clipping a bright star and losing faint details below the noise level). As such, I standardize on memorable exposure times: 900, 600, 300, 120, 60, and 30 seconds. I restrict my CMOS cameras to three gain and matching ofset settings and operate at -15 °C. Tis implies 18 exposure settings with matching calibration frames. Not all combinations of gain and exposure are helpful. To reduce complexity, I only use short exposure times for bright objects with low-gain settings and 600+ second exposures with medium- or high-gain settings for dim targets. Each year I spend a few summer days generating a new library of dark fles for ten useful exposure/camera setting combinations for each camera. Tese are binned 1x1 and taken at -15 °C, achievable in the UK. Te fat frames and fat darks use the same sensor settings before the optical confguration changes afer sequence completion. Tis is acceptable practice for a refractor. However, those open refector telescopes that are more prone to contamination require fat frames afer each imaging session.

101

The Astrophotography Manual

Hardware Setup A little forethought makes hardware setup fast and trouble-free.

T

his chapter concerns itself with the physical assembly of the imaging system with an updated emphasis. With care and forethought, it is entirely possible to have an imaging system assembled and cooling down afer 15 minutes. Some of the principles are universal, but many vary according to the equipment choice. To complicate things further, it is seldom the case that one makes the right purchase choices at the outset, and it usually takes several attempts to fnd a system that works for you. Tere is a compromise between swapping out selective items and endlessly changing equipment in a futile search for the holy grail. Each change can potentially introduce a hiccup in your workfow and require fresh learning (a.k.a. debugging) to make the system operational once more. As product life cycles shorten and there is a race to introduce new models, early customers too frequently fnd themselves doing the fnal product validation. Earlier chapters have suggested some of the foundations (literally) of best practices. Tese include the essentials of stability, rigidity, and siting. Tis chapter and the few that follow build on these assumptions and recommendations. Setting up the hardware and sofware is an engineering task, and I frmly believe in the 80:20 rule. Tis philosophy attempts to get 80% of the way there with 20% of the efort. Let me share an example; while it is possible to use advanced multi-point sky-modeling to improve slewing and tracking accuracy on a mount with poor polar alignment, in practice, it is easier to generate and more efective if the mount has reasonably polar aligned. Te same applies to guiding; pragmatic polar alignment and periodic error correction remove the bulk of the tracking errors, making it easier for the autoguider system to correct what remains. Tis is not an obsessive-compulsive disorder, just pragmatic engineering. In recent years modern telescope mounts and sofware have improved dramatically and diversifed in architecture. To an extent, this has changed how you set them up. For instance, it is not as essential to have highly accurate mount alignment and slews if the image capture applications employ plate-solving to center an image. Similarly, several mounts now use optical encoders on one or more axes and employ ad-

vanced sky models to track unguided. While some concepts remain constant, others vary wildly with equipment. For example, of the 8 mounts I have owned, one worked best with no attempt to balance, two others needed precise balancing and the others were tolerant of a slight imbalance. A similar story plays out with polar alignment accuracy; a mount with appreciable backlash may guide better if there is a slight drif in DEC. In short, to maximize the performance of your equipment, it is helpful to be familiar with its characteristics and foibles. Basics Given the previous discussions about stable and solid supports, I use a couple of ideas to reduce setup times. Tese include accurately aligning any pier/tripod to the North and making it level (to within 2 arc minutes). While neither is a necessity for a German Equatorial Mount, they have long-term advantages, and it is just an example of aligning to known and repeatable references. All my pier/tripod adapter plates are close-tolerance, as are their couplings to the telescope mount. For all but the most critical applications, once I accurately polar align a mount, I can precisely re-assemble it on another occasion without further adjustment. Consistent tripod placement is made trivial by placing the tripod’s spiked feet onto three substantial, long metal rods or coach bolts that have been hammered into the earth. I initially set the tripod orientation with one leg facing North. Tis requires an accurate compass or using the shadow of a plumb line on the ground at the moment of the Sun’s opposition. Afer these preparations, I reliably achieve repeat assemblies within 5 arc minutes. Te Paramount and Rainbow mounts make this easy, as they have integrated azimuth adjusters within their baseplates. I use a diferent approach for those mounts that have two opposing adjusters, that push against a tripod-mounted spigot. For these, I thread a nut onto one of the azimuth adjusters, and once the mount is polar aligned, tighten the nut to lock the bolt position. Tis becomes the mechanical reference, and, at the end of each session, I slacken the adjuster on the other side to facilitate disassembly and, conversely,

Setting Up

the next time the system is put together, the only one that is adjusted. Mount Handling Mounts are a dichotomy; on the one hand, they are generally heavy, unyielding objects, yet, at the same time, they are delicate and precise instruments. Tese contrasts make transport, handling, and assembly a challenge. Te aim is always to protect the mechanism from undue stresses and prevent moving parts from crashing into something or causing injury. Some mounts have clutches, which are usually disengaged for storage and balancing, others have levers that disengage the worm gears, and others have a combination with locked positions that prevent the mount from fopping about or damaging the mechanism during transport and assembly. Te best advice is to do the unthinkable… and read the handling instructions frst (they are usually placed strategically in the shipping carton or at the beginning of the manual). Initial Fact-Finding Setups Several one-time setup processes establish the nature of your system. Tese include measuring periodic error, backlash, setting approximate focus position, and adjusting camera spacing for feld fatness and tilt. A clear night with a full Moon is an ideal opportunity to understand the strengths and weaknesses of the mount and the imaging system as a whole. I run through a checklist with a new mount before attempting to image. In the list below, those marked ** are likely to be model specifc: 1. 2.

Fix the mount to a level and aligned tripod. Set an approximate altitude by adjusting mount’s saddle inclination to the current latitude (using a digital level or smartphone app). 3. Assemble the imaging system using a guide camera as the main imaging camera. 4. Bundle the various cables, and sleeve and route/ support them for minimum drag. 5. Rotate the camera so its long axis is aligned to RA axis (for PE analysis).** 6. Establish an approximate balance on both axes.** 7. Polar align and secure Alt/Az adjusters. 8. Find rough focus on a bright star using a focus mask or capture application utility. 9. Check the collimation of a refector telescope using star testing.** 10. Calibrate the autoguider near the meridian at low declination using short exposures.

102

11. Use an autoguider utility to evaluate DEC backlash and adjust/compensate as required. 12. Disable the autoguider output and monitor the tracking error for half an hour.** 13. Calculate the periodic error correction and apply it to the mount/controller.** 14. Slew to diferent parts of the sky and evaluate default guiding parameters. Tese assume a host of other setups, including accurate time and location settings in the mount and a functioning imaging system. Setups are ofen initially iterative to dial in a fnal solution. If the backlash is unexpectedly poor, it may indicate a problem with worm engagement and alignment. Continuing the example, I would then propose to: 1.

Reconfgure the imaging system with an imaging camera, flter wheel, of-axis guider, guide camera, etc. 2. Focus the imaging camera and note the position for future reference. 3. Adjust any guide camera focus position. 4. Evaluate feld fatness and tilt with short exposures. 5. Adjust the fattener-camera spacing to minimize feld curvature. 6. Re-balance the system and mark the balance points for future reference. 7. Run autofocus with each flter and note their position ofsets from luminance. 8. Re-calibrate autoguider and measure/consider backlash compensation 9. Evaluate the tracking error with guider output disabled and periodic error enabled. 10. Use tracking error’s rate of change to suggest guider exposure and aggressiveness settings. 11. Set the mount’s park position and safe tracking/ slew limits. One should now have a functioning system that is ready for imaging. It usually takes a few sessions to settle on optimal operational settings or, conversely, if the plan is to image unguided, construct a comprehensive pointing/tracking model. Some of these are worth considering in more detail. Worms and Backlash Some mounts will always have some backlash due to their gear design, but unexpected amounts may indicate an underlying mechanical issue. Several autoguider applications evaluate backlash during their cali-

103

The Astrophotography Manual

bration routine of the DEC axis. Its efect causes the DEC axis to be unresponsive over multiple guide commands when the guiding polarity changes. Te performance of a mount using a worm/gear is particularly sensitive to the assembly tolerances. Te worm should be perfectly tangential to the worm gear, without any lateral play and with just the right engagement force. Te ability to make adjustments depends on the manufacturer and the design and is sometimes discouraged by the manufacturer’s warranty policy. In the case of Sofware Bisque (Paramount), they realize that the micro-settings for their mounts may change during shipping or afer extended use and provide detailed instructions for user adjustment. Other designs are not as accommodating judging from some how-to YouTube videos, and adjustment is more drastic than tweaking a few bolted assemblies. In the case of the Paramounts, I clean and replace the lubricant, and check the cam stop, worm spring force, and belt tension every few years. Te efect is usually noticeable in my before- and afer-guiding performance, with less cross-axis interference. It is good practice to have a clean drive mechanism. My Meade SCT looked like a vacuum cleaner had been emptied inside.

Polar Alignment More words are wasted on polar alignment than any other topic, so as not to disappoint, here are more. It is easy to become obsessive and collect gizmos and apps like confetti. So, in a moment of boredom, I compared half a dozen and tested their relative accuracy, speed, and repeatability. First things frst, though, there is no such thing as perfect polar alignment. As a result of atmospheric refraction, polar alignment is a compromise between pointing accuracy, drif, and feld rotation, whose optimum positions lie within an altitude range of a few arc minutes. Tere are several alternative approaches: polar region star alignment, sky modeling, and drif alignment. It is worth considering each in turn. Polar Region Star Alignment Tis is the foundation of optical polar scopes, in which a short telephoto lens focuses the stars around the celestial pole onto a calibrated reticule, the QHY PoleMaster, which works out the center of rotation and correct star positions, and SharpCap Pro’s and NINA’s polar alignment utilities, which use plate solving to calculate the center of rotation, in relation to the computed pole position.

fg.1 This is a selection of some of the more common polar scope reticles from a polar alignment app. Most now use a clock face of some kind and require alignment so that the vertical cross-hair aligns to the mount’s altitude axis. The one above is inverted… something to watch out for. The bottom left image is from an older mount and is difcult to use without accurate hour-angle settings on the mount.

