Prime focus digital imaging

• see also:
  • CCD Astrophotography
  • astrophotography with digital cameras
  • deep sky astrophotography
  • Jim Solomon's cookbook for prime focus astrophotography

Introduction:

• prime focus is when a camera is attached directly to the optical path of a camera lens or telescope without an eyepiece although a camera tele-converter lens or a Barlow lens may be added to the system to increase magnification.
• most telescopes use a T2 mount which is a screw mount similar to the original T mount (the Pentax M42 mount) but the thread is 0.75x that of the T mount thread and the flange to image plane is 55mm compared with 45.5mm on the M42.
• you can buy T2 adapters for practically all 35mm and digital SLR mounts
• the main advantages are:
  • image quality as there less optical elements to add to aberrations as with eyepiece projection
  • shorter exposure durations needed as faster f-ratios
• exposures longer than 1/30th sec will need tracking with a motorised RA drive:
  • images will be improved further if there is some form of accurate guiding:
    • via a 2nd telescope system attached to the mount:
      • manual guiding via an illuminated reticle eyepiece
      • auto-guiding via a dedicated auto-guider
    • via the same telescope optics:
      • manual guiding or auto-guiding as above but via an off-axis guide that splits the beam to some is visible in an eyepiece right angles to the camera path
      • some dedicated CCD cameras have a built-in auto-guider
    • NB. auto-guiding systems require a telescope mount that will accept the auto-guider's input
• focusing can be problematic see focusing a telescope
• digital options are:
  • digital camera - for a digital camera to be able to do this, it's lens must be removable & thus it must be a digital SLR
    • eg. Canon Digital Rebel 300D
    • NB. need to ensure you can set the mirror to be up before the self-timer starts to avoid camera shake from the mirror.
    • Canon EOS cameras have a reputation for noisy mirror slap and apart from the 1Ds, what is more difficult to deal with is their shutter shudder even with mirror up (may need to use a manual shutter such as a cover over the telescope). This is why the Olympus OM-1 was so popular.
  • webcam - usually modified but tend to have low pixel counts
    • great for high power imaging of moon & planets with eyepiece projection as take hundreds of images for stacking & the resulting images rival if not better than most DSLR images eg. 500 frames at 5fps although total duration is limited by object rotation esp. jupiter as the clouds will appear blurred in > 90sec.
    • eg. ToUCam $US150
    • eg. Quickcam 2000 $US40
  • dedicated astrophotography CCD camera
    • these cameras are especially designed for high sensitivity, high dynamic range (often 14-16bit) and low noise via cooling to minus 30-45degC
    • eg. S-BIG's range
• the best set up for deep sky objects if you have lots of money is a high-performance mount, CCD plus autoguider with adaptive optics, and sharp/flat optics.
• maximising the signal to noise ratio:
  • readout noise is not constant. By summing images this non constant noise (poisson distribution) is averaged out over the entire image. Yes it will bring up the background brightness if stacking many many images, but for the most part summing many images increases the signal to noise ratio by the square root of the number of images summed. There is a point of diminishing returns as the background level increases but for those who do not have mounts that can track for 3 to 4 minutes or do not autoguide, summing images works quite well. Noise that is constant such as bad/hot pixels or readout defects due to charge transfer efficiency will however degrade the image if summing. This is where darks, flats and bias frames come in.
  • whichever option is used, you will still have to weigh up multiple short exposures vs one or a few long exposures:
    • This debate over 1 long exposure vs. summing has been going on for quite sometime. They both have there benefits and downsides. There is a sweetspot that every image train and setup has where you are taking the longest exposure you can and have to sum images beyond that. You just have to find that spot.
  • when effective focal length is long such as 2000-4000mm, most people will not have the equipment to track accurately to allow more than 10-15sec exposures.
    • bright globular clusters such as M13 and Omega Centauri can be imaged at 1600ISO, 14" f/11 4000mm with stacking of approx. 30 frames of 10-15sec exposures.
• some maths for digital prime focus:
  • image scale = 206 * (pixel size in microns) / (focal length in mm)
  • the Canon 300D pixels are about 7.2 microns so assuming an LX200 10" f/10 telescope you would get:
    • 206 * 7.2 / (25.4 * 10 * 10) = 0.58 arcseconds per pixel. 
    • So your field of view would be:
      • 3072 * 0.58 / 60  by 2048 * 0.58 / 60 = ~ 30 arcminutes x 20 arcminutes
    • If you put your 26mm eyepiece in the scope and assuming an apparent field of view of 50deg and again an f/10 telescope:
      • Magnification is: 10 * 10 * 25.4 / 26 = 97x
      • The actual field of view is the apparent field of view divided by the magnification: 
        • 50 / 97 = 0.51 deg or about 30 arcminutes
      • So the long side of your ccd chip is going to give you about the same field of view as your 26mm eyepiece assuming its a 50 deg APOV
        eyepiece.

