need help and ideas for calibrating a UNI-T UT(D)2042C scope (11 years old)

[Pfriemler]

Hi folks,
this is my fist post, so it has to be to the beginners, hasn't it?
Should it better be placed in "Metrology"?

I am looking for a way to improve the disastrous basic accuracy of my UNI-T UTD2042C scope.
After fresh internal calibration (Utility -> Self Adj.) both probes show in both inputs and in 1:1 as well as 1:10 at 10 V DC only "9.20V", which is an error of -8%. This is not dramatic for an ugly thing like that but increasingly annoying and I'm tired of the conversion now.

Reichelt sold this device 10 years ago (it is named UTD2042C, but I wonder if there are any mean differences to UT2042C).
I sent it back for replacement and got an equally inaccurate one in return (it was a different, thanks to docu-photos). At that time I shrugged it off as a serial error.

Of course, I did not found any adjustable screws inside.
What bothers me: The 1kHz 3V test signal (for probe calibration) is shown as a 2.96 or 3.04 amplitude, which fits well. An old Voltcraft DMM750 scopemeter indicates 3.25V and half a dozen of DMMs get it as an 1.58 V average. So if the test signal and the internal calibration are derived from the same source, this can be the point where I have to start, because if it calibrates to a higher voltage it will measure lower values.

I simply tried to pull down the test signal with an external 10k (lowers signal by -5%) during calibration, but without success.
I have simply no idea how calibration works and if there are any procedures to do like for some DMM (key combinations and calibrating with external references and so on).

These are my questions:
1. to all UT-2xxxC owners: Are yours also so inaccurate or ok?
2. has anyone ever found any successful help regarding this?

Any hint is welcome. Service manuals or a circuit would be the best, but I found no single hint that something in this way exists at all. Only a German Site did offer a calibration for this thing some years ago, so in principle there must be a way.

Before I get a lot of well-meaning advice:
a) of course I know about the 8-bit-limits and that the displayed jumps are 20 mV when the voltage changes. But already 9.8 or 10.2 V would be a significant improvement.
b) yes, I know that the whole thing is not very good. It will be replaced soon. But this is a different matter and should not discussed here.

btw: Thx to DeepL :-)