Our STX16803 cameras running with SkyX or the Universal Driver in Linux reach saturation at about 46000 digital counts. Since the full well of the CCD is ~100,000, is the actual gain set at >2 e/ADU? SkyX writes 1.31 into the FITS header. If the gain is really 1.31, why does the camera saturate in ADU short of 16-bits full dynamic range?
The factory measurement agrees with 1.31. Can you post a couple of FITS images? Bias frame and light frame showing saturation. You may have to dropbox them due to size.
Here are two in the original SkyX format unprocessed. The target image has saturated stars. We see the same behavior with an older STX camera running with the Universal Driver under Linux. https://www.astro.louisville.edu/john_kielkopf/documents/sbig_tests/
Using the bias, dark, and flat frames you provided, I have manually calculated the gain at 1.4 e-/ADU. That is pretty consistent with the automated factory measurement. Your camera is saturating at about 48,000 ADU, which given the 1000 ADU bias level comes out to a well depth of 65,800 e-. The datasheet specification has a minimum well depth of 85,000 e-. So this is modestly lower. We often adjust the clock levels to optimize other performance measures, such as trailing charge. Those adjustments can modestly impact the well capacity, but it is better to have a clean image than slightly more well depth.