Keithley 2000 Self diag faults 400.2 401.2 601.2 (error @ 100mVAC range)

[pelule]

Hi,
I like to ask for some start-with tips from the more experianced Keithley heroes here in the blog.
I never repaired a Keithley meter yet, just Fluke and HP/Agilent systems.
Now I need to repair a potential defect Keithley 2000. So this will be my first Keithley.

The self diagnostic reports following faults:
  400.2 (VAC Non-Inv Path)
  401.2 (VAC Inv Path)
  601.2 (Amp shunt).

So I measured the values to check the performance and got following result:
  ok - VDC, IDC, Ohm/2W and Ohm/4W in all ranges
  ok - VAC in 1V range and higher
  stupid values only in the VAC 100mV range.

IAC I am not able to test, because I have no source.
But I assume, as the IDC >= 1A and VAC >= 1V is ok, it may ok also.

Meter was build end of 2013 (IC date codes 2013/wk32 to 2013/wk48), first & last cal was in November 2013.
Everything is looking quite nice & new, no visible damages on the board.

I have the repair manual and the schematics.

I assume a problem in the switching path to the AC amplifier circuit (U102) for the 100 mV range (U103-d, DG211).
Are there any other suggestions or tips?

Any help/tips welcome.

BR
PeLuLe

[TiN]

Can also show us new meter internals? I wonder about ADC, if it's still custom gate array chip.

[pelule]

I will post some tear down picture soon, it was quite late yesterday, so had to get some sleep (I guess you have experiances on that problem too :-)

[pelule]

Can also show us new meter internals? I wonder about ADC, if it's still custom gate array chip.

Here are the photos.
Seems, its using an integrating ADC from Analog Devices.

[3roomlab]

did you have a chance to trace a small signal thru the AC paths? using this pdf? page 74 = block diagram
http://circuitslab.case.edu/manuals/Keithley_Model_2000_Multimeter_Repair_Manual.pdf

and this schematic
https://dev.xdevs.com/issues/874 (TiN's website)

the schematic is for older revision PCB, but the general paths are very similar.

the key items to verify i think is the relay K101--U102, U105 switching into the opamp U112 (if can locate table 2-6, verify the switching vs range)?

looking at your picture, my K2015 is year 2007, and some sections are updated in your version. the DCV sections are newer in yours.
( in any case, i busted my ohms section, waiting for parts, again hahaha! i forcibly overloaded it)

[pelule]

Problem solved - I am no Keithley repair newbie anymore. 
 I started to follow the signal path with a 80mVAC input signal (followed 3roomlab's tipps), measuring the RMS-AC voltage at pin ACV with an additional  control meter (Fluke 87V).
When I measured with the probe at DG111/U103 pin 1 (expected a low), Suddenly the correct value was shown on my control meter.
Reason: a bad soldered pin, loosen over the time. I re-solder U103 and now all works fine. For security a did a check of the other ICs in that area, if they are potential bad also, but none found.
Tool runs again perfect, Self diag runs through, measured values in the low AC volt ranges are correct now. I checked also the AC current ranges - ok.
As the pin was the MUX control pin and not at a part of the signal path, I have the strong hope the calibration is not risked.
But I will verify in the next days.

@ 3roomlab: thanks for the experianced tips.

PS: I found Keithley maintanance description very good and sufficiant, especially the manual stepping through the diagnostic is very helpful.

BR
PeLuLe

[Macbeth]

Wow, first time I've seen this new board. First differences I spotted are the tiny U165, the R110 & R113 10k dividers and the soldered on rather than socketed firmware PLCC's. Perhaps they are now flash instead of OTP?

Latest known firmware for K2000 is A20. I have flashed A19 because it doesn't need a DC recalibration. What is your firmware?

Please, more hi res photos. Nice would be if you can get a backlight to show through the traces 

[Macbeth]

Can also show us new meter internals? I wonder about ADC, if it's still custom gate array chip.

Here are the photos.
Seems, its using an integrating ADC from Analog Devices.

AD637 is a well known TRMS IC, I think TiN is referring to U165 which is labelled U165A on this new board? (Bottom right of your first photo)

[TiN]

Good job on repair. It's great when it's something simple as bad soldered pin 

Yes, ADC is new, U165A in TSSOP. Did not see that one before. Perhaps a closeup pic would be possible?

[pelule]

Just have closed the meter and have started the "noise measurement" under following conditions:
- Keithely 2000 (build ~eof 2013)
- loggin data via GPIB in Fluke 8842A mode
- meter runs at 230Vac / 50Hz
- short on inputs
- no filtering or math activated
- fix range 100 mVDC (for my personal interest)
- fix range 10VDC (most precise)
- the other ranges will follow later
- measure in "slow" mode / NPLC 10 (my GPIB logging program is not able to do faster or I will get sync errors)
- started logging after 3h meter warm-up to get system temperature stabilized)

After the measurments I will open the meter and will (try) do some better pictures as requested.
My DigiCam is not the best.

BR
PeLuLe

[azer]

Latest known firmware for K2000 is A20. I have flashed A19 because it doesn't need a DC recalibration. What is your firmware?

If I squint enough it looks like it says B01 on the chips.
I saw on Keithley page there is a C01 firmware for the K2015 which is not compatible with the K2015 that run the B firmware http://forum.keithley.com/phpBB3/viewtopic.php?t=123600
So the C01 is then probably for a K2015 using this new K2000 board.

