Help me repair a Keithley 2000

[mimmus78]

I scored a very good shape Keithley 2000 for 89€ (last calibrated in 2009).

The seller stated that everything was working except ohm in 2 wire mode.

I checked repair manual and this convinced me to bet that was K102 that has gone "marenghe".

Today the meter arrived and after a fast check it seems everything was as described.

With meter disconnected by mains, I opened it and checked K102 with its neighbour K101. The relay states was different confirming me that K102 was gone.

How much replacing K102 will effect calibration?

Another problem is front panel buttons are very stiff, do you think it worth to clean the membrane with some isopropyl alcohol, or just using it will improve?

[plesa]

There is thread https://www.eevblog.com/forum/testgear/keithley-20002015-thd/

The replacing the relay should not affect the calibration, but on repaired instruments you cannot be really sure what has been also affected.
So for serios measurement ask someone to check the accuracy or buy at least few precision resistors to check it.

Do not forget to check all capacitors on the power supply.

[ddavidebor]

You must ceck the schematics to be sure, but substitution  usually don't affect calibration.

Maiby some milliohm in 2ohm mode, but who cares.

[mimmus78]

Davide,

it seams schematics does not (officially) exists for this multimeter.

I read of your hakerspace  ... which city in Italy?

Domenico

[ddavidebor]

padua

[mimmus78]

Bel posto ma un po' lontano da roma :-)

[ddavidebor]

eh si.
roma è proprio una bellissima città, anzi LA città, il problema è che fare la spesa richiede 3 giorni e un elicottero.


just some random chatting about our country guys.

[mimmus78]

Ohh this puppy is driving me crazy!!!

When the meter arrived in my hands it showed the following faults: 302.1 302.2 600.1 600.2 601.2 (the latter maybe because I does not short the amp probe).
I decided to change the relay k102 (because? read first post) but after this the problem was not solved. So I put back original relay in place.

At this point I decided to follow what was stated in manual for fault 302.1 but some shit happens. After a while the 4 wire ohm started to read higher.
I run autotest again and now I have also error 100.2 (and many others related to this).

After this "shit" the 4W ohm is reading higher:

• 100 and 1k ohm: +5%
• 1M ohm: +30%
• other range: no consistent reads

Strange is that DC is still OK.

So I decided to go ahead and try to solve the 2wire ohm problem when at a certain point it started to measure similar values to the 4W ohm.
I just had time to run the 300 series tests with no fault and probe some voltage before the miracle to vanish.
During this time I notice that I have expected voltage (~ 0.490V for ~500 ohm resitor) to the input/output of U113. This is almost the same value I have with 4 wire and same resistor now.

Don't know ... any idea?
It's very hard without schematics to figure what's happening.
Nobody has any sketch of analogic section of this meter?

[mimmus78]

I finally found what was originally causing 30X faults (a cold solder joint on resistor R108). I was thinking this resistor series was used also by DC pat so I initially did not checked it.
I should be happy of this if it was not that I screwed everything else!!!

So now I still have this 100.2 fault, high and noiser readings on ohm range.

DC is still spot on (with others meters I have) so I do not lose the hope to repair it. I notice that if I set "fast" or "med" mode, with a 100 ohm resistor, switch to ACI and than back to 2 o 4 wires ohm it takes more than a minute to recover. During this time it starts reading something like 1M or 200k and counting down until 109 ohm (according to the other meter, resistor is actually 100.6 ohm).

I now this is more harder problem than the original. I would appreciate if someone have any idea?

[Galaxyrise]

I had a problem with a 34401a where DCV was working, but only on the lower ranges. This turned out to be an open resistor in the input pathway. The lower ranges still worked because what's another GOhm when you've got GOhm input impedance anyway!

Have you tried higher DCV ranges on your 2000? Running through the whole "performance test" section of the service manual may show some other failures, and they might be easier to troubleshoot.

[mimmus78]

Thanks Galaxyrise,

I'm also thinking to another open resistor (or better FET) because of the slower recover on ohm.

DCV is noisier but linear on all the ranges, so if some open path is the cause of both problems (noise and high/non linear measurement on ohm) has to be something on the ohm path ...

I used a "spring loaded" probe to check reference and reference opamp buffer. I measured a strange noise after opamp, something like 10mV (it has a 100 Hz component, but a lot noisier).

The reference too have some noise but so lower that I don't know if trust readings of 600uV with my Rigol 1052.

Don't know if OP is the cause or if the noise is injected by some other component on his input/output.

Any idea how to check from origin of this noise?

