Restoration glory of Keithley 2001 DMM

#700 Reply
Posted by connor on 29 Jun, 2018 14:41
Hello,My 2001 have a problem，when the power on，The ARM is not display，so it can not test anything。what is the problem with it ?

#701 Reply
Posted by TiN on 29 Jun, 2018 15:26
Likely no problem. You just need to read manual about triggering meter.

#702 Reply
Posted by Mr. Scram on 29 Jun, 2018 16:31
At least have the decency to put some NSFW tags on this thread. You're going to get me caught watching electronics porn on the job with threads like these.

#703 Reply
Posted by connor on 01 Jul, 2018 12:37
Hello,I have read it，config-trig-Arm ,but it can't save the settings, when i setting in confirm page,the display will freeze, the same as the test page and the Reset page,is the digital board have same problem?thanks.

#704 Reply
Posted by TiN on 01 Jul, 2018 14:22
Could have problem. If meter is old, taking it apart and replacing all electrolytics is mandatory. There is enough information and photos in this thread to understand why is it so .

#705 Reply
Posted by connor on 01 Jul, 2018 15:29
thanks,this is a very old meter ,the first calibration is 1994 ,I have read you blog,and i have changed all the Electrolytic capacitors, so is there any suggestions?
The TP601 in digital board is AC 2.5V or DC 2.5V? The other voltage is normal except this one.

#706 Reply
Posted by TiN on 01 Jul, 2018 16:30
I didn't ever test TP601 before, you can check schematics for what it does.
Bad VFD DC/AC/DC module can be also reason for digital board to hang randomly.
I had to replace those in few meters before.

#707 Reply
Posted by Robert763 on 30 Sep, 2018 18:58
Well threads been quiet, so here goes,
I just aquired a 2001 unexpectedly. It was on a test equipment dealers "spares or repair" pile at a Hamfest. Marked as "dead" and £25 I got it for £20 ($25). It's a early one '92 date codes and A06 firmware. Investigation sshows it's not too bad visually but C624, C630 and the VFD inverter are missing from the digital board.
I've ordered a full set of electrolytics and will lash something up for the VFD supplies. Further reports when the capacitors are replaced. The 2001 is a a lot smaller than my other 7.5 digit meters (Solartron).
I note that a couple of contributors have commented that C116 and C117 are underrated at 35V as they are part of the 38V bootstrap suppy. This is not correct, the capacitors are part of a voltage doubler and are in series with C114/C115 so see about 1/3 of the supply. Even if they saw half the supply it would only be 20V so 35V is OK. Nothing wrong with using 50V capacitors if all other ratings are similar, but lower voltage capacitors in same size and range tend to have higher ripple current ratings.

#708 Reply
Posted by Kjelt on 14 Jan, 2019 16:21
Yes! I got a Keithley 2001 !
Now I have to get it to work again

So I got a nice Keithley 2001 from a friend: status broken does not work.
The unit came from a testrack and thus misses its hinge and feet, not that important to me.

Mains fuse was blown, but before replacing and trying first inspect what could have caused this.

Ofcourse I was curious to see what was going on under the hood.
Knowing the horror stories of leaking caps with these beauties I was afraid what to find.
On first short inspection not that much seemed to be going on.
No obvious oxydation or blueish copper salts on the components but there was this oily stuff... hmmmmmmmm

#709 Reply
Posted by Kjelt on 14 Jan, 2019 16:22
Desoldering the el-caps reveiled the real damage, except one all capacitors were leaking and these were real Nichicon, so far for japanese quality
Bad series probably or thermal abuse over the years.

Most terrible experience thus far:
1- soldering this gunk releases an odor similar to a 25 year old cat peeing on a hotplate.
2- pulling the capacitor slowly and carefully while heating both pads and still find the oxydized/rotten vias attached to its feet.

Next steps is to clean up the mess with losts of IPA and a toothbrush and see how far I can get with replacing the caps and bodgewires.

#710 Reply
Posted by Kleinstein on 14 Jan, 2019 16:28
For the initial clean up of the electrolyte water (e.g. deionized) is more effective than IPA. Pure IPA is not good in solving salts.

#711 Reply
Posted by TiN on 14 Jan, 2019 17:01
Based on photos you must remove ALL components at 5cm radius from electrolyte pond , clean board very very well, and replace parts back in. I would recommend just ordering new parts, most of them are just cheap ICs , resistors and caps.

