Keithley 2001 - AC Self calibration failure

[dhillman]

I am still working on an older 2001 (1992 date code) that passes all self tests and otherwise functions, yet will fail AC self calibration with +415 error.
This error resolves in the calibration book as: Pos x10 peak offset out of spec  <- I do not know what this means.

All electrolytic capacitors have been replaced, and thankfully none were leaking.  Still the areas were cleaned with ethanol.
The main DC supply rails are solid and within specifications.

I have noticed that the +/- 8VF supply voltages are in 7V range, and not perfectly balanced. 
Running the thermal camera over the board, it seemed that a voltage follower (U331, LF351) was running warm (70 degrees C), so it was changed without result.
Also of note is U514, an AD847 that is running around 70 degrees C. 

TIN had some thermal images of a 2001 on his website, but I was not able to ascertain if they were of a good unit.

Is it possible for a bad low-level calibration to make the AC self calibration no longer effective?
I plan on dumping the calibration ROM to see if anything is deemed an outlier.



 

[Kleinstein]

The AC part has an peak detector circuit around U512 (LM311) and U534.
So I would imagine an error in this area could cause such an error message.
Too much leakage from U511 (DG408) could also contribute.

U331 is related to the Ohms circuit and might be driving quite some load - so probably normal to be a little warm.
U514 is somewhat related to AC signal zero crossing detection. Its a fast OP and sees some load, so a little more power is likely also normal.

I would expect that one could reasonable probe the function of the AC amplifier and rectifier with a scope.

[dhillman]

It certainly could be a peak detect issue as you suggest.

Some further details:

In AC, while measuring frequency, the voltage is displayed correctly, but the frequency is not always.
Additionally, the unit is not able to calculate crest factor; always reading ‘over’.
With a low voltage input (<2VAC @60Hz), the readings are correct.  From 2VAC to ~80VAC K500 begins cycling at 2Hz, four pulses at a time, then rest for equal period.   The displayed frequency is 2Hz as well.
Above 80VAC, the meter reads correctly. 

I can see that REL4 goes high through this period, and U503 (HC123) provides the 2Hz pulses.

—
Aside, I did change out U534 (AD711 or LF351 depending on how old the unit is) as I could see an incoming sawtooth on pin 3 but nothing leaving the follower.   This made no change.

Time to acquire more parts and keep probing.

EDIT:  I will qualify that the above test was performed with a variable transformer.   I just checked with a HP33120A function generator set to  7VAC @ 60Hz, and the 2001 reads correctly, less the crest factor.   

[Kleinstein]

The cycling of K500 could be an indication that auto ranging (or maybe protection)  kicks in.  As auto ranging might use the peak reading, this could be a consequence of a not working peak detector. So for tests it might be a good idea to turn off auto-ranging.

With U534 already changed, there is not that much left in the peak detector. Also probing should be relatively straight forward.
One should see the rectified AC signal before the peak detector, and than a reasonable accurate (e.g. +-10 mV + 1%) peak values.

[dhillman]

Poking around the peak detector has yielded some interesting discoveries.


It seems that the caps C411 and C586 are charging to a much higher value that expected.
The yellow trace is on the anode of CR517, and the green trace (for holding capacitors) is on the cathode of CR517.
Also of note is that the peak detector reset is asserted partway through the reading - almost looks like a glitch (D0 trace).

To see if the U511 (DG408) was leaking back to the capacitors, I lifted pin 4 from the pad.   No change in the charging of the capacitors, but there is now a VPeak reading on the meter, though it is a garbage value.

Next up will be the replacement of U512 (LM311).

[dhillman]

To put closure to this topic, replacing U512 (LM311),and U511 (DG408) repaired the peak issue and the associated AC calibration failure.
Thank you, Kleinstein, for the hint.