Keithley 2002 repair help

[nikonoid]

I finally assembled the instrument back together, gave it 24 hours of warm up and now started to verify performance before sending it to be calibrated.

I would like to thank TiN, Le_Bassiste, Kleinstein, Samogon and others who helped me though the process .

Performance evaluation in NORMAL speed mode (NPLC1, I think):

Copper short on 200mV range after long warm up fluctuated a bit between -1.3 and -1.6uV.  StDev=93nV over 1000 readings.

5 volt reference from DMM check plus ("5.0000" as of 2 years ago when I bough it) gave me 5.0000295 or 5.9ppm out.  StDev=1.3mV (0.26ppm) over 1000 readings. So pretty good for the reference and a meter.

Bias with 10MOhm resistor on 200mV range measured  120uV (for some reason it was a bit less when meter was not fully assembled and not completely warmed up). So 12pA of leakage

Resistance worked fairly well comparing to my other meters and standards, however noise on resistance was somehow higher than what I would expect, especially at 1MOhm:

Code: [Select]

Nominal(Ohm)    Measured      ppm      StDev     StDev(ppm)   Readings
10,000,000.00   9,999,711    -28.9     15            1.50     1000
 1,000,000.00   999,942.6    -57.4     1.5           1.50     1000
   100,000.00   100,000.89     8.9     0.074         0.74     1000
    10,000.00   10,000.131    13.1     0.0038        0.38     200
     1,000.00   999.98898    -11.02    0.000131      0.13     200
        10.00   10.00081      81       0.000010      1.00     200

Please let me know if these look ok before sending for calibration an if there are any additional parameters/test I can ran before shipping it out. I am considering using 4GTE company. They have Fluke 5700a they can calibrate it on. If you have other recommendations for good and reasonably priced calibration lab in US, it would be great to know.

On the strange side, I found two things: meter is reporting internal temperature 27F above calibration temperature. How could it be? The calibration date is from 1999 (about a year after the chips date).

Another peculiarity is that at some point meter gave me very jittery Ohms reading. When I went to General menu to check Line Sync is was displaying 400Hz despite my power being 60Hz. Single power cycle of the meter fixed it and now it properly shows 60Hz. Have you seen something like this?

[Samogon]

What 4GTE quotes are?
Simple cal without cert quoted 600 from Tektronix

[nikonoid]

I thought 4GTE was $240. I will have to call them tomorrow to confirm if it is with data or not. They previously did a good job for me.

Any other suggestions?

[MadTux]

Someone lucky scored another crusty 2002 for a spectacular price.
Of course I missed i t ;-(
http://www.ebay.com/itm/KEITHLEY-2002-Digital-Multimeter-8-Digits-Factory-Calibrated-/282569341996?epid=1701817052&hash=item41ca71fc2c:g:AkkAAOSwWdZZZsb0

[nikonoid]

That was actually me. I could not believe it when I saw it by pure chance and pulled the trigger right away.

It looks bad, but it will be fun to restore it. Arriving today. I will post some pictures.


Sent from my iPhone using Tapatalk

[lowimpedance]

That was actually me. I could not believe it when I saw it by pure chance and pulled the trigger right away.

It looks bad, but it will be fun to restore it. Arriving today. I will post some pictures.


Sent from my iPhone using Tapatalk

Lucky indeed 
Will be interesting to see the before and after pictures 

[nikonoid]

Ohhhh HORROR!!!!!

The package arrived today. I was very surprised that it was not a box, but bubble wrap blob. Oh horror!

I started opening and realized that few layers of thin bubble wrap were put right on the meter. No cardboard or peanuts anywhere. No other cushioning at all. Oh horror!

When I took the bubble wrap off the meter inside did not have the metal case! Oh horror!

More over it did not have a analog board shield at all. And apparently it was stored somewhere without the shield as the analog board is full of dust. Oh horror!

Looking the the board closely revealed few.... mice droppings on analog board! Oh horror!

Upon quick inspection I found at least two places where the analog board was apparently burned through. It appears that internal traces overheated and charred the PCB above them to charcoal. Oh horror!



I uploaded high resolution pictures to TiN, to I can post them.

Seller was able to power up the DMM up, but I am not as brave. I am going to follow Dave's typical advise "don't turn it on, take it apart".