Maybe you spend 50$ on parts and wait few weeks, but its better than powering meter with almost 80 volts floating supply and making a nice electrolysis station under not cleaned parts instead of working DMM.

Also blown fuse indicates that mains section need same treatment, and perhaps big ceramic resistor is open. Perhaps my meter can act as example.

Sent from my Lenovo K900_ROW using Tapatalk

#712 Reply
Posted by macboy on 14 Jan, 2019 17:13
Follow TiN's advice.
If it isn't obvious, when cleaning, hold the boards so that the dirty runoff flows away from the sensitive parts of the board. Don't just lay it flat, soak it and scrub away. You'll contaminate the entire board with conductive electrolyte, effectively ruining it. This thing is supposed to be able to measure G\$\Omega\$ or nA, and it can't do that if the pixies can flow anywhere they want to over the surface of the board.

#713 Reply
Posted by Kjelt on 14 Jan, 2019 17:53
Thanks for the advice will sure follow it.
Unsure if all components are easily found but start with clean water then desolder the parts then scrub with bioethanol a d finish with IPA.

#714 Reply
Posted by Kjelt on 26 Jan, 2019 12:27
The exposed area is too large to replace all components It is the entire red area.
I am now starting from basics, I have cleaned thoroughly and will repair pcb traces and replace capacitors and then see what the status is.
If the power supplies are ok and I have some life I may continue.

For now I am looking at the capacitors. I saw on the xdevs site they used chemicon capacitors.
At my sources Farnell and TME these are not available, I am thinking about replacing them with the same brand and series, so Nichicon UVZ(M).
At least I think this was a bad batch or was the capacitor choice poorly made by the original designers ?

I also see that the 470uF/63V caps at xdevs have been replaced with lower ESR types, any reason for that, high ripple ? Or just prevention?
Any advice on other brands if the Nichicon are out ?

#715 Reply
Posted by coromonadalix on 26 Jan, 2019 13:54
For a good cleaning, at my job we use VIGON SC 202 or Atron (dont recall the number)

We use an 5-7% concentration and spray it with deionized water and it clean pcb's very very well, anything greasy, flux etc .. i found out it remove the leaking caps mess too ... we rince with deionized hot water and put it in a heated chamber 110-120 degree C for 15 minutes.

I dont think all the surrounding parts needs to be changed, but while cleaning, its important not to push the electrolyte further into the surrounding parts.

I had exploded. badly leaked caps on pcb's, with a good cleaning/neutralisation, surrounding parts never corroded and no ic's where damaged.

my 2 cents

#716 Reply
Posted by TiN on 26 Jan, 2019 16:34
I would still go with removing all "electrolitified" parts and doing thorough cleaning under packages. If you just clean without removing, electrolyte that remains under package will cost you days and weeks or senseless troubleshooting for leakage and drift issues. Been there, done that, never again. So it's just tedious and time-consuming at start, but after that part is done you will be saving days of time in further repairs (or maybe no need other repairs at all, like my second K2002 which use same components/circuit for marked red area).

#717 Reply
Posted by Kjelt on 26 Jan, 2019 18:16
Quote from: TiN on 26 Jan, 2019 16:34
I would still go with removing all "electrolitified" parts and doing thorough cleaning under packages.
You warned about the heat sensitivity of the Keithleys pcb's. So how do you suggest removing esp the ic packages without destroying them by for instance cutting the legs.
I am to afraid to damage pcb traces with heat.

The packages look clean though. The pcb traces affected are all around the caps location, further away I can not see any sign of erosion.

#718 Reply
Posted by ddcc on 09 Feb, 2019 10:17
I recently picked up a Keithley 2001M (advertised as 2001), which came with firmware B16, and decent-looking Nichicon capacitors without any sign of leakage. Thanks to the information on TiN's website, I upgraded the unit to firmware B17, and added the MEM2 option. Everything seems to function correctly, including the self-test, but when I perform the open circuit step of user-level comprehensive calibration, I'm getting error 363, which is described as "200uA gain out of spec". The closest corresponding self-test seems to be 309.1, which routes the 89uA ohms source through the 200uA current range and checks for 89mV at A/D input. Now, I'm a bit stumped where the problem is, as I'd like to avoid desoldering any of the precision resistors, e.g. R344 (900, 90, 9) and R215 (24.5k, 500), the ADC board is mounted upside down over the A/D amplifier/ranging area, and I haven't found a reference locator for the layout (e.g. where is A/D IN?).

