The Role of CCD Cameras in Amateur Astronomy

Presented at NCRAL 2000  May, 6  2000 by R. A. Greiner

Abstract: This presentation will include specific discussion of middle and advanced level CCD equipment commonly used by amateur astronomers for digital imaging. Both the camera equipment and the control software will be discussed. The specifications of the equipment required to accomplish specific guiding and imaging goals will be emphasized. Current available hardware and software and how they interrelate are of concern and will be detailed. The presentation will include discussion of complete setups, from camera to telescope to software, which need to be coordinated to successfully accomplish specific imaging goals. Speculation on the state of the current art and the future of digital imaging will be presented.

0. Amateur and Professional Cameras Compared

One needs to be careful when using the terms amateur and professional. In this paper, the distinction is made without any denigration of the amateur in any way. It is made only to narrow the field of concerns and applications regarding CCD cameras. Generally the discussion here will be for cameras that are applied to smaller telescopes, say under 16 inches and in a price range that is within the reach of a serious amateur.

While amateur and professional cameras have mainly similar goals. The amateur camera is most often used to make pretty pictures while the professional camera is used for scientific imaging, quantitative measurements, spectral studies and the like. That is not to say that many amateur astronomers do not do science of a high order. Indeed, this is especially possible with some recent cameras and accessories. There is no doubt that good science can be done with what is generally considered amateur equipment.

The main distinctions between professional cameras and amateur cameras are usually the size of the CCD chip, its freedom from defects and the amount of cooling provided to the chip. The distinction of size is not accidental at least partly because the field of view of the imager is determined by the size of the chip and the focal length of the telescope. The cost of the chip is directly related to its size and its perfection and is often a significant part of the total cost of the camera.

Thus we have modest chips, say 4 by 5 mm, in modestly cooled cameras that cost $ 2500. We have 12 by 12 mm chips in better cooled cameras for $ 7,000 or so. Both of these cameras have been popular with amateur imagers. These cameras are suitable for use with 10 inch Schmidt Cassegrains at $ 3000, 5 inch Apochromatic refractors at $15,000 or 15 inch Ritchey-Chretien Cassegrains at $ 25,000. Such telescopes are also popular with many amateur astronomers. Attention will be directed to cameras in the under $10,000 category. Their characteristics, features and design problems apply, generally, to all CCD cameras.

Typical amateur cameras considered here would be those with the Kodak 400 series chips, like the SBIG ST-7 and Meade 416XT, at the low end and those with the Kodak 1600 series chips, like the SBIG ST-8 and Meade 1616 in the middle. Higher end cameras using chips like the Kodak 1300 or various SITe chips are available from Apogee Instruments and Finger Lakes Instruments. They will be discussed briefly as well.

In addition to chip size and quality, a major factor in camera quality and usefulness is the depth and stability of the cooling provided for the chip. The cooling factor in camera design is both important and complex and so will be discussed at some length. It will be seen that a temperature differential of 20 to 30 degrees C. below ambient is relatively easy to attain. Steady temperatures of -40 degrees C. are much more difficult to attain. These require sophisticated cooling systems and elegant camera body design.

The electronics associated with the downloading of the information on the chip is also an important design factor, but as will become apparent is of secondary difficulty compared to other physical factors. That is to say, the design of the electronics is relatively easy and the quality of the electronics can be made high enough to be a minor problem in the overall camera design picture. Note in passing that ordinary photographic cameras are available for under $1000 with pixel arrays of 1500 by 1200. The large differential cost between these cameras and astro-imaging cameras is quite striking and seems to be increasing. This can be explained by the very different requirements of these cameras. The arguments justifying the cost difference must be differed to another time and place.

1. A. CCD Cameras for Imaging - Overview

Chip Size

Let us look specifically at some of the chips sizes available which are commonly used in CCD astro-imaging cameras. The size of the CCD chips is one of the most obvious factors which differentiate CCD chips.
 

This table is shown in graphical form below where the impact of chip size is even more striking.

Chip size determines the imaging field of view of the telescope and thus the actual angular size of the object which can be imaged with a single image. There are several other factors that are just as important. Among these are the size of the pixels, the quality of the chip and the spectral sensitivity.

Chip Quality

Chips also come in a variety of qualities ranging from 0 to 4. The rating depends on the number of point, cluster and column defects. In general, the 0 quality chips have no defects and are the most expensive. Astro-cameras usually will have 1st or 2nd quality chips but often the purchaser has a choice. The general feeling among amateur imagers seems to be that quality 2 chips are satisfactory for pretty pictures because the images can be repaired, if necessary, with digital image processing programs. Quite a few more perfectionist imagers go for the quality 1 chips. Few use 0 defect chips because of the very high cost.

Defect frequency grading is listed for chips of various qualities below.
 

In general the quality rating of the chips used in amateur imaging cameras is not widely advertised. However, a few companies, especially Apogee Instruments and Finger Lakes Instruments, list quality as a feature and have a price list which reflects the various quality levels available. The chips are steeply graded in price as quality improves.

As is clear, the chips vary widely in both size of the chip and the size of the pixels. The array of chips available from several manufacturers is way too long to discuss in detail here. The approximate characteristics of film are given simply so a comparison of size and resolution to typical color film can be made. As is clear, that most CCD chips in amateur cameras are relatively small compared to the 35 mm frame. Those with smaller pixel sizes provide resolution similar to film.

