Uni-T UTD2202CE Compensation issue

[Illukas]

Hi everybody!

Am having a bit of trouble getting the compensation right on the 10x probe. After initially running self calibration and looking at the built-in square wave generator, everything looks good at 1V/div. Looking at the same waveform at 20mV/div, there seems to be quite a lot of over-compensation. After I've fiddled with the capacitor a little bit and everything looks good at 20mV/div and then looking at 1V/div, there's some under-compensation happening.

You can fiddle the capacitor to your heart's content, there doesn't seem to be a setting that works both at 1V and 20mV/div. Now I do hear a relay clicking inside the scope when switching up to 1V/div, so that must mean there's a separate vertical deflection circuit for higher voltages, right? Since the under-compensation can only be seen when going up higher from 1V/div, that must mean the seperate circuit is at fault, rather then the input or probe? (I tried 3 probes on both channels, issue is consistent.)

I also took some snapshots of the problem, these can be found on imgur here: http://imgur.com/a/lO3ed
The scope itself is brand spanking new. Hasn't been dropped or mishandled as far as I can tell. No other obvious signs of malfunction either.

Can you give me any advice what else to try? Or should I just send it back as faulty? Or am I just showing my stupid here?

[electr_peter]

I think you are trying too hard. You should calibrate scope at 0.5V/div or 1V/div (so that all signal fits on the screen in vertical direction) and it will be compensated properly.

When you set scope to 20mV/div, signal overloads input circuitry and weird effects happen (and signal you see on the screen is essentially invalid and wrong, so any compensation to this signal is invalid as well). Other scopes also  do this with overloaded input.
This is not the right way to do compensation - if you want to check 20mV/div level, provide good quality square wave at ~50mV Vpp (use attenuators).