Technical content, feel free to ignore...
I'll suggest, especially for EAA deep-space dim objects use cases, ....
1. make sure it has a very good "Live" processing software with features beneficial for astro use
2. look for the detail spec on its A/D converter resolution, if possible, get its AFE's (A/D converter and other components) part number
e.g., AD9949A is an AFE (analog front-end) that has a 12-bit A/D converter in it (this model is popular in high end analog video cam as well as low end "image head")
e.g., the same firm's AD9824 is a 14-bit AFE typically used in a DSLR or a mid-range astro "image head"
e.g., same firm (analog devices) ADDI7013 is an AFE that has 16-bit AD converter (of course it's for an even higher end) <-- more costlier part
Please note some mfgs use these terms which are not good way to describe the real spec:
- 8 bit output (sensor has 12-bit A/D) <-- fuzzy uncertain spec, many CMOS image sensors can operate in 8, 10, 12 bit A/D mode, depend on setting
- 16-bit data format (no mentioning on A/D converter's actual spec.) <-- it is possible to fit 12-bit digital data in a 16-bit data word
3. If the AFE part number is known, one can always look up its capability on analog signal gain (some model can be set very, very high)
<-- the built-in internal gain-stages (in the form of CDS gain, pre-amp gain, and VGA amp gain) are usually much cleaner than the ad-hoc bolt-on LNA
Some AFE parts have very high analog signal gain capability prior to its A/D converter
4. Why I mentioned (2) and (3)? It is about the topic on how to get the CCD signal extracted in a digital form while keep the "Noise" components at bay.
If the astroCCD image head mfgs know how to squeeze out the best possible CCD analog output signal (and contain the N part of distraction at the same time),
select the most optimal AFE part (more A/D converter bits and higher possible gain-stages), and
let the gain-stage GAIN setting be user configurable,
then it can be a winning formula.
P.S. In many scenarios, A/D converter resolution is not the bottleneck, the CCD analog signal's own S/N (too much noise) is the problem.
Theoratically, a 16-bit resolution can represent (6.02N + 1.76 db <-- N=16) 98.08 dB while most of CCD cannot be easily squeeze out such a high quality signal source in the first place.
Edited by ccs_hello, 24 May 2015 - 05:13 PM.