PS: if I set meter to 100M ohm (withour probes) it reads -15M

[Galaxyrise]

Tests 200.1 and 200.2 are for the voltage reference.  If you set it the manually run 200 and step, then at 200.1 you should measure 1V between the shield and TP105.  (I measure .983 on mine right now, and the manual says +-0.1)  At step 200.2, you should measure 10V (I get 9.73 and the manual says +-1V is acceptable.)  Both look "clean" on the oscilloscope.

If I run test 600.1 with an ammeter between HI and AMPS, the ammeter reads 0.9993ma (ideal=1ma), and I get 12mV at TP105 (ideal=10mV.)  Again, both look "clean" on the oscilloscope.  (I mean, I can zoom way in and get MHz noise, but they're basically flat at 10mV/div.)

Incidentally, since you have your case open too, I have a question:  Does your power supply make noise?  Mine makes some white noise, as if I'm hearing the switching frequency.

[mimmus78]

So on 200.1 and 200.2 it's OK (0.990V and 9,990V) and checking with scope both are clean (1mV and 4mV rms noise/ripple).

I run this tests many times and it's strange that sometimes both test did not failed.

Test 600.1 fails every time even if I see 1.063mA on the fluke 287 (maybe too much?) and a clean 11mV on TP105.

600.2 seems not to fail.

Relay works because I measured 10 and 0.1 ohm during tests. Anyway the meter measure current "correctly" on all its ranges.

Where did you get this manual? Mine does not explain what, and how to use this test points (or I'm missing it?).

Thanks,
Domenico

PS: yes also may power supply make a very high frequency noice ... it's not normal?

[Galaxyrise]

Where did you get this manual? Mine does not explain what, and how to use this test points (or I'm missing it?).

The manual I'm using doesn't mention the test points, it just says "at AD_IN".  I found that TP105 corresponds to AD_IN by experimentation.

Test 600.1 fails every time even if I see 1.063mA on the fluke 287 (maybe too much?) and a clean 11mV on TP105.

The manual says 9.5mV +- 1, but mine passes even though I read 12mV.  So your still failing self tests? Which ones?

PS: yes also may power supply make a very high frequency noice ... it's not normal?

I didn't know if it was normal or not, that's why I asked   Since both our units do it, I'm guessing it's normal.  (phew!)

[mimmus78]

So your still failing self tests? Which ones?

- 100.1 (50% fails - 50% is ok)
- 100.2 fails 100%
- 201.1 and 201.2 (80% fails - 20% is ok)
- 401.2 402.2 403.2 fails
- 500.1 500.2 fails 100%
- 600.1 601.2 fails 100%

I'm still thinking its only problem is noise ... so the 100.2 failure.

Since both our units do it, I'm guessing it's normal.

Yes I think almost every transformer make some noise.

[mimmus78]

I add some oscilloscope waveform if someone wants to help or in any case for future reference.

All capture made with spring on the probe to avoid common mode noise and with following settings:


Analog Ch  State   Scale    Position   Coupling  BW Limit  Invert
CH1        On     2.00mV/   0.00uV     AC        On        Off

Analog Ch  Impedance   Probe
CH1        1M Ohm      1X

Time    Time Ref    Main Scale    Delay
Main    Center      2.000ms/      -327.5600ns

Trigger  Source      Slope    Mode      Coupling     Level    Holdoff
Edge     CH1         Rising   Auto      DC           12.0mV   500ns

Acquisition    Sampling    Memory Depth    Sample Rate
Normal         Realtime    Normal          100.0kSa   


So this if first buffer (U139 LT1124) after the reference (LM399) and I'm trying to measure noise on the input of the opamp (pins: 2 and 6):



Don't know but it seems to me too much (my "cheap" non reference bench power supply can do it better).


Here instead is the reference out (pins: 1 and 3 of the U139):



Here still some noise, if I remember datasheet set 40uV as max. But considering 2mV is the limit of my scope don't know if this has any meaning (DCV is 6.969V).


Here is the most strange capture:



This is buffered reference at pins 6 and 3 of U133 during test 301.1. The buffered output from U139 go to U133 that is switched around the analog part of the meter and finally arrive to AD-in.

Don't know but this looks quite strange to me ... there is also common mode 100Hz noise.

I checked how signal arrive to to the AD-in and it's something similar to this: filtered but with the same (rms) of the original 100Hz common noise.

[CompElitePC]

I am sure you have already tried this, but what the hell,   did you try freeze spraying those areas of the board while trying the self test?  worth a shot anyway,   as for the noise it does look odd  I wouldn't let the 100Hz  distract you too much, even though its your first error i would seriously look for simple things... if you found one cold solder joint, possibly more opposite side of board may be?

[mimmus78]

Yes I "freeze" both side ... and it's this way I found first fault on ohm range.

