Finger Lakes Instrumentation

CCD FAQ's and Myths

Myth: Small pixels are better than large pixels for achieving high resolution images.

Myth: 14-bit digitization is enough / 16-bit digitization is MORE than necessary.

Question: Do I need a shutter on my CCD camera?

Question: How important is cooling?

Question: How important is a low noise in a CCD camera design?

Question: Why does FLI offer a 2 inch nosepiece?

Question: What is involved in purging a CCD camera?

Fact: A primary factor in achieving high resolution images is to match your telescope's focal length to your CCD's pixels size. All things being equal, a long focal length telescope will give better results with large pixels and a short focal length telescope will give better results with small pixels.

Under steady skies, the best combination of telescope focal length and pixel size will result in 2 arc-seconds of sky per pixel.   For example, an ideal match for a 9 micron pixel is a telescope of 36" focal length (such as a 8" f4.5, 6" f6).  An 8 inch Schmidt-Cassegrin telescope, even with a focal reducer, is oversampled (too many pixels per star). Note: there are custom focal reducers that will get Schmidt-Cassegrin telescopes out of the oversampled range but they are very expensive and camera-specific.

The following graphs depict the "ideal" telescope focal length range for arrays with 9, 16, 20 and 24 micron pixels.  It's important to keep in mind that the values do not need to be met exactly.  Larger CCD arrays tend to be more forgiving in "focal length mismatches" than smaller CCD's.  A good example of a mismatch producing good results is this IMG1300 image of Hale-Bopp taken with a 200mm Olympus lens.

Fact: The optimum level of digitization required depends on the electron resolution of the CCD (derived from the RMS noise of the CCD's output stage) and the full well capacity of the CCD. For example, the Kodak KAF-0400 and KAF-1600 CCD's have a full well capacity of ~80,000 e's and an output stage resolution of ~15e's RMS. Therefore the required level of digitization is ~5000 counts. It would seem that 12-bits (4,096 counts) is sufficient, however there is always error in the resolution of the last bit and it is generally good practice to adjust the camera signal output to a resolution of at least 7 e's per count or one half the 15 e RMS output noise of the CCD. Therefore 13-bit digitization is better. Given that 13-bit A/D's are rare, a 14-bit camera will do the job and 16-bits is more than enough. However, most of the newer generation CCD's have larger pixel wells and some have lower output noise. For example, the new Kodak KAF-0261E has a pixel well of 500,000 e's (when configured in low gain/ high dynamic range mode) with an output specification of 30 e's RMS. This requires 15-bit resolution, therefore16-bit's is enough. As a final example, SITe's TK1024 CCD has a full well capacity of 150,000 e's with an output stage resolution of ~6 e's RMS. This requires a 16-bit A/D to handle the dynamic range of 50,000 with an optimum resolution setting of 3 e's per count!

Answer: A shutter is required when using a full frame CCD's such the Kodak KAF series. Otherwise, light falling on the CCD array following an exposure will cause smearing in the image. Interline CCD's such as the Kodak KAI371 have a "digital" shutter that can shift the image to a protected area in about 50,000th of a second. Some type of shutter is required for lunar and planetary work.

Answer: Cooling is extremely important to minimize noise due to dark current growth in the CCD. All CCD's experience gradual filling of the individual pixel cells due to dark current infusion. The rate of dark current growth is related to the temperature of the CCD and roughly doubles every 6 degrees C.  CCD cameras with undersized heat sinks and/or simple single stage cooling are unlikely to maintain a stable temperature below -5 C when the ambient air temperature is 20-25 C. Regardless of quantum efficiency, these camera's will perform poorly with long timed exposures of faint objects. A well designed camera with a proper size heat sink, properly mounted low vibration fan and a dual stage TEC will easily maintain a temperature of -30 C, lowering the dark current by a factor of 16 times (4 times less noise). Another important factor is the ability of the the camera to return exactly to the same temperature settings at some future date so that ones research is reproducible.

Dark Current Growth for the KAF CCD Arrays:

Answer: Low noise is important in resolving fine detail. Electronic noise is always present and "creeps" into the system while measuring the individual pixels during the CCD readout phase. Electronic noise can be minimized by using double correlated sampling or dual slope integration techniques. The best cameras utilize dual slope integration when measuring the value of each pixel element during the CCD readout phase. This is analogous to properly "zeroing" your measuring device each time before measuring the next pixel. Properly designed electronics which incorporate dual slope integration elements will certainly out perform simple sample & hold based correlated double sampling designs when digitizing to 16-bits.

Answer: A T-mount thread is 1.654” in diameter and has 33.866 threads per inch. The inside diameter is typically 1.615”. That is further reduced by the tube wall thickness making the actual clear aperture about 1.45” If you use a large CCD array and a fast telescope system, vignetting becomes unavoidable.

A large CCD array (such as the SITe TK1024) is 1 inch square and has a diagonal dimension of 1.41”. If you could place your eye at the very corner of the CCD and look at the primary mirror, you would see the edge of the T-thread cutting off the edge of the mirror (75% or more of the mirror would remain visible). The vignetting would not be a sharp shadow, but a gradual intensity falloff. In many applications it's effect is insignificant.

For the FLI IMG series cameras using the 1024 x 1024 SITe array, our color filter wheel, and a 2” adapter tube, we’ve determined that vignetting isn’t an issue with f/8 and slower scopes. For faster scopes we’ve introduced a 2 inch nosepiece to avoid any vignetting issues. This nosepiece uses the industry standard 2”-24 coupling thread, as found on the rear port of Meade, Celestron, and many other scopes.

Answer:  Finger Lakes Instrumentation has researched purging extensively and we have prepared a thorough explanation to this question (which is too lengthy to place here in its entirety).  Please follow this link to get the scoop on purging.  Note:  a left mouse click will open the Word file directly; a right mouse click will allow you to save the Word file directly to your hard drive.