CCD astrophotography

• CCD basics
• CCD image processing
• CCD imagers for telescopes
• see also:
  • astronomy index and links
  • astrophotography
  • using a digital camera for astrophotography
  • advanced astrophotography
  • prime focus digital imaging
  • CCD websites:
    • SBIG (US)
    • Starlight Xpress (British)
    • Yankee Robotics Triffid Cameras
    • K3CCD software
    • using a simple webcam:
      • web-caddy
      • Quick-cam and unconventional photography group
  • Ron Wodaski's book

Disclaimer: I am not an expert in this field, the following are derived from internet forum discussions:

• The introduction of Charge-Coupled Devices (CCDs) has dramatically changed the methods astronomers use to view objects. The remarkable sensitivity and dynamic range of CCDs has made them the detector of choice.
• A CCD is an array of photosites (pixels) on a silicon substrate. The pixels are arranged in rows and columns. Light falling on the CCD is recorded as an electrical signal by converting the photons to electrons.
• The sensitivity of a CCD is ten times faster than the fastest films and is linear in response (which films are not). The number of photons converted to electrons, referred to as Quantum Efficiency, is in the order of 50%.
• These advantages as well as the digital nature of the data make CCD devices ideal for astronomical imaging.
• A general CCD imaging rule is to take the longest exposure before star blooming and telescope mount tracking cause problems in the image. To image fainter objects you stack multiple images.

• wide field astrophotography:
  • Tak FSQ 106 refractor with SBIG STL-11k ($US10,000 or more in total)
  • Tak Sky90 refractor with reducer/flattener and SBIG ST-2000XM or ST8XE
  • Stellarvue SV80S or Orion80ED refractor and SBIG ST-2000XM or ST8XE ($US6000)
• planets:
  • consider a modified webcam to allow hundreds of images to be stacked while excluding the transitory poor seeing images
    • Philips Vesta or ToUCam
    • SAC (US) / ATIK in Europe
    • Meade:
      • Lunar and planetary imager (LPI)
      • Deep Sky Imager
      •  
• compromises:
  • LX200 SCT with f/6.3 reducer and/or AO-7 (AO-7 is $US1000):
    • notoriously difficult to use for CCD imaging without an AO-7
    • SBIG ST2000 may be possible with binning
    • SBIG ST-7 or ST-8 may be better
    • avoid  ST-9 as pixel size too large
  • LX200 SCT with .33 focal reducer and the ST-7E camera
  • digital SLR but this is a compromise given their inability to be cooled to reduce noise and their lack of H-alpha sensitivity, but despite this they make a reasonable compromise for versatility as a day camera.
  • digital prosumer cameras (most not suitable for planets unless can be used afocally)
  • webcam:
    • A typical way to obtain a deep sky image with such a camera is to take 80 - 100 unguided 40s shots, select the better 50% of them and stack/align them in K3CCDTools  
• guidescope:
  • a 60mm refractor at 900mm with Barlow does well if the webcam is SC1 modified for longer
    exposures (lower magnitude stars).
  • for guiding, using an unmodified webcam, like the Philips Vesta or ToUcam is sufficient. Adapters, which replace the webcam's objective by a 1.25“ barrel, are available from various sources such as http://webcaddy.com.au/astro/

(2003)

• CCD vs DSLR:
  • MX716:
    • Monochrome chip→narrow waveband imaging +++
      Cooled chip→longer exposure +++
      Deeper imaging +++
      narrow field of view (could go either way)
      useful for Astro work, not for family use
  • Canon EOS 300D:
    • Chip is twice the size→field of view larger (+/-)
      One shot color with 6 MP +++
      Can be used for family use +++
      Exposures <5 mins (+/-)
      Size of chip not compatible with some focal reducers
  • “I have had a MX716, a ST-2000XM with color filter wheel and I have an EOS
300D.
The SBIG and MX716 were more sensitive, but at f6.3 and below I get much
better results and have more fun with my EOS. I sold the other cameras after
I wrote my Maxim/DL driver for the EOS.