[pelule]

Hi,
have checked the meter details:
  Keithley 2000
  Ser# 4049298
  Firmware B01/A02
  Build ~eof 2013/~bof 2014

Did the noise measurements for 0.2VDC, 2VDC, 20VDC, 200VDC, 1000VDC.
But as my Excel is German based, it uses the comma as decimal point, thus the CSV format would ruin the data.
Need to check how to convert the data...
HRes pictures will follow, interested in any special details?

BR
PeLuLe

[pelule]

Hi
as requested, I made some high-res pictures of the board, also with backlight.
This is a huge amount of data.
I don't want to blow-up Daves database, thus I uploaded it to my dropbox.
You have access to it via below link:
https://www.dropbox.com/sh/sndkvfbgw70fnbw/AADvmtLe7YFSI_RjMUjY303ka?dl=0

NOTE: I will delete that pictures in 2 weeks, as I need the storage for a joined project.

BR
PeLuLe

[TiN]

Thanks for photos. They are safe here now.
You can drop your data onto my serv, as listed in here. Format does not matter, I can bodge python script to convert commas into dots, etc.

[Macbeth]

Great work Pelule! ... and of course thanks to TiN for re-hosting on the best Keithley site on the internet! 

[pelule]

@ TiN: great thanks for re-hosting the data.

I have detected my data is useless and I have to do again.
My selfwritten labview sw (for Fluke 8840A) formats and cuts the data to just 6 digits (ok for 8840A). I need to do change the sw to not-format and do again. No big issue, but need more time for.
I do not want to open a new discussion, but I got the feeling it would be a quick & easy job with Python.
I have no Python experiance nor any Python sw enviroment. (Simply means I have to add time for getting it running and learning to use).
I check a bit the WWW and found many recommendations for many different several IDE/GUIs. Thus become over-informed and postponed the decision.
BR
PeLuLe

[TiN]

I mostly use plain python on Raspberry Pi with GPIB, and can help with an app. You don't quite need IDE or anything, just your favorite text editor, python interpreter and few sleepless nights to get it all up
If your host is Windows, EZGPIB is easiest to run, as already have scripts for 2000 on it (using RS232 VISA).

[pelule]

Have just tried with EZGPIB and your K2000 script.
Could not get connected (Debug Windows shows presence of the GPIB card), thus I need to dig into it a bit more.
Only difference I found yet, is your K2000 has address 15, mine has 16.
BR
PeLuLe

[TiN]

That EZGPIB using VISA RS232, not GPIB.

[pelule]

Yes, I have seen. GPIB VISA driver is present and detected by EZGPIB, thus should work in principle.
But I need to look more detailed into the PASCAL code.
BR
PeLuLe

[pelule]

you are using RS232?

No, I am using the GPIB card. EZGPIB detects the card and the VISA driver. I also had the GPIB tracker running in the background and there was commands send out by EZGPIB, but generating only timeouts .
Thus the VISA driver is working and EZGPIB tries to communicate with the K2000, but gets no response.
So I assume, the commands are not correct for GPIB. I had no time yet to dig into, but will do next weekend.
Didn't also look in detail into the K2000 manual for the correct commands/programming.
Let's see on weekend.

By the way:
I tried to step through the K2000 script using the STEP (F7/F8) commands in EZGPIB.
EZGPIB steps (stepping changing outputs on the Console Output, but you can't see the actual code line highlighted.
Is that normal for EZGPIB or is there soemthing wrong?

BR
PeLuLe

[pelule]

Found some time to dig a bit into.
EZGIB is now communicating with my K2000 via the VISA/GPIB bus.

But there seems to be an error in TiN's script (or it may caused by the faster bus speed of GPIB against RS232/USB)
Initialisation works.
But
- the "TRACE:DATA?" query breaks with error -420 (Query unterminated).
- the shortly following "SYST:ERR?" query breaks with error -410 (Query interrupted).
(See attached screen & the VISA bus trace data - txt file)
So need a bit more time...

I will next try to use the Agilent USB-GPIB interface (to slow down response time).
If it works - I will do the noise measurements.
If it doesn't work - I will implement the GBPI-SRQ processing to get the time dependencies removed.

PeLuLe

[pelule]

Finally got the script running at EZGPIB.
Needed some optimizations in the timing as the GPIB bus is much faster than via RS232.

Works now via VISA @ my GPIB-Card and via VISA @ my Agilent USB/GPIB Adapter.
Actuall measurements are running. When done, I will upload at TiNs webside (incl. the changed script, someone may use)

I did a corrected the Voltage-Range values also. The K2000 ranges are 0.1V, 1V, ... (not 0.2V, 2V...).
Script is still not perfect (K2000 shows time to time error -420, query interrupted), but good enough to do the measurements.

PeLuLe

[pelule]

Just uploaded the data to xdevs.com via FTP.
https://www.eevblog.com/forum/metrology/dmm-adc-noise-comparison-testing-project/275/
BR
PeLuLe

[dagema]

Hi,

Just to add my experience with Keithley 2000 repair, I bought a broken one on Ebay with 400.2 error. After testing the voltage values on R271 pins, the unit began to pass all tests, so I simply reflowed all of the pins of R271 and everything works fine now.  The unit was built 1995, time taking it's toll on solder joints.