Some forum members have reported that the Cal Offs and Cal Gain interact to some degree. I have also noticed this and it can be very pronounced. I have found a plausible method that may eliminate that interaction.
I understand that full scale Cal Gain adjustment is always better, but in this case, the half-scale Cal Gain seems to be adequate.
If (for each V/Div range) you place the Gnd Pos in the center of the screen (half way up) and then adjust and save Cal Offs, you now have a constant baseline for easy Cal Gain adjustment without Cal Offs interaction (for that V/Div range).
With 1x probe, 5V/Div range, Gnd Pos set to mid screen (Cal Offs = 0), and with ME set for Vavg, I alternated both plus_9.53VDC and minus_9V.53DC (using a digital voltmeter for monitoring) and performed and saved Cal Gain adjustments by averaging the plus and minus voltage differences with the Cal Gain. The Cal Gain I saved was +0.43%.
Then I set the Gnd Pos at the bottom of the screen, and applied 36.52VDC to the Nano and measured 36.34VDC on the Nano (once again, digital voltmeter monitoring). This results in an accuracy of 36.34/36.52= 0.99507 or just 1/2 of 1% error. For a non-lab calibrated instrument this accuracy should be more than adequate for most users, and it makes the calibration much faster and simpler with no diddling around going back and forth.
It may also be possible that if you just double the Cal Gain that you think it needs using this method, then you may get results similar to full-scale calibration. I say this because because my Cal Gain was +0.43% and may measured error was -0.493%. It will take more testing to confirm this, so maybe some of you forum members can test out this theory across various V/Div scales and report your findings.