Sick Keithley 2001: Errors 200.1, 200.6 200.7 and 201.2

[Krampmeier]

I was able to get my hands on an old Keithley 2001 DMM which failed calibration and could not be brought back into spec by full adjustment.

All electrolytic caps in the unit were replaced, but that did not help.

The following steps of the self-test fail:

• 200.1 A/D "No A/D comm, noisy A/D"
• 200.6 "A/D U808 malfunctioning"
• 200.7 "A/D U808 malfunctioning"
• 201.2 Testcal "Cannot measure 7V at A/D"

The meter displays reasonable values in all ranges, but the noise is higher than I expected, and there seem some strange non-linearities.

I connected the 2001 to 10 V from a calibrator, parallel to a Keithley 2010 which should be OK, took 500 readings at 10 PLC, and got the attached curves.  k2010_k2001.png (88.62 kB. 831x602 - viewed 77 times.)
I also went through the voltage range from 0 V to 5 V using a DAC (not the calibrator) and recorded the deviation from the set value. The general curve shape is the fault of my DAC, but you can see the noise level of my K2001.  k2010_k2001_DAC.png (50.2 kB. 792x631 - viewed 61 times.)

I have already built an adapter cable to get the ADC board out of the way, and probed around a bit, but did not find anything that looked obviously wrong to me. I don't have a deep understanding of what is happening inside the meter, and what I should be able to measure when single stepping through the self-test steps. The unit also does not seem to provide any more details about what it thinks is wrong...

Does anyone here have a suggestion on how to get this poor thing fixed?

[picburner]

I also had a K2001 with capacitors that have leacked electrolyte on all pcb.
In addition to replacing them, it is necessary to decontaminate the PCB several times from the electrolyte and then dry the PCB very well.
Furthermore, it is necessary to check that the electrolyte has not corroded tracks or vias (even under the ICs).
The measurements of the instrument will not be reliable as long as there are errors in the diagnostic test.

[Kleinstein]

The measurement looks really noisy with some periodic background. This could be something like excessive ripple maybe in combination with a wrong detection of the mains frequency.

The "noise" / low frequency signal could explain the 1st error.

A failed main frequency detection should however not explain the 200.6 and 200.7 errors. So there must be something else wrong.

My suspect would be supply ripple and maybe a failed CMOS switch. The ripply on the ADC signal may be to small to see it with a normal scope. One could still check it, in case it is 50/100 Hz ripple that just is not so visible for the ADC.

[Kjelt]

If you still smell the electrolyte you still have work to do.
Also check in the mains entry, below the switch there are hidden components, also check all the smd ic's around the power supply elcos if they have goo under them.
It really is a mess. I have never been able to solve all errors, the pcbs are ok when swapped with another K2001 so if you get steady readings then be happy, till then clean , clean etc.

[Krampmeier]

Thank you for the suggestions!
The original capacitors had leaked a bit, but the electrolyte had not spread far yet. The affected areas of the board were rinsed with isopropanol, but no ICs were lifted for that. There is no smell left.
All tracks under the caps were inspected and looked OK, according to the trusted person who did the recapping. Of cause, there may still be some gunk trapped somewhere...

I remember I had difficulties interpreting the voltages I measured at the interconnector to the ADC board, and IIRC I have already replaced one of the analog switches about a year ago (this meter has been sitting here for a while already, and it seems like I lost the notes I took back then  ). No luck with that though.

I also don't understand yet how test 200.4 and 200.5 can pass ( x10 line cycle integration and x.1 line cycle integration), and 200.6 and 200.7 fail (x0.02 line cycle integration and x0.01 line cycle integration). I have checked all the logic states in the service manual and I think they were correct.

I will open the unit again, measure the supply ripple, and poke around some more...