Field of View

The purpose of the telescope optics and the CCD camera together is of course to image a celestial object. With any camera, to photograph an object, the object must fit on the film. In the case of the CCD imager the objects must fit on the chip. For this discussion we are baring doing a mosaic because it is another level of complexity. Thus the combination of the size of the chip and the focal length of the telescope must be selected to allow the angular extent of the object to fit the chip dimensions. Since the angular size of desirable objects varies from a few arc-seconds to several degrees, it is not possible for one chip size and one focal length to do a good job of resolving every object. Some images will fill the chip, some will be very tiny and some will overrun the chip. Ideally, the image of the object would be large enough to more or less fill the chip and thus use the pixel resolution available to its best advantage.

Here are three examples of what one might get with three different telescope focal lengths but the same CCD chip size. The first image is rather pleasantly framed since it shows M51 in relation to the surrounding stars. The second image shows M51 filling the CCD chip frame fairly tightly. This framing is good as well and will give the best resolution since the image utilizes the pixels effectively. The final image will give even better resolution of the object, but it is clearly not esthetically satisfying.
 

The point of this discussion being that the individual imager must try to choose the right setup for the particular objects to be imaged.

For a given telescope focal length and camera, the angular object size that can be encompassed by the chip is usually adjusted by the use of a focal reducer or focal extender. Such reducers can be made in strengths of 0.50 or more commonly 0.68 without serious optical compromise. Focal expanders (Barlow lenses) are also available in similar expansion strengths. Focal extenders are required for planetary imaging, where the objects are smaller than 40 arc seconds. It is up to the user to decide by the choice of focal length of the primary instrument and auxiliary lenses which classes (sizes) of objects will most effectively fill the chip area..

With this information in hand, it can be determined what focal length and chip size will encompass the objects of interest. Since interesting deep space objects range from a few arc minutes to a degree or so, it is clear that with a given chip it is nearly a requirement to have telescopes of several focal lengths. If planetary imaging is included where the objects are only tens of arc-seconds It is even more clear that image magnification is required. Fortunately there are a variety of ways to get high magnification while maintaining high optical quality. Similarly, shorter focal lengths than a typical telescope has can easily be obtained with standard camera telephoto lenses. Some specific camera/telescope combinations will be discussed in the following material.

Resolution Within the Image

The resolution attainable is controlled by the size of the pixels. More specifically by the angular field of view of each individual pixel. With pixels of size 9 microns or less, the resolution is similar to that of photographic color film. The angular resolution required of each pixel should be compared to the quality of the seeing. It is not sensible to try to have each pixel image a fraction of an arc-second of the sky since seeing is rarely that good. On the other hand, if a pixel images too many arc-seconds, the angular resolution of the image will be poor compared to the seeing. For example, it is often possible to resolve the four stars in the double-double in Epsilon Lyra. When this is possible, the seeing is about 1 arc-second. Then to image these stars, a pixel resolution of better than 1 arc-second is required.

On the other hand, even when the double-double is quite clearly separated to the eye, it might not be perfectly steady. Since the exposure for the imaging is bound to be time integrated over at least several tens of seconds or even more, the apparently good seeing may well translate into time integrated resolution of several arc-seconds. Experience has shown that the best resolution of a star image will almost never be better than 2 to 2.5 arc- seconds. Allowing about 4 or 5 pixels to generate a nicely resolved star image suggests that one might set 0.5 to 1 pixels per arc-second as a suitable resolution target. Most experienced imagers would find this resolution to be somewhat too great because the price paid in the sensitivity of the total imaging system becomes too high.

The literature is replete with discussions, heated at that, of the number of pixels required to image a star. Theoretically, the star image is a point the size of the diffraction limit of the telescope optics. This is a finite size which must be captured by the pixels. Some argue, wrongly I think, that only two pixels in orthogonal directions are required to image a star by using some mysterious application of the Nyquist criterion. In fact such stinginess with pixels will result in square looking stars. It has been more sensibly argued by the best astrophotographers that 3 to 5 pixels across the diameter of the star image will give reasonably good looking (round) stars. When two few pixels make up the star image, it is said to be undersampled and will appear to be made up of individual squares. With more pixels than are necessary for a good image are used, the stars are oversampled. Oversampled images tend to have a nice photographic look, but precious pixels, field of view and total system sensitivity may be wasted as described later..

The following two images show the problems of under-sampling on the left and slight over-sampling on the right. The star image is approximately 3 arc-seconds in diameter. On the left the selection of 1 Pixel per arc-second gives a star image made up of about 3 pixels over its diameter. On the right the selection of 4 pixels per arc-second gives a much smoother image. The images are greatly enlarged to show the effect. Only two or three pixels per star makes the stars blocky and unpleasant looking. The slightly oversampled star on the other hand will look quite round. But there is a high price to be paid for this high resolution. The better resolved image requires 16 times the exposure.
 

When the seeing is less than very good, taking less resolution is usually a better choice. Many amateurs find that time integrated viewing is almost never better than 4 to 6 arc seconds. In that case a pixel resolution of 2 arc seconds might be a good compromize. An informed choice must be made by the person doing the imaging which balances field of view, resolution and overall sensitivity of the imaging system. This decision depends on a variety of factors, many of which are outlined in this paper.

An image of Jupiter emphasizes the point. On the left the resolution is about 0.3 arc second per pixel, in the center it is 1 arc second per pixel and on the right about 2 arc seconds per pixel. Jupiter was about 30 arc-seconds at the time. It is clear that for planetary imaging the optical gain of the telescope has to be increased greatly over that which works well for deep space objects. Fortunately this can be done since the planets that can be imaged are very bright and thus the exposure times short. It is good practice to sample generously so that the digital images have a more photographic quality. Again, it is not wise to significantly over-sample but to take whatever resolution the seeing at the moment will allow.
 

