Keithley 2010 Repair

[Kleinstein]

The change from +-0.03 to overload is huge for 305.2 step. This does not looks like noise, but more like a relay not closing or similar. Without more information on what the test actually does, the information it gives is pretty limited. My guess would be something like a bad relay contact or maybe a cold solder joint / loose connection or bad front rear switch.

As far as I understand it, it is testing the 4 wire ohms low side. So the question is if the normal 4 wire ohms mode give correct / stable readings.
This would be something like testing a 4 wire short and switch between the 4 wire and 2 wire ohms mode several time and check how stable the zero readings are. Similar switch between front and rear to look for changes.

The 400.x tests should be for the AC section. It is a bit surprosing that the readings are chaging quite a bit. Does one only get a single value, or is there a contineous reading that changes over time ?
Where did your get the partial info on the tests from ?

[KungFuJosh]

2W/4W seems pretty stable, especially 4W.

Front:
-00.010181 2W
-00.000611 4W
-00.009837 2W
-00.000527 4W

Rear:
-00.009351 2W
-00.000359 4W

Front again:
-00.012645 2W
-00.000248 4W

I tested a lot more than that, but only wrote down the more significant changes.

Thanks,
Josh

[KungFuJosh]

The 400.x tests should be for the AC section. It is a bit surprosing that the readings are chaging quite a bit. Does one only get a single value, or is there a contineous reading that changes over time ?

The values written are what is displayed on the screen when you press the shift key after the test is performed. It's always a single value shown.

Where did your get the partial info on the tests from ?

Which partial info? The basic descriptions of the error are shown when the fail summary lists each error. For example, if it says 400.2 on the screen, press shift key, and it will say NON INV PATH on the screen.

That matches the list from the 2000 service manual. Unfortunately, the 2000 manual doesn't have 305.x. As @Cyclotron noted, 305.x wasn't added until a later fw version on the 2010.

Thanks,
Josh

[Kleinstein]

To me the 4 wire ohms part does not look that stable: With a 10 mA test-current in the 10 ohm range 0.1 mohm change still corresponds to 1 µV or 10 ppm of the range.  The specs call for 9 ppm as 24 hour accuracy. So there may be some issue, possibly from not fully warmed up as the drift is in one direction.
The front / back part looks OK for the 2 wire mode. At least in those tests no visible contact problem.
The K2000 plan has an extra buffer for the low side sense part - this could be a source of drift. The low side sense path would be a part to check.


It is interesting to see that the ohms part still looks very much the same as for the K2000. Even the part numbering seems to still largely match. Only the FET for the protection seems to have changes from Q120 to Q153 (TO220 case) - may well be needed for the higher current. There is however also an extra 10 mA current source added somehow.
So the description for the 400.2 test may indeed still work with the 2010.
The question of the inverting / non inverting path should be for the AC input amplifier, not the RMS converter or test signal.
It is a bit unclear what could cause so much variations. Maybe some intial charge in a capacitor or a relay at the AC input.

[KungFuJosh]

Yeah, I guess it really isn't that stable. It was slowly drifting down overall, but maybe a little weird after switching between 4W/2W. It goes back up slightly after switching, and then resumes its descension.

4W: 0.000099 to 0.000112
2W: 0.000896 to 0.000904
10 minutes later:
2W: 0.000833 to 0.000845
Switch to 4W then wait 10 minutes:
4W: 0.000112 to 0.000128
A few more minutes...
4W: 0.000100 to 0.000118
90 minutes later:
4W: 0.000093 to 0.000108

[KungFuJosh]

Short-term stability is about 15uΩ. I'm guessing it will continue to drift down, shifting the 15uΩ window. I'll run TC overnight and see where it's at in in the morning.

Thanks,
Josh

[KungFuJosh]

It drifted down over night as expected, but the short-term stability was worse after running all night.

[Kleinstein]

The curves look like they use something like 10 PLC. AFAIK the specs are for 100 PLC and thus with some additional filtering (like 10x average) and this may bring the noise just to the spec limit. The occasional larger jumps and drift (likely with temperaute) are still not great. Still there is no large error that should trigger a self test to fail.

[KungFuJosh]

Correct, it is 10 PLC. 15 PLC is the max.