Digital SLR vs dedicated CCD camera:

• digital SLR's:
  • see also http://www.pbase.com/mataylor/image/39066227 for a method of using a modified 300D with filters for prime focus.
  • pros:
    • versatile & can be used for other photography
    • cheaper (although still cost $A2000) => great bangs for bucks
    • easy to use
    • noise is becoming less:
      • you can shoot with a Canon 10D at 10°C for 10 min and get almost zero noise.
    • easy one-step colour photos but not as good a quality as more complex 3 or 4 photo colour-filtered B&W CCD's
    • great for bright objects such as moon, jupiter, sun even in light polluted areas
    • good for the brightest 10-20 Messier objects (eg. M45, M42, M31) away from light pollution, but not as good for the dimmer ones as these need long exposures (eg. several minutes) for each frame and noise interferes.
  • cons:
    • not good in light polluted regions for deep sky objects
    • dynamic range is still not as good - usually only 12bit per channel
      • with only 12 bits, and with gains dialed in optimized for daylight, it is incredibly difficult to overcome light pollution, which shortens how long one can expose.  The dedicated astro camera has electronics optimized for scraping the electrons from the dimmest signal off the noise floor.  That's what you're paying for.
      • with dark skies and a fast scope, you gain signal as fast as possible before noise can take over to begin with.  In these conditions, the differences narrow significantly and come down to more RBG Bayer Matrix limitations vs. color filtered Mono with Luminance ability.
    • in-built fixed IR cutoff filter limits sensitivity:
      • you will always have the limited sensitivity due to the Bayer filter and will have the red-cut off filter to preserve a natural color balance in daytime use.
      • it captures H-B (which occurs with H-alpha, but in smaller amounts) quite well, which produces those very lovely magenta emission nebula (like most 10D lagoon images).
      • the purpose in trying to hit the h-alpha line, and the sulfur line above it, is to bring in detail that wouldn't ordinarily be seen by the eye.  This is largely why all the great astrophotography films are disappearing, because the wavelengths that make for detail in emission objects are NOT the same as those in portrait photography, for example.
      • these can be removed but then may make camera unusable for daylight photography
        • see Hutech's modified Canon 300D
        • see http://pacificheight.com/canon/canonDR.htm
    • the problem of colour sensors vs B&W CCD's:
      • D-SLR will be close to cooled one shot color cameras, but never close to b/w CCD cameras in performance. This is a physical principle from the design - getting all photons to all pixels.
      • imagers in light polluted environment, the narrowband imaging gets more and more essential and this is a specialty for b/w imagers.
• dedicated astrophotography CCD cameras:
  • pros:
    • can produce the best quality images of deep sky objects, esp. in light polluted regions, in experienced hands
    • can image dimmer objects
    • use of narrow-band filters allows imaging even in light pollution or the full moon
    • low noise - cooled to minus 45degC
    • B&W sensors give higher resolution and sensitivity
      • high quantum efficiency (QE) - but lower if ABG or if have color filters placed in front of the sensors as is the case in most one-shot colour CCDs
    • higher dynamic range - usually 14-16bit
    • sensitive to H-alpha region of IR light
    • often incorporate auto-guiding correction - but need compatible telescope mount
    • often are designed to limit "blooming" effect of bright stars ie. have an anti-blooming gate (ABG)
    • the better ones use Kodak KAF scientific sensors
    • tri-colour imaging using a monochrome CCD minimises chromatic aberration in refractors so may be able to go for a cheaper non-APO refractor
  • cons:
    • expensive, especially the larger pixel versions
    • require use of a laptop to focus and capture image
    • use water to cool them down to minus 30-45degC - tubing, pumps, etc and when get cold, are subject to dew condensation if high relative humidity
      • see how much trouble it is here: Yankee Robotics Trifid Camera
    • older ones are small image size and slow image transfer to computer via parallel port - avoid these - choose USB2.0 for better speed
    • can only be used for astrophotography 
    • for colour images:
      • B&W sensors:
        • need to take 3 photos ("tri-colour imaging") - one with each of the 3 colour filters in place:
        • the normal practice is to take tri-color images with each filter at a specific length given a particular ratio.  This ratio is determined by the length of three images with RGB filters that produces TRUE color in an object we KNOW to be a specific color, in most cases, on a true white G2V star. From there, we just let the colors fall as they may.  This balance is indeed affected by the extra information from h-alpha sources, for example, but it's also affected by light pollution and atmospheric dispersion.  So, when this happens, the idea is to balance the image on a known constant, where white stars are indeed white in the image. 
      • colour one-shot sensors:
        • all Kodak color sensors have an IR blocking coating on them (that makes them TRULY ONE SHOT!), however this coating can be polished off.
  • consider as a starter kit:
    • astronomical filters + filter holder, and either:
      • used SX HX-916 camera - nice field of view, great sensitivity, doesn't need dark frames & is very lightweight
      • SX V-H9