Cooling of the Chip

Cooling of any chip used for astronomical imaging is essential. The reason is that the brightness of the image is exceedingly low. As we all know, with our first look at a deep sky object, even through a large amateur telescope, most of the objects, so gorgeously pictured in astronomy books, appear to be not much more that gray smears. This is a terrible disappointment for most amateurs when they take their first look at a "faint fuzzy." But then we realize that the eye has difficulty seeing color and contrast at very low light levels. One of the major reason for using CCD chips is because the light level is so low that photographic film exhibits severe reciprocity failure. In the same way, at normal ambient temperatures, the electronic noise level in the pixels is so severe as to wipe out the image of faint deep sky objects. No amount of additional exposure time will overcome this problem. Fortunately, it is relatively easy to cool the chips so that the electronic noise becomes almost negligible. The noise decreases by half for each 6 degrees C. of cooling. This is shown in the graph. below.

Thus if the chip can be cooled to -20 or better to - 40 degrees C., the electronic noise becomes an almost negligible factor for amateur imaging applications. The more modestly cooled cameras, those with a single stage of TEC cooling, will attain a differential between the ambient temperature and the chip of 25 to 30 degrees C. But note that the ambient temperature is not the temperature of the air, but rather the temperature of the heat sink. The heat sink at maximum cooling effort will be easily 10 degrees C. above ambient air temperature. Thus a cooling effort of 30 degrees C. will give a chip temperature of only 0 degrees C. Also, this temperature will depend on the ambient temperature if the cooler is already exerting maximum effort. This amount of cooling is considered adequate by some imagers, but quite marginal by many imagers. In more sophisticated cameras, sufficient cooling is provided to hold a given low temperature, regardless of the ambient temperature. (within reason) Such cameras will generally cool to a fixed temperature of -40 degrees C.

These results are dramatically shown in the graphic below. The patches show pixel noise for a dark exposure of a fixed time. From the left to right are shown cooling of the chip to -10 C., -20 C., -30 C., and - 40 C.
 

The cost of top line cameras that cool to the lowest temperatures in the face of summertime ambient temperatures is quite high. This is caused by several factors including the cooler equipment and the more elaborate camera design that must be effected. Even with increased cost, cooling should be considered carefully when choosing a CCD camera.

Spectral Sensitivity of the Chip

A factor of great importance is the spectral sensitivity of the chip, more accurately, the quantum efficiency of the chip, with respect to wavelength. This property varies tremendously among chips as can be seen in the following graph.

The standard front illuminated chips all have sensitivities similar to the red line shown in the following graph. The new KAF 0400E chip and similarly the KAF 1600E have the response shown by the green curve. They have greatly increased sensitivity in the blue region of the spectrum and are thus are good choices for CCD cameras Almost all newer cameras use the type "E" chips. They have sensitivity of about 60% from 550 to 750 microns. Also shown is the sensitivity of the back illuminated chips made by SITe. These are much more sensitive than any of the front illuminated chips at all wavelengths. They are also significantly more costly because of the great difficulty in their manufacture.
 

The issue of the sensitivity of CCD chips cannot be over emphasized because both overall sensitivity increase and more uniform response over the spectrum greatly shortens the exposure times required. Exposure time is a vital issue because it greatly speeds up the entire imaging process and reduces the demands on guiding. This can be seen by briefly considering the procedures for CCD imaging in color.

Deep sky objects have a brightness per square arc-second of about 21st to 22nd magnitude. Thus, depending on the telescope speed, a typical monochrome CCD exposure time might be 10 to 30 minutes. In order to do color imaging, it is necessary to take three exposures through three color filters and possibly a luminance exposure as well. Additionally, the blue exposure with the original chips had to be 3 to 5 times as long as the red and green exposures. It could take a nominal time of at least 2 hours to get a complete set of images for a color image.

With the newer "blue" chips and especially the back illuminated chips, this time can be reduced to an hour orso. Many of these factors will be discussed in more detail in Section 2 of this paper.

1. B. CCD Cameras for Guiding

One of the very important uses of CCD cameras has been for automated guiding of telescopes. Anyone who has guided photographic exposures of 30 minutes or longer using a guide scope or an off axis guider (OAG) knows the true pain of astronomical photography. Stories abound of astronomers getting their eyelashes frozen to the eyepiece. Manual guiding was a tedious task. Hubble tells the story that he knew who was working the 100 incher just by the rhythm of the control relays clattering away, and in the dead of Winter feeling sorry for the person doing it. The electronic autoguider puts an end to this task and the CCD chip is paramount to making accurate automatic guiding possible.

This can now be done on any telescope at a reasonable price. The most popular of these guider cameras was and is the ST-4 from SBIG. It uses a tiny chip, the TC 211, which is only 2.6 by 2.6 mm and has a pixel array of 192 X 164. In the ST-4 the chip is cooled but not regulated and it has a high dark current by any imaging chip standards. Never the less, it is a good choice for guiding because it can be read out rapidly and thus provides fast correction information for on the fly guiding. To do the guiding, a guide star is centered on the chip through a separate guide scope or through an off axis guider, OAG. The CCD is read out several times per second and any movement of the star off of the original pixel set is detected immediately by the digital electronics and the necessary guidance commands are sent to the main telescope. Guiding to sub-pixel accuracy is possible.