So if this isn't the source of the issue, should I follow the suggested path in the repair manual for 400.2, or is there something you suggest I check first?

Thanks,
Josh

[Kleinstein]

I don't think the errors are related as 305.2 is ohms (low side) and 400.2 is about the AC mode.
The ohms mode still seems to work at least reasonable.

The rather different readings for the 400.2 mode looks a bit like a poor contact at K101 or K102. One could check the contact resistance.
Another point would be to test if the low AC ranges work. E.g. use a stable 1 V AC signal for the test and switch between AC and DC  and the 1 V and 100 V range. To operate the relays. A bad contace should also show up in thsese test, at least from time to time.

[KungFuJosh]

I pulled the short a while ago. In DCV, the meter is at -11.906V with the last 3 digits after that unstable with nothing inserted.

[KungFuJosh]

Looks like that's a decent clue. I think your theory about a bad switch or relay is looking more and more likely...now, which one? 🤔

I power cycled the meter to remove the case. The voltage was immediately closer to 0. It didn't change much for a few minutes, so maybe not heat related.

I switched back and forth between 2W & 4W, and then back to DCV. -11.9V is back now.

Power cycled again, started immediately climbing back to -11.9V.

Switched to ACV, audible relay click, now the meter won't go back to -11.9V in DCV.

U139 is still running hot at around 92C. I don't think it's the cause of the issue, but it might be a symptom.

[KungFuJosh]

20 minutes later and it's back to -11.9VDC. I wasn't paying attention, but I heard what sounding like a couple relay clicks on its own.

I clicked through all the mode buttons and back to DCV, and it immediately went back to -11.9V (starts near 0, then -1.x and climbs back up to -11.9V).



Thanks,
Josh

[KungFuJosh]

On a separate note, Phoenix Contact 1674833 is too loose on the pins, so they won't be a good replacement option. I thought the pins were 2.5mm, but it looks like they're ~2.35 (though if I squeeze my mediocre caliper harder I can get it down to 2.3). 2.38mm is 3/32", so maybe that's what it was if they weren't metric. 🤷

[KungFuJosh]

It appears that it's easier to get the -11.9VDC to show up if I manually switch to the 10V range.

To rule out the new U138, I connected a meter to the output (pin 6) and it never went higher than -000.6302mV (cold) and -000.5562mV (warmed up).

Thanks,
Josh

[KungFuJosh]

The amplifier seems to somewhat work, but there is a huge offset and for some reason the voltage at the input is much smaller than it should be, like quite some current flowing. It looks  like there is something wrong with the amplifier (e.g. Q193, Q194, ...U173U177) or U173U176.
The part around Q184 still looks wrong in the plan. The link to U174 pin 3 makes no real sense. That pin should have a resistor to GND or maybe +15 V.

One could check the voltage over R370 - this should ideally be very small (sub µV range), in case of a fault could be up to 1 V or so.

One way for current flow could be via U173 U176from pin 3 to pin2. The switch should be off, but could be bad. The voltage at pin 2 could give a hint.

Pin 7 of U173 U177should be the other input of the main amplifier and should be identical to TP105 if things are correct.

ps: fixed chip mixup

I think I might need to revisit this. 🤔

But I think I will also look at the -12V supply near U139. Since U139 is running hot, it seems logical that area might be a potential source for this garbage.

[Cyclotron]

You are putting up an amazing fight. Thanks for all the details you've recorded.

[KungFuJosh]

Thanks, I'm very stubborn. Sometimes that works to somebody's advantage. 

-12 not present at U139.

-12 only present on R271 where expected at pins 3 and 5 (-12.56V).

Allegedly, I might also need to check Q104, Q105, Q108, and Q113.

The easiest path for me to get -11.9V to show up on the display is:
Switch to 4WΩ.
Insert short (relay clicks).
Remove short.
Switch to DCV.
Manually set range to 10V.

So, the "DCV & Ohms Switching" seems relevant. Still dunno about the leak source. 🤷

Thanks,
Josh

[KungFuJosh]

-11.9V was present at:

U163 pin 7

U176 pins 3,6,7,10,14,15

-11.9V was not present at U130.

Should I replace U176?

