Agilent 34970A: Error 609 and 610 but no 611! Why?

[matemathieu]

This story started 5 years ago, I found a DMM card for a 34970A frame (with option1) for very cheap, knowing it was not functional, I wanted to get it to experiment and get better at fixing electronics. It was the time when money was low.
It was showing all kinds of errors that I don't recall, and looking at the schematics it looked like the IC U101 1SK6-0001 was the root cause of the problem. Luckily I found two of these chips from Taiwan. Replacing one 1SK6 chip was hard enough, but came through fine.
Then self-test returned only two remaining errors 609 and 610. I was getting on another job, I left it on a shelf.

Now, there is an urgent need to have this unit work again, hence my request for help.
Errors 609, 610 and 611 seems to be very similar and not having the last one might be a simple matter of a voltage source drift (the LT399H). Do you think it is plausible that for a 1V, 10V the drift is too high but for a 100V it is not?

I feel on doing more diagnostic and it would be nice to have your advice on what and where to get it.

Docs: https://www.keysight.com/us/en/assets/9018-04219/service-manuals/9018-04219.pdf
page 194 for list of components
page 240 for schematics

[matemathieu]

Here are the description of the errors:

609
DC gain x1 failed This test configures for the 10 V range. The dc
amplifier gain is set to X1. The measure customer (MC) input is
connected to the internal TSENSE source which produces 0.6 volts.
A 20 ms ADC measurement is performed and checked against a limit
of 0.6 V ± 0.3 V.

610
DC gain x10 failed This test configures for the 1 V range. The dc
amplifier gain is set to X10. The measure customer (MC) input is
connected to the internal TSENSE source which produces 0.6 volts.
A 20 ms ADC measurement is performed and checked against a limit
of 0.6 V ± 0.3 V.

611
DC gain x100 failed This test configures for the 100 mV range.
The dc amplifier gain is set to X100. The measure customer (MC)
input is connected to the internal TSENSE source which produces
0.6 volts. A 20 ms ADC measurement is performed and checked for
a + overload response.

611 says that it only checks for an overload response.
What does it means?

[Kleinstein]

With 0.6 V at the input and a gain of 100 the ADC should see an overload condition (some 16 V from the amplifier). So the check would only be for seeing the expected positive side overload.
The test for the 100 mV range is less stringent than the tests for gain of 1 and 10.

A possible explaination to see the errors 609 and 610, but no 611 could be if the amplifier allways gives to high a voltage. Similar a gross error in the gain (though very unlikely) would not be detected.

The error 608 metioned indicates a problem in the communication to set the configuration (signal switching and gain settings). So if this error comes up chances are high that the same defect also causes other error mesages. The 608 error would be the first to tackle - this may also solve other problems.

Drift of the LM399 is something the meter can not really detect, as there is no other reference only close in accuracy. For the reference it would be only a total failure (e.g. get less than 3 V instead of 7 V) that may cause trouble. The less than some 3 V limit would be because there is a 5 V supply for the HC4053 that is derived from the / V reference that should than cause trouble.
The usual first points to check are the usual supplies (2 x +5 V , +-18V and local voltage at some chips). Another good point to check would be the amplifier output (ADIN signal, e.g. at pin6 of U106) with a shorted input and in amps mode.

It is possible to get problems in the 100 mV,1V and 10 V range but not in the 100 V range, as the protection part is different.

[matemathieu]

Dear Kleinstein,
Thank you for your answer.
About the error 608, I made a mistake, it is 609 and 610 that are showing up.
The text is corrected.

The error listing from the previous attempt of fixing it has resurfaced from the notebook, there was the same as in this post:
https://www.eevblog.com/forum/repair/hp-34970a-self-test-fail-ideas-needed/msg942341/#msg942341
Errors 609, 610, 611, 617, 618, 621. Changing U101 fixed 617, 618 and 621.

Today I tried to replicate the Selftest measurement that WanQi did in the above discussion and that you mention (I found Pin6 of U106). I was able to see the initial voltage changes but it was not sustained, there was just a peak. This is so strange, I must have made a mistake on the setting of the scope.
It looks like your advice is what should be done: "check the amplifier output (ADIN signal, e.g. at pin6 of U106) with a shorted input and in amps mode."
Can you explain to me how would you do this?

Then, I started to verify the voltage rails on the DMM board, but I was not able to see any.... 
Finally time ran out and I had other priorities to fulfill.

Unfortunately, I have to go for 2 weeks to another location for my work. So I will resume the checkups when I come back.

[Kleinstein]

The ADIN signal is the signal to the ADC. One check this with a few possible conditions e.g. :
Input shorted and voltage reading (1 V / 10 V range), should give a zero output with possible short spikes from auto zero switching in between.
Amps mode with no current should also read rear zero (e.g. < 20 mV) like before, but this tests a different input path.
With a well defines input signal (e.g. 1.5 V or 9 V battery in the 10 V range) one should see the alternating 0 V and input signal from the AZ cycle. So one 200 ms or zero and 200 ms of the signal.

The test at ADIN should tell it the problem is more with the front end or with the ADC. One can than decide what to check next.

A first fast test is to check the readings one gets in the voltage mode (e.g. 10 V and 100 V range and  0 V  and +-9 V (e.g. battery)). Both ranges use different input paths and thus make sense to test both.