So what I did, seemingly back in the distant depths of time, was to take a bunch of bias frames with different offsets. This showed me that at the default values in the presets weren't to my liking...the histogram was all bunched up at the left hand edge (zero end).
What I was looking to see was the Gaussian distribution of the bias frame to pull its extremity away from the zero end, such that the bias/dark frames would subtract from my light frames correctly.
I did this for several different gain values...notably gain 75, 139 and 200 on my ASI1600. In the end, as much out of laziness/convenience, I settled on the offset value of the highest gain I would practically use, where I felt I could calibrate accurately. This worked out to be (drum roll) an offset of 50. This, coincidentally, is the value ZWO set by default in the ASCOM driver of the ASI1600 Pro and later models.
Using a little magic maths in Shiraz/Ray's thread (
http://www.iceinspace.com.au/forum/s...hlight=asi1600 ) for exposing , I "calculated" target ADUs for different gain values I might use. Due to the relationship of gain/ADU and read noise, the target ADU increases with gain. I'm sure there was a formula in there somewhere but I don't see it right now. It was something like:
Target ADU = bias + (read_noise^2 * 10) / gain
[Note: ensure all your values used are of the same word length, i.e. 12-bit or 16-bit]
I haven't tried it with my ASI462, but I've used that "technique" with every other camera I've used subsequently, and it's my guiding light for achieving sky-limited exposures. Keeping exposure times to a minimum helps maximise the dynamic range, but also compensates for mainstream mount performance
So I think, to answer Marc's original question, the maths is the same as with a CCD, only the numbers are different