If you just want to have fun and make pretty images, the DSLR cameras can't
be beat. If you are imaging really fair objects and slow f-ratios then you
need a cooled CCD camera and will have to spend 10x the price of a 300D for
comparable images.

The ST-2000XM is not quite a fair apples to apples comparison since it has
an auto-guider, but the chip is so small that it was a pain to find a good
guide star anyway for some objects, a separate guide scope, or wide field
imaging is a great solution and still much cheaper.

If I had all the money and time in the world I would probably have a ST-10
with A07 and a Color filter wheel as well as a DSLR, but given I don't I
pick the DSLR for the fun of it.

I have a friend who has a 30k+ imaging setup. He bought a 300D and now he is
gaga over it and has hardly touches his sensitive, expensive setup all
summer.”
• it depends a lot on the telescope:
  • resolution:
    • with perfect tracking and perfect seeing conditions and perfect optics the resolution will be
      206 * pixel_size(microns) / focal_length(mm).  So if your using a Canon 10D on an 8” f/10 telescope thats 206 * 7.2 / (8 * 10 * 25.4) = 0.73 arcseconds/pixel.  If you put a Canon 1Ds on the same scope thats 206 * 8.8 /(8 * 10 * 25.4) = 0.89 arcseconds/pixel.  So the 10D will give better resolution if your seeing is good enough.
  • field of view:
    • field of view you get with each camera  is approximated by just multiplying the arcsecond/pixel value by the number of pixels…..the 10D will give you a field of view of about 37'x25'…..and the 1Ds field of view would be 60'x40'
    • One thing to keep in mind is that unless you have a high performance scope.  Your scope will probably not have good enough edge of field performance to take advantage of a larger chip.  You will start to see
      distortion or coma in the stars at the edge of the field.  I personally didn't think the difference in price was worth it for a larger field of view so I got the 10D.  I would rather save the money and get a 300mm f/2.8 lens.  That's a perfect lens for doing some astronomy shots and has a faster f/ratio than most telescopes you can use.
  • LX200 (2004 comments):
    • seems you need a ST-7, f/6.3 reducer and an AO-7 to use with an LX-200 and mount on a high quality German Equatorial
    • see http://www.geocities.com/samirkharusi/m42.html for Samira's foray into imaging with a SCT.
    • “The LX200 is notoriously difficult to use for CCD imaging without an AO-7, and I've read some opinions that the 12” mount is undersized for the load it carries.
      However, some people get excellent results with an LX200 and AO-7 combination. You might want to check out Mark de Regt's site (http://home.earthlink.net/~akilla/MAD/); Mark is a steadfast proponent of
      the LX200 for imaging, and his photos stand as testament to that.
      OTOH, I had a 10“ LX200 and was constantly frustrated trying to get it working well, to the point of replacing it with a Celestron CGE equatorial mount. Maybe I didn't try hard enough before giving up on the LX200, but I find the CGE to be a MUCH easier mount for imaging”
    • “The FOV of a 35mm film frame is much larger than an ST-2000 CCD chip, so everything will be much smaller. For example, M51 pretty-much fills the ST-2000 frame, but consumes less than 1/15 of a 35mm frame. What happens to your stars if you enlarge one of your film frames 15X? Are they still round
      and tight?
      I'd say that using an ST-2000 with your f/10 12” LX200 would be very difficult without an AO-7. And your budget seems to be the limiting factor - the AO-7 costs about $US1,000.“
    • “With the LX200GPS - 12, unless you also purchase an AO-7 unit, I would suggest you go with a .33 focal reducer and the ST-7E camera.  You can pick up a good used ST-7E on Astromart for about $800 to $1,000.  The chip size of the ST-7 goes well with the .33 focal reducer.  You don't even have to go with the CFW8 right now. and the parallel interface is, IMO, not a problem with the file size of the ST-7.  If you are new to CCD work, it is advisable to tackle the basics first.  Go for color after you master good luminance imaging.  If you are an old hand at CCD work, get the CFW8 and dive in.  The ST-7 will hold its value fairly well and you can upgrade later should you decide to.  The AO-7 is almost a requirement for using a LX200 mount at anything like F:6.3 or above.”
    • “The ST-2K's small pixels are not a particularly good match for the long focal length of the 12” LX200, so you would always have to bin, even with the AO-7.  OTOH, the pixels of the ST-9 are so large that you will not be able to image at very high resolution, even with an AO-7.  I have a 10” LX200 and, when faced with exactly the same question three years ago, I went for a used ST-7 using a .33 reducer for f/5.  It's cheaper than an ST-9, gives somewhat better resolution and, I think, larger FOV than the ST-9, and, with an AO-7 once you get a bit better at this, will allow very high resolution imaging.  At the point you get much better, and frustrated by the limitations of the ST-7, you can start trolling Astromart for a used ST-8
      and an AO-7.
      I think that there is a very high probability that you would quickly become dissatisfied with either an ST-9 or the ST-2k.  With the ST-7, which go for something just over $1k on Astromart, you save quite a bit at the start, and you will be free, when you know more about your equipment and desires, to
      upgrade appropriately.“
    • “The pixel size of the ST-7 is a better match for the 12” LX200. At 9 microns across, each pixel will see a section of sky approximately 0.62 arcseconds across. That's extremely high resolution and requires very good seeing. If you don't have exceptional seeing at your location, use of an f.6.3 reducer or an AO-7 or both will be mandatory for most nights. See the SBIG website for info on the AO-7.
      For the ST-2000, the image scale will be even higher resolution, 0.5 arcseconds per pixel. That is a very demanding resolution.”
  • Tak Sky 90 or Televue:
    • I'm assuming that A) you have chosen to start with widefield imaging (good choice), and B) you have made a definite decision on the camera..ST2000XM.

