That's an interesting beast John - should be a really good match for your scope.
the specs are quite different from those published for conventional 16 bit CCD cameras, presumably to allow for high speed 12 bits operation. Looks to me like your strategy is most sensible: get the best dynamic range with DSOs, by working down in the 0-100 gain range. Because the ADU values will not be padded out to 16 bits, the signal will be stuff all (maybe a few ADU), but the read noise will also be minimal at around 1ADU, so you will still get a good image. Even though you cannot get fractions of an ADU in a sub, you can after stacking with a floating point operator, so quantisation noise should not be an issue if you have enough subs.
I think that the equation here will still apply for determining sub lengths
http://www.iceinspace.com.au/forum/s...d.php?t=117010
eg, at a gain setting of 100, the calculation suggests that you aim for sky ADU values of ~90 + whatever the bias is. When stacking subs with such low ADU values, you could consider using additive stacking rather than average, simply to make the data easier for the display algorithms. I don't understand the implications of the published specs sufficiently well to offer an optimal sub time, but as a general observation, a fairly low noise camera on a reasonably fast scope should require maybe 5-10 minutes per sub. Be very interesting to see what you find.