A few nights ago, August 25th, I finally had the first clear sky in a good while, so I went out and did some imaging.
After all the discussion we had and finding out that the blue channel is way behind the red and the green, I decided to activate the debayer function in KStars/EKOS, to see that the blue would at least have a healthy ADU count for the background mean.
The lens of choice was the 50mm, stopped down at f/5.6.
An exposure time of 90s gave me a mean ADU of 840, for the blue channel. Not optimal, if we consider our 10*RN2 calculations, but I was planning to do dithering and 90s was already quite a short exposure time and I didn't want to end up with hundreds of frames.
The same 90s exposure, undebayered, meant a mean ADU of 940. So, the blue channel is about 100 ADU behind.
What's really interesting is the following experiment. I decided to monitor how the ADU was varying as the target was climbing. I was going to be beside the laptop and my rig for the whole imaging session, like I always do, anyway. So, as the images were downloading, I kept an eye with FITS Viewer on the mean ADU.
The target was a widefield shot of the Heart and Soul Nebulae and part of Cassiopeia, approximate coordinates of the center of the frame: RA = 1h 58m, DEC = +61°32'. The camera was tilted almost 90° with respect to the mount saddle, so that the long side of the sensor could be parallel to an imaginary line going from the Soul Nebula to Caph.
I started imaging at 10:20 PM, the center of the frame was at an altitude of roughly 36°, the bottom (left of the frame when viewed in landscape orientation) of the frame was at 22°, the top (right, in landscape) of the frame was at 50°, so about a 28° variation from bottom to top. Bottom of the frame was still in the more light polluted region of the sky.
In those conditions, at the start of the session, as I previously said, 90s gave me a mean of 940.
As the pictures were coming, about 100s apart (counting download time and dithering), the mean ADU was decreasing. About 20 minutes later, the mean ADU dropped to 900.
I interrupted the scheduler and decided to up the exposure by 15s (guessed estimate) and I scheduled 10 shots. The first 105s exposure put me back to a mean of 950 ADU (overshot 940 by a little). The last one, 20 minutes later, was at 920 ADU.
I upped the exposure by 10s and took 10 more shots. The first 115s exposure had 945 ADU. The last one, about 20 minutes later, 920 ADU.
I went on like this, until clouds covered up the region I was imaging, with increments of 10s. First 125s exposure 945 ADU, last 920. First 135s exposure 935 ADU, last 920.
The target, at the time of the first 135s exposure (00:05 AM) reached an altitude, in the center of the frame, of about 46°. Bottom and top, 32° and 60° respectively.
So, between the first 90s exposure (10:20 PM) and the first 135s exposure (00:05 AM), 1h 45m later, the target rised about 10°, and the exposure had to increase by 50% to give me the same ADU with which I started (around 940).
Granted, the 50mm covers a lot of sky (about 28° wide, paired with the D5300), so the light pollution gradient still manages to get in the bottom (left, landscape) of the frame, even as the object climbs higher. But the difference, as the "higher" sky is darker, is meaningful enough to impose a 50% increase in exposure time, in less than a two hour period, to keep ADU as constant as possible. Had I used 90s for the whole session, the last shots would have probably been underexposed. I had to cut the session short, so I couldn't monitor any longer.
I suspect the narrower the field of view (longer focal lengths), the less this phenomenon is likely to be meaningful (ADU won't change as quickly), since the frame would only cover a handful of degrees. For example, the same camera paired with a 300mm covers 4.5° with the long side of the sensor versus almost 27° with the 50mm. So, within those 4.5° the light pollution gradient would have been more uniform, for longer periods of time.
It is still an interesting experiment - to me, at least - and it's worth knowing that the exposure could need adjusting from time to time, as the target rises in altitude, away from light pollution and toward darker regions of the (local) sky.