Both Meade and SBIG make cameras suitable for guiding. These are priced in the range of $400 to $900. These guiders, with their very tiny chips, are generally not used by serious amateurs for imaging. Some of the cameras do have imaging capability. The SBIG ST-4 system is shown on the left and the Meade 216XT is shown on the right. The ST-4 has a separate box used to control the operation of the guider. It can also be computer controlled. The 216XT can be used without a computer through a rather arcane series of menus and button presses in the back. It can be better controlled with a computer via a serial connection. Adequate software is provided in both cases.
 

An exciting new guider camera has been announced by SBIG, the STV. They call it a Digital Integrating Video Camera and Autoguider. It has particularly interesting characteristics. It provides a real time video signal that can be viewed on any TV set.

Additionally, it can integrate the image for up to 600 seconds. Thus it acts like a small chip digital camera. It uses a TC237 chip which has 656 X 480 pixels of 7.4 micron size. This is a small chip, but has TEC cooling and good low dark current. It would not be considered suitable for high resolution imaging as are the other cameras under discussion here. But for a TV display, it is adequate to show images in real time. As an auto-guider it has excellent characteristics. It is significantly more sensitive than earlier guiders, has a full time and very fast guiding capability of up to 30 corrections per second and the control electronics are optimally tuned to do guiding. The unit can be remotely controlled from a computer through a serial connection so that complete guiding functions can be effected remotely. The chip image can be simultaneously viewed on any standard TV screen. While the STV is self contained, including an LCD display, it has control software that allows full function of the unit from a remote computer. This should make finding suitable guide stars and remotly controlling the guiding process much easier than ever before. The STV system is shown below.

The stand alone guiders described above can be used in several ways. For film or CCD imaging they can be used on a separate guide tube to guide the main OTA or they can be used on the main OAT to guide the telescope with the imaging camera on a piggy back arrangement. For greater precision and longer focal length imaging, the auto guider can be used on off axis guide equipment. (in place of the human eye and hand)

Obtaining rigidity between the main OTA and a separate guider scope is not a trivial matter but can be done. Similarly, setting up an OAG is not trivial. The choice of operating method is up to the personal taste of the operator and will depend on the final imaging goals and the operators expertise and experience. Guiding to a few arc-seconds is in any case not a trivial pursuit.
 

Directly below, is shown a typical separate guide scope on the left and the components of a typical OAG on the right. The guide scope is chown with a flip mirror for easy finding of a guide star and centering it on the guiding camera. The camera in this case is a Meade 216XT. The OAG is shown with some accessories that might be used including a large focal reducer lens. This OAG is large enough to be used for medium format film photography as well as with a CCD camera.
 

One cannot terminate this topic without mentioning the self guiding CCD cameras. SBIG has provided, for some years, high quality imaging cameras with built in guiding. The most notable cameras are the ST-7 and ST-8. They differ only in the size of the imaging chip. Each contains a second chip of small size which is used as a guider chip. This chip, since it is a part of the camera structure does not suffer from mechanical problems of rigidity or flexure or even mirror flop. This camera design has greatly increased the usability of CCD imaging for the amateur and has been a strong driving force in popularizing CCD imaging among amateurs. This camera should be seriously considered as a first CCD camera for anyone interested in CCD imaging because it solves so many of the fundamental and persistent problems of imaging and guiding. The ST-8 will be used as the camera of choice for the more specific example in Section 2.

Guiding with a CCD camera is a great advance over manual guiding. Such guiding is relatively easy to set up, guides to sub-pixel accuracy and best of all never gets tired or distracted. Automated guiding is now almost a necessary technique for any serious imager.

2. CCD Cameras for Digital Imaging - Specifics

Specific Example of SBIG ST-8 Camera and LX200 Telescope

At this point several cameras, their specifications and application will be considered so as to firm up the general overviews outlined above. Among the most popular CCD cameras used by amateurs are those made by the Santa Barbara Instrument Group, SBIG. They make a wide array of cameras for all aspects of amateur CCD imaging. Their most sensational cameras of the past several years have been the ST-7 and ST-8. In reality these two cameras are based on a single basic design with two standard chip sizes. Several recent modifications and very useful and innovative accessories have extended the life and range of application of these cameras. Since both the glories of use and the problems with CCD imaging are manifest in these cameras, they will be used as examples.

A unique and significant feature of the ST-7/8 cameras is that they are self guiding. (as described above) That is, they contain a second small CCD chip which is near to and par-focal with the main imaging chip. This chip is used for guiding and since it is firmly connected to the imaging chip, it gives guiding accuracy that is as perfect as is possible. Any flexure or motion of the optical tube which would move the image, is immediately corrected by the guide chip.

This is an advantage over other guiding techniques, especially for the amateur who may not be conversant with the many mechanical problems of establishing a solid, flexure free guiding system as is required with either an OAG or especially a separate guide telescope. That is not to say that the latter techniques are not effective but that they require special care and experience. A discussion of these mechanical factors would take another paper or two and so will not be pursued here.

The ST-7/8 cameras come with either the Kodak KAF 0400 or KAF 1600 chip. These two chips have recently been upgraded to the E series chips which have better blue sensitivity as described previously. The chip sizes, pixel array and pixel sizes are:  6.9 X 4.6 mm 765 X 510 pixels 9 micron square,  and 13.8 X 9.2 mm 1530 X 1020 pixels 9 micron square.

These are not large chips, but are of a size common in amateur cameras and are priced within the reach of most serious amateurs, at about $ 2500 to $7500.