Camera telephoto lens vs Refractor telescope:

• the main issues:
  • if your mount and guiding is not high-end & critical then you must use short exposures and this then needs fast, flat field optics such as f/3.3.
  • optics must be high quality to ensure minimal aberrations and a flat field (ie. stars all over the image are sharp) as any problems become very obvious in astrophotography. Zoom lenses are not usually adequate.
  • temperature instability may change the focus point during a long exposure.
  • does the image circle cover your camera sensor - this can be an issue with larger sensors such as the 35mm full frame sensors.
  • the more elements in a lens, the lower the contrast and the poorer the image.
  • for colour images, the lens must have minimal if any chromatic aberration, and thus all glass should be APO not just ED.
  • ability to use for visual use &/or cooled CCD cameras with filters
  • cost and versatility
• camera telephoto lens:
  • the resolution problem:
    • lenses with f-ratios smaller than f/8 result in poor resolution on digital cameras
    • if you put the x2 teleconverter on the 400f5.6L you cannot get even the Canon 10D resolution out of it, while the 300f4L is a borderline case, with barely 10D resolution
    • only the f2.8 telephoto lenses seem to have more resolution with a x2 teleconverter than 10D has.
    • a 105mm aperture f/5.8 Traveler refractor w/flattener outperforms a Canon 500 f/4 and 300 f/2.8 for astro imaging with a 1Ds.  Stars are smaller/cleaner and contrast is greater in the scope.  The 500 and 300, however, focus much easier and are easier to use under the stars than the scope. 
    • Canon L series IS lenses tend not to be as sharp as non-IS lenses for astrophotography but still acceptable.
  • examples:
    • see super tele
    • Canon 600mm f4.0 L IS USM - 133mm aperture:
      • ~$US7000
      • Samir Kharusi feels that this lens competes well with modern APO refractors in 2007 see here
    • Canon 500mm F4.0 L IS USM - 125mm aperture:
      • lighter than the 400mm f/2.8
      • approx. $US5000 ~ same price as Takahashi TOA 130 f7.7 but half the weight
      • used with the Canon 10D and the 2x + 1.4x teleconv. results in an effective focal length of 2240mm.
    • Canon EF 400mmf2.8L II USM IS - 130mm aperture:
      • heavy for a telephoto (6kg) - perhaps too heavy for wildlife use
      • 11 lenses incl 1 flourite and 2 SD lenses
      • high resolution, perhaps not as much contrast as a APO refractor
      • equates to 142mm aperture & gives good results with 2x teleconverter (ie. similar to 5.5" APO with focal reducer, but lighter)
      • can be stopped down to 100mm aperture at f/4 for even sharper images, thus similar to a 4" APO with focal reducer
      • autofocus works on bright stars
      • even at f/2.8, it is sharper than Borg 101EDf4 with focal reducer, but the Borg costs $US2490 while the lens, second hand, is around $US4300.  The IS model costs about $US6500 but is still not as sharp as the Borg refractor.
    • Canon 300mm f2.8 IS:
      • great for wildlife as only ~3kg
    • Canon 300mm f4 IS:
      • slightly less sharp but cheaper than the Borg 77ED
    • Canon 200mm f/2.8 II:
      • a favourite for many astrophotographers
      • slightly softer than a good APO but is slightly better color corrected
    • Olympus C8080ZW with 1.4x teleconverter = 196mm f/3.5:
      • cheap, great quality optics, no problems with dust getting on sensor
  • pros:
    • lighter, cheaper, more versatile and more readily available than APO refractors at focal lengths 300m and less.
    • auto-focus removes headaches
    • can be used for general photography
    • for most amateur astrophotography these are adequate as seeing conditions and guiding problems usually cause more problems than the L series optics, particularly when talking about focal lengths of 135-400mm. 
  • cons:
    • very few have the diffraction-limited optics of APO refractors (which are designed to work mainly at infinity focus), unless stopped down.
    • cannot use for visual use with eyepieces unless get special adapters, and even then you cannot use a diagonal with the lens as there is not enough back focus available (although medium format lenses do have sufficiently long back focus).  Hence, for visual use a scope is always much more versatile.