Setting Up

104

fg.2 The popular QHY PoleMaster is a good way to align a mount electronically with its own application and its 25-mm CCTV lens (left), or with SharpCap Pro, replacing the lens with a QHY mini guide scope (center). The camera base may also be redeployed as a guide camera, using PHD2 autoguider software (right).

It is easy to dismiss the polar scope as low-tech. However, used correctly, with good (valid) reticle centering, they are easily capable of a respectable 5 arc-minutes alignment error and, with care, 2 arc minutes in no time. Tis is sufcient for most setups using an autoguider; the worst-case drif rate is a tiny 0.5 arc seconds per minute, and plate-solving centering techniques quickly fxes any slew errors. Even so, they are slowly losing favor as “low-tech” as they need careful handling to preserve centering and their inconvenient ergonomics. Te electronic methods accommodate the centering issues that afect polar scope accuracy. Te most popular is probably the QHY PoleMaster, which comprises a short 25-mm focal length CCTV lens mounted on a guide camera. Its companion application requires you to manually identify Polaris and line up a template (rather than plate-solve) before and afer two RA slews. Tis process establishes the center of rotation. Afer rough centering to a target crosshair, the fnal alignment is sensitive to the Alt/Az adjustments and routinely achieves an accuracy of 1 arc minute or better. SharpCap Pro's polar alignment utility uses plate solving to replace the manual labor of matching a star template. It ideally images through a short guide scope and reliably fnds the refracted pole within 15–60 arc seconds. In summary, the three techniques are similar in approach with the fundamental diference that a polar scope requires accurate centering (i.e., zero cone angle) whereas the two electronic methods cancel out any cone-angle misalignment and do not need to know the

hour-angle. Te QHY PoleMaster camera does not have an ASCOM driver, but SharpCap Pro, EKOS, and PHD2 autoguider applications directly interface with it, creating an opportunity for several hybrid confgurations with the camera base (fg.2). Sky Modeling As used by 10Micron, Rainbow, TeSkyX, and MaxPoint, modeling compares multiple assumed and actual sky positions in all directions (from manual alignment or plate solving). It builds a mathematical model from these errors, including the derived polar misalignment. As the method does not directly image the celestial polar region, it requires dozens of points, or more, to isolate the polar error from atmospheric refraction, fexure, cone angle, and tracking errors. With hundreds of points, the resulting comprehensive tracking and pointing model is the ultimate for polar alignment and unguided operation. However, it takes time to generate, and if you move or change anything, it needs repeating, so it is most suitable for permanent setups. Pointing and tracking models are discussed at length in a later chapter. Drift Alignment Tis method measures the physical error mode, i.e., star drif, to facilitate alignment. It is evaluated with the autoguider/tracking model disabled. As mentioned in the earlier analysis, a modest misalignment has negligible drif. When using a sensor with a typical pixel scale of 0.5–1.0 arc secs/pixel, it takes several

105

The Astrophotography Manual

minutes to be confdent of any star drif. For that reason, drif alignment is assessed through a high-power eyepiece. Since the drif analysis only examines along the DEC axis, it is unafected by RA tracking errors arising from periodic error. Tis, is not the case with some electronic drif alignment methods. One method captures and platesolves any point in the sky and calculates the drif over several minutes. Since it knows where it is pointing, it can relate the drif on both axes to polar misalignment. Tis method, however, is adversely afected by periodic error. On my Paramounts, with just 1 arc second PE, it takes 10 minutes for a stable reading. Te same method is unusable on the Rainbow RST-135 because of its larger periodic error. Fortunately, the polar region methods that determine a center of rotation are unafected by periodic error and deliver a more accurate result in a fraction of the time. Like many mounts, the Paramount and Rainbow models have an accessory QHY PoleMaster adapter, or failing that, one can use a universal adapter that fts a tube ring or the scope’s dovetail plate. Polar Alignment Best Practices Te following is a summary of my polar alignment best practices designed to be pragmatic and meaningful at the same time: •

•

•

•

•

•

Be realistic about the level of alignment accuracy you need. In many situations, periodic error and seeing conditions will be more signifcant than drif. Polar align with the mount loaded with the imaging system. Tis takes into account any compliance in the mount, support, and ground. Short unguided exposures and short focal lengths are more forgiving of drif (and hence alignment accuracy) than long ones. Unguided tracking, with a low PE mount, benefts from good polar alignment and atmospheric refraction correction if you are imaging near the horizon. While some believe unguided operation (afer sky modeling) does not require accurate alignment, in practice, the tracking models on several premium mounts work best with a good initial alignment. Tere are several polar scope reticle designs; those with a simple centering bubble (e.g., older SkyWatcher EQ series) require an independent means of setting the hour angle. Tey are almost certainly incorrect for the current epoch and, ideally, should

fg.3 The results from a night of comparing polar alignment techniques on a Paramount MyT. The circles represent the range of reported errors after several attempts with each method. The StarAid results were made with their original drift technique, which has now been superseded by a polarregion method that is less sensitive to periodic error. The lower green dot is an alternative position suggested by TPoint for minimum feld rotation.

•

•

•

•

be replaced. More recent designs use a clock face or align on multiple stars, eliminating the need for fddly setting circles. Polar scope centering is conveniently done in daylight, aiming the RA axis at a convenient terrestrial target. Tis may be a TV aerial or a distant lamp post. Te three small grub-screw adjusters around the polar scope eyepiece are gradually loosened/ tightened in pairs until the target remains perfectly centered when the mount is rotated. Replacing the grub screws with longer cap-head bolts permits convenient hand adjustment. When centering the reticle, center it in one orientation, rotate it, and then remove half the centering error with the adjusters. When centering a polar scope that is on a swing arm (e.g., Fornax/AstroTrac TT320), it is best to secure the polar scope in a fxed and repeatable orientation and rotate the arm to confrm the centering. Tis has the beneft of canceling out the reticle centering error and any cone angle introduced by any tilt in the pivoting arm. An of-axis, polar scope shifs position as you rotate it, introducing a parallax error. (At 300-feet

Setting Up

distance, a 1-foot lateral displacement introduces a cone error of 11 arc minutes!) To remove this error, use a target on the horizon, or preferably a star, as the target point. Balancing Acts If a mount is balanced about its two axes, the drive mechanisms only have to overcome residual friction and inertia. During tracking, the RA worm gear always rotates in one direction, reducing the potential for any unwanted backlash efects. If the RA axis is not balanced, other forces may prevail unpredictably. On those mounts with a little backlash, one recommendation is to ensure the RA drive is “pushing” the entire time. Tis requires a deliberate imbalance on the telescope end or the counterweight end, depending on the telescope’s side of pier. Some imagers pause proceedings afer a meridian fip to make a small weight adjustment, but this is not feasible for remote operation. I have also seen Heath-Robinson contraptions with elasticated cords around the telescope mount to tension the RA drive mechanism in one direction to avoid backlash. Tese are ofen a workaround for those older mounts that use spur gears between the motor and the worm. Many of these models have been replaced with designs that use toothed drive belts, with less backlash and gear noise. Several third parties sell upgrade kits to replace the spur gears with a toothed belt, and a better quality worm, to good efect, for some popular mounts. Balancing Methods Balancing methods vary between mounts on account of their diferent designs. Tose without clutches are difcult to balance; if balancing is essential, usually rely on electronic methods. In these, the controller drives each axis in both directions and compares the motor currents to indicate balance. When the current in both directions is the same, the mount is balanced on that axis. Tose with clutches still vary in approach on account of the friction of the bearings. A signifcant imbalance will cause the mount to swing, and it is important to support the mount before releasing the clutch. Static friction will usually prevent any motion when the system is approaching balance, and it usually requires a gentle nudge by hand in each direction and observing if rotation in one direction is “faster” than the other. I usually lock the DEC axis and balance the RA axis frst. Inertially speaking, having a larger weight close to the mount is better than a smaller one at the end of the counterweight bar. I rotate the counterweight bar hor-

106

izontally and assess the balance in this (more sensitive) orientation. With the weights secured, I lock the RA axis in the horizontal position and carefully support and release the DEC axis. I carefully slide the dovetail bar fore-af for rough balance and then tighten. Afer focusing, which changes this balance, I repeat the procedure and mark the position on the dovetail bar for later use. Complications may arise when the imaging system is lopsided and requires balancing lef-to-right (as with a dual camera setup). Similarly, the mass of the camera system may require a telescope fore-af position that fouls the focus motor/flter wheel or ofaxis guider with the saddle plate. Workarounds include fipping the focus mechanism and adding spacers between the telescope tube rings and the dovetail plate. Similarly, of-axis cameras (for example, on Newtonian telescopes) may require rotation to move the camera into a balanced position or need a small counterweight on the opposing side (which may require additional RA balancing). However, there are exceptions to every rule, and if the worm drive has some play, perfect balancing may not always be the best thing for easy guiding. Cabling and Connections Balancing is assessed with the fully assembled system, without the lens caps but including the wiring. Poor cable management ofen causes both electrical and mechanical issues; for instance, I have seen a single trailing USB cable ruin the guiding performance of a premium 40-kg mount. It seems unlikely until one considers a sub-arc second tracking accuracy is less than 1/5,000th degree. Several strategies to manage wiring include wiring looms, interface boxes, and through-mount cable routing. Dealing with the last, frst, some mounts pass cables through the chassis or, in the case of the Paramounts and others, have a USB/ power hub directly on the saddle plate. Tis is usually USB 2.0 and sufces most needs. Modern CMOS cameras ofen have USB 3.0 interfaces, and an entirely USB 3.0 connection from camera to computer is noticeably faster with large image fles and most useful for speeding up autofocus runs. Additional through-mount connections usually include a shielded cable for the dew-heater power and a cable for the focuser motor. Cable looms serve several purposes; they are a permanent assembly of the right cables (for easy and repeatable assembly) and, when sleeved, form a bundle that is less likely to snag on the mount or tripod. Physically bundled together, there is also less opportunity

107

The Astrophotography Manual

mini USB connector. Finally, for quick assembly/disassembly, I use Velcro® cable ties and self-adhesive Velcro pads at strategic positions.

fg.4 The convenience of a telescope-mounted control system is hard to overestimate. There are no trailing cables, the cables are short, and the whole assembly lifts of the mount with just two connections, ready for the next night.