The cameras come with either anti-blooming gate arrays (ABG) or standard, non-ABG arrays, of pixels. Which chip type to use has been discussed over and over among users. The trend seems to be toward the NABG chips since the ABG type has its sensitivity reduced by about 30%. The NABG chips have the disadvantage that some blooming takes place on bright stars. But with modern image processing techniques readily available, these artifacts can be removed effectively. Improved sensitivity is traded for a bit more digital image processing.

The Implications of Cooling the Camera

Both cameras come with a standard single stage thermal electric cooler, TEC. This design is capable of cooling the chip to a differential temperature of about 25 to 30 degrees C. below the ambient temperature of the heat sink. Of course the heat sink has to be well above ambient in order to dissipate the total heat load of the camera. This might easily be 10 degrees C. When working near the maximum differential temperature, the chip temperature becomes dependent on the heat sink temperature. This situation for cooling has to be considered modest and in some cases inadequate. It might be fine in the Arctic but not too good in the Mojhavi. After some extended discussion of the cooling problem last year, SBIG made available a retrofit cooling modification which includes a second TEC stage and a fluid cooled heat exchanger. It is now possible to cool the chip to temperatures which are limited only by other factors in the design of the camera such as the chip chamber thermal characteristics, the design of the body of the camera and the design of the electronics.

Without going into excessive detail, the ultimate chip temperature is limited by frosting of the chip, dewing of the camera window, dewing of the internal structure and electronic circuit boards and ultimately dewing of the entire camera body to the point where it becomes dripping wet. All of these elements depend on the atmospheric conditions, especially the ambient dew point temperature.

The ST-8, with the added cooling option, is shown on the back of a Meade 10" LX200 telescope below. The optics have a focal length of 1600 mm and a focal ratio of f 6.3. The liquid cooling tubes are apparent. The camera is coupled to a JMI focuser. Two inch tubes are used to insure minimum flexibility of the mounting. The focuser might be considered excess baggage, but it is required since the LX mirror is locked down to prevent mirror flop. It seems to be common practice to use a good linear focuser instead of depending on focusing with the mirror on SCT type telescopes.

Cooling directly affects the dark current which is an important factor in determining the signal to noise ratio, S/N, in the image. For the 0400 and 1600 series chips, which have 9 micron square pixels, used in this example, the dark current is about 1 e per second per pixel at a temperature of - 10 degrees C. (e is the charge on an electron) There is some variation of these specifications in the literature. The values used here are conservative. If a S/N ratio of 30 is desired with a half full pixel well, the dark current accumulation should be less than about 1000 e. This allows for an exposure of 1000 seconds. (16 minutes) This is an adequate exposure time to capture a monochrome image of many deep space objects. But when color filters are used, only a slice of the spectrum is allowed to reach the chip. Additionally, the infra-red part of the spectrum is eliminated entirely. This greatly lengthens the required exposure. Exposures of 30 to 60 minutes are not unusual. Then the dark current produces an accumulated charge of 1800 e. When compared to a half filled pixel well of 30,000 e this dark current is not negligible. Thus, dark current noise is an issue until the chip temperature gets well below -10 degrees C.

Chip Size and Field of View

At prime focus, a 1600 mm focal length and ST-8 (1600 size) chip gives a field of view of 19.7 X 29.5 arc-minutes and a resolution of 1.15 arc-seconds. With a focal reducer or a focal extender these values can be changed significantly. The following table shows a variety of possibilities using a basic 10" f 6.3 telescope. The last column is for a typical star image of 2.5 arc-seconds time integrated size, which is typical under very good seeing conditions. A time integrated star size of 4 to 5 arc-seconds might be more typical of average to good seeing.
 

The fields of view can be directly related to the classes of deep sky objects that can be squeezed on the chip by considering the objects listed in the "List of Nice Objects" which is summarized here.
 

Clearly no single focal length is good for all objects since they range so greatly in size. But with a wise choice of the prime focal length and with good basic optical quality, the focal length can be adjusted optically to cover a broad range of objects with good framing and good resolution.

For planetary imaging a 3X extender could be used to give a satisfactory size image for the larger planets. For example with a 3X extender: Jupiter would have an image diameter of about 1 mm or 100 pixels. Saturn and Mars are half that. Venus varies greatly in size but on the average is in the same range. Thus it is apparent that a camera and telescope selected for deep sky imaging is not a good choice for planetary imaging.

Focal lengths over 10,000 mm would be a better choice. Since not too many amateurs have several telescopes, though some do, it is more likely that most persons will use optical attachments to change the effective focal length of their instruments. Optical elements like the focal reducer/field flattener will generally allow for shortening the focal length by a factor of 0.63 or even 0.5. These devices can be effective with smaller CCD chips, but vignetting and optical distortion are apparent with larger chips. Focal extenders (Barlow lenses), on the other hand, work well up to magnifications as high as 5 times. The extender lens expands the circle of illumination and so there is no problem covering the larger chips. But, this means that the light is also spread out more and the sensitivity of the overall imaging system is greatly reduced.

Taking these matters into account, it is wise to choose a telescope with a basic focal length that is shorter rather than longer. That is, a focal length of between 1000 and 2000 mm is about right. Many 4 to 6 inch refractors fall into this class as do the faster SCTs in the 8 to 12 inch size. With a telescope on the shorter end of this range but with a fairly low focal ratio, the basic optics can be shortened by a factor of 0.5 and extended by a factor of 5 to give coverage of a large selection of deep space objects as well as bringing planetary imaging within reach. The 10 inch f 6.3 is a good choice in many ways.