Thanks,
Josh

[Kleinstein]

The case with an open input in DCV mode is not a good testpoint. The voltage is rather random from a very high input impedance and a tiny bias currents that both have only limits and can change with temperature. +-11.9 V may also be at about the limit what the ADC can still convert (it only gets a +-12.8 V reference and a few percent are lost to fixed parts in the pattern).

U139 (LT1124) is running hot by design. It is a rather fast amplifier with a high power consumption. I consider it a poor choice, but the alternatives with the odd pinout are limited. The LT1013 is also not ideal (more white noise, though less 1/f noise).
A point to check with U139 is whether it oscillates - as a fast amplifier it is somewhat prone to oscillation.

Keep in mind that the partial schematic is not free of errors. The area around U176 - Q184 - U174 is wrong. So it would be more about updating / checking the schematics than actually replacing U176.
3/4 of U176 set the gain of the amplifier and the 4th switch seems to connect one more input.
The self test does not list an error for DCV, so this part should be largely working - at least as much as the selftest can easy check.

The switching part that could have issues is the JFET base part at the input. This part may very well follow the K2000 schematics, but we don't know for sure. I would especially suspect the mechanical front rear switch and relays.

The

[KungFuJosh]

The case with an open input in DCV mode is not a good testpoint. The voltage is rather random from a very high input impedance and a tiny bias currents that both have only limits and can change with temperature. +-11.9 V may also be at about the limit what the ADC can still convert (it only gets a +-12.8 V reference and a few percent are lost to fixed parts in the pattern).

It's not random. There's an issue with a specific sequence that causes ~-11.9VDC to appear in the circuit as described, and be displayed on the front panel. This is not normal behavior, and is clearly the result of a fault.

U139 (LT1124) is running hot by design. It is a rather fast amplifier with a high power consumption. I consider it a poor choice, but the alternatives with the odd pinout are limited. The LT1013 is also not ideal (more white noise, though less 1/f noise).
A point to check with U139 is whether it oscillates - as a fast amplifier it is somewhat prone to oscillation.

Yes, but we know from other users that mine is hotter than average. 60 to 70C is more common. Mine runs at ~92C after replacement, and was at 102C before replacement. The one I replaced it with is spec'd to 125C, so I'm not worried about the part itself. However, it seems like a potential clue if a part is running 20C above average.

Keep in mind that the partial schematic is not free of errors. The area around U176 - Q184 - U174 is wrong. So it would be more about updating / checking the schematics than actually replacing U176.

I will check them again today when I look for the voltages at U177 and Q156 to see if that leads to more clues regarding the -11.9V source.

3/4 of U176 set the gain of the amplifier and the 4th switch seems to connect one more input.
The self test does not list an error for DCV, so this part should be largely working - at least as much as the selftest can easy check.

Largely working is not necessarily fully functional.

The switching part that could have issues is the JFET base part at the input. This part may very well follow the K2000 schematics, but we don't know for sure. I would especially suspect the mechanical front rear switch and relays.

Yes, I assume you're referring to Q104, Q105, Q108, and Q113? That's on my list to check stuff later on if I don't find the fault in the section mentioned above. I'm working on the assumption that the faults might be related, and that is the part in the block diagram closest to the INPUT that would affect both ACV and DCV. X1 Buffer and BUFCOM might be worth checking out too, but it seems prudent to review the A/D MUX & Gain section first to at least rule it out.

This is from the K2010 service manual:


Thanks,
Josh

[KungFuJosh]

Assuming 2N4392 is the correct part number for Q184, the pin numbering in the schematic (which was not visible before) is wrong. That doesn't mean the connections are wrong though; it might just be a pin labeling issue copypasta thing. I will check them out.

Thanks,
Josh

[KungFuJosh]

Here's the correction for Q184. I will share the schematic updates when I'm done verifying that section.

Thanks,
Josh

[KungFuJosh]

Attached is the output of U139 (pin 7). First cold, then switched to ohms, then back to DC with the range changed to 10V and the -11.9V bug active. The last screenshot is the same with a 20M filter applied.

Thanks,
Josh

[KungFuJosh]

Does anybody know where Q156 is located? It appears to be another proprietary chip.

Q156: TRANS, DUAL N-CHAN JFET, SNJ450H99 Keithley Part #: TG-326

I can't find the bastard anywhere though. Maybe it doesn't exist on my meter, but I do wear glasses. 🤷

Thanks,
Josh