When buying a CCD camera (yes, I am still using a CCD)... the focus is on stats like full-well-capacity, quantum efficiency, dark current, readout noise, dynamic range, ADC count (16 bit or less), and linearity.
The CCD chip is what it is. The chip is as good as e2v (or other manufacuturer) provides for the particular model and class you buy.
However, beyond all that is the ADU hardware and software. That can determine how accurate and stable your camera is in reporting the actual flux of a star to your computer. I have not found anywhere that dicusses ADU stats that should be considered when evaluating the stability of the ADU. This information does not seem to be publicly available. One has to ask the manufacturer for it (if they even know).
Does anyone have a recommendation on the ADU subsystem and what separates a good ADU from a bad one? Any stats that should be considered? Any architecture that is better or worse than others? etc. How does one test an ADU subsystem? And what benchmarks should be used to guage a good ADU from a bad one?
Yes, no ADC is perfect. It is common knowledge, but unlikely to make a difference in variable star measurements.
I like 16 bit ADCs. That and speed is about the only specs I look at for cameras.
But, if you really want to get into ADCs, you could do your master's thesis on this stuff:
So you can start with the basics:
This one has a nice Guide for the Troubled:
Here is another 1100 pages, with Chapter 5 covering testing of converters: