Keithley 6517B Electrometer, Bad OHMS reading

[HighVoltage]

This 6517B Electrometer seems to work perfectly in the voltage and current measurement ranges but can not read OHM correctly.

100 V Source  gives 9.69 nA on a 10 GOhm resistor
100 V Source  gives 0.107 nA on a 1 TOhm resistor
So, the current readings are correct.

But when I try to use the OHMS reading, it shows something stupid and negative and fluctuating.
Like -600 MOhms or so.

The instrument was bought used from private without possibility of return.

May be someone can give me a pointer, of where to start the repair.
Thanks
 

[TiN]

Did you try manual OHMs mode? That is with fixed current range/voltage source levels?

[Kleinstein]

I first test with trouble in ohms mode is to measure the test currents with a 2nd meter. This should at least work for the lower ohms ranges.

The high ohms ranges may well work different - like apply a constant voltage and measure the current.

[HighVoltage]

Thanks for the feedback.

TiN
I followed the OHMS verification in the manual (See attachment)
And I get total garbage reading of like -32 MOhm to -500 MOhm for any ohms input.

Kleinstein
I did a manual and simple cable setup and removed the TRIAX cable.
Used a TRIAX to BNC and then to 4mm banana plug connector

Results:
0 test current in second test meter.

Then I added a ground lead between Vsource LO and Triax LO and suddenly I get a source current.

Hmm...
Tested with a simple 1 MOhm resistor and get a good reading.

So why do I not get a good OHMS reading in my TRIAX shielded box?
Need to do more investigation.

[HighVoltage]

I think I found the problem.

The 6517B has a feature, called "METER-CONNECT"
<Config><OPER><METER-CONNECT>
This option needs to be turned ON, even when I had a manual ground cable hooked up.

I am surprised this is not turned on by default.

[HighVoltage]

And here are two pictures from the top PCB.

I will do some more tests but it seems to be working now.

[TiN]

Older 6517A had this function too. If you use guarding on triax - you must route LO from Vsrc terminal instead. If you use "legacy" unguarded configuration, then inner shield of triax is carrying LO signal.
It's interesting to see that 6517B is completely redesigned vs 6517A. If you can snap more photos, that would be greatly appreciated. Looks like femtoamp pron is well sealed with soldered in shields? 

[HighVoltage]

That is the interesting part.
I had the inner shield of the Triax connected to LO manually (inside my TRIAX-shielded box) and the OHMS reading was still weird.
Only when Meter Connect was ON, did the OHMS reading made sense.
I need to investigate the setup again.

Yes, allegedly the "B" version was completely new designed.
The big metal shield in the pictures is indeed soldered to the PCB.
 
OK, will have a look what can be taken out to take more pictures.
But I will not de-solder the metal shield.

This 6517B was built in 2016.

[TiN]

I had the inner shield of the Triax connected to LO manually (inside my TRIAX-shielded box) and the OHMS reading was still weird.

Connecting guard externally to LO would make nothing useful (and actually dangerous as you can damage guard driver), because you don't have signal path closed to LO anyway, unless you provide direct LO connection to front-end input (either thru proper configured triax port or external Vsrc LO wire).

OK, will have a look what can be taken out to take more pictures.
But I will not de-solder the metal shield.

Thanks again. I wouldn't expect you to teardown/break fully working expensive meter, so you are the boss on this one.  If you can get better camera, that would be wonderful too. (yes, I know, shouldn't complain too much...).