If one wants to do wider field imaging, a very short telescope is necessary. At some point, it is more effective to use a good telephoto lens such as one of the 300, 400 or 800 mm f 2.8 lenses commonly available for 35 mm cameras. These lenses cover a large field with high resolution and can be easily mounted on a larger telescope in a piggy back arrangement as show. Shown below is a Canon 500 mm f 2.8 lens piggy backed on a 10" f 6.3 LX200 telescope. The telescope does all the work of guiding with a 216XT camera mounted at prime focus. The lens is shown with a film camera attached but a CCD camera could be used just as well.

Next are photographs of two typical CCD imaging setups. It starts to look like a lot of "stuff" on the back of the telescope and it really is. On the right is shown an SBIG ST-8 with cooling tubes mounted on a JMI focuser. In this case an EXT is used for guiding with a Meade 216XT CCD camera attached. The ST-8 could also be used in its self guiding mode. At the right is shown the Lumicon giant OAG with the Meade 216XT CCD camera used as a guider. The imaging in this case is being done with a Canon 35 mm camera. The camera has a high powered magnifier attached for precision focusing.
 

The SBIG cameras are excellent cameras and would probably satisfy many if not most amateurs as their only cameras. They work well, have adequate software and are reliable. They have excellent upgradability. Additional cooling is now available and the chips can be upgraded. There are a number of very fine accessories such as a color filter wheel, an adaptive optics (OA) unit and now even a spectrograph.

Other Top Line Cameras

A number of top line cameras are made by both Apogee Instruments and Finger Lakes Instruments. They have in some cases better thermal design than the SBIG cameras. They are available with a wide variety of larger and more sensitive CCD chips, including Kodak, Thompson and SITe chips. These companies also make cameras comparable to the SBIG cameras, at competitive prices. They have in the past concentrated on making cameras with larger chips which would qualify as high end amateur cameras or low end professional cameras. These cameras are typically in the $7000 to $12,000 range depending on the chip quality selected.

The most attractive Apogee Instruments camera for the amateur is probably the AP-7p. It has an array of 512 X 512 pixels which are 24 microns square. The chip is a SITe back illuminated chip and thus provides the high sensitivity described above. It has a complete array of attachments including, a color filter wheel and a liquid cooling accessory. The price at this moment is somewhat less than the ST-8E camera. Another of their cameras carries the KAF 1300 chip which has been mentioned above. It is a larger chip, 16.4 by 20.4 mm with 16 micron pixels in an array that is 1024 by 1280 pixels. It is mounted in a very high quality camera. Many CCD users consider Apogee Instruments cameras to be the gold standard for both amateur and professional applications.

Apogee Instruments has a very detailed web site which includes specifications about their cameras but also a large series of articles and papers which describe characteristics of CCD chips in general. It covers most of the topics discussed in this article and a few more specific to CCD chip characteristics. Reading their site provides a whole course in CCD camera properties. It is at www.apogee-ccd.com.

Just a few weeks ago, Apogee Instruments made an striking announcement about a new series of lower priced cameras designed specifically to meet the needs of the amateur imager. It is called the Lisaa system (Low cost Imaging System for Amateur Astronomy) Because of the newness of this system, it has not yet been evaluated by the imaging community. It certainly deserves investigation and consideration.

Finger Lakes Instruments also makes cameras of high quality which are used by many professionals. They make a complete line of cameras using both the Kodak and the SITe chips. The camera is well designed, has excellent cooling and of course all of the accessories one would expect. They too have recently announced a very attractive camera, the IMG1024S. This has a SITe back illuminated chip with 1024 by 1024 pixels. The size of the pixels is 24 by 24 microns. This is a very large chip, 24.6 by 24.6 mm which gives it an area of about 80% of a 35 mm frame. It is priced at only $ 7500 and for good reason is called the Dream Machine. Finger Lakes Instruments also has an excellent web site with extensive information about CCD imaging in general. It is at www.fli-cam.com.

A photo of one sample of each of their cameras is shown below. The Apogee camera sports with a regular camera lens and the FLI camera sits naked with its cover off. It is interesting and promising that several top line camera manufacturers are going after the amateur market.
 

Exposure Times and Procedures

One of the raves about CCD cameras is that they are so wonderfully sensitive that they can take images in just a few minutes where film takes an hour. This is true for simple monochrome images. For monochrome images the entire spectrum of the star is captured including the infra-red portion of the spectrum. The CCD is especially sensitive to infra-red light and has no reciprocity failure. Thus many deep space objects can be captured in 5 to 15 minutes. But, and it is a big but when one wants to do color imaging, it is another story entirely. While one can do many objects in 30 to 60 minutes with film, it takes as long or longer to do these same objects with three or four color CCD imaging.

A principal reason is that three separate exposures have to be taken through filters that each pass only a small fraction of the spectrum. Additionally for realistic color the infra-red is thrown out completely. So exposures that give good signal to noise ratios are often 30 to 40 minutes or longer for each filter. Additionally, dark frames and flat fields have to be taken. All of these images and auxiliary frames are then combined in a digital image processing program. This latter step may take another hour or more.

