Keithley 2010 Repair

[KungFuJosh]

One more connection added between Q185, Q186, and U188 p1.

Confirmed voltage at U174, corrected voltage at U190.

[Kleinstein]

The TLP626 is an opto-coupler with AC input and thus back to back diodes at the input side.

The area around Q184 is still wrong. The gate of Q193 should not be connected to the gate of Q184.
It would be more pin 2 that connects to Q193 - gate, as there must be a path connect the signal from U176 pins 10+15 to the gate of Q193.

Chances are R427 is somehow connected towards U174 output.

There is likely a resistor from U190 pin 3 = LM431 to ground (maybe the -15 V ?). This for the current for the LM431.

[KungFuJosh]

The TLP626 is an opto-coupler with AC input and thus back to back diodes at the input side.

The area around Q184 is still wrong. The gate of Q193 should not be connected to the gate of Q184.
It would be more pin 2 that connects to Q193 - gate, as there must be a path connect the signal from U176 pins 10+15 to the gate of Q193.

You are correct. I checked, and it's pin 2 to pin 2. I (obviously) haven't tested every connection on the schematic, so anything else I should review, please let me now...within reason, I don't want to go through all 10 pages.

Chances are R427 is somehow connected towards U174 output.

I think you're right.
There is 1kΩ between U174p6 U188p1. There is about 99.7kΩ between U174p6 and Q184p3. So this looks correct.

Unrelated side note, R427 is noted as 4.99kΩ in the service manual, but it is correct in the schematic as 1kΩ.

There is likely a resistor from U190 pin 3 = LM431 to ground (maybe the -15 V ?). This for the current for the LM431.

U190 p3 measured to ground is 15kΩ. Assuming we're lucky enough that it's a single resistor, and measures accurately in circuit, then we're limited to R371, R392, or R421. R371 is in the right area. I just checked, and it looks like the via connecting to R371 is the same one on the trace from U190p3 to U185p1. I'm still not taking the PCB out, but I'll put this in the schematic until proven wrong.

Thanks,
Josh

[Kleinstein]

I would consider the signal marked as "In Amp " more as the feedback input (inverting amplifier input). The amplifier signal input is with R370.

U176 pin7 should go to U193 gate  - that was the original plan and would make much more sense for the function. Likely no more link the U184 pin3.
For better readability it may help to move U174 (and the related JFETs) to the lower right in the drawing.

[KungFuJosh]

You were correct, the wrong pin of U184 was connected.

More corrections, some additions, and lots of readability improvements. Please verify that my readability improvements didn't screw anything up.

U176 pin 2 connects to 10kΩ R319. The other end of R319 connects to the via I pointed out on the PCB. I lost the trace somewhere around U120 (unless I lost it earlier lol). If somebody else can track it down, that would be helpful.

Thanks,
Josh

[KungFuJosh]

I'm ordering more parts, and want some spares. Is STP7NK80Z a good drop-in replacement for 2SK1412? Google and bots are giving me mixed info, and the datasheets don't seem to agree. 

V(BR)DSS doesn't look good at all by comparison.

Thanks,
Josh

[Kleinstein]

The 2SK1412 is a small 1500 V MOSFET.
The STP7NK80 is more a higher current, 800 V type.
1500 V type in a TO220 case are hard to find however. One could maybe accept a lower voltage (1200 V) type, like maybe
IXTP1N120P. It would be at least OK for a first test for used with slightly lower voltage (e.g. < 800 V) only.

For the target of R319, there should be a 1 V reference votlage level, like the point between R185 and R189.

It is only a small detail, but the ACA signal at U163 should not be for current but like with the K2000 is more like a differently filtered output of the RMS-DC converter.

[KungFuJosh]

The 2SK1412 is a small 1500 V MOSFET.
The STP7NK80 is more a higher current, 800 V type.
1500 V type in a TO220 case are hard to find however. One could maybe accept a lower voltage (1200 V) type, like maybe
IXTP1N120P. It would be at least OK for a first test for used with slightly lower voltage (e.g. < 800 V) only.

Thanks, that's what I figured. For now, I'll just hope I don't ever need one.

For the target of R319, there should be a 1 V reference votlage level, like the point between R185 and R189.

That specific point doesn't make sense since the trace for R319 goes in the wrong direction (I did check the closest via I could find, and the resistance values don't make sense either).

It is only a small detail, but the ACA signal at U163 should not be for current but like with the K2000 is more like a differently filtered output of the RMS-DC converter.

I don't understand U163 at all anyway. How would I test that?

I was curious about how different modes effect the voltages there, and they're mostly the same.

U163 Voltages by Mode:
DCV:
p01: ~4.97V (unstable- signal?)
p02: +4.97V
p03: -15.82V
p04: +108mV
p05: +1.000V
p06: +6.28V
p07: (unstable- signal?)
p08: (unstable- signal?)
p09: +0.78mV
p10: Signal?
p11: +0.06mV
p12: +6.9251V
p13: +15.164V
p14: +0.36mV
p15: (unstable- signal?)
p16: (unstable- signal?)

DCI is same as DCV.

2W, 4W same as DCV except:
p07: +6.28V.

