Keithley 2000 repair after aplied high voltage

[rehakmatej]

Dear friends,
I did big mistake in my measurement procedure using Keithley 2000. I did some measurement of high voltage power supply (-830 V DC) used for photon multiplier tube. This voltage is OK for K2000, I have done lot of measurements of this high voltage using K2000. But last time I did mistake. I have forgotten to disconnect test leads from high voltage power supply. Next day I have turned high voltage power supply first. K2000 was turned off. Then I have turned the K2000 and suddenly I heard blow up sound and burning smell so I quickly turned the high voltage power supply off. Then I decided to open K2000 and see that R115 is burned. So I have replaced them but it probably did not help. The measurements are very inaccurate. Then I did some thermal imaging and observed following temperatures:
U123 (AD706JR) - 83 °C
U133 (DG444/SO) - 65 °C
U139 (LT1124CS8) - 67 °C
Q119 (J270) - 81 °C

I have removed U123, but still waiting for new one so it stays unplaced, then I have replaced U133, but temperature was same. Then I have disconnected Q119 and then U133 stopped heating. This leads me to conclusion that some transistors are probably shorted causing the heating. So I will take a look into schematic and will measure transistors connected to heating components.

All suggestions to help repair my K2000 are welcome.

I will periodicaly update this thread to report my progress of repairing K2000

Thank you.

[Kleinstein]

If R115 is blown, there is a good chance that Q101 and/or Q102 are blown. More of the input protection (e.g. U107) may be damaged as well.
The ohms current source part may also have more damage (Q120, Q123-Q126).
For R115 it may be a good idea (may also in the specs) to have a fusible resistor, just in case the same happens again.

It is nornal that U139 runs quite hot, it has a relativel high supply current.

Q119 may still be OK, just seeing excessive current flow. One may still chose to replace it.
A broken U123 may cause to much current for Q119 and U133.

[rehakmatej]

Thank you very much. I have continued my investigation and this is my actual state of K2000 about desoldered components: Q111, Q119, Q120, Q123, Q124, Q125, Q126, Q129, Q130, U123, VR109. Replaced U133 for new one. With all these components desoldered there is no excessive heat from any other components, except U139, which should be normal. Then I have measured Q101 and Q102 in circuit. Q101 has drain-source resistance > 60 MΩ, gate-source resistance about 773 kΩ, gate-drain resistance > 60 MΩ. Q102 has drain-source resistance about 3.69 MΩ, gate-source resistance776 kΩ, gate-drain resistance about 4.65 MΩ. From measured values my opinion is that they should be OK (from datasheet VDS breakdown voltage is 1500V). In this state (with desoldered components) I have performed built in self test which failed on error 100.1. I have expected this result with so many parts desoldered. From repair manual to this error there is sentence "The MUX word is applied to register U130 which sets lines A0, A1 and A2 of U163 high". I have measured lines A0, A1, A2 and they are really in high state (measured 5.032 V) so it seems that U130 works properly. Further measurements will be perfomed after the required parts will be delivered. Finally I have found one hidden blown part which is R283. Through this resistor, the AGND is connected throu 1.5 kV gas discharge tube and varistor to case which is probably connected to protective earth. Blown resistor is shown in the attached image.

[Kleinstein]

The 100.1 test is for the ADC part and the removed parts are for the input and ohms part.  So the meter should still pass that test (read 0 V with ADC).
Also many other part of the front end that may be damaged (e.g. JFET Q104 - Q109) should not effect that test.
The ADC input (TP105) and pin 8 of U163 would be good points to test - maybe with some of the first self tests in manual mode.

[rehakmatej]

I have measured voltage on TP105 and pin 8 of U163 during 100.1 error and it is 0V with respect to AGND. Then I have measured bit pattern from repair manual on page 2-23. There are some disagreements with pattern in repair manual. U106 has correct pattern and on bit marked as "V" (Q5) is value 0. U109 has on Q6 value 0, based on bit pattern it should be 1. U134 has on bits marked as "V" (Q7, Q1) value 0. On Q8 there is 4.52 V which is correct logic value but little bit lower voltage, othervise it is correct. U130 has all bits correct. U121 has bit Q7 measured value 0, value from repair manual is 1, othervise it is correct. Should I check communication of MCU with ADC with oscilloscope or with logic analyser?

Just for interest, on the pictures bellow is visible burnt PCB around R283.

[Kleinstein]

The damage around R283 looks more like an arc over from the resistor to the case. R283 may even still be OK.  So the failure they would be too much voltage to the case, which is seprate from the circuit GND.

0 V at TP105 would be OK for the amplfier.

For the digital signals, both scope and logic analyser should be OK, depending on the scope. There are 3 or 4 signals (data, clock and 2 Strobe signals) that may be of interest. The first point is to see if there is anything happening at all.