Keithley 197 gone crazy

[MathWizard]

Theory of op. and block diagram page 36, schematics on page 66
https://download.tek.com/manual/197_901_01A.pdf

It was working fine until I had it apart to check the caps, and add a buzzer. Then I thought I had gotten solder into the Amp SW, and was reading -4V w/ no I/P. I cleared it to the point that it was reading only a few mV or 50-150megaohms on ohms, w/ nothing attached. So I took out the whole 11pcs SW assembly, and the Amp SW seemed ok.

Now after I put it all back in, the DMM really went crazy. With nothing attached, it can think there's 100's of Volts, and go OL most of the time. At best it thinks a 9V battery is 17V. But only draws 1.2uA

On ohms, it seemed to think there's 600k on the I/P. I have assembled 1R, 10R, 100R, etc to test properly. But when I put a 300k on, it read ~220k.

On amps it thought there was +6A w/ no I/P, yet the DMM is powered from my PSU, and the overall current draw of the DMM stayed as is should, 86mA on +15V, ~10mA on -15V.

All DC voltages are fine, I never checked all the logic chips, to make sure they are all still ok.

When powering up into the diagnostic mode, there's no error and I can get into the Dmode on page 50. I just started checking the 20V range, and getting familiar with the TofOP. and the schematics.

The main buffer op-amp seems to be working, at least it's not outputting gibberish. I'll get some screen shots later, when I properly test it.

As for actual damage i might have done, the battery charging PCB, was dangling, and I had part of it taped over, but not all, so that PCB might have hit something I never heard/saw/smelt anything at all while working on this.

The LCD also seems to noticeably cut out at a periodic rate, I better check that too.

Is it the SW's gone crazy ? Or the ADC ? Overall this circuit is over my head still, but it's a good way do dive deeper into it all.

I can systematically apply voltages, currents, resistances, I have more DC points to check too.


What does this sound like ?

[Ordinaryman1971]

Clean the contacts on your range switches... deoxit or even WD40 (contact cleaner)  or what you have on hand should be helpful.
Clean also the selection switches for your measurements... and power switch... just because it's right there.

[trobbins]

My 197 has a smallish metal enclosure just behind the input terminals, that mounts on top of the main metal shield cover that fits over the front section of the main pcb.  That small enclosure can be a pain to disconnect and reconnect without causing some kind of accidental shorting of leads.

Operation is also very sensitive to power supply rail noise, so you may benefit from swapping out all power supply e-caps, and removing a GPIB accessory board if it is fitted.

[MathWizard]

I checked the ESR, 1 was a bit high, the rest were pretty good. But for now I'm running it off +/-15VDC. I did a few calculations and the pwr control op-amps U109 and U116 are fine.

I haven't figured out what U101 is doing, I guess on volts its for shifting the zero point, but I don't "see" the circuit for that yet? And on ohms to sample the Rref and then sample Rtest ? Is the MCU just sampling 1 value at a time, then does math on them ? But I just skipped the whole ADC. I barely read that part yet, it doesn't seem anything like I expected, so it's all new.

Here's a scrshot of the 200R range, shorted, and it calling it about 14-15ohms.  Yellow is on the far side of R116, which is on the NI of U102 (main buffer). Blue is on the output of U101. Green is on the OP of U102

then the same thing with 99ohms attached, it calls it 74R

pic's are attached

I don't get when the multiplexer is supposed to be shifting. The big peaks are at 1.1Hz, and the shorter scale is 3.33Hz (for a short step and a longer step if u follow).

I need to look at the signal going over limit to see what the signal voltage actually is. I'm learning it.

[ceoxrad]

Clean the contacts on your range switches... deoxit or even WD40 or what you have on hand should be helpful.
Clean also the selection switches for your measurements... and power switch... just because it's right there.

I don't think WD40 is suitable as a contact cleaner, I'd stick to a product labeled as such, but maybe I'm wrong

[Gyro]

Clean the contacts on your range switches... deoxit or even WD40 or what you have on hand should be helpful.
Clean also the selection switches for your measurements... and power switch... just because it's right there.

I don't think WD40 is suitable as a contact cleaner, I'd stick to a product labeled as such, but maybe I'm wrong

This has caused confusion here before, and is an example of the stupidity of naming your company after your first product - the WD40 water displacer!

There are a whole bunch of [EDIT: make that 35] WD40 products today, including dry lubricants and a contact cleaner.  You need to be very careful when talking about them - just telling someone to use WD40 isn't sufficient, and can cause irreversible damage.

