Fluke 8508A Weird Problem w/ HP 432A

[Hawaii596]

Measuring DC levels while doing 0 dBm 50 MHz on an EPM442A (not really too important to the story, why I put in DC/LF section).  I have BNC to single banana adapters on VComp and Vrf outputs of 432A (using HP 478A-H76).  I have been doing this one for years and have never seen this issue (I think I know what the problem is - my guesses at the end of post)....

So I connect HI and LO volts inputs of the Fluke 8508A input to the 432A as detailed above (those of you who have done this check should be familiar).  The 432A needle pegs in the negative.  I go to COARSE ZERO and can't get the needle out of the negative (no RF input).  I go to zero dBm range on the 432A just to check.  Still pegged negatively.

I tinkered with everything (adjusting zeroes etc.) for a while, thinking I had a problem with the 432A.  Finally, I disconnected the Fluke 8508A from the setup and the 432A began working normally (coarse zero worked fine, set to minimum range, depressed the zero momentary switch and it worked fine, set back to zero dBm range and turned on the cal output from the meter being cal'd and got an approximate zero dBm reading).

Then I even connected the 8508A to the 5720A and checked DC reading levels approximating what I want from the 432A in the calibration and it worked fine.

But when I connect the 8508A up to the 432A, it pegs the 432A negatively.  Finally, in troubleshooting mode, I set the Fluke 8508A to External Guard setting.  Once that was done, it worked fine.

Suspecting ground loop problems, I replaced both power cords and made sure they are both connected to the same exact circuit (adjacent outlets). 

The Fluke 8508A was recently damaged in shipment while sending to Fluke for primary lab calibration.  They repaired it.  I am suspecting there may be something amiss in how they reconnected the LO input terminal or somehow a missing connection in the LO circuitry causing an artificial ground loop problem.  I ran into a similar problem a couple of weeks ago (right after it came back from Fluke repair) measuring a high accuracy low resistance value.  There was a weird offset that corrected once I used External Guard.

Anyone have similar issues that could provide some wisdom?

UPDATE:  I just did some more troubleshooting... I have a Keithley 2700 DMM on the bench.  I connected the Vcomp and Vrf from the 432A to the Keithley DMM and get nice normal numbers >> Vcomp vs Vrf w/ no RF = ~0.02 mVDC (good/normal).  Vcomp vs Vrf w 0 dBm = ~89 mV (good/normal).  Vcomp vs Chassis GND = ~4.48 VDC (good/normal).  Precise #'s not an issue.  I get the good, appropriate #'s when doing the 0 dBm calculation.

So after this check with the Keithley DMM, I put the Fluke 8508A in parallel to see what is happening.  If I turn on EXT GD using the 8508A INPUT key, I get good numbers.  When I turn OFF the EXT GD on the 8508A, the 432A pegs negative and I get a reading of about 4 VDC.  Meaning, I think, the 8508A INPUT LO terminal is inducing somehow, about 4 VDC.  I think it is improperly floating or there is still a bad component.  I'll have to check on my other 8508A when I get back to lab (on the road now).  I also tried using front and rear terminals on the 8508A with same results.

[atcwhite]

What is the serial number of your 8508A?
I work on these for Fluke. I haven't heard of this problem. I will look into it.
Mike

[Hawaii596]

I was pretty tied up when this issue happened.  So I didn't get a chance to post an update.  It ended up that the Fluke 8508A had a loose screw rattling around inside it.  I opened it up, removed the loose screw and it fixed the problem.  The screw must have been on just the right spot to short something.  It was a very strange problem.