First, don't worry about your image scale sampling with either
scopes, because they both have relatively short focal lengths and
with either scenario, you are more toward the undersampled scenario
vs. oversampling.  For example typical “hi-res” with the larger
scopes/longer focal length is around .60 to 1 arcsec/pix!  Remember
that if you have typical seeing (including atmospheric, mount &
guide issues, etc.) of around 3 arcsec, your _best_ sampling is that
number divided by 3.3 =  .90 arcsec/pix.

Now, with widefield, you are obviously aware that you will be
sampling differently…no problem, the images still come out great.

So, I guess it's just a matter of which optics you prefer..the SKY90
or the Televue.  I can't speak for the Televue, but I've seen great
images with Televue products.  However, some people complain that
the Televue focuser is not quite robust enough.  The SKY90 will also
produce great images…you MUST use the field flattener (another
expense), but most people love the Tak focuser.  Of course, with
either setup the Robofocus will work great and is a great tool.

Therefore, if I were you, I'd consider two things before making a
final decision.  First, check out how Televue users feel about the
stock focuser and make sure you feel it's acceptable.  Secondly,
determine how wide a field of view you want.  As you said, the TV
will offer a smaller FOV but with slightly more detail, the SKY90
would offer the larger FOV and the detail would still be great. 
Personally, I like the TAK products and purchased a SKY90…never
used it because the flat field wasn't large enough to cover the
large chip in my STL6303…replaced it with an FSQ.“
    • “The ST-7 chip just barely works with the Sky90 without the flattener. That
chip is 4.6 x 6.9mm. The ST-2000 chip is quite a bit larger, around 9 x
12mm. So I would expect the reducer to be essential.

In my experience, you can't use the image circle on film to determine the
useable image circle on a CCD chip. Roughly speaking, film has a resolution
of around 25 microns (but depends on grain size). CCDs are in the 7-9mm
range for the most part, and thus will show optical issues much more easily
than film will.

In answer to your original question, the Sky90 will be a joy to use. It is
easy and forgiving. The FS-102 would be excellent as well, but of course
guiding and tracking will be more challenging.