for ground/induction loops. It is sometimes necessary to disconnect/reset a particular USB device, and I mark any similarly-looking USB cables with colored electrical tape to distinguish them apart. Power connections can also be ambiguous and increasingly use the same 5.1-mm plug for anything between 5- and 48volt inputs. Terefore, it is important to clearly label or have a scheme (like wiring looms and unique connectors) that prevent accidental insertion into the wrong device (a.k.a. Poke Yoke). Tere are several ideas that avoid cable drag; starting at the top, support the loom close to the middle of the saddle plate for minimum drag and then form a loop with the loom and support again near the top of the tripod. Some other schemes attach the loom to the top of the counterweight bar, close to the mount. I also bundle cables from my interface box to the Paramount base, as the more robust power cable connectors provide mechanical support for the more fragile

Interface Boxes In prior years I conveniently housed all the electronic modules in a steel enclosure and used high-quality connectors for power and communications. I called this my master interface box and described its construction in earlier books. It still serves in the observatory, and I constructed a second one for portable setups in the backyard. Tese are nothing more than a metal DJ enclosure into which the various USB, power, dew-heater, serial, and focuser modules are placed. Te pre-punched XLR holes make it easy to use a variety of robust connectors for diferent purposes and prevent accidental connections. Te svelte Pegasus Powerbox has recently revolutionized my portable system and replaced the interface box. It lacks a 48-volt power line for the MyT, which is accommodated by a 12- to 48-volt 5A DC-DC converter module housed in a weatherproof enclosure. With the module and NUC (or RPi) PC on top of the telescope, cable lengths are shorter and easier to manage, with less need for complex cable looms. Used on the Rainbow mount, it only requires a single power cable to ground level, and on the Paramount, a single USB cable runs back through the mount into its USB connection and a short power cable to the saddle plate's 4-pin power connector (fg.4). When I teardown the system the following day, I leave the Pegasus/ NUC assembled to the telescope/camera, ready for the next session. A label on the NUC reminds which power and USB connections are for which devices, which corresponds to the labels in the Pegasus controller application, for robust power and USB control. Good Connections Connector and cable reliability are always on my mind and even gold-plated contacts are not immune to environmental contamination. On the permanent setup, I periodically pull out and insert each connector a few times to clean its contacts (with the power of). On that last point, connecting all the power and control/communication cables is important before turning on any power. If you do not, there is the potential for static discharge and circuit damage when you plug in a cable. Tis is because some power supplies are “foating” and have no ground reference. Te cables form the electrical connection to provide a common reference and must be in place frst. When it comes to the metalwork,

Setting Up

108

one might think that the aluminum construction would create the circuit, but the oxide formed naturally on bare metal or by an anodizing process is a perfect insulator (a fact exploited in semiconductor fabrication). Flattener/Reducer Spacing Many simple refractor and refector telescopes focus on a curve plane. While the human eye accommodates this during visual astronomy, a fat sensor is not as forgiving. Te solution is to place a weak negative lens (usually a doublet or a triplet) directly in front of the sensor. Tis extends the focus of the image margins further out. Te practical upshot is that stars are uniformly focused across the frame and, signifcantly, are circular. Without a fattener, the stars in the corners are elongated, radiating from the image center. Field fatteners and reducers share many characteristics, with the main diference being that a reducer lowers image magnifcation (reducing the efective focal length and improving the f/ratio). Tese optical devices are generic in so much that they will work acceptably well with telescopes of similar f/ratio and focal length. In practice, it usually requires some adjustment of the sensor-fattener distance to achieve the best results. As the spacing increases, the fattener efect increases and eventually over-compensates. Shorter focal lengths typically require a larger spacing; many fatteners and reducers assume a minimum spacing of 55 mm (the T-thread standard), which works well for longer focal lengths, and requires further spacers for shorter focal lengths. Determining the correct spacing may be as simple as following the product specifcations. Further improvement is ofen achieved by altering the spacing by a few millimeters. Evaluating its efectiveness or setting the best spacing from scratch is ofen a case of trial and error. One can examine the star shapes in the image corners to evaluate feld fatness or use a utility program (e.g., ASTAP or CCDInspector). Tese provide relative numerical and graphical feedback, but sometimes comparing diferent telescopes or making a qualitative assessment is tricky. Manual Evaluation To evaluate and adjust fattener spacing, it is best to focus carefully at the image center and take several short exposures (of up to 10 seconds) in a star-rich part of the sky. Tis short exposure minimizes star trailing from polar misalignment or periodic error. For mounts with a high periodic error, it is best to autoguide during the process. Once the image downloads, increase the screen magnifcation to 100 or 200% and examine the star shapes in each corner. If they are elongated along a radial axis, try increasing the spacing by a few millimeters, re-focus and re-evaluate. At the best setting, the star shapes at the image corners will become circular and, if the spacing is either too short (or long), will be elongated, as in fg.6. If the star shapes in each corner are quite diferent, this may indicate a tilt issue, focuser droop, poor collimation, or a combination. I initially set up the system to the specifcation, marking the sensor position with a Plimsoll line on the camera barrel and using calipers to check the spacing. Any system always has optical and mechanical tolerances that will afect performance. Comparing star shapes over a range of fattener spacings of ±2 mm about this recommended value is a sensible precau-

fg.5 OK, you may be wondering why there is an image of two Allen wrenches in the hardware chapter. Well, this is a salutary warning about messing things up. Equipment manufacturers use either metric or UNC/UNF threads and in some cases, a mixture of both. While some threads are very obviously diferent, some are similar enough to mix up, with the result of cross-threading and thread-stripping a soft aluminum housing. M6 and ¼inch UNC are favorites for mix-ups but it is possible to detect an incorrect selection as a bolt locks up after a few turns. The smaller sizes are more tricky. My advice is to hand tighten bolts and only use a tool for the fnal tension. The same issue is true of Allen cap head bolts. Without care, and especially on the smaller sizes, it is easy to use the wrong size and mess up the head (or wrench) with tricky consequences.

109

The Astrophotography Manual

fg.6 The appearance of these bottom-right corner stars give an indication of any spacing issues. The fattener spacing is good enough (left) and too small (right)

tion. It is ofen possible to detect the diference (especially in the case of a reducer) of a 0.5-mm change in spacing. Another way to see if the spacing needs to increase or decrease is by altering the focus position. Here, one carefully focuses on the central stars, takes an image, and then compares the corner stars before and afer shifing the focus position. If the stars improve with a slight outward movement, increase the fattener spacing and vice versa. Te William Optics’ range of refractors includes several fatteners and reducers, most of which work with several telescope models. Several of their latest updated designs have adjustable couplings within the fattener/ reducer body (fg.7), and the company website usefully documents recommended settings for each fattener/ refractor combination.

fg.8 The assembly of the QHY cameras, with their flter wheel and of-axis guider (OAG) requires a little forethought. The OAG pickup requires a minimum distance of 46 mm to the sensor and there are many spacers and adapters. Once established, I make drawings for the diferent systems to achieve the right spacings to the feld fattener and with the right thread coupling. It seems like a chore but it saves time in the end and ensures repeatability.

fg.7 This WO fattener’s built-in adjustable spacer is convenient and rigid, though it does rotate the camera..

For the Next Time Many activities are designed to “learn”, so determining the detailed equipment setup is only a one-time efort. Tis learning exercise is supported by good record-keeping. For example, while some camera systems have integrated flter wheels and of-axis guiders, my QHY system comprises many separate components and a comprehensive set of adapters. As each of my feld fatteners and reducers has a slightly diferent spacing and coupling, I make drawings for each confguration and label the spacers with their depth so that I can quickly recreate the same assembly conditions (fg.8). As we shall see in the next chapter, setup conditions continue into sofware setup, with even more parameters to set and remember to ensure the acquisition process goes smoothly.

Setting Up

110

Software Setup The devil is in the detail, and there are a lot of details.

E

qually important and just as unique, sofware setup is essential to make image acquisition work seamlessly and without any hitches. Tis is made all the more challenging as astronomy sofware has evolved even more rapidly than hardware in the last few years, increasing in complexity and sophistication. In this period the choice, depth, breadth, and platform support have mushroomed and continue to do so. Te market has changed considerably, too, with established, mature (and expensive) applications challenged by innovative, inexpensive (or free) alternatives. Te quality of these new titles is similar or better too. At the same time, afordable applications for tablets and smartphones ofer useful portable utilities. Tis explosion in sofware titles presents a fantastic opportunity and a bewildering dilemma. I hope the chapter on imaging hardware simplifed some aspects; here we have a high-level guide to installing, calibrating, and using astronomy sofware for automated image capture. As you can imagine, no two installations are the same but at the same time, common elements apply to most. Tere are several steps to setting up and making the imaging and system sofware run smoothly. Tese are: 1. 2. 3. 4. 5.

core operating system (and updates) application and drivers device setup application setup fne-tuning

Core Operating System (Windows) Windows and Linux operating systems have diferent approaches, starting with Windows. I quickly found a dedicated computer set up to run lean and mean hugely benefcial for imaging, in terms of reliability, speed, and efciency. Tis involves disabling powersapping themes, animations, notifcations, and social media (though it does make some of the screen-grabs look old-fashioned). I recommend a 64-bit version of Windows 10/11 that supports remote desktop. Windows 7 was stable but Microsof and some recent applications no longer support this OS. In the case of the NUC PCs, I make a fresh installation of Windows and then remove all the superfuous applications to reduce

power consumption and Internet usage. I also create a power profle in advanced power management, which prevents the computer and peripherals from going into standby or low-power sleep modes. Tis includes ensuring the screen, cooling strategy, and maximum processor usage are scaled back as far as possible. For example, for a 2.5 GHz Core I5 processor, I set 90% max processor usage for image capture. On a laptop, I recommend similar advanced power settings for both battery and mains operation, as a computer may assume it is on mains if connected to external batteries. While it is an indulgence to have a separate PC, malware and virus attacks are less likely if you do not use it for email or browsing. In my home system, I connect through a hardware frewall and disable the Windows frewall and power-robbing drive scanning/encryption. My sofware, utilities, and drivers are downloaded via a Mac and stored on a USB drive and an application archive. I keep the archive up to date with the last two versions of any particular driver or application, just in case the latest version introduces a bug. I install these applications from a memory stick or for those applications that self-update through their websites. I store image data on a removable drive and copy it to a large-capacity external drive. When everything works smoothly, I create a complete system backup to an external caddy. Application and Drivers Tere is no universal installation sequence that is without issue. I generally start with hardware support and conclude with the highest-level applications. I fnd the following sequence installs with few hiccups, though it takes several hours to complete, most of which were consumed by innumerable Windows updates. Te following sequence concludes with a system backup: • • • • •

hardware drivers (e.g., PC system, cameras, flter wheels, focusers, USB-serial converters) ASCOM platform (from ascom-standards.org) ASCOM device drivers (from ascom-standards.org or the manufacturer) image capture applications autoguider application

111

• • • • •

The Astrophotography Manual

utilities (e.g., focusing, plate-solving, polar alignment, modeling) planetarium, catalogs planning and automation applications image processing applications (if required) time sync utility

plate-solving, in the PinPoint application folder (and set the path for each in the application’s preferences). Some planetariums have an associated catalog compiler that converts the otherwise diferent formats into a compatible version. C2A and others have extensive catalog management support, as does TeSkyX. ASTAP has replaced PinPoint on all my Linux and Windows systems. It additionally has two star catalogs on its download page, based on more recent Gaia astrometric data, and for diferent limiting magnitudes. NINA and TeSkyX also have available visual catalogs to display representations of deep-sky objects, which are helpful during image planning and framing.