Galaxies and nebula are typically 21st to 22nd magnitude per square arc-second. For a given scope using a focal reducer increases the light flux on each pixel and thus reduces the required exposure time. Binning the pixels also increases the sensitivity of the chip but reduces the resolution. For this reason, some imagers bin the color images and obtain the resolution in a monochrome image. The monochrome image is then combined with the three color images to create the final image. This is known as the LRGB color imaging technique. Some imagers consider it less satisfactory that the normal RGB technique

Most of the well known imagers use scopes that are, or are reduced to an f 6.3 focal ratio. While some of these telescopes are Schmidt Cassgrain type, many are also refractors. But because it is important to get as much light flux as possible to the chip, an amateur might well consider a 16 inch f 4.0 Newtonian with a good coma corrector. Such an instrument would give shortened exposures and good results for larger deep sky objects.

Most deep sky objects are only a few percent brighter that the sky. This means that it is necessary to distinguish very low contrast objects. The CCD camera is very good at this since the image can be selectively stretched about the average value of the image. This differentiates small brightness differences and creates quite good images where almost nothing existed. Ultimately, the sky brightness limits the exposure and the ability to image very low contrast objects.

Color and Color Filters

Significant additional equipment is required for doing color imaging. It all starts with color filters and color filter wheels. To help automate the process as much as possible, camera manufacturers provide accessory color filter wheels which hold the required filters. They are computer controlled so that the filters can be shifted and a succession of exposures taken. This whole process is generally automated under control of a computer program. The basic filter wheels provided by camera manufacturers look similar. An example is show below on the left. Another form of filter attachment is the Optec filter slider. It is shown on the right. It has some advantages in that it uses larger filters and they are further away from the chip. This allows them to cover larger chips and prevents any dust from casting a sharp shadow on the chip.
 

Filter sets come in many, many varieties. There is constant argument over the best filter set to use for color imaging. Each imager has a favorite set of filters. Only a sample of the filter spectra are given here to show their general nature. A long study of filtering techniques must be undertaken to become expert in this area. The filters shown are for normal RGB imaging at the top left. This set of filters has infra-red rejection built into it. Filters without this characteristic come back up in the infra-red region of the spectrum and would need to have a rejection filter added. If no rejection filter were used each color image would be contaminated by the strong infra-red light from the object.

As can be seen in the other views, there are filters made for capturing the image in the H alpha emission lines and filter sets to match certain professional standards. Many imagers use sets of the standard Wratten color filters. There are at least a half dozen sets described on my web site. It is necessary to study the color imaging literature at length so as to make and informed decision about filter selection.
 

There seems to be no absolutely correct filter set. The choice depends on whether the imager wants to make pretty pictures, do scientific imaging or do photometry. Generally manipulation of the digital image afterward using any number of image processing programs alters the content of the image significantly in any case.

4. Processing Digital Images

Digital processing of the images is a forgone conclusion. The raw digital images are usually full of problems ranging from bloomed stars to streaks and non-uniform illumination to dark noise. These artifacts can and must be removed prior to full scale image processing.

Often multiple images will be taken and added to improve overall signal to noise ratios. This can amount to combining two or three exposures to combining dozens of short exposures. The latter technique is often applied when the tracking ability of the telescope is deficient. It is then called track and accumulate imaging. Sometimes, shorter exposures are taken when there is a unusually bright star in the field. This tactic prevents the bright star from overloading a set of pixels and causing blooming on the chip.

Dark frame subtraction is usually done to remove bright pixels and the basic noise associated with the dark current in the chip. If the dark current is kept small by deep cooling of the chip. Sometimes a bias count is removed from all pixel readings to do away with any residual noise in the chip and its readout electronics. This usually amounts to only 100 counts or so.

A much more persistent problem arises from one of the great advantages of the CCD and the processing of the digital image. The digital image can be stretched over a relatively small count range so as to exaggerate or distinguish very low contrast objects from sky background. This makes it possible to "see" very faint and low contrast objects with the CCD camera. In fact, this is a principle advantage of the CCD camera. However, this also has a consequence which arises this way. If there is any non-uniformity of the illumination of the chip caused by the optical system of the telescope, including things like dust, fingerprints and smears that affect the image focused on the chip, they too will show clearly. Sensitivity variations of the individual chips also contribute to a non-uniform image. To control these problems, a so called "flat field" exposure is taken of a uniformly illuminated "gray" field. Any optical non-uniformity across the chip area will be detedted. This "flat" is than multiplied into the image and a corrected image results which has the non-uniformity removed.

This is a good idea in principle, but the flat field must be taken with exactly the same optical setup as the image. That includes the same vignetting and dust and at the same temperature of the chip as the desired image. It turns out to not at all trivial to take these flat fields. Ideally, different flat fields are taken to match each exposure and each filter. This greatly lengthens the time required to get a set of exposures that can be used for processing the final image. Absolutely clean optical systems, freedom from all dust and absence of vignetting are important so as to reduce the task of flat fielding.

So, one of the great advantages of CCD cameras, great sensitivity and enhancement of contrast is also one of the headaches of the process.

Software for Capturing and Processing Digital Images

There are as many types and qualities of software as cameras and then some. Only a very few will be mentioned here. These fall into three basic categories. There is considerable overlap among them. The three basic types are planetarium programs, telescope control programs and camera control programs. For some time these programs were rather exclusive. But in the past few years the camera manufacturers have seen that releasing their camera control codes was an advantage since then others would write programs that included their cameras.

Among planetarium programs, The Sky from Software Bisque and Starry Night from Sienna Software are well known. These programs contain the data bases and the command codes required to point the telescope. The control program code for the Meade LX series of telescopes has become an almost de-facto standard. Many telescope manufacturers make mounts that use this code structure.