The issues around the TV focusers are real; I had a devil of a time imaging
with the NP-101, for example. They are designed for visual work. The optics
are first rate, however, and if you don't mind a bit of machining you can
probably rig up something that will be more robust.”
    • “get the SKY90 (with the reducer/flattener) along with the ST2000…you'll be extremely happy” be aware that you will spend many hundred dollars for so-called “accessories” for the Tak, to make it work for imaging, but they do work, and very well.
      • Takahashi 35MM WIDE MOUNT CAMERA ADAPTER FOR FS-78 SKY 90, FS-60C, AND FC60E $49.00
        Takahashi 7×50 Finder Bracket  $59.00
        Takahashi Adapter Q  $49.00
        Takahashi Camera Angle adjuster for FS78/Sky90/FS60C and FC-60 E(NZ) $179.00
        Takahashi CCD adapter wide mount to t-thread for CCD  $79.00
        Takahashi Extender-Q 1.6x  $269.00
        Takahashi FCL 90 Sky 90 II refractor OTA Only  $1,849.00
        Takahashi Reducer/Flattener for SKY-90 f/4.5  $339.00
        Takahashi TUBE HOLDER FS78/FCL90 $129
    • “The Sky90 might well satisfy your needs in the near future, but if you wish to quickly move to a larger format detector then you will have problems with its flat FOV.
      • The Tak Sky-90 image scale would be forgiving of guiding errors and would have nice wide-field views and seem to be a reasonable match with the ST-2000XM with 3.0 - 3.75 arcseconds/pixel
         depending on use of the focal reducer. This would also be very tolerant of seeing conditions and require shorter exposures (might be important here where the number of  clear nights is not  large). Robofocus would seem to be important since the zone of sharp focus is very small with this focal length (I noted an image in the files section of someone who had successfully mounted a Robofocus on the Sky 90).
    • My suggestion–look seriously at the FSQ106.  Huge, flat FOV, similar
focal length, all threaded interface–an imager's telescope. Look at
the spectacular stuff being done right now with the STL11K and the
FSQ106.  The 106 isn't that much more money than the Sky90 (not like
the 11K), but you won't outgrow it.” “With all of its specialized adapters, I don't think it is a visual observer's telescope.  For that I would recommend the TV/NP101.  But for imaging, the choice is different.”
      • The TV-102 or FS-102 would give a bigger image scale and allow more objects to be imaged with some detail. Not sure whether the ST-2000XM is a good match with 1.74 to 2.17 arcseconds/pixel - perhaps could bin 2×2? Seeing conditions would be more of a factor but perhaps not too bad especially when using the focal reducer (f7 with the TV-102, f6 with the FS-102). Exposures would be 2-4(?) times longer. Mount requirements would increase but focal lengths are still relatively short and shouldn't be too much of a problem (famous last words).
  • CGE1100 and Orion 80ED with ST4 autoguider:
    • some thoughts:
      • With the CGE1100, you have the potential for mirror flop and this will be an issue if you don't use a self guiding camera (one of the SBIG models).  Therefore, I would also consider selling the ST4 and considering one of the following options:
      • The ST2000XM with a CFW8 would be a good combination for the large scope as well as the 80ED.  It would allow you to image in color and should fit your budget..especially if you find a good used combo.  I would avoid the one shot color more for reasons of sensitivity than resolution.  Also, you can't bin the one shot color and get color images.  Finally, you limit your ability to do narrow band imaging with the one shot (Ha filter for example).  Do yourself a favor and go with the versatility of a monochrome camera + CFW8 option.
      • If you want more sensitivity and a slightly larger FOV, you can consider a good used ST8XE/CFW8 combo.  The ST8 will have an overall higher QE allowing more efficient use of your imaging time…more data per hour imaged, plus a MUCH higher QE in the RED/Ha spectrum. (very helpful when imaging objects in this spectral area..and there are _a lot_ of such objects…
      • if buying an ST8, make sure it's the 8XE…meaning that it's the USB version with the larger 237 guide chip.  Also make sure that the CFW8 is a fairly recent model with filters that have the built in IR rejection in the color filters.
      • in mid2004, many are upgrading from ST series to STL series and the XE models are being replaced by ME versions.