Many installers are distributed as zip fles and require unpacking before running. Some applications and occasionally the installation programs themselves need to be run as an administrator or using an administrator account. Using administrator privileges, in general, may cause problems. For instance, some device drivers will only permit multiple ASCOM connections if the administrator privileges are the same for Device Setup each application. I previously selected “run as adminis- Te initial setups follow a ground-up logic, starting trator” as the default, but I now do the opposite. To add with the network, general communications, and virto the confusion, TeSkyX and the AAG Cloud- tual COM ports before moving on to the specialist Watcher must be run once as an administrator, to com- hardware. Device communication comes in several plete the installation process. forms, including serial, USB, and Ethernet, usually Other applications require additional sofware, in- wired but increasingly wireless. Most devices, other cluding the Windows .Net frameworks and Visual Ba- than cameras, do not need fast or large data support, sic, although the installation programs normally link and use short serial strings for communication. In to these downloads automatically. Other utility pro- Windows, this is done through a virtual COM port, grams, such as Adobe Acrobat Reader and Apple which allows serial communication through alternaQuickTime, are free downloads from the Internet. At tive interfaces, such as USB. Here, the application the time of writing, astronomy applications are evenly thinks it is sending/receiving serial strings, but in divided between 32-bit and 64-bit, with some having fact, it is physically using USB or WiFi. Some USBdual versions. While a 64-bit operating system does to-serial adapters include virtual COM port drivers, not require 64-bit applications, a 64-bit application re- or you can use a free utility like HW VSP3 from quires 64-bit device drivers, which are not universally www.hw-group.com. available. I run 64-bit versions of NINA, SGP, and TeSkyX, as I am fortunate to have full 64-bit support COM Port Assignments for my particular hardware. One issue with setting up communications is the inSeveral applications have a fnite number of activa- creasingly present Windows security settings. For extions and can catch you out if you do not manage them appropriately; for instance, deactivating before an OS upgrade or switching between computers. Installing catalogs may also be confusing. For imaging work, catalogs are used principally for plate-solving (astrometry) and target planning. PinPoint uses the venerable General Star Catalog (GSC) for plate-solving and several fg.1 For Windows, applications connect to devices directly, using the devices application interface (API) or through an ASCOM driver, which standardizes the interface. For planetariums for displaying stars. those devices that need to connect to multiple applications, they need an ASCOM Tese use the same astrometry data hub, or a device driver that has a built-in hub. In an increasing number of cases, the but require diferent formats. I put host computer is controlled remotely from a client PC or tablet via an Ethernet or WiFi one version in a “Catalog” folder in connection. The Linux equivalent is not too dissimilar. My Documents and the other, for

Setting Up

ample, if you have enabled a sofware frewall, it may be necessary to grant safe passage for your ports in the frewall settings. COM port settings ofen catch out the unwary and there is a one-time-only technique for defning the hardware settings. In Windows, powering up the system with connected serial hardware assigns COM port numbers to each serial device. An assignment links the device and the USB port, and plugging it into a diferent USB port ofen causes a duplicate assignment. Te applications and ASCOM drivers store the initial COM port selection for each device and fail to connect if a COM port is re-assigned. ASCOM Confgurations Tere are a few approaches to ensure consistent port connections. One is to use a fxed hardware confguration (made easier with a labeled interface box), and the other is to manually set up the COM port assignments. I use both to be sure; I fx the COM port assignments in Window’s Device Manager/Serial/Com Port/Advanced menu. Te trick is to plug in each de-

112

vice in turn, manually set its COM port, note it down, and repeat for the next device. In that way, these devices will always have the same COM port assignment, regardless of the order or the socket used, and the ASCOM device driver properties will not require random updating. With the COM ports fresh in your mind, setting them up in their ASCOM device driver properties is a good idea. Tis is a confusing subject for many as it is possible to daisy-chain programs through others to a fnal piece of hardware. Fig.2 shows an example of the ASCOM connectivity between devices and applications. Tose ASCOM drivers that accept multiple connections are called hubs. Many modern mount ASCOM drivers are also hubs, permitting several applications to monitor and control the mount. Te standard ASCOM platform also includes an observing condition class and hub, which combines multiple monitoring devices and shares, e.g., mount, dome, and focuser controls. Tese confgurations are a one-time-only setup but one needs to take care; the daisies in the chain sometimes have to be in a specifc order. For ex-

fg.2 A fully-automated image acquisition system, capable of unattended imaging, requires planning. The above are the majority of settings required to ensure the correct operation of the acquisition, guiding and focusing applications. Some are device specifc (e.g., camera settings); others are associated with the system’s operational logic.

113

The Astrophotography Manual

ample, to reap the beneft of accurate pointing for all connected programs, MaxPoint should connect directly to the mount driver and not be further up the daisy chain. Tese program links also trigger multiple programs to start up and run when you connect, say, Maxim DL to a mount. Some of these chains cause time-outs (fault tolerance is not astrophotography’s strong suit) and then require a Windows Task Manager intervention or a cold reboot to fx. One of the limitations of an ASCOM driver is there can only be one instance, which prevents multiple connections to identical devices. To address this, several focuser and camera manufacturers create a duplicate driver for a second device with a diferent ID. Optec Inc. released an all-purpose hub called the Optec ASCOMserver which additionally allows two connections to like devices. Unlike some of the original ASCOM platform utilities, this hub is transparent to all commands and, therefore, can serve specialist equipment and dual rigs. Application Settings Afer all the hardware is physically connected and checked out in Windows Device Manager, the next task is to initialize all the application settings. Tis is a lengthy task and it is helpful to list the common parameters for handy reference. I have captured many of the common ones in fg.2. It is a case of patiently going through each application in turn and flling in the relevant information in each dialog box. Tere will be a few instances where programs read data from other sources, but having a crib sheet is helpful. Te good news is that most modern applications have a feature to save general settings and equipment information. In the case of several similar imaging systems, one can usually create a confguration, duplicate it three times and then modify the few values related to the feld of view, focal length, camera, and angular resolution. A few require experimentation, and it is good practice to update afer a few sessions; for instance, guiding and focusing parameters. Tere are many settings to consider; some, in particular, are responsible for causing the most havoc. Tese baddies are the ones to watch for: • • • • • •

USB trafc setting on the camera driver Sleep settings on USB ports COM port assignments and settings incorrect pixel scale setting for plate-solving plate-solve exposures too brief settle/centering error threshold is too small

• • • • • • • • • • •

meridian limits not set or incompatible with mount limit settings diferent mount and PC time/location setting conficting pointing model with plate-solve synchronization incorrect latitude/longitude settings incorrect/difering epochs for mount and capture application daylight saving time/time zone setting autofocus step size/count too big/small no backlash compensation on the focuser collimation issues with a refector telescope DSLR mirror lock-up feature incompatible with image capture settings camera USB trafc control setting (to avoid conficts and reduce image anomalies).

One area that is sometimes problematic is autofocus. I set a medium gain, 2x2 binning, and a 3-second exposure in NINA/SGP. Both apps evaluate multiple star diameters at several diferent focus points. Other applications dynamically change the exposure according to a single star’s intensity. In both cases, set the step size and count so the half fux diameter “V” curve is over a 2:1 to 3:1 diameter range. Scripting and Templates Apps such as Voyager and NINA ofer facilities to store complex instruction sequences and recall them later on for reuse. For example, I update the stored template version as I refne my NINA imaging sequences to accommodate safety/recovery management, dual camera synchronization, and conditional imaging. Tis captured experience saves time and helps imaging sessions become more consistent. Linux Setups Te equivalent of ASCOM on the Linux OS is INDI (and INDIGO). It consists of several distinct program types: clients, servers, and device drivers. A client is typically a high-level imaging application or planetarium. It communicates through an INDI server, the central hub for communication between clients and device drivers, and permits local or remote operation. Linux setups are a mixed bag. It is possible to download a complete Linux OS, packaged with all the likely device drivers and applications as a single distribution fle or “distro”, save it onto an SD card, plug it into a Raspberry Pi, and turn it on. On the other hand, the Astroberry and StellarMate distros include all the INDI content, including the device drivers for the various device types and high-level applications.