[MathWizard]

[ Specified attachment is not available ]Ok after taking some measurements with a resistor sub-box, and doing some calculations, the buffer op-amp U102 is outputting just fine and the MCU is reading what's actually there. For example, on the 200k range, when shorted right now, the NI pin see's
OhmHi-Ref=2.60V
OhmLo-Ref=31mV
OhmHi-Sense=12mV
OhmLo-Sen=0V
And so that works out to 467ohms and that about what it's reading, ranged correctly.

And at 160kOhms, its
OhmHi-Ref=2.58V
OhmLo-Ref=1.188mV
OhmHi-Sense=996mV
OhmLo-Sen=1.1mV
which is 71.5k, that's about what it's reading, which is around 50% of expected.

But then the equation stopped working, on other ranges.

I  took a lot more data, then I got hung up on seeing 8-10k between the ohm's current source diode and VHI_I/P. So I reflowed some switches again, and cleaned it all pretty well.

The precision multi-range resistor was also bent a few times, so i reflowed it.

Anyways I read in another thread about taking the switches apart, so I took them apart and cleaned some residue off them, IDK if it's flux or de-oxit from me, or grease or something from before.

With some switches out, I was still getting either leakage through a mechanical varicap, or more gunk that needs cleaning off the traces.



But now with it back together, on the 200k-range, it reads 43R shorted, into my substitution box, and zeroed, I can go up to 9R and it counts to 7 of 220k, and I can dial it up to 219,401, before my Sub-box fails me and I can't reach 220k to roll it OL. So thats fine.

The other ranges seem ok, but the lower 1 ones are probably off a bit, but now I'm back to like it was before, it reads 46Mohm open circuit. At the start of all this, I had it reading 50-150MO O/C by focusing on pin 9 of the Amp SW.

So somethings leaking, I was for instance measuring a voltages that would indicate 90nA going through C103, or by-passing it somehow. But trying to measure across R104-1.111M directly with my other Keithley DMM, didn't work, I guess that's not high enough I/P resistance, and the measured voltage dropped by 100.

No J-fet's are open at least, but as for how they leak and act when I'm trying to measure them in circuit, IDK......


But voltage is broke, S/C read as 2.5V on 20V, 3V on 200, more fun yet



So later I find that on the VDC ranges that use 10x gain on the buffer U102, there's 180mV on the N.I. pin of U102, and that's the 1.8V on the shorted output. Thats coming through R107-106, but from where ?


I found 6mV across R104, when AC/DC pin 3 goes nowhere, so I removed the SW springs, jumpered it, and still same problem. I already had C102 out, so I don't get it.

So I resoldered AC/DC SW pins 1 to 4 again 2x times, and now with my PSU set to 100mV, and GND attached to pin9 of the ohms SW, I'm going around 1 at a time, testing currents, through my good DMM on 20uA ranges.


I don't get where the 180mV is coming from. I see U101 only had 0 to 1mV on it's OP and Inv at the time, so basically GND. How the J-fets measure/adust for the errors, IDK, and IDK if that has failed.

Those currents, most all are very stable and repeatable, so not phantoms, and barely changed over the testing period.

This is fun tho, as long as it's fixable.

[trobbins]

I used the spray, flush and toggle technique to clean the front panel switches - with the board standing just over vertical so that the flush was unlikely to drip on the pcb surface.  I recall there was one switch whose rear was obstructed by a ceramic package.  I didn't want to dismantle any of the switches, so just used an electronic contact cleaner spray with extender thin tube to apply just to the switch openings.  The cleaner supposedly leaves some oil film to suppress oxidation and provide some lubrication - so not exactly IPA - but that exorcised a few quirky meter responses and I get OL on all ohms ranges.

[MathWizard]

Well now on volts I have -110uV stuck on pin 3 of the AC/DC SW, and somehow that's getting onto pin 4/5 which should be GND. And I had the same few counts negative on all ranges, scaled accordingly, that shows up by pulling U102-NI to negative 0.51V

So again I too out the AC/DC SW, and U104 is not the cause, I GNDed R113 on U104-NI, and it acts as it should, and it's not those 2 protection diodes, and only picoAmps are going through R113. But before, there was 200uV across it, and 5nA heading to pin 4, and to pin 3.