TEMP same as DCV except:
p09: +0.07mV

ACV same as DCV except:
p01 and p02: +0.04mV
p07: +0.07mV
p09: -0.05mV
p16: +0.04mV (not very stable)

FREQ same as ACV except:
p08: +1.5mV

Thanks,
Josh

[KungFuJosh]

I made some more readability edits, and couple other things. Hopefully I didn't screw anything up.

Thanks,
Josh

[KungFuJosh]

This might be interesting? I'm reviewing switches and parts on the analog front end that might be relevant to my meter's problems. K101, 102, and 104 are listed as TQ2E-L2-5V, but at least my K104 is EA2-5TNJ.

[Kleinstein]

There should be very little difference between the TQ2 and EA2 relays - mainly a different manufacturer.

The high voltage MOSFETs, if they fail, they usually fail short and one can check this in circuit if needed.


At least the K2000 the ACV and ACA singals are both from the RMS converter chip output. The ACA signal has an additional low pass filter and would thus be the path for slower readings. Both paths should be possible for voltage and current - it is more a question of the reading rate.

Overall the K2010 looks very similar to the K2000 in the front end. The main concept is as follows:
Relays and JFET switching (Q104,Q105,Q18,Q109,Q113) are used at the very input to choose between the voltage input, the high side and the high voltage divider.
Than come a zero drift buffer. In the K2000 this is an LTC1050 (may be a different type in newer meters) with a bootstrapped suppy.
In the K2010 there is an additional pair of JFETs to support the AZ OP-amp for the higher frequency part. This is to get lower noise and maybe allow for a bit more filtering.

The multiplexer chip U163 (+1/4 of U176  or U129 for the K2000) is than choosing which signal gets to the main amplifier stage with gain. Signals here include the buffer, low side sense, the current shunts, the RMS converter output, ground and reference levels to check the gain.
In the K2000 the amplifier is with 2 OP-amps, with the K2010 the main amplifier is with the discrete JFETs, the LM394 and more parts.

The auto zero loop would normally keep the JFET part fixed and switch U163 for the signals (input, ground and reference voltage - depending on the speed).

The idea of checking if R319 does to 1 V was the analogy with the K2000. As the K2010 has 2 references, it could be that this would be the 6.4 V reference, though I don't see much need for this (mainly a self test thing).

So far it looks the issue with the meter in question here seems to be the JFET switching part seems to be open. This than cause the hum for the signal from the AZ buffer. My prime suspect would be the comparator (LM339) that controls Q104 to be struck at low and thus never connecting the input.

[KungFuJosh]

So far it looks the issue with the meter in question here seems to be the JFET switching part seems to be open. This than cause the hum for the signal from the AZ buffer. My prime suspect would be the comparator (LM339) that controls Q104 to be struck at low and thus never connecting the input.

That's exactly my plan. Replacing the LM339s is the easiest part to access related to these problems, so that's the next step.

I'm also getting a bunch of the other switches, including the TQ2-L2-5V relays just in case. I don't know how much it matters, but there are timing and current draw differences between those relays.

I dunno if this is AI slop or not, but this is what deepseek said about the two relays:

The Critical Difference: Coil Characteristics: The most significant difference is in the coil. The EA2-5TNJ has a resistance of 178Ω and draws 28.1mA, while the TQ2E-L2-5V series has a resistance of 125Ω and draws 40mA. The Keithley 2010's driver circuit is likely designed for the specific current draw of the official relay. The EA2's lower current draw could cause it to latch unreliably or not function correctly.

Is that bullshit, or a legit concern?

Thanks,
Josh

[Kleinstein]

I don't know how the relays are driven, but my guess would be some ULN2003 or simular drivers that are good for some 300 mA (and more pulsed).
Relay drive is rarely made for a very specific current. Even 28 mA would be too high to drive vom a normal logic chip. So the difference in the coild would not be significant.

For the JFET drive part one could just measure the gate voltage to see of the drive side is the cause. There could also be a few other causes (e.g. fusible series resistor that is open)

[KungFuJosh]

I figured it wasn't likely as dramatic as the sloppers said it would be.

I'm not aware of any fusible resistors in this. Are there any? I think it has regular replaceable resistors for the protection stuff, like R373.

[Kleinstein]

It is hard to see which resistors are fusible and some resistors may only unofficially work as fusible. It could be R115, R368.
It should be pretty easy to follow the signal how far it gets.

[KungFuJosh]

It is hard to see which resistors are fusible and some resistors may only unofficially work as fusible. It could be R115, R368.
It should be pretty easy to follow the signal how far it gets.

Any resistor is a fuse if you try hard enough.

[mawyatt]

It is hard to see which resistors are fusible and some resistors may only unofficially work as fusible. It could be R115, R368.
It should be pretty easy to follow the signal how far it gets.

Any resistor is a fuse if you try hard enough.

Yes, and Murphy's Law teaches that any circuit fuse will be protected by the most expensive semiconductor which always blows first

Best

[KungFuJosh]

R373 is specified in the sm as protection. R368 is a standard CMF65 1% 169Ω.

R115       RES, 1K, .1%, 1W, WIREWOUND          R-249-1K
R366,373       RES, 5K, .1%, WIREWOUND             R-249-5K
R368       RES, 169, .1%, 1/2W, 350V, METAL FILM       R-94-169

[KungFuJosh]

Any resistor is a fuse if you try hard enough.

Yes, and Murphy's Law teaches that any circuit fuse will be protected by the most expensive semiconductor which always blows first

Well, obviously.