Setting Up

Tese two distros are for the Raspberry Pi, but installing the INDI drivers, server, and clients onto any standard mainstream Linux system is possible. In the case of my existing Intel NUCs, used M2 SSDs are inexpensive, and rather than implement a dual boot, I swap drives to use an alternative operating system. However, I have to install the OS and applications separately in this case. Te Astroberry and StellarMate distros for the Raspberry Pi include KStars and EKOS as the imaging client applications. Te distinction between them is that StellarMate supports client OS versions of KStars/ EKOS, which connect to the devices using the INDI server protocol. Te Linux version of these applications will also run onto an Intel hardware platform. Other popular INDI clients are PHD2, Cartes du Ciel, CCDciel, TeSkyX, ASTAP, HNSky, and Stellarium. To set up a device, you frst select the driver in the INDI server utility, start the server and then connect to it from your client app(s). For example, to start the INDI server to run the Rainbow telescope on localhost and listening on default port 7624, type the following in Terminal: $ indiserver indi_rainbow_telescope Alternatively, one could run the graphical application INDIstarter to bring up the INDI server control panel and click “Add driver”, and select the Rainbow mount. Afer setting up all your other devices, you then press start and wait for the server indicator to turn green. You can then click on the INDI client and select the device tab and adjust the connection details and any options. For convenience, you can also save a set of devices as a server profle and, in the client application, with the server running, you connect to it and populate all the devices. Tere are many favors of Linux and even more distributions. Some are very pared-down for a job, and others have extensive graphical interfaces (like Windows) for more general use. Many use Debian/ Ubuntu. I use Linux Mint Mate (64-bit) OS. I fash an OS image onto a memory stick using a utility like balenaEtcher and then instruct the NUC to load it (afer pressing F10 at boot). You will need to install the INDI library, INDI server, and preferably a utility to set up the server (for example, INDIstarter). Ten, the applications follow, with the guider, planetarium, plate-solver, image capture applications, etc. Loading applications requires some accurate typing in the form:

114

sudo apt-add-repository ppa:folder/fle sudo apt-get update sudo apt-get install application Fortunately, the application providers usually spell out the exact install command lines and it is best to copy-paste into the terminal window. For example, I initially tried KStars/EKOS (the basis of Astroberry and StellarMate) as my capture system and then discovered CCDciel/Cartes du Ciel. Tese are easier to use, equally capable, and capable of complex sequences. Te Linux hands-on approach is not to everyone’s liking. Increasingly, more user-friendly applications package these instructions into an install script. In addition, many common devices now have Linux drivers, which are added to the INDI library and refreshed with a sudo apt-get update command in a terminal window. Te Linux device hardware interfaces are unchanged and in the case of virtual serial ports, Linux is less sophisticated than Windows. It helps to keep a fxed USB connection confguration and use any available utilities to set up the ports. You will likely be using a remote connection and may need to install and enable a (free) remote protocol. I use xrdp, which works with the same Microsof Remote Desktop application I use to control my Windows PCs. Alternatives include x11vnc, TeamViewer, NoMachine, and noVNC. Te Internet is the best resource for the most recent installation information. Te image capture application apps in Linux (several of which are cross-platform) have the same requirements for setting up and similarly store equipment confgurations for easy switching. Fine-Tuning It usually takes several sessions to fgure everything out, establish some of the remaining settings and iron out the operational wrinkles. Tese include: • • • • • • •

camera – gain, exposure, and ofset settings autoguider – calibration and initial settings centering – plate-solver setup, pixel scale, error thresholds, pointing and tracking sky models mount limits – tracking limits, meridian fip points, and imaging horizon focuser – initial focus position, flter ofsets, stepsize, and AF range feld fatteners – collimation, tilt, and spacing image fles – folders, FITS header, and fle naming

A few of these activities are worthy of more discussion.

115

The Astrophotography Manual

fg.3 The various meridian fip settings within the acquisition software and the mount setup can be confusing. To avoid collisions, one must not fip too early or too late. Things become clearer if you lay it out, similar to the diagram above.

Camera Ofset and Gain (CMOS) As part of the initial sofware setup, settling on a few gain and ofset settings for each of your cameras is useful. I typically choose zero and a medium gain setting (which is a good compromise between read noise and well depth). For these, I take several 10- and 300-second exposures and use the capture app’s statistics to fnd the minimum pixel value. Next, I set the camera ofset so that the minimum pixel value is about 100 units and varies for each exposure (i.e., it is not a stuck pixel or a black pixel with an uplif). Once established, I save it as a preset in the ASCOM driver and record it for future reference. I then repeat for the guide camera and establish the ofset for a medium-high gain setting. Centering Plate-solving is the must-have utility to efect accurate centering. Shareware apps like ASTAP, PlateSolve2, and astrometry.net work well. Most work faster if they have an idea of the approximate location and pixel scale. If not, a blind solver, like those available online at astrometry.net, work it out for themselves. Te GSC star catalog was the standard go-to catalog for astrometry but more detailed ones from USNO, UCAC and Gaia, with dimmer magnitude limits, are increasingly used for narrow felds of view. I use 2x2 binned exposures of about 2–5 seconds for speedy and reliable operation. Tis reduces the efect of hot pixels and almost guarantees the detection of sufcient stars. Plate-solving applications exist in their own right but are mostly called from image capture applications for target centering. Some key setups change hot-pixel removal, error, and timeout thresholds and specify a backup “all-sky” routine for blind solves. Te astrometry.net site is one such resource for blind-solving and feld use; it is also possible to download the catalog fles beforehand and run the blind solver on a local server from: (https://adgsofware.com/ansvr/)

Mount Limits and Meridian Flips Automatic meridian fips are a liberating thing that permits unattended all-night imaging. But, unfortunately, they are also a source of things going wrong, and correctly setting up the system makes all the diference. First, set the mount slew/tracking limits so there is no chance of crunching a camera or flter wheel into the tripod or pier. Te worse-case slew limits are best set by the mount and using the handset controls. Mount limits are specifed in degrees or time past the meridian and stored in the handset, mount, or its parent application. Tey are likely to occur just past the meridian line. Extending this beyond 16 minutes (4 degrees) in each direction has little advantage. Assuming we set this as our limit, the following setting, the meridian fip point, is specifed in the capture sofware. Tis is the point at which the program fips the mount, which should occur before the telescope reaches its limits. A possible setting might be 12 minutes (3 degrees). Tere are some further options to prevent the mount from fipping too early and wait until the fip point. Tese manage the imaging sequence during this tricky period. A good capture program checks for sufcient time to complete the next exposure before the fip point. If not, it should wait for the telescope to pass the meridian and reach the fip point. In the ongoing example, if a 10-minute exposure concludes just as the telescope passes the meridian, it will complete the following exposure with a few minutes to spare. It might waste a little time, but I suggest instructing the application to wait for the fip point rather than fipping immediately and potentially causing a collision on the opposite side of the pier. Another recommendation is to avoid fipping or setting the limits precisely on the meridian; any timing or position ambiguity between applications is sufcient to prevent fipping. Te better mounts place the object in the frame again and rotate 180° afer a meridian fip (unless one has a rotator). Tere will usually be some error arising from fexure and alignment errors. Te automatic centering methods now common place in image capture applications quickly resolve any initial alignment error. Once the telescope has passed the meridian, they generally follow the following automatic process: 1. 2. 3. 4.

establish the last image center (plate-solve) or use the original target coordinates instruct the mount to slew to that position or fip wait for the slew to complete take a short exposure and plate-solve it

Setting Up

5. 6.

sync the mount to the plate-solved position or establish an ofset instruct the mount to slew to the coordinates or by a new ofset

Te image should now be precisely centered as before, only rotated 180°. Te mount and the RA autoguider polarity fip, and it may be necessary to enable the RA invert polarity option in the autoguider sofware. (Some mounts sense which side of the pier they are on and do this internally.) To fnd out what works, choose an object in the south just about to pass over the meridian, calibrate the guider system and run the autoguider. Once the mount fips over (either automatically or manually), stop the guider, select a new guide star, and start guiding again. If the guide corrections have the wrong polarity, their error trace in the autoguider graph will rapidly disappear of the graph. Autoguiding Tose systems with good tracking and alignment ofen do not require guiding with less demanding optical systems. For example, some premium mounts use an internal control system and accurately track using both RA and DEC motors in conjunction with a sky model based on multiple star alignments. Te rest require an autoguiding system. However, even a perfect mount with perfect alignment will exhibit tracking issues as the object’s altitude moves closer to the horizon due to atmospheric refraction. At 45°, the efect is sufcient to cause a 5 arc second drif over a 10-minute exposure. A sky model is designed to remove this efect, assuming average environmental conditions (temperature, pressure, and humidity) or, more accurately, from sensor measurements. Te sofware setup for guiding is ofen complex though some programs, notably PHD2, do their best to keep things simple. It has to accommodate mechanical, optical, dynamic, and atmospheric factors, and that is before it tries to work out which way to move! For well-behaved mounts, a few button presses are all that is required. When that does not produce the desired result, more analysis is required. Although some mechanical aspects, for example, balance and alignment, have been already covered, this might not be sufcient, and, for that reason, autoguiding, model building, and tracking have dedicated chapters later on that explore the many factors that adversely afect guiding performance and the potential remedies.

116

Image File Formats Tere are various camera fle formats, the most common of which is JPEG. Tis is the standard output for consumer cameras and mobile phones and is a compressed, lossy 8-bit format. Most astrophotographs require extensive image manipulation, and 8-bit fles are too coarse as a quality source fle but great for sharing a completed image on the Internet. Te highquality option on any camera is now the RAW fle, an “unprocessed” fle. Tese are normally 12-, 14-, and increasingly 16-bit images stored in a 16-bit format. TIFF fles are less common as a direct camera format but usefully stored in 8-, 16-, 32- and 64-bit lossless formats. Dedicated astro cameras output uncompressed fles. Dedicated image capture programs save images in a 16-bit FITS format. Te “Flexible Image Transport System” is an open format extensively used in the scientifc community and works up to 64-bitdepth. Just as with the other image fle formats, the fle contains more than just the image. It has a header that contains useful information about the image and the acquisition details. In astronomy, this includes things like place, time, exposure, equipment, sensor temperature, and celestial coordinates. A typical imaging night can capture hundreds of fles. Te image calibration process uses the image’s FITS header to automatically sort and order the fles into separate objects, equipment, and flter selection. During the sofware setup, it is a good idea to fnd the part of your image capture program that defnes custom felds in the FITS header and check it is adding the information required to group those exposures that share a common calibration. Some image calibration apps do not use the FITS header but expect like-exposures in separate folders. PixInsight defaults to saving calibrated and edited fles in XISF format. Tis specialized scientifc format has its roots in the FITS format, and has benefts within PixInsight, but has not yet been widely adopted by other applications. It is tempting to tinker and update programs once you have everything working. However, unless an update has a specifc beneft, it is sometimes better to resist the temptation. As a rule, I keep all my applications up to date. Still, with the complex interaction of hardware and sofware, history has taught me to avoid early adoption of signifcant app upgrades.