For decontamination I am considering a large ultrasonic bath with distilled water and IPA. Or possibly using syringe to power wash it. 

Suggestions and condolences are welcome (;



Sent from my iPhone using Tapatalk

[TheSteve]

It's beat to hell - looks like a good challenge that needs a lot of TLC. If you're not up for it there are many of us who are. I'd love a go
Paypal ready and willing 

[TiN]

At great price comes great responsibility  .
I don't ever remember 2002 going for that low.

New analog board cover kit is ~150$ from Keithley/Tek.

[alm]

Looks like an interesting repair .

I would remove anything that can trap water (e.g. switches, non-sealed relays) or that you will replace anyhow (electrolytics?). Then rinse with demi water (to get clean water in all the nooks and crannies), then demi water with some mild soap, then demi water again and then IPA to get rid of the water. Then try to figure out what blew the traces (short on the analog board? excessive voltage from power supply?) and go from there.

[nikonoid]

Here are more images. TiN, thanks for hosting.

The digital board looks excellent and has MEM2 option installed while analog is burned and dusty. The date codes are from 1994 and firmware looks like is A05.


Here are some details of bad parts. First image is intentionally overexposed to better show charred parts.


Few more pictures of the bad section. It looks like the electrolyte spill. It also appears that one of capacitors short circuited and that started chain reaction.




I already have few pars ordered from Digikey.

Would anyone know what kind of diodes this section is using? The repair manual lists them as "DIODE, CONTROLLED AVALANCHE, BYD17GSOD-87".

Also, the CR113 rectifier is listed as "DIODE, BRIDGE, VM18". I could not find it and ordered DF01MDI-ND instead. Would that be a good choice for replacement?


Tomorrow I will call Keithley for replacement cover and metal casing. I still cannot believe I got it without the case, but with plastic handle!

At great price comes great responsibility  .

Love the quote.

[saturnin]

Tomorrow I will call Keithley for replacement cover and metal casing. I still cannot believe I got it without the case, but with plastic handle!

If I were you I wouldn't complain. You got exactly what was shown in pictures in the auction. One can always ask a seller about further details, but in such cases (extremely low price), there is a risk someone else would buy it before you get answer It is always lottery, if it is repairable or not. I wish you it was. Looking forward to repair details 

[TiN]

WYSIWYG of T&M world.

I better buy third 2002 pronto, or Keithley freak title might go to nikonoid soon
Even if you decide to give up, just voltage reference parts alone would cost more than what was paid .

[macboy]

Tomorrow I will call Keithley for replacement cover and metal casing. I still cannot believe I got it without the case, but with plastic handle!

If I were you I wouldn't complain. You got exactly what was shown in pictures in the auction. One can always ask a seller about further details, but in such cases (extremely low price), there is a risk someone else would buy it before you get answer It is always lottery, if it is repairable or not. I wish you it was. Looking forward to repair details 

Exactly. I did not for even a second assume that the cover or shield were included. Nowhere to be seen = not included. Dirt and corrosion were assumed to be free though.

[Kleinstein]

The burned trace seems to be in the power supply part. So worst case one could isolate that part and have a new power supply board. Also leakage is not that critical in that area. Chances are that after replacing the caps, a few diodes / rectifiers and with maybe a bodge wire to replace the burned trace it might actually work.  A shorted power supply is more like a good sign.

Really bad would such a burned trace in the Ohms or low current area, where all the guard traces are.

The analog part cover and the case is more like one of the last parts to fix.

[nikonoid]

I am certainly not complaining. This was a great purchase. My reaction was more to a condition of the unit being an insult of a fine instrument.

I like the challenge and am looking forward to restoring it, hopefully with the group's help. The help and advise I got while repairing my first K2002 was invaluable.

For now my plan is this:
1) Remove analog board
2) Identify and replace burned parts and repair traces. I expect replacing bridge rectifier, diodes and caps.
3) Power up the meter and self test with analog board outside of the case
4) Address self test issues
5) Decide on overall cleaning and decontamination procedure
6) Recap, new shield, case

[alm]

I would probably clean the board(s) after removing suspect parts, and before mounting new parts or testing. The leakage may very well mess up the self tests anyway, so I see no reason to test it before hand. It may also make it easier to see defects and nicer to work on.