    • zoom lens are not usually good enough and an IS lens is not as good as a non-IS lens.
    • can only be used with one style of SLR camera mount
    • stopping down the lens, while producing sharper images and perhaps less aberrations, loses resolution and light-gathering as the effective aperture is reduced.
    • for critical use, even the best camera lenses are NOT adequate as:
      • limitations on back focus
      • inability to use modern filtering techniques & cannot use filters as too expensive to fit on the front aperture, although fortunately the non-IS supertelephotos take 48mm filters - exactly the same as 2" eyepieces take. 
      • cannot use cooled CCD cameras with them
      • lowered contrast because of the large number of optical elements than the optimal 3 or 4 that are in an APO refractor
      • focusing problems
      • temperature instability affecting focus
• refractors:
  • depending on the scope, it either has been designed for astrophotography or not, and it might need an additional field flattener for that, and it might not be diffraction-limited all the way to the corners of a full-sized sensor (NP-101 is not, while the FSQ-106 has been designed for medium format photography)
  • in general, APO refractors are less expensive than high-quality L series Canon lenses when considering focal lengths greater than 300mm
  • the shortest APO refractor is 300mm
  • the best refractors for astrophotography are those with true triplet or Petzval (4 element) designs are true APO and include:
    • TeleVue                127 mm aperture     660 mm focal length      f  5.2 $US6500
    • TeleVue                101 mm aperture     540 mm focal length      f  5.4 $US3600
    • Takashi  FSQ       106 mm aperture     530 mm focal length      f  5.0 $US3700
    • Vixen  APO          130 mm aperture     860 mm focal length      f  6.6 $US4400
    • TMB  APO          105 mm  aperture     650 mm focal length     f  6.2 $US4400
    • TMB   APO         130 mm aperture      780 mm focal length     f  6.0 $US5500
    • StellarVue APO    115 mm aperture      805 mm focal length    f  7.0 $US4000
    • StellarVue APO   130 mm aperture      780 mm focal length    f  6.0  $US6000
    • William  APO       110  mm aperture      715 mm focal length    f  6.5 $US3300
    • William Megrez 80 Triplet APO 80mm 480mm f/6.0
    • see william optics 
  • other options:
    • the new Takahashi  Epsilon   180 mm aperture    500 mm focal length   f  2.8   $US3900 covers 35mm full frame DSLRs well.
• hyperbolic astrographs:
  • Newtonian-configured reflector telescopes with hyperboloid rather than parabaloid primary mirror. 
  • 3 or 4 element Ross corrector element at the output end.
  • Imaging at f/3.3 is really great!  One can take centering shots and bin 4x4 or 6x6 giving another factor of 16 or 36 in sensitivity.  If one can't see a comet or galaxy or nebula in a 5 second exposure, it simply isn't there!  Even with an H-a filter that cuts down the transmitted flux by a factor of about 2000, one can usually see the target with a ten-second binned centering shot.
  • cons:
    • not useful for visual use as very few eyepieces will come to focus but excellent for prime focus astrophotography.
    • central obstruction from the diagonal secondary mirror reduces contrast.
    • thick spider arms result in square stars and fairly prominent diffraction spikes.
    • Collimation is exceedingly difficult, and most owners send the scopes to TNR if they have to be collimated.  If your new scope is out of collimation from the factory, you may have additional costs to get it right. Finally, Tak accessories are expensive. 
  • see http://dpersyk.home.att.net/
  • Takahashi E-160 160mm aperture, 532mm focal length hyperbolic astrograph with its f/3.3 and no perceivable CA from 420 nm to 1060 nm, meaning one can take luminance frames without killing off the NIR with a blocking filter. Has a 43 mm image circle and costs $US3300.
  • Takahashi 180ED hyperbolic astrograph that is 500mm focal length and f/2.8, $US3900.