117

The Astrophotography Manual

Automation and Remote Control Hands-on is great, until it requires continual monitoring in more challenging or remote environments.

B

eyond the most simple hands-on methods for image capture, astrophotographers increasingly use varying degrees of automation and remote control to make the most of favorable conditions, all in the quest for more exposure and a good night’s sleep. Both have diferent levels of sophistication, ranging from the simplest exposure sequence to managed multiple target acquisition at a remote observatory. Automation and remote control go well together, with obvious synergies of convenience and interdependency. A common combination uses automation to do the capture work and remote control for monitoring progress and intervention, as required.

Automation Once things are stable, even basic automation (and control from somewhere warm and mosquito-free) make long-duration sequences a practical and sociable reality. Apart from convenience, automation can potentially make acquisition processes more reliable through consistency. However, things need to work in the frst instance. It is important to be familiar with and ensure the individual methods work reliably and without constant attention (e.g., focusing, flter changes, centering, guiding, etc.). If not, automation makes any problems and user interventions occur more frequently! Automation opportunities arise in several distinct arenas: • • •

target planning (single and multiple targets) capture sequencing and observatory control remote control (local and far-fung)

Target Planning Target planning determines what to image and when. At its simplest, it will consider the visibility of a single target and its optimum season and imaging times. More sophisticated examples manage multiple targets (that rise and set during the night) to ensure we use every clear hour, or for creating an image mosaic. Te most complex use a master database to track exposure plan completion for a set of targets. Te extreme case briefy images many targets in any session to search for supernova or minor orbiting elements.

For deep-sky imaging, the long exposure runs lessen the need for more than a few live targets, and it is more of an assist than true automation. Some target planning utilities list objects from a catalog based on type, visibility, and so on, and export their details into a text fle format. As we have discussed, the object’s center coordinates are a useful starting point but not necessarily the best for composition; the composition tools within SGP and NINA usefully show the object’s altitude throughout the night, darkness, and Moon details. Tey both have mosaic planning tools that overlay the camera feld of view over a downloaded sky survey. Tese either show a tiled overlay over the target area for a mosaic or a single frame and store the center coordinates and rotation angles in the imaging sequence. Once the coordinates are known, subsequent tools permit imaging schedules based on object altitude on a given night. Details vary between the applications; NINA has an instruction that waits for a particular time, altitude, or elapsed time and SGP has a planning tool that permits a manual selection of the imaging span. Both accept an external target list that will auto-populate an imaging sequence. Capture Sequencing With the target(s) defned, image capture takes over. Sofware sequences are a combination of several setpiece modules, such as focusing and centering, and a sequencer that directs a number of exposure events of a certain duration, count, binning, and flter. Tese basics are common to many capture applications, though they may not be immediately obvious. Sequencing is the backbone of automation, and with the changes in ASCOM 6.5, camera gain and ofset are added to the list of selectable options. From here, to meet the requirements of extended sessions in real conditions, sequencing diversifes to include other activities. Tese include target scheduling (as above), target acquisition, centering, triggered autofocusing, dithering, autoguiding, meridian fips, and monitoring. In a few button clicks, what started life as a basic set of exposures is now an interaction of multiple device controls following a prescribed logic. A typical imaging sequence follows a familiar pattern, shown graphically in

Setting Up

fg.1. Te implementation varies considerably on the image capture application. In the case of SGP, the logic is a combination of hard-wired and user-set options distributed across various dialog panels (fg.2). In stark contrast, fg.3 shows the initial part of an imaging sequence in NINA’s advanced sequencer. Here, the control logic is laid bare; full user customization comes at the price of a visually more complex sequencing screen. To manage this complexity, it is a good idea to create logical groups of instructions and controls, and then save them as templates. For a new sequence, these are recalled and collapsed (to save screen real-estate), and, once partnered with an equipment profle, usually need editing for the new target and exposure plan. Unattended operation in many climates requires external monitoring of weather conditions and taking appropriate action. If the guider system has difculty fnding a star or settling, SGP enters a recovery mode. Tis attempts to re-center and restart the guider and image acquisition several times for a specifed period. Similarly, if the autoguider reports a signifcant tracking error during an exposure, the current exposure is abandoned and restarted. SGP has several logic options, so, at the end of a sequence, or if the ASCOM safety monitor indicates poor weather and, depending

fg.1 A schematic of a typical automated imaging sequence for unattended operation.

fg.2 The sequencer and control logic in SGP is a combination of hard-wired code and various options, distributed across several control dialog panels. These take a little time to become familiar with but are mostly one-time settings and, at the same time, they make the main exposure sequencer screen much simpler to understand.

118

119

The Astrophotography Manual

fg.3 NINA uses a diferent approach to managing sequencer logic. In their radical advanced sequencer, the user determines the logic by dragging operators from the right-hand “Instructions” palette. These include device instructions (e.g,. Take Exposure), conditional instructions called triggers (e.g., AF After Filter Change), and loop conditions (e.g. Loop Until Altitude Sets Below). There are multiple instructions for most devices, which can be arranged to operate in sequence, or in parallel (e.g., cool camera and autofocus). To manage the complexity, I organize mine into four groups which, once debugged, are saved as templates for redeployment: start up observatory, calibrate guider, image acquisition, and shut down observatory. The screen-grab above shows the frst two. Templates can be nested, which permits dividing complex tasks into more manageable units. For example, my image acquisition template is made up of four templates: wait for target, center target, RGB exposure sequence, and NB exposure sequence. Each contains multiple instructions and their own triggers and loop conditions. In practice, it is easy to rearrange, delete or duplicate these with a few mouse clicks.

Setting Up

on the selected options, it parks the mount, closes the observatory roof, and warms the camera to ambient before disconnecting the hardware. NINA does similar, but the logic is determined by the user, which is excellent for custom setups, but not without the risk of making a mistake. SGP has a simple shut down/recovery protocol, and NINA has the tools to do the same, with more sophistication. Several imaging applications provide direct access to some of their subroutines or methods, which permits additional customization using short programs called scripts or plugins. Voyager, another imaging application, goes one step further and relies more heavily on user-supplied scripts for its operation. Tere is something for all abilities and preferences. Observatory Control Te established high-level applications TeSkyX and Maxim DL publish an application interface (API) that permits external control from a high-level control application. In the past, this would likely have been CCDCommander, ACP, or CCDAutopilot. In some professional observatories, these are still the setup of choice. In the last 5 years, however, imaging applications, especially NINA and Voyager, include most imaging and small observatory control features that fulfll the needs of amateur imagers. In addition, their graphical scripting engines permit endless customization and there is less need for external scripts or applications. Best Practices Te trick with all these systems is to reuse things that work. When the logic is built-in mainly, it is ofen only necessary to store and recall a few equipment settings. For NINA users, once the logic of the assembled triggers, loops, and instructions is proven, it can be stored as a template or sequence. A stored sequence remembers the equipment setup, target, and imaging progress and is simply rerun (like SGP). (Templates store the sequencer logic and require partnering with an equipment profle and target to run.)

Remote Control It takes a certain type of person to do extended imaging and the occasional image rewards patience and perseverance. Even so, the novelty of fumbling around in the cold and dark quickly recedes, and the allure of operating one’s system from indoors grows stronger. Remote control of astronomical equipment has similarly advanced and diversifed as every other aspect of the hobby. Ten years ago, a remote system

120

could be controlled from a client machine running the same operating system. Today, those restrictions are no more, with apps and protocols which allow one to mix and match at leisure. So, what is remote operation anyway? At the simplest level, it is a convenience that allows the user to operate their backyard system indoors, using wired or wireless connections. At its most advanced, it distributes computing resources between far-fung sites and permits full control of a remote observatory over the Internet. Basic Systems First attempts invariably consider USB extenders that move the imaging computer into more convivial surroundings and leave the USB hub at the telescope. Implementations vary, but the forums indicate many are unreliable. My best results were using a USB extender that comprised two modules; a powered USB 2.0 hub and a transmitter plugged into my laptop’s USB socket. A Cat 5 Ethernet cable linked the two and could operate up to 100 m apart. Tis was reliable for home use with all but high-speed cameras that required full USB 2.0 bandwidth or USB 3.0 (fg.4). Tis confguration uses one computer that it powered for the duration. Te Cat 5 cable was unruly and a potential public trip hazard, and I upgraded to a wireless system between a small host computer located at the mount and a remote client computer. Remote Desktop Systems Tis is the next evolutionary step afer a USB extender. Here, the image capture computer is located at the mount and is controlled remotely by a second computer (PC, Mac, or iPad). Tis is not as extravagant as it sounds and works well; leaving a laptop out in the open is not ideal; they are too fragile, have limited battery life, and require a table and chair for ergonomics. A better alternative is to use a small, lowpower stick or brick PC loaded with Windows (or Linux) and the essential image capture applications. Tese sit close to (or on) the mount and conveniently powered from a 5- or 12-volt DC supply. A core i5 Intel NUC easily copes and stores images on a USB 3.0 memory stick. Tese units are run “headless”; that is, they do not have a connected monitor or keyboard and are entirely operated remotely, typically connecting to a WiFi router and to a client computer on the same network. Te client computer does not necessarily have to be running the same OS and maybe Windows, Linux, OSX, iOS, or Chrome. As far as the user is concerned, the client machine is a remote

121

The Astrophotography Manual

monitor, keyboard, and mouse for the imaging computer. Tere are several confgurations and protocols.

fg.4 USB 2.0 (and now 3.0) over Ethernet cable is a robust method of moving the imaging computer away from the mount. These premium units are in metal enclosures and have a powered 4-way hub. Some later models also transmit through existing LAN networks or use fber, but their cost exceeds that of a modest NUC computer.

fg.5 A small heavy-duty industrial-quality USB hub is essential for remote use. This USB 3.0 model usefully accepts a 12-volt DC power supply and has USB-C and A ESD-protected inputs.

fg.6 Some Raspberry Pi boards and computing sticks have weak WiFi. A travel router is an ideal solution. Set as an access point, it is great for Windows ad-hoc networking in areas without an Internet service.