[TiN]

Yea, item number 5 should be after number 2, not the other way around.
However PSU section for 2001/2002 analog domain is bit tricky, because of capacitor bootstrap doubler. That's exactly what two of the big caps near transformer are for, doubling unregulated +18/-18 to +36/-36 BS rails, which float around input to provide high impedance on main ranges. These rails power many opamps on frontend, and risk is high that one or more of those are taken out.
Schematics for that part almost same as 2001.

[nikonoid]

Thank you for suggestions.

On another note, today I got a call from local calibration lab (New Jersey) telling me that my first Keithley 2002 failed the calibration.

The tech said that meter was in spec everywhere but in DC Volts that were out of spec by a bit. However his attempts to calibrate it were not successful. His exact words were that "unit does not hold cal on DC Volts". He tried several calibrators but could not get new constants to store.

It sounds like EEPROM problem or new constants are too much outside of normal range and firmware is refusing to use them.

I will have it back on Thursday. I am considering doing a dry run calibration myself. Not that it would be a good calibration, but just to see what exactly is not working. I got calibrated K2000, EDC 520A, GenRad 1433-B for resistances and Agilent 6611C for 2 amps range.

I might also consider changing calibration parameters manually though secret menu.

Have you ever had a similar situation with Keithley 2001/2002? Do you have any other suggesting on what to try?


Thank you.

[Zucca]

Have you ever had a similar situation with Keithley 2001/2002? Do you have any other suggesting on what to try?

One fo my 34401A  has same issue... fully story

https://www.eevblog.com/forum/repair/hp-34401a-drift-in-the-2w-r-mode-in-the-100-ohm-range/25/

At the end it was a bad cell in the EPROM that would not get written... and this drive you crazy because you triple check all the analog circuit, but in the end is a tricky digital failure. Fortunately your cal lab pointed you in the right direction, mine was just saying 2WR 100Ohm is out of spec... good luck.

[TiN]

2001/2002 are not friendly for hobby level calibration, meaning that you must run ALL calibration steps without error to allow it to store data into NVM (which is pair of 24C16 EEPROMs in case of 2002). 

Also since you had unit repaired and also it's an old one, you must run low-level manufacturing calibration, otherwise your current, resistance and AC functions will not work/calibrate correctly. For this you need Fluke 5700A/5720A or Wavetek 4808 (due to 200V 30kHz requirement) + HP 3325/3245/similar signal generator for 2V 1Hz AC. I think I have the footage recorded of both 2001 and 2002 cal procedures run, will need to find it and upload to YT, if you interested. 

Then user-calibration or point cal adjustment can be ran to tweak ranges/functions. They still run in groups, so you cannot calibrate single range in traditional manner. GPIB control is required for easy calibration process and advanced error reporting. 

There are few "self-calibration" steps during which meter measure internal signal path sections, comparing to externally measured references, and if any of them failed for whatever reason (gain error or offset error) than meter will throw whole calibration data away and you must start over from the beginnig. It took me a while to sort all this stuff out, and it's definately not an "few hour playjob with access to friend's calibrator" sorta thing. 

[Le_Bassiste]

For decontamination I am considering a large ultrasonic bath with distilled water and IPA. Or possibly using syringe to power wash it. 

Suggestions and condolences are welcome (;

Sent from my iPhone using Tapatalk

no condolence but congrats, nice find!
as for the suggestions: clarify first with component manufacturers as to whether the ultrasonic can harm any components. if in slightest doubt, don't use ultrasonic. just, don't.
also, the shroud over the analog board (or, to be more precise  , the absence of it) will really bite you when trying to debug the analog section (fingers crossed that you don't have to do that) , as it is ultra sensitive to any temperature fluctuations across the board.


...my 0.02.

[Dr. Frank]

..
For decontamination I am considering a large ultrasonic bath with distilled water and IPA. Or possibly using syringe to power wash it. 

Suggestions and condolences are welcome (;

Suggestion: Never use ultrasonic bath on assembled PCB.
This will probably kill many/most of the semiconductors.

Frank

[WastelandTek]

OK, I am subbing to this thread

this is getting really interesting

[Samogon]

Probably it depends on what ultra sonic cleaner to use. Crest for instance is proved to be good for Mac motherboards and cell phone pcb cleaning.