So I cut the topside trace to pin 3, and THERE's a current going through R104-1.111Meg, but pin 3 is no longer neg., now it's +70uV, but it should be floating.

On Ohms, I get the same 430 count error scaled across the ranges. And with 4.27V on R104-1.111Meg, and C102 out and the trace cut, calculations say there's 3uA still going through R104.

At least all this seems to be mechanical and PCB related, and not an actual fault. I'll pull the AC/DC SW alone, and flush it and pull R104. After cutting that trace, it only possibly touches the paint, on an old style brown epoxy resistor. Then there's C103, a 250femtoF-1.5pF varicap

Or am I tracking ghosts dealing with such small signals across large resistors

[MathWizard]

Ok I fixed it, it was the trace from R104 to AC/DC-p3, something under the SW, so I drilled the THH/pad, and jumpered it, and it's back to reading 0.5R shorted on all ohms ranges, and 50uV on 200mV range when shorted

2V range is around 0.1807V shorted,, I read 16mV on R106/7 that is still off maybe, maybe it's normal drift. Lots of old style resistor in here, I could replace some.

But for now anyways, GOOD ENOUGH

[trobbins]

Are you saying there was leakage along that pcb trace to other nearby traces, or the switch enclosure, or ?

[MathWizard]

Well there was DC and AC leakage from AC/DC-p3, into pin4, on the PCB tho, because when I pulled the pin out, I could still get the crazy readings touching the pad. I killed the worst problem by drilling the whole pad/via out, under the SW.

And I cut the trace, but STILL have AC current across the 1.111Meg and 2 caps, that all terminate on pin3, which is meant to be a deadend.

And I still had a 10uVDC across it on ohms, so I removed the wire, that I had attached to the TOP of the pin, that sits over the hole I made. Now it's only 1uVDC.

I measured the VAC that my other DMM says is 60Hz, and with the 2 caps, they are on the order of GigaOhms at 60Hz, at latest calculations I had 181nArms in that 1.111Meg.


Anyways tho, checking the 20V range, there's 15mV stuck on the bottom of R106/R107, and that screws up the 20V range, the lower ranges are very good.

And on 2Meg and up range, the reading drops down like the current is being split., 10Meg turns into 3-4Meg

So back at it.

[MathWizard]

On voltage mode test 0u3, here's whats across the discrete 1.111Meg R104, which I cut the trace off of, and there's nothing near it to touch at all. Yellow is on the lower potenial deadend side, that used to go to pin3, grn is the unwanted voltage, white is their diff

But where is this voltage coming from, because it's supposed to be at zero volts, or much much closer, otherwise the DMM would never have worked right on 20V, but it did.

[MathWizard]

So , with the schematic on page 66, look at the tank circuit R104 C102/C103, the side on pin3 again. the trace leaves R104, goes about 1-2cm...then I cut the entire rest of the trace off, so it's miles away from pin 3, and there's no other trace near the ends of either C102/103. And the PCB is cleaned.

Yet with the K197 unplugged* and with a PSU, when I attach a positive 2V to the OTHER side of R104, and the GND to the GND of the K197, and the Input is shorted, I still get the following, which I don't get.

*(it's -15V rail stays alive when turned off btw for any one using 1, it's for the field-battery pack charger)

2.00009V on that side, and on the "cutoff" side I get 1.81686V, so 183mV across what I measured to be 1.1108Meg gives 165nADC.

And when I removed the resistor and the fixed cap, I forgot to check the fixed cap, I better do that, but the other one  in circuit, was over 200Megohms, so OL.

I'm sitting with 60 Hz AC and SMPS all around me, are they know to cause 165nADC in anyway ? Is there some AC thing I'm missing ? Looking at the AC w/scope on the problem side of R104, it looks pretty normal with a Vrms of 385uV (my PSU is good, sure tracks well at low currents anyways). I'm new to FFT, but it looks all super low level, I can see the radio stations are among the strongest signals, but I'm not familiar enough with the units or settings yet. 60Hz looks like -140dBm, so isn't that 10AttoWatts, 10x10^-18 ? And isn't 183mV*165nA 30nW and -45dBm ?

Then there's the PCB material, I barely touched that area to remove those 2 parts, so it should be as it always was. I have to pull that resistor again. Maybe this is ust what high impedances do. I suppose I should check it on my LCR meter, since I never checked what a resistor looks like to it on L/C before.

It's a 1980s THH PCB, with traces on both sides, again with the closest traces 5mm away, so that's pretty far.