Keithley 2001M - Self test errors - ask for help

[JK21]

Now I have to ask the community for help.

I bought a K2001M as defective with the description: Does not pass the self-test.

Externally, the device is complete. Like the fan, the inside is very clean.
It looks like someone has tried to fix it though.

What have I done so far:

- Visual inspection inside: U316 has burst and has a hole and U317 seems already exchanged
- all electrolytic capacitors renewed - no leakage to be seen. The extracted C were also OK. Only C114 and C115 had 30 ohm ESR.
- the instrument can be switched on - error 514
- production in 2006 ?
- analog board revision 04P (-P5)
- Firmware B16 A02
- last calibration 2012
- all voltages are within tolerance
- DCV, some DCI and some 2W and 4W is working
- I read and saved the calibration data using the I2C C341A Flasher (see xdevs.com|TiN)
- 20 self-test errors:
   
   301.1 - 2
   302.2
   303.2-3
       304.2 - 7
   305.2
   307.2-3
        308.1
   309.1 - 3
        309.5-6

- U316 renewed, then there were the following errors:

   304.3-5       
   309.1

Now I tried to debug these errors using "TEST, BUILT-IN-TEST, MANUAL, STEP" starting by
   304.3:
   U332/10 is not = 1, U321 exchanged (also has a hole on the underside), installed an CD14504 due to lack of an MC14504B.
   
Now the following errors appear:

   304.2   
   307.2
   308.1
   309.2 - 3
   309.5-6
   310.1 and
   412.1

Removed the analog board and so far I have found Q311 to be defective (it is currently removed), CR305 and CR309 OK.
Then I replaced the U321 with a MC14504B and renewed the U332. No change.

I don't understand why the levels at U332/9 (OHM FB) are still wrong. This signal goes but only on U321/4 and U332/9.   
It stays at ~14.95V even though U321/5 gets the correct Low level from U307/5 (slightly increased with 0.245V).
I have already exchanged the U321 twice :-(
The input U332/9 to +15V power has ~4.9 MOhm in both polarities.

I can't get any further here. Is there another connection of OHM FB?
Does somebody still have an idea?

[JK21]

Hm, I don't quite understand how the complete reference block U330 works.
I measure 600Hz Ref Clk and -10.354V as -Vref. The reference current is set via R368. For my 0V that gives about 2mA Iz.
With 0V at U328/3, U328/2 must also be at approx. 0v. This then results in 7.028V at pin 14 U330. The 1.75V at the output U327 are available.
So far this part is OK. This 7.028V is also present at PIN13 via Q540.
But what voltage should come out at pin 13? Only around 0V can be measured here. Is that right?
Thanks in advance.

[Kleinstein]

The 1.75 V reference is for measuring the ADC gain in the 2 V range and has nothing to do with the ohm functions.  -10.3 V look about right for - Vref.
7.028 V for the reference looks OK.

U330 works as a switched capacitor, transfering the 7 V reference voltage also to C324. So one could consider C324 as a kind of floating copy of the voltage reference.

For the voltage at pin 13 Q540 sets a lower limit of some 6.8 V.
However the OP-amp U324 would normally drive the voltage higher via R271. So the voltage there depends on the DUT. In most case I would expect something the the DUT voltage plus some 7.1 V.
So U324 should be regulating the more positive voltage (e.g. +9 V) at the ref. resistors.
U 330 should be regulating the more negative voltage ( e.g. +2 V) at the ref. resistors.
Under normal conditions the opto coupler U537 should be on, it can turn off if there is an external negative voltage applied. For the analysis if the normal operation I would consider it constant on.
As far as I understand it, the 2nd switch chip U325 is only used for tests / calibration and not used (set to pins 4 and 11)  in the normal ohm source operation.

The normal main current path for the current shource should be from Q317 trough pin 13 of U332 through one of the 4 ref. resistors, through Q320 and than the protection part with the high voltage  MOSFETs to the DUT and than to ground.