Currently, I suspect one of the analog switches for the current ranging, and I'm thinking about replacing all of the ‎DG211BDY/DG411DY parts in this area (U323, U317, U320). I don't think the problem is with the main opamp U322 (LT1007) or the gain selection resistor R215 for 1x/50x, because the unit works fine otherwise, and I haven't received any errors about DC gain, which appears to be calibrated first during open circuit calibration. Rather, while probing the 200uA and ACA switch inputs for current ranging, I see the switches actuate, but there's also a slight ~2mV offset between the input and output of one of the analog switches. Does this seem reasonable?

#719 Reply
Posted by ddcc on 16 Feb, 2019 06:44
Update: I swapped out the analog switches, but the problem still remains. I've also verified all of the current sources (e.g. 9.2mA, 0.98 mA, etc) in two-wire resistance mode, so that's not the problem either. Since I know all the switches are good now, I plan to focus on checking the calibration-related components around U322, the main A/D amplifier, including R340 switched through /ZERO, the 1.75V reference switched through /2VREF, etc, before I consider touching any of the precision resistors.

#720 Reply
Posted by TiN on 16 Feb, 2019 06:59
Did you check front/rear switch? Check current with scope and actuate switch few times to see if you get same stable behavior every time. AD IN is pin 4 at ADC board. Also run self test many times to see if any rogue errors still might be hiding.

Sent from my Lenovo K900_ROW using Tapatalk

#721 Reply
Posted by ddcc on 16 Feb, 2019 07:06
Quote from: TiN on 16 Feb, 2019 06:59
Did you check front/rear switch? Check current with scope and actuate switch few times to see if you get same stable behavior every time. AD IN is pin 4 at ADC board. Also run self test many times to see if any rogue errors still might be hiding.

I haven't touched the front/rear switch, everything I've been doing has been with it set to the front. Since it's a physical push switch, can the rear affect anything when it's set to the front? I will check it tomorrow. I'll also check the current sources with a scope, so far I've only measured them with another meter. I've probably run the self-test at least ten times now, but unfortunately it's never found anything.

#722 Reply
Posted by ddcc on 20 Feb, 2019 06:29
Thanks for the suggestions! I checked the output of the current source using the front panel in resistance mode, and something definitely seems weird. With 1M $\$\Omega\$$ load, the 200 $\$\Omega\$$ range shows a bit of a glitch in the output voltage, perhaps it's just a bit of noise... But, as I move into the 20k and 200k ranges, it gets significantly worse, with the 200k range showing a drop of ~2.4V at around 2.4 Hz. Unless this is intended behavior, there's definitely something wrong with the current source!

Aside from that, everything else looks fine. Nothing seems wrong with front/rear measurements, and the resistors around the A/D amplifier seem fine, though they're not easy to measure in circuit. I don't see the 1.75V reference, but I'm probably not in the right mode for it to be generated.

#723 Reply
Posted by ddcc on 03 Mar, 2019 01:17
I spent some more time measuring various components, but unfortunately I think I might have been misled by the behavior of the input buffer when it is overloaded. Because I was looking at a 1M $\$\Omega\$$ load in the 20k $\$\Omega\$$ range, the voltage was around 12.76V at OHMS ISOURCE, and at 5.7V going into the input amplifier U322, which pushed up +8VF to 12.77V and -8VF to -1.41V. I wasn't able to figure out exactly where the glitch was coming from, but I'm pretty confident the problem isn't with the ohms current source, because everything looks correct during both self-test and normal non-overload conditions.

On the other hand, I'm still no closer to determining the cause of the calibration failure. I've measured and reviewed a lot of the ohms circuitry, but I still don't quite understand how some parts of it operate. For example, I'm not sure what the purpose of comparator U333 and buffer U331 are, the function of the nearby OHMOLD / 10mA BIAS section, and why U331/U333 are connected to the switched capacitor block U330. Also, given the 7V output of the LM399 and the 1.75V reference voltage, I'd expect U330 to be configured for dividing-by-four, but the schematic doesn't match that of the example given in the LTC1043 datasheet.

#724 Reply
Posted by openloop on 03 Mar, 2019 03:41
Quote
and why U331/U333 are connected to the switched capacitor block U330

[My understanding:] They just move the reference over to different DC offset. And U331 buffers it.

2.4 Hz pulses are probably from autozero. Switch it off to check.