Keithley 2001 high offset/auto zero problem?

[laosiji141]

I recently bought a Keithley 2001 for 2500CNY，the self-test showed no errors.But it has a strange problem: 20V range has significant drift and takes a long time to stabilize (more than an hour)，other range drift seems reasonable.If turn off the auto zero, 20V range has the same drift rate compare to other DCV range.

20V range is *0.5 gain range,maybe it has a different auto zero path? I check the K2001 repair guide but have no idea to handle the problem

The ACI,ACV(include 20VAC) have a constant zero point but have no drift,maybe a low leave cal can solve the problem.

[laosiji141]

2V range test

Also I have replaced the fan with a nidec 4015 5V fan and powered by the digital board 5V LDO(need change connector pin),it is really quiet and has the same air flow

[Kleinstein]

The 20 V range uses a quite different signal path. The main amplifier is operated at a gain of -0.5 and the signal goes through the buffer amplifier U341.
The AZ steps would also get zero in a different way.
The output of U341 would be an intermediate point worth looking at. It should still show the AZ steps. This may need the K2001  to operate rather slow (e.g. 10 PLC ) and use a fast 2nd DMM (e.g. 1PLC non AZ) to record.
The errors is relatively large (a few 100 µV range) and thus should be visible, though tricky for a scope.

One could check the AZ loop steps with a defined small input signal (e.g 10 mV or 100 mV) and than use a scope to look at U341 output.
AFAIK the AZ loop in the K2001 is not that straight forward as with most HP meters. It includes extra steps, measuring a ref. voltage.
The kind of oscillation looking parts could be thermal oscillations and thus a result of temperature changes or it could be an effect of the gain measurements and drifting ADC gain that is slowly corrected in steps.

[laosiji141]

Thank you so much for the reply
This reminds me of when I was repairing a DMM7510 (which had high drift and INL on the current range).  I spent a long time looking for the i-v path and finally I found that the problem was because an OPA2277 which still had "opamp function ."
I will check the sus part and continue to share the repair process of K2001

[laosiji141]

After set 10 PLC(AZ ON) and short-circuit the input or apply 10mV on the input，the Output of U341 is reasonable,it has constant ΔV (20V range ΔV=400uV , 2V range ΔV=100uV) when AZ ON or very low drift when AZ OFF
So the input path has no problem?
By the way The Q532 was missing when I first got it , I don't know what happen on it, but i have repaired before testing and the voltage rails seem ok.

[Kleinstein]

Q532 is used to limit the current of the bootstrapped supply. So more like a part to limit damage if things go wrong, but not essential for normal operation. The missing transistors indicate some prior repair attempts.

Low drift with AZ off suggsts that the input part (e.g. the protection around Q337-Q340) is OK or at least not causing the offset. Also the ADC seems to be OK. The main suspect would thus be the AZ switching around  Q527 and U323.

It is still odd to have the AZ mode to work with the 2 V and 200 mV range, but not the 20 V range. The issue may be the part that is isolating the lower voltage amplifier around Q522, Q523 and U339.

A point to check can be the input current in both the 20 V and 2 V range. A problem with AZ switching could result in extra input bias current.
Another point to check could be gain drift of the ADC. In AZ mode there are some background corrections for this, though I don't know the details. So this would be the non AZ mode with a relatively large voltage in the 2 V and 20 V range.