The most useful planetarium programs also contain the telescope control code and thus overlap the second category of programs, telescope control programs. There are however programs that are designed to control the telescope and use other planetarium programs as the source of their pointing information. One which is becoming well known is Astronomer's Control Panel from DC-3 Dreams. This type of program is more versatile than others since it can import a great variety of data bases for different purposes. This program, ACP, is especially interesting since it is part of and compatible with the new standards which are being established, by the ASCOM Initiative (Astronomy Common Object Model Initiative) for general control of everything related to automated astronomy including telescopes, cameras and domes. The purpose is to bring scientific digital control to the amateur community. This group now encompasses a considerable number of the major suppliers of amateur software.

Each camera manufacturer will have programs which control the actual taking of images. These usually include automated taking of dark frames and flat fields. Some also include programs for automated guiding, centering of objects, focusing the telescope and even doing mosaics of larger objects. SBIG, Meade, Apogee and others have camera control programs. Providers external to the camera manufacturers provide such programs as ACP, MaxIm and Sky to mention just a few which overlap still more of the process. That is they include telescope control. camera control and varying levels of image processing.

In the image processing area the best known programs are MaxIm and Mira. Though other programs are used. Very few experienced digital imagers use only one set of programs. While astronomical images have special properties since they are bright spots on a black background or very low contrast images, some persons use photoshop-like programs to manipulate their images. In general the specialized astronomy programs are more suitable for digital astro-images since they are designed to handle the appropriate file types and do specialize digital processing functions on the images.

Of considerable interest recently has been an attempt to integrate all categories of programs into a unified system which enables the compatible programs to talk to one another easily. This effort, carried out under the auspices of ASCOM, results in the ability to script large control scenarios for total automation of what are otherwise tedious tasks. A discussion of these new directions can be found at: www.dc3.com.

6. Speculation On the Future of Digital Imaging

This is a great time to be interested in astronomical digital imaging. A number of advances are on the near horizon.

Digital photography, ordinary photography that is, is just coming into its own. More and more digital cameras in both still and video formats are appearing, almost daily. This means the chip manufacturers are revving up their ability to make better and better chips. This has already affected the prices of astro-cameras, with larger and better chips dropping in price by factors of two in the last year.

Astro-imaging is also coming into wider use. This has resulted in several digital camera companies recognizing a profit center and getting into the amateur camera business. This has and will continue to result in better cameras with more features and at a lower price.

Several entirely new concepts mentioned above will continue to pur the pressure on camera companies to meet the needs of amateurs with a broader spectrum of cameras. Camera prices will drop significantly and/or twice the camera will become available for the same price. Rumors abound about a $1000 camera coming out soon with capabilities similar to one that last year might have cost $ 3000.

The really big move will be to full color astro-cameras. These will come about with the improvement of chips used in video cameras to the point that they are suitable for astro- imaging. Indeed, it is a great time to be interested in astronomy and digital astro-imaging.

References:

Most of the information in this paper can be found on the web at a variety of commercial sites. These sites are listed in addition to a series of books because they are really the best way to get the most up to date information in this rapidly changing field.

CCD Astronomy, Christian Buil, William-Bell 1991

This is about the best of the early books about CCD cameras and image processing. It covers everything from building a camera (out of date) to discussing the principles of imaging and image processing.

The Art and Science of CCD Astronomy, David Ratledge (ed.) Springer 1997

This book is a collection of CCD imaging experiences related by a dozen amateur imagers. It is rather inspirational to see what some amateurs have accomplished. The book is quite over priced in my opinion. $40 for 150 paperbound pages. I found most of the articles of some value but a bit disjointed because the several authors have not been edited together carefully. There are several interesting appendices. Probably a book to get, for the tidbits scattered about in it, but then again somewhat of a disappointment.

A Practical Guide to CCD Astronomy, Patrick Martinez and Alain Klotz Cambridge University Press 1998. This is a recent book with detailed discussion of the design of CCD imagers. The imagers discussed seem to me to be somewhat arcane. There is also a long discussion of image processing. There is a certain sense of authority in the writing. The images used as examples are quite terrible. The book gives the overall feeling of and old book in a modern binding. I found it useful but not exciting to read. Lessons given on the SBIGUSER group are much more up to date and applicable to use of the SBIG ST imagers.

Astronomical Image Processing, Richard Berry, William-Bell 1991. A nice pamphlet on the basics of image processing. This material is most readable and has some nice examples. Additionally it has some software for the PC. Unfortunately, it is well out of date.

Electronic Imaging in Astronomy, Ian S. McLean, Wiley 1997. This is an astonishing collection of about everything you could want to know about electronics as applied to astronomical imaging, photometry and measuring techniques. Written for the professional. The book is essential to those who want to know much about the practice and theory that underlies electronic astronomy. Still after a thorough reading I found that it did not help me much with understanding how to use my current equipment to do imaging. I felt that is was useful to know all of this stuff but I still am looking for a book that helps with the day today problems of imaging with a modern CCD imager.

Web Sites of Interest:

Santa Barbara Instruments Group: www.sbig.com
Meade Instruments: www.meade.com
Finger Lakes Instruments: www.fli-cam,com
Apogee Instruments: www.apogee-ccd.com
Astronomers Control Panel: www.dc3.com
MaxIm control and signal processing software: www.cyanogen.com
Mira image processing software: www.axres.com
Optec optical attachments: www.optecinc.com
The Sky planetarium and control: www.bisque.com

In addition to these references, a great deal can be learned from the manuals that come with CCD cameras and the software discussed in this article. Many of these are downloadable from the manufacturers web sites.

Top of page