Because the subject was intriguing, I took the opportunity last night to rebuild my dark/flat library (after a full refresh of my fixed brand new observatory), and I also took bias. For a blurry reason, I never captured bias data on this camera, and I was really surprised to find a mean ADU of 210 on them today. The DMK I was using previously had nearly zero bias, and probably I assumed it would be the case for the Atik also, or there was another reason, I don't remember. But now I find that bias is the main source of noise in my low-SNR frames. Has it grown with time I have no idea. Embarrassing
HEQ5-Pro - Atik 314E - Orion ED80T - DMK21 on Orion 50mm
DIY 3D-printed Moonlite and FWheel RGB/LPR
KStars and indiserver on two Atom 1.6GHz 1GB RAM Linux, VPN remote access
This is also the case for mine. Mean ADU of one frame in kstars viewer is 215 ADU
The master bias is showing a mean of 3.24e-03 in PixInsight which after some math equal to 212 ADU
The most important is not the mean of the bias but the average deviation. In my case the bias is 0.32% of the max ADU with an average deviation of 0.006%
A big average means that you crop the full range ADU of your camera but if the deviation is near to zero, you can live without the use of the bias in you processing.
Unfortunately, in the case of Atik 314E we need both of them. I don't know what is the delta temperature reached by the cooler but what I'm sure is that night temperature in winter and summer is not the same in France. I have to use at least 2 set of master bias. I think I will perform in the near futur some experiment in my fridge and my freezer just to know the effect on the mean value of the bias.
Linux Mint 19 Cinnamon using PPA for latest KStars-bleeding and INDI
Raspberry Pi 3 B+ with StellarMate
Atik 314E - ASI120M - CG5 with Celestron C9.25