Dont' be afraid of CCD

Integration times (II)

Integration times mainly concern deep sky astrophotography because in planetary photography, excepting Pluto, asteroïds and comets, all celestial bodies support snaphots.

Typical integration times to record a deep sky object on a CCD are usually set between 1 to 5 minutes operating the camera in 1x1 or better in 2x2 binning (merging two pixels reduces the resolution of 50% but increases the CCD sensitivity). This way using a 4" f/6.3 scope you can reach magnitude 18 in only 5 minutes of integration and magnitude 20 using a 14" at f/6.3 !

In worst cases, the integration can last 30 minutes on prime focus of 16" scopes to record faintest galaxies (or faint spectra) and 18 times more if you want to produce tricolors composites. Why 18 times ? Because using a colored filter the incident light passing through the filter is reduced and the CCD camera requests 6 times more exposure to get the same result than an unfiltered image. Then you must take at least 3 exposures, in red, green and blue channels (some add a fourth channel, the luminance with is a grayscale image). Add to this constraint CCD registration offsets that can occur with moving objects (Jupiter, etc). Therefore color CCD is another challenge reserved to the experienced amateur. This is however the best solution to records sky colors but at a time and financial cost. 

Another solution is using a "one shot" color CCD like the Astrovid Starlight Xpress MX5c sold $1300 (Astrovid also provides several other models : MX7c, and MX25 and many B/W CCDs and videocams). MX5c uses a color matrix filter over the pixels composed of "secondary color" dyes in a grid of Cyan, Magenta, Yellow and Green (like the popular CMYK technique). The filters are arranged in such a way that the Luminance component of the image can be extracted with high definition. Ideally the light reduction is only 33%. The resulting color image is fine but at the expense of lowering the resolution and longer exposures. Also, the MX5c is cooled, but not regulated. That means you have to take a dark frame at each ambient temperature level each time you use it. Of course you can get this dark frame by image processing too.

Today the only competitor of MX series is the "Deep Sky Imager" (DSI) from Meade released in 2004 and sold... $300 only. Time passing, the DSI comes with several automatic features not available on the Astrovid : auto exposure settings (optimization), auto-stretch (to bring out dim objects), automatic dark substraction (dark frames are stacked, averaged then substracted automatically), auto-align and stack images, magic eye focus (focusing), etc.

The overall resolution of a color CCD is lower than a B/W due to having two arrays together. Although this physical limitation the results are not bad at all but if you are really into critical observing then you have to choice a B/W CCD camera and acquire a color filters wheel. This  accessory will allow you to adjust the tricolor channels (RGB or CMY) integration times for the sensitivity of your chip, do astrometry and photometry works and even study any object using a single bandpass.

Common problems

CCD cameras performances and their defaults go two of a kind and is it in vain to hope getting good images if you do not control all factors that enter in our equation. So let's describe clearly problems that you can encounter in using a CCD camera and methods to avoid them or to reduce their effects

Focusing

Focusing problems and fine focus adjustments are emphasized using electronic camera with a telescope. Since the CCD field is very small, especially in planetary imaging in wich one works with an eyepiece projection (or a Barlow), a special attention must be given to stabilize image shift and temperature changes. Typically a shift is equal to the square of the amplifying factor of the secondary mirror, which is 5x on a f/10 SCT. Thus a mirror shift of only 0.001 mm creates a focus shift of 0.025 mm which is easily recorded by the CCD detector. Then the focal point positionning varies roughly as the square of the f/ratio. So using a scope at f/6.3 requires a focusing 2.3 times more accurate than at f/10. Therefore an zero-shift electrical focuser is highly recommended (like the JMI NGS-F for SCT's) as a locking mechanism to tighten the all focusing system when heavy accessories are attached.

RGB image of M81 and M82 recorded by Robert Gendler with a 12" f/9 Ritchey-Chrétien scope fiwed on AP1200 GEM equipped with a SBIG CCD camera. Image postprocessed.

Thermal noise

In an ideal CCD camera each pixel would give a brightness level of 0 when there is no light, and a value increasing perfectly linearly with increasing light until it became saturated. In addition, the reponse of every pixel would be identical. Actual CCD cameras are far to reach this ideal objective.

The first reason is that the electron count for a pixel is a function of the number of photons that strike it plus the number of electrons due to "thermal noise". In using electronic components, for lack of an ideal efficiency (100% of the inpout energy should be converted without loss), they dissipate some heat that generates a thermal noise that is reduced by half for every 5°C decrease in temperature. Like with photoamplifiers, a CCD sensor is thus very sensitive to infrared, ambiant temperature and temperature changes. For years detectors are cooled (the best are regulated) to around -30°C and keep the temperature changes below 0.5°C. But external infrared sources can potentially be seen by the detector although invisible to your eyes. So remove all bright accessories in the neighbor of the detector which emit infrared light or are not black anodize (or painted flat-black) which is also a good infrared absorber. Such sources are eyepieces holders, digital clocks, digital circles, dew heaters... 

To avoid any parasitic noise, it is advisable to place the CCD camera at ambiant temperature and to wait for about half an hour after to have switched on to take your first pictures.