Remote Protocols Tere are several methods to connect two computers, each with pros and cons. Te three main types are remote desktops (RD), Virtual Network Computing (VNC), and Virtual Private Networks (VPN). Of the three, VPN is probably the least suitable (on its own). It is designed for highly secure Internet connections between a commercial intranet (computers and resources) and a remote client computer on the public Internet, for instance, to permit working from home, rather than in the ofce. When used in combination with a remote desktop protocol, it ofers a broader capability with a remote observatory. A VNC ofers a remote desktop experience, allowing client and host computers to interact. It shares graphical data over the connection, which makes it easier for cross-platform use but less efcient than a remote desktop protocol, which shares resources. For example, the Linux RPi Astroberry and StellarMate distributions use noVNC as their default remote access client. Other popular VNC implementations include RealVNC and TightVNC. Te most common remote desktop protocol is Microsof’s MRD. It is supported by Windows Professional, Enterprise, and Ultimate operating systems. Tere are free client apps for all major operating systems, which usefully support multiple monitor confgurations. It is not the only remote desktop system; RemotePC, NoMachine, AnyDesk, TeamViewer, and Chrome Remote Desktop ofer similar products, usually with a free version for non-commercial use. Tese operate over a wireless or wired Ethernet connection and permit the imaging computer to be entirely controlled by a remote tablet, Mac, or PC connected to the same router and running the client application. It helps to set up a static IP address for your observatory computer and then it just needs that address, your username, and password to log in. Minor features vary between the implementations, including fle sharing, copy/paste, and security. Te latest operating systems provide access to the many OS menus, allowing remote shut down and restart. Some Windows confguration settings cannot be changed remotely for security reasons, and require the user to directly attach a keyboard, mouse, and monitor to the host computer to set it up, as they do to alter the BIOS settings at boot-up. One of the advantages of Microsof Remote Desktop is its low data requirement. It is entirely possible to link an iPad running MRD to an ad-hoc WiFi network created on a mobile phone and use the phone’s 3G cellular network for the Internet data connection to the imaging system computer. Alternative Confgurations Depending on the host and client resources and their relative locations, there are several physical remote confgurations. Te choice has expanded with the introduction of new operating systems, astronomy protocols, low-cost computing units, and the increasing use of WiFi connections between devices.

Setting Up

122

Tese include: • • •

locally or really remote distributed systems isolated networks

Locally Remote Tis is the most common confguration where the client computer is efectively used as a terminal on the same local network to monitor the imaging computer. It is the simplest to set up, and due to the proximity of the imaging system, it is easy to fx hardware issues, reconfgure the system and make darks and fats with ease. With this confguration, one can log in and out of the host computer at will and use the host’s automation features to automatically take care of sequencing and start up/shut down procedures and react to changing weather conditions. To set up: 1.

2.

3.

Enable PC Microsof’s remote desktop access, or install a third-party utility on the imaging computer. Note down the PC’s MAC address for its WiFi (or LAN) and set this up in your home router to have a fxed IP address, e.g., 192.168.0.12. On the client, install Microsof’s remote desktop application (on a PC, Mac, or tablet).

Te IP address identifes the imaging computer (it helps to give it a friendly name). Tis and your normal username and password are all that is required to connect. I share a local disk to allow fle transfer and set up two side-by-side remote desktops. If the client computer is a laptop, I connect a monitor to permit dualmonitor operation of one or two systems. With the sparsity of clear nights in the UK, I initially ran a second system in the backyard in addition to the one in the observatory. Similar FOVs allow simultaneous imaging of a common target, or the systems and targets may be entirely distinct. With dimmer targets, I struggled to acquire sufcient exposure. My recent preference is to image with dual imaging systems on one mount, using dual instances of NINA running on the same NUC computer. In this way, the system automatically protects from adverse weather conditions.

Really Remote If you know a router’s IP address and allow access through its frewall, it just needs a suitable Internet connection to operate a remote system. Tis has encouraged the popularity of geographically remote sites.

fg.7 The lessons learned in a local observatory are essential for trouble-free operation at more remote locations. These include roof and mount position sensors, weather sensing, protected electrics/electronics and a dehumidifer unit (which some mount manufacturers insist on for warranty purposes).

Imaging from a dark feld site, with many clear nights, and lower humidity, is pure joy. Tese locations are usually less obstructed and, if at a latitude between ±40°, will deliver all-year-round imaging and access to targets inaccessible from your backyard. Remote control has many advantages but also has its share of challenges. My friend, Peter Carson, remotely controls his observatory in Spain from his home in the UK and generously shared his experiences. As he so aptly observes, when working remotely, little problems become big problems; things that are easy to fx at home now require a plane journey or relying on others on-site, who may speak a diferent language! Planning a really remote site should consider: • • • •

location observatory style setting up running

Location Choosing the right location is key. Tere are many things to weigh up, including the reliability of the power supply, Internet, and accessibility. Sites are ofen shared with other like-minded astronomers, solo ventures, or as part of a managed commercial activity. Tose managed sites with a cluster of observatories ofen beneft from the availability of on-site help. Practically, the best sites are ofen located close to a small village in an otherwise rural area. While operating at a high altitude ofen has benefts from reduced atmospheric, turbulence, and transparency, at the same time, it may also have more challenging and

123

The Astrophotography Manual

fast-changing weather conditions. For example, if it is on the wrong side of a hill, there is the potential for more cloud/precipitation, because the air is forced to rise and cool. Te best locations are ofen in the Goldilocks zone. Spain and France are common remote locations in Europe. Australia, South America, and some southern states are also popular. Observatory Style Protecting your equipment from the elements is paramount, and the diferent observatory styles have pros and cons. Te geometry of a classical domed observatory is ideal for accessing lower altitudes (valued by comet hunters). Te roof and shutter both require motorization and careful synchronization with the mount and the optical topology. Tis is not a trivial task and many opt for a simpler roll-of roof observatory design. Te smaller ones restrict the imaging altitude to ~40° and require the mount to be parked carefully before roof operation. In the larger communal commercial roll-of roof sites, the increased wall distance lowers this horizon and usually does not require the mount to be parked. Tese sites, however, expose your equipment to the elements, whether you are imaging or not, as they do not give you control over the roof position. Some other sites have individual sliding roofs without intermediate walls and only partially protect your equipment from the elements. One should also consider security and insurance; commercial sites ofer varying levels of protection against 2- and 4-legged intruders and some hosting facilities include insurance in their monthly fee. Most of us cannot aford a total loss of our equipment, and the running costs should include a single- or groupinsurance policy. Te availability of other facilities plays an important part in weighing up one’s options. Tese include local accommodation and all-year-round economic transport. Low-cost airlines are great unless their service is seasonal. Lastly, the availability of local support is essential for those situations that cannot be recovered remotely. Tings go wrong, fuses blow, and adjustments are required. Setting Up Te key to successfully setting up is to ensure your setup operates reliably in a potentially hostile environment. Unfortunately, expenditure is not a guarantee of reliability and the best way to ensure a smooth installation is to initially run-in the entire system (including cables and power supplies) at home and debug it be-

fore transporting it. Debugging includes deliberately introducing problems and checking the fault tolerance and remote recovery process, all without cheating! An essential consideration is the choice of the telescope; we have not mentioned it before but the likelihood is that you will be imaging with a single focal length at the remote site and only you can make that choice, based on your imaging needs. However, agreeing on a range of optics between the members may be a consideration in a group. As mentioned in earlier chapters, all power supplies and electronic modules must be industrial quality and capable of extended use in an outdoor environment. Tey need to be weatherproof or housed in a suitable enclosure. Cables should be high quality and gold plated to prevent corrosion. In addition to the items you regularly use in a home observatory, remote operation requires additional equipment: • • • • • • • • •

electric telescope/mirror cover fat feld light panel robust focuser and dew control environmental sensing CCTV/all-sky camera web/USB/power control roof control uninterruptible power supply humidity control

Electric Telescope/Mirror Cover Several manufacturers ofer servo-controlled covers and some double up as fat light panels with in-built LED or EL light. Tese ofer protection from the elements and creepy crawlies, enabling fat and dark calibration exposures. In addition, many are USB-controlled, and others are Ethernet controlled, which ofers further protection should the imaging computer crash. Flat Field Light Panel Tere are numerous electroluminescent and LEDbased fat panels. Tey can be mounted over the end of the telescope on an articulated arm or arranged on the observatory wall at which the telescope is aimed (with the roof shut). Some have USB-controllable intensity and operated from the capture application. Robust Focuser and Dew Control While both are normal requirements in any setup, these systems beneft from further robustness. In particular, on focuser mechanisms, to avoid slippage, it is better to use a rack and pinion focuser mechanism

Setting Up

and connect a geared motor directly to the focuser shaf rather than the reduction drive shaf, which is friction-based and may slip under load. For dew control, many inexpensive dew-heater controllers are set and forget. In a remote environment, electronic control is useful to set the heating level according to the environmental conditions. Environmental Sensing It is essential to monitor cloud, humidity, and temperature to anticipate good and bad imaging conditions and, for those mounts that compensate their tracking for atmospheric refraction, temperature, humidity, and pressure. An anemometer is a useful addition too. Te popular Lunatico CloudWatcher system and an environment sensor, like the MBox range from astromi.ch are popular choices. For rain detection, I recommend employing dual systems. My backup is a Hydreon optical rain sensor. It is very sensitive, and its relay output connects to the Arduino roof controller. CCTV/All-Sky Cameras Having eyes on the ground (and in the air) are very useful and is the ultimate confrmation of mount/roof position and sky quality. Te usual setup is to have an IP camera monitor inside the observatory and ft a USB CMOS video camera (or guider camera) with a wide-angle CCTV lens and mount them in a clear dome outside the observatory and connect it to the imaging computer. Some applications, such as Tektite Skies, monitor an all-sky camera and calculate a clear sky indicator based on star count and sky intensity. It can usefully work remotely, too, by monitoring a cloud-based folder for fresh images. Web/USB/Power Control Without a physical presence, web control of power and USB communications is essential to start up, shut down, and re-boot selective parts of the observatory system. Tese, in turn, require a high-quality Internet switch/router capable of working continuously in demanding conditions. Web control is benefcial as it does not rely on the imaging computer and is the fallback system for re-booting. A controller will link to the router (via a cable) and have multiple relay/power outputs suitable for individual control of the observatory and imaging systems. USB control is a godsend when a single device fails to initialize or crashes and requires re-booting. For example, several astro modules integrate a resettable USB 3.0 hub with full power, focuser, and dew control.