• when using a monochrome CCD camera, one needs to take several photos with different filters to eventually create a colour image
• to avoid having to take photo system apart when changing filters, and have to re-focus and risk dust exposure to the CCD, etc, one can use a filter wheel which holds a number of filters that can simply be rotated into place sequentially - one of these can be black for taking dark frames if need be.
• SBIG CFW8A wheel:
  •  
• Finger Lakes FLI CFW-2 wheel:
  • allows you to have 8 filters ready to go. That is the only way that one can have LRGB plus Ha and OIII in a single session without switching out filter wheel carriers.
  • can then use Astrodon filters which are are closely balanced in efficiency and matching parfocal Ha Astrodon filter.
    • compared with the SBIG filters, the Astrodon filter set, including Ha, are really are parfocal, which the CS filters really weren't; the blue filter is wonderful, with very low noise in the images and not losing guide stars, and the Ha filter really is parfocal, so that you can focus with the clear filter and image with the Ha filter.
  • NB. the Custom Scientific Ha filter is not even close to being parafocal with the C.S. LRGB filters.
  • is lower profile and therefore more forgiving with tight backfocus scopes and lenses (FSQ 106N and Nikon lenses).
    

• Raw CCD images are exceptional but not perfect. Due to the digital nature of the data many of the imperfections can be compensated for or calibrated out of the final image through digital image processing.
• Composition of a Raw CCD Image:
  • A raw CCD image consists of the following signal components:
  • IMAGE SIGNAL - The signal from the source.Electrons are generated from the actual source photons.
  • BIAS SIGNAL - Initial signal already on the CCD before the exposure is taken. This signal is due to biasing the CCD offset slightly above zero A/D counts (ADU).
  • THERMAL SIGNAL - Signal (Dark Current thermal electrons) due to the thermal activity of the semiconductor. Thermal signal is reduced by cooling of the CCD to low temperature.
• Sources of Noise
  • CCD images are susceptible to the following sources of noise:
    • PHOTON NOISE - Random fluctuations in the photon signal of the source. The rate at which photons are received is not constant.
    • THERMAL NOISE - Statistical fluctuations in the generation of Thermal signal. The rate at which electrons are produced in the semiconductor substrate due to thermal effects is not constant.
    • READOUT NOISE - Errors in reading the signal; generally dominated by the on-chip amplifier.
    • QUANTIZATION NOISE - Errors introduced in the A/D conversion process.
    • SENSITIVITY VARIATION - Sensitivity variations from photosite to photosite on the CCD detector or across the detector. Modern CCD's are uniform to better than 1% between neighboring photosites and uniform to better than 10% across the entire surface.
• Noise Corrections
  • REDUCING NOISE - Readout Noise and Quantization Noise are limited by the construction of the CCD camera and can not be improved upon by the user. Thermal Noise, however, can be reduced by cooling of the CCD (temperature regulation). The Sensitivity Variations can be removed by proper flat fielding.
  • CORRECTING FOR THE BIAS AND THERMAL SIGNALS - The Bias and Thermal signals can be subtracted out from the Raw Image by taking what is called a Dark Exposure. The dark exposure is a measure of the Bias Signal and Thermal Signal and may simply be subtracted from the Raw Image.
  • FLAT FIELDING -A record of the photosite to photosite sensitivity variations can be obtained by taking an exposure of a uniformly lit 'flat field”. These variations can then be divided out of the Raw Image to produce an image essentially free from this source of error. Any length exposure will do, but ideally one which saturates the pixels to the 50% or 75% level is best.
• The Final Processed Image
  • The final Processed Image which removes unwanted signals and reduces noise as best we can is computed as follows:
    • Final Processed Image = (Raw - Dark)/Flat

• the correct imager depends on requirements such as:
  • telescope characteristics
  • computer - PC vs Mac
  • remote computing - does telescope have a serial port?
  • CCD pixel size and imaging area - what due you wish to image - deep sky objects vs planets vs wide field searches for asteroids/comets
  • is there a need to take photometric measurements of variable stars or determine precise asteroid positions
  • do you need to automatically guide the tracking for long photographs eg. dual-CCD self-guiding
  • do you need it to be Fastar Ready - designed to work with Celestron's Fastar lens assembly to allow par focal operation at F/1.95.   
• fastar-ready single CCD cameras start at $US1400
• dual CCD self-guided cameras that attach to telescope eyepiece holders start at $US3000.