Alistair. Me too. I think I have a grip and...
You know these sites
http://starizona.com/acb/ccd/advtheoryexp.aspx
http://www.samirkharusi.net/sub-exposures.html
I get from this... basically speaking
Background flux is less at a dark site and object flux is greater - e/sec
Less integration time, but longer subs to increase background adu - is this correct?
Conversely, background flux is greater with light polluted sky, object flux less.
More integration time required and shorter subs because backround/skyglow flux is greater.
What is the noise if background adu is not adequate? Aside from fixed pattern, thermal noise?
Next installment... read noise, which is considered differently CMOS vs CCD.