124

Roof Control Similarly, the roof and mount systems will require automation, preferably using sensors (rather than a dubious sofware status) for reliable parking and control logic to prevent collisions. Some roof systems already have of-the-shelf controllers, or a confgurable system like the Lunatico Dragonfy will control a roll-of roof in response to various sensor inputs. Others may wish to design their own, as I did with the Arduino-based project shown in previous books and still on the support website. In these remote systems, you are not around to press “on” buttons and the imaging computer should be confgured to boot upon the application of power and ideally, if secure, to log the user in automatically. On Intel machines, this is confgured in the BIOS settings, accessed by pressing Fn2 during boot and disabling the login requirement in the Window’s “netplwiz” command dialog, respectively. Again, Intel NUCs or ruggedized PCs are a good choice, using solid-state drives and housed in a protective enclosure. UPS Uninterruptible Power Supplies (UPS) are an intelligent emergency power source that kicks in when the mains fail. Tey detect a mains failure, and their battery has sufcient capacity for the system to shut down the observatory safely and wait for the return of power. Power outages in the UK are rare, but the same is not true in many remote locations. So, making sure your system reacts appropriately is one of the essential diagnostic tests that should be tried at home frst. Finally, try to think ahead… for instance, are the main sockets diferent from those at home! On a pragmatic note, one needs to ship the equipment safely to the remote site. If this entails freight, you must pack up the entire system carefully, insure it, and include all the tools required to re-assemble it at the other end. It is sensible to pack spare cables, fasteners, and adapters, too, just in case. Re-assembling will take more time than you estimate (especially if that includes collimating a refector telescope), and if the prospect is too daunting, several astronomical equipment retailers worldwide ofer a remote observatory installation service. It comes with a cost, but it may be a false economy to try it yourself. Humidity Control Weather shifs and sudden temperature swings ofen cause high humidity and sweating, where condensation forms on circuitry and causes them to fail. Te

125

The Astrophotography Manual

dehumidifer in my roll-of roof observatory has an adjustable humidity control. It is plugged into an intelligent web switch that is disabled during the summer, while imaging, and on a timer during the winter months. I also place desiccant bags in the IP-65 enclosures that house the power supplies and observatory electronics to prevent condensation from forming on sensitive circuits (typically when turned of). Operation Running a really remote system is subtly diferent to one at home. Tere are a few more control panels to consider, and multiple monitors are useful. However, you are entirely reliant on the various monitoring systems. For a while at least, it will be difcult to resist the temptation to shut the system down when you hear it raining outside. One of the pleasant surprises will be the amount of data the system will collect, and the likely bottleneck will soon be one’s ability to download and process it.

Distributed Systems An earlier chapter mentioned distributed computing resources, in which a typically low-powered computer operates at the mount and the client machine does the demanding activities using the same or a diferent operating system. Tis is ofen associated with miniature Linux computing modules (e.g., RPi), to relieve them sufciently to make the user interface more responsive. A specifc protocol is used in these systems to pass device-level controls between the computers. For Linux, this is INDIserver and Alpaca/ASCOM remote for ASCOM-connected devices. Tese systems also pass the image data back to the client (say for focusing analysis or plate-solving) and require a fast, reliable network connection to run smoothly. As such, a distributed system confguration is not ideal for really remote use.

One of the advantages of distributed systems is that with a static hardware confguration, there is little need to continually update the remote PC’s sofware and potentially introduce issues. Te continual churn of mainstream OS updates is limited to the accessible client computer. As image capture device miniaturization continues and mobile devices increase in performance (and size), I predict this will be an increasingly popular format for general use.

Isolated (Local) Networks By their nature, dark-feld sites are ofen remote and without Internet service. However, it is still possible to use the same confguration as in the backyard by using an ad-hoc network or a wireless access point. An ad hoc network is a temporary local area network, typically created by a computer rather than a router. It is the default confguration for a Raspberry Pi astro distribution but, while it is possible to create an ad hoc network on a Windows PC, the connection is slow, even if you use a utility application like Connectify. Connecting the imaging computer’s Ethernet port to a travel WiFi router, like the one in fg.6, creates a faster connection and usually with better signal strength too. I set mine to Access Point (AP) mode and assign a static IPv4 address to the MAC address of the astro computer’s LAN connection (one time only) in its confguration menu. It is then an easy task to use this as the IP address in Microsof Remote Desktop and confgure a robust connection. Tere is no Internet connection in this arrangement and as it is not possible to sync the computer’s clock with an NTP server, it requires a small GPS unit (or manual entry) to correct for any error in the PC clock.

fg.8 In full remote mode, multiple screens are useful for monitoring all the systems and confrming weather and observatory status. The power control will typically run on the client machine, independent of the imaging computer.

A detail from the Pelican Nebula

Image Capture

127

The Astrophotography Manual

Exposure Planning Putting theory into practice and learning from the experience of others.

T

he equipment is selected, bought, assembled, and aligned, and all appears good. Te imaging computer is loaded with the applications, drivers, and appears to connect and function. Tis is the point when one capitalizes on the efort that has gone before, for while it all should work, many things prevent image capture from going smoothly or cause sub-optimal results. Over the years, astrophotographers have developed a sixth sense to anticipate many issues and close in on the best settings. However, learning never ends, and new technology, concepts, or misconceptions are just around the corner. Keeping up-to-date is part of the allure and a continual risk. Best Practices Tis chapter, and the two that follow, concentrate on three essential activities; exposure, focusing, and tracking (including guiding). Focusing and guiding discussions are prevalent on forums and for some reason, exposure is less so. Tat is changing as techniques and recommendations accommodate new technologies or the shifing signifcance of various aspects of image capture, including lucky imaging and photometric analysis. We know that increasing

exposure improves image quality, but there are many other things to consider.

Exposure Plan Te plan precedes the detailed exposure execution. It considers simultaneous aspects to reach the best compromise, including the target, environment, technology, processing, and pictorial requirements. New technology, and the increasing use of CMOS cameras in both cooled and uncooled versions, have caused a reevaluation of several established best practices. For example, my observatory is semi-rural, and I mostly do narrowband imaging to block light pollution. My narrowband nebulosity images comprise narrowband and RGB image stacks; the narrowband exposures are optimized for the “faint fuzzies” and the RGB exposures for realistic star color. All use one of several standard exposure times, allowing me to reuse a dark-frame library for convenient calibration. I image with narrowband flters when the Moon makes its presence felt and broader RGB flters when it is darker. Te narrowband exposures run at medium/ high gain to reduce read noise over a standard exposure of 300 or 600 seconds. Tese long exposures may

fg.1 Updated to consider the additional controls and defects of CMOS sensors, the signal path in a modern sensor has several noise sources before and after photon conversion and sampling. Together with the characteristics of the object, conditions and artistic vision, these afect the choices we make in our exposure plan.

Image Capture

clip a few pixels in bright star cores, but it is not a concern since I suppress or remove stars during image processing. Te stellar images, however, use a low camera gain to maximize full-well depth and shorter exposures to ensure the brightest stars do not swamp the sensor. Although the sensor read noise is higher at the lower gain setting, sky noise through these broadband flters exceeds it by some margin. Te combined noise is not a concern, however, as the stellar image is stretched much less to preserve star color, and the background (and its noise) is removed before combining with the narrowband image. On another night, I might image a galaxy in "real" color, using LRGB flters and a diferent strategy altogether. Expose for Object SNR Te overriding principle of a good exposure plan is to give oneself options, which might include lowering noise to reveal faint nebulosity, yet restrain exposure highlights, to retain detail and color in galaxy cores and stars. It is a careful balancing act that becomes easier with practice and observation. Regardless of the camera technology, I generally choose an exposure for pictorial work to make the best use of the sensor's well depth. Tis might clip a few bright star cores, but nothing excessive. (It is possible to restore star cores with a bit of re-scaling and using the color information of neighboring pixels or another image during image processing.) For photometric work, all exposures operate in the linear operating region of the sensor, and peak intensities are usually no more than 75% of full-well capacity. Te outcome optimizes every imaging hour to improve object SNR without distorting the peak signal. Te keyword here is object and an illustration of the principle considers the selective flters used in narrowband imaging. It is common practice to have an imaging sequence that cycles through Hα-SII-OIII or R-GB flter exposures. Many objects have red nebulosity associated with Hα emissions. If the object does not have SII or OIII emissions, one does not waste imaging hours exposing it through the respective flters. If the object has strong Hα transmissions but less intense OIII and SII emissions, an alternative strategy might take more OIII and SII exposures to improve their SNR. Discussing alternative exposure and fltering strategies for diferent object types and treatments is useful. Before we do, however, it is helpful to understand the science behind these recommendations.

128

Exposure Bookends Conceptually, an optimum exposure duration is a compromise between over- and under-exposure. However, that is not the only variable; CMOS astro cameras have a sensor gain setting and occasionally alternative readout modes too. A sensor captures more sky background photons with longer duration exposures, and the associated shot noise increases. Te skylimited exposure is the point at which the shot noise from the sky background is signifcantly greater than the sensor noise and usually assumes the sensor read noise in each frame is 5–10% of the total shot noise. Tis is the quality threshold Q, in this case, 0.05–0.1. Te sky-limited exposure time is afected by aperture, fltration, sensor gain, light pollution (sky fux), and sensor read noise. If the sensor read noise reduces, the sky-limited exposure shortens and if the sky becomes darker, the sky-limited exposure lengthens. Calculating this exposure requires a little math, some knowledge of the sensor characteristics, and the background light level. Tese simple equations work well in a spreadsheet and are conveniently incorporated into several image capture applications. Te sky fux is measured in electrons/second. It is calculated by comparing the sky background pixel level with that of a dark frame (using the same exposure duration T and camera settings). A simple subtraction gives a pixel value for the sky background and a photosite value in electrons once sensor gain and ADC bit-depth are accounted for. Armed with the sky fux, read noise, and a quality threshold Q, the equations below calculate the sky-limited exposure duration t. (In the equation below, K is the sensor’s internal multiplier from the ADC to its 16-bit output. It is 16, 4, or 1 for 12-, 14-, or 16-bit ADCs, respectively.) skyfux (e-/s) = gain∙(background-dark frame) / (T ∙ K) When Q is small (