Readout noise This degradation is caused by statistical errors in reading out the number of electrons per pixel (photosite). This relative sampling error decreases inversely with the square root of a pixel brightness level (DN factor). Then the final problem is that the pixels are not equal in their light sensitivity, with typical variations of 1-2 percent among the photosites in an array. Resolution and binning mode  Deep sky imagery with a CCD camera requires preferably larger pixel dimensions because large pixels simply collect more photons than smaller. This is what we call the binning, a mode that offers the possibility to sum signals from several adjacents columns and rows of pixels (binning 1x1, 2x2, 4x4, etc). But, drawback, in binning mode 2x2 the resolution drops of 50%, but the sensitivity is improved. Conversely, in lunar and planetary imaging the amateur searches for the higher resolution and does not need so much light sensitivity. In this case a smaller binning mode is preferable. At last, there is a tendency toward CCD cameras using larger array sizes. A 16-bit "depth" (65536 brightness levels) is preferable to 12 bits (4096 brightness levels). But all these aspects affect the size of the image, the download time and processing as well as the disk space need to save this file. As for the color, this is no more with 12 or 16 bits that we work, but at least 24 bits. In this case the file size is practically no more managable by amateurs standards and users of scanners know very well that problem. This is for this reason and for an image quality question too that the usage wants that amateurs work from LRGB images to get color composites instead of using color cameras.

M16 in Serpens. This is a composite of three RGB images recorded with three Astro-Physics scopes : a 180 mm f/7 EDF refractor, a Maksutov-Newton 235 mm f/4.3, and a Maksutov-Cassegrain 250 mm f/14.6. These scopes were equipped with CCDs SBIG ST-8 and ST-10 with a CFW-8 Color Filter Wheel, FLI MaxCam CM10-2E with a Custom Scientific Hydrogen Alpha Filter. Images were processed with CCDSoft, Mira Pro, Maxim DL/CCD, Sigma Beta, and Photoshop. Document published with the courtesy of Philip Perkins, Trent Kjell, and Roland Christen.

Mounting

Using such an accurate device as a CCD camera, you dramatize the stability of your instrument and it is mandatory to use a sturdy mount. Remind you always that the mount is more important than your optics; you will get easier fine pictures with an ordinary optics fixed on an excellent mount than the contrary.

From a pure photographical aspect, an Alt-Az mounting will display the problem of field rotation while you will track an object across the sky. Field derotators made for the popular SCT's are just adding one more mechanism and a new freedom axis to drive. The Alt-Az mounting can only be use for Lunar and planetary imaging since the exposure time are short (faster than 1/10^th sec or so). 

For all pictures of DSO's in high resolution there is no compromise. You have to use an equatorial mount. There is only one exception; scopes of 10" and larger can use an altazimutal  mount but have to use a derotater to avoid stars trails in the corner of the image. This heavy and cumbersome accessory is not recommanded for smaller scopes. 

Your mount must be well polar aligned and able to be driven by a step motor which error is limited to a couple of pixels. Any larger guiding irregularity in the drive gears will be recorded in the image. Therefore you quickly note that many advanced CCD users fixed their scope on robust and heavy german mount like Astro-Physics 1200GTO, Byers retrofit, Lichtenknecker Optics M145, Losmandy 300, Takahashi EM200, etc, mounts which weight is over 30 kg, electronically driven with PEC (Periodic Error Correction) and always fixed on a pier. In such working condition, the periodic errors can be reduced to a few seconds and exposures over 1 minute without guiding are then possible, but this system has its limits. For longer exposures you will need an accurate guiding system, typically a second and parallel CCD device which microcontroller corrects the alignment drift in real time.

Astro-Physics mounts At left 1200GTO suitable for to 8" scopes and larger. At right an AP 130 mm EDFS f/6 and ST-4 CCD mounted on a 600E GTO mount. Documents AP and Mike Cook.

Guiding 

As the size of a CCD pixel is very small, some 10 microns, and the complete chip up to 10 times smaller than a 35 mm film, you now understand why a sturdy equatorial mount and an accurate focusing system are required to get good results. 

Many accessories can help you guiding your instrument, beginning with a secondary CCD camera (SBIG ST-4 or ST-5) that, fixed on your off-axis guider system, drives alone your steps motors while you take your snapshots with your main camera. Optionally you can use an adaptive optics borrowed to professional astronomerx to correct effects caused by the turbulence.

But how to assembly all these parts together ? The best solution is using a special CCD instrument combining a camera and a detector for auto-guiding. SBIG ST-7EA (camera TC211 and KAF-0400 detector with antiblooming) uses such a technology. Astrovid Star 2000 (MX5 or MX9 camera) on the other side is not a full frame CCD as the Kodak device. It uses a special structure that splits each pixel in two independently readable halves, one integrating long exposures while the other selects a star in the field up to magnitude 11 for guiding the telescope. Whatever the solution you select, these combinations eliminate the need for an optional off-axis guider.

Michel Peyro's Hisis 22 camera. Pédro Ré's SBIG ST-7 camera.

If you don't use a CCD auto-guider, you will have to find a solution to aim your telescope on a specific object without removing the CCD camera focused with accuracy. How to do ? It is out of question to remove the camera and checking the field. The most elegant solution is to add ahead of the CCD a True Technology or similar flip-mirror system (Lumicon Giant Easy-Guider, Murnaghan or Meade flip mirror, etc). This unit has 2 outputs and allows you to use a powerful reticle eyepiece at 90° with the camera in order to ensure corrections on tracking from your remote controler. However this is not the best way of working with a CCD which, as you understand, requests much attention to get good results.

At left a True Technology 2" flip mirror; at right a wireless guided eyepiece from Orion T&B with its light source.

Once all these problems solved you are ready to picture your favorite celestial object. The good news is that after looking at some CCD images produced by amateurs and their equivalent taken by POSS (Palomar Observatory Sky Survey) or Hubble Space Telescope, subtle is the one that can say who is on first and who is on second ! Of course, getting closer, the bigger always wins, but do not forget that image processing can cover one's tracks... Let's compare for example M42 pictured with the HST and by Jason Ware using a RCX400 12" f/8...

Next chapter

The Image processing

Page 1 - 2 - 3 - 4 -