Just to make sense of word noise: it means DCV reads jump "randomly" from +0.1XXX mV to -0.1XXX mV (100mV scale) when in fast mode, and +0.0020 to +0.0040 in slow mode (with input shorted).

I should be something trivial, I'm sure, because during some tests noise went away for 2 minutes, and even fast mode was very stable (up to last digit). I've tried to replicate it without success.

[Galaxyrise]

100.1 fails! I can see why you're looking for noise, then.  The repair manual says that 100.1 sets the ADC to measure 0V. 

I checked how signal arrive to to the AD-in and it's something similar to this: filtered but with the same (rms) of the original 100Hz common noise.

That does seem like a problem.  When running 100.1, I don't see any low-frequency noise at TP105.  Just 4mV of high frequency stuff.

Let me know if you want me to do any other comparisons for you, I still haven't calibrated mine and I'm happy to pop the cover and jam a probe in there from time to time.

[mimmus78]

@galaxyrise

100.1 fails only 50% times. 100.2 fails always.

Anyway this is waveform during 100.1 and 100.2 tests at TP102 (shield) and TP105:



and it seems less noiser than yours (less than 1mV).

The common mode noise at ADC happens on test 301.1 ... if you can check during this test.

Thanks,

Domenico

[Galaxyrise]

On 301.1, I see mains frequency (not 100Hz) at around 3mVpp, along with a LOT of noise.  Given the high impedance pathway the 7V takes in that test, noise is not really surprising.

If 100.1 and 100.2 are failing, then every test after them is near useless, so you're probably back to figuring out what's upsetting 100.1 and 100.2.

[mimmus78]

So some updates guys!

Yesterday I brought another K2K for parts/not working and it seems I managed to repair it.
I found the usual cold joint on C150 or R182 that are on the path from ADC to pin 1 of U130. This pin is the strobe signal of the shift register connected to the A/D MUX. The missing signal was causing the A/D MUX (before ADC) to never switch the various signals (reference, input, etc) needed during ADC measurements.

So now I have a working 6.5 digit multimeter (finger crossed) ... and something to compare to for not working one.

Let's back to the not working one ... 100 Hz is double of 50Hz that is our mains frequency in Italy. Don't know why I have double frequency ... there should be some explanation.

I compared output of the reference with the working K2K and noise seems to be pretty the same, except that on the working one I don't have the glitches you can see in the waveform of the not working one. I don't think this glitches can cause all the noise I have so I will ignore it for now.

Something interesting happens when measuring a 120 ohm resistor with meter set to 1Kohm scale. In this configuration I expect a very clean DC signal (coming form some buffer of the reference) on the resistor. And this actually happens on the working K2K, I have almost flat trace with less than 1 mV high frequency noise (almost like 100.1 test). On the broken one  I have instead a huge 15mV common noise. I found also that the common noise gets smaller as I increase ohm range, and disappear completely when meter switch to 10M or 100M ohm.

I will think tomorrow if this make any sense to me ...

[PedroDaGr8]

So some updates guys!

Yesterday I brought another K2K for parts/not working and it seems I managed to repair it.
I found the usual cold joint on C150 or R182 that are on the path from ADC to pin 1 of U130. This pin is the strobe signal of the shift register connected to the A/D MUX. The missing signal was causing the A/D MUX (before ADC) to never switch the various signals (reference, input, etc) needed during ADC measurements.

So now I have a working 6.5 digit multimeter (finger crossed) ... and something to compare to for not working one.

Let's back to the not working one ... 100 Hz is double of 50Hz that is our mains frequency in Italy. Don't know why I have double frequency ... there should be some explanation.

I compared output of the reference with the working K2K and noise seems to be pretty the same, except that on the working one I don't have the glitches you can see in the waveform of the not working one. I don't think this glitches can cause all the noise I have so I will ignore it for now.

Something interesting happens when measuring a 120 ohm resistor with meter set to 1Kohm scale. In this configuration I expect a very clean DC signal (coming form some buffer of the reference) on the resistor. And this actually happens on the working K2K, I have almost flat trace with less than 1 mV high frequency noise (almost like 100.1 test). On the broken one  I have instead a huge 15mV common noise. I found also that the common noise gets smaller as I increase ohm range, and disappear completely when meter switch to 10M or 100M ohm.

I will think tomorrow if this make any sense to me ...

Full wave rectification results in a 100Hz signal (because both peaks end up on the same side).

[mimmus78]

Tada!!!

I found cause of the 100Hz common noise (and maybe noise in general)! VR109 is gone.

Without this zener (voltage regulator) U134 is directly connected to the capacitor after the bridge rectifier ... here where this noise is coming from.

So now problem is to find this part without spending 20€ in shipping consts ... or maybe find similar part.

[ddavidebor]

What part it is?