Pin 13 of U332 and U325 should be at a higher voltage, like some 7.5 V for shorted terminals for the more normal ranges (not the 200 M range).
For the start I would use something like the 1 K range, not be too sensitive to leakage / load from a meter or scope.

[JK21]

Kleinstein, thank you for the explanations!

That gives me a better understanding of the basic function. Unfortunately, I have a typo in my question: Of course, it wasn't about Pin13 (7.028V measured here), but about Pin7 from the U330 and the further control U331 / U333. But your explanation of the C324 gives the answer.
That means U330 is working properly.
Due to the incorrect activation (OHM FB) of U332 by U321, there is no further progress in debugging self tests for the time being.
I check in 2W with short at the inputs in the 2k range: There is also only 7.033V at U332/3 + 13 and no higher voltage.

A 10M resistor at hand is measured nearly correct to 9,97 M in 20M, 9,65 M in 200M and 0,00928 G in GOhm range.
The voltages at pin 13 U332 in these ranges: 7,73V, 0,488V and 0,488V with 10M on input.

I have to keep thinking about that.

[Kleinstein]

The lower voltages in the 200 M and 2 G range could be OK, as there is than an additional divider active (U323 pins 2  to 3).
With the Ohm FB signal actice the current source may never swich to the 2 K range.
For a test one could check the currents with an external 2nd meter.

The Faulty FB signal is strange - there may be short on the PCB (e.g. burned PCB, leaking capacitor or maybe a tin wisker). Given that is is a digital signal it should not be that difficult to find the culprit.

[JK21]

In the meantime I have swapped the U321 and U332 again and cleaned everything meticulously.
After switching on, U321/4 (FB output) was high again! The trace between U321 and U332 showed
overflow towards 0V and +15V with removed IC's when measuring the resistance with an external
multimeter.

In the end I renewed U307 in addition to U321 and U332. Now the low output level at U307/Q2 is also
normal at approx. 0V and no longer 0.245 V.

The error numbers now start from 309 (309.1-2-3-5-6, 310.1 and 412.1).

So there is progress! 

[JK21]

Unfortunately, after other commitments, I was able to continue troubleshooting today.

I have programmed the firmware to version B17.

In the first post I wrote that the Q311 seemed faulty to me in the simple tester (2 diodes with Uf 0.7V are displayed).

In test 309.1 the voltage at the ADC was only -0.043V instead of 0.098V +-8mV. So there had to be one larger part of the test current can be misdirected.
For the Q311 I installed a J177 due to the lack of a replacement "TG-198".
Now after removing the J177, the self-test runs with the only error 309.3 

Unfortunately, a JFET 5432 / TG-198 doesn't seem to be available anywhere at the moment.

What else can this JFET be replaced with?

[Kleinstein]

The J177 is a p channel FET, while Q311 is supposed to be a N channel low resistance one.  A J108 may be a reasonable good match for the 2N5432 as a low resistance switch.  For a first test a J111 or PN4391 may also do, and 2 of these in parallel may be an option.

Many of the cheap component testers have trouble with high threshold JFETs and show them as diode and DS connected.

[JK21]

Many of the cheap component testers have trouble with high threshold JFETs and show them as diode and DS connected.

Hm, so may be a chance that the old Q311 isn't bad! But the place around the Q311 at the analogue PCB  has only 1 free side for manipulation of the part.
For every JFET to exchange the PCB must be removed 

I will order a J108 now and up to the delivery I can check the old one.

Thank you, Kleinstein! 

PS:
It's possible to reset the error 514 simply by change the cal date and save the new date without destroying the remaining cal values.
-> Calibration -> Calibration-Dates  -> Change ...Unlock Switch.... Change Date

Edit: This runs without an error message, but does not actually change the saved Cal date!

[JK21]

I have now reinstalled the old Q311 (5432) and checked the control again.

There are no more self-test errors. Not even in a continuous run test.

After that I did a comprehensive cal (with my modest means only) and saved the values.

My K2001M is now completely back to live!   

Thanks and Best regards