Hi Dan,
MasterFlatDark (0.21 s)
April: mean: 153.1 DN, median: 152.6 DN
August: mean: 761.5 DN, median: 761.1 DN0
MasterDark (300 s)
April: mean: 154.2 DN, median: 152.9 DN
August: mean: 761.7 DN, median: 760.2 DN
The difference between the calibration frames taken in August and in April is about 608 DN. This compares favorably with the value (605 DN) that I evaluated with PixelMath (post #43).
My reasoning was and is: during capturing images in April, a low offset value was set in the capturing software or in the camera driver. In August, a much higher offset value was used. A difference of 608 DN is like day and night. It is all-important that the light frames and all calibration frames are captured with the same gain and offset value. According to the FITS header, the gain values (EGAIN = 1) are matching. Unfortunately, despite of the update of the SGP version there is no OFFSET keyword in the FITS header. Admittedly it is not only the SGP version that matters, the ASCOM platform version and the camera driver version have to support setting the offset value as well.
Which ASCOM platform version are you using?
Are you using the QHY ASCOM camera driver or native camera driver, and which version?
Your light frames of the Iris Nebula (captured on July 16, 2021) have a mean value of about 170 DN. Therefore they are matching the old calibration files of April. If you try to use the new MasterDark for calibration of these light frames, the difference (light - MD) becomes negative except for stars and brightest regions of the nebula. This result also suggests that the change of the offset value occurred after July 16, 2021).
If you want to check the compatibility of light frames and dark frames or flat-darks, please use the Statistics process for mean and median values and the histogram (HistogramTransformation) with an appropriate horizontal zoom (a value of about 40 might be right). If the mean value of a dark frame or flat-dark is greater than the mean value of a light frame, they are not compatible.
Bernd