Author Topic: Keithley 2100: How does one identify a burnt SMD resistor without service manual  (Read 18402 times)

0 Members and 1 Guest are viewing this topic.

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
FYI, mine is stamped OCT 2008 on the outside of the metal chassis, under the outer beige cover.

For you volt nuts: It has the VRE310J reference.
« Last Edit: September 03, 2016, 12:06:36 am by FlyingHacker »
--73
 
The following users thanked this post: technogreek

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
BTW, you might want to post the error codes on the Keithley forum (and then tell us what they say  :-+). The seem to offer up some technical info when you ask.
--73
 
The following users thanked this post: technogreek

Offline adilmalikTopic starter

  • Contributor
  • Posts: 32
  • Country: gb
I bought this meter on ebay, no clue what happened to it. From the first insepction, i saw the two burned resistors so it seems it was some kind of over voltage fault. This is confirmed by the error messages that point towards faults on the high voltage measuring circuitry.I get these errors:

612 Ohms 500 nA source failed: This test configures to the 10V DC range with the internal 10M
100:1 divider R204 connected across the input. The 500nA ohms current source is connected to
produce a nominal 5V signal. A 20ms ADC measurement is performed and the result is checked
against a limit of 5V ± 1V.

614 DC 1000V zero failed: This test configures to the 1000V DC range with no input applied. A
20ms ADC measurement is performed and the result is checked against a limit of 0V ± 5 mV.

618 DC high voltage attenuator failed: This test configures to the 1000 V DC range. The 500nA
ohms current source is connected to produce a nominal 5V signal. A 20ms ADC measurement is
performed and the result is checked against a limit of 5V ± 1V.


The thing that relates the errors is the 500nA current source.

Thanks for the de-shielding instructions!


 
The following users thanked this post: technogreek

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Did you check the power rails? Just a good practice.

Maybe finding the 1000V attenuator will take care of two issues. I will be in mine again shortly. I will see what I see, but I am no expert.

Also, check the font/rear panel input switch (the long white gang switch). I had problems with mine. A little contact cleaner fixed it.
--73
 

Offline adilmalikTopic starter

  • Contributor
  • Posts: 32
  • Country: gb
Did you check the power rails? Just a good practice.

Maybe finding the 1000V attenuator will take care of two issues. I will be in mine again shortly. I will see what I see, but I am no expert.

Also, check the font/rear panel input switch (the long white gang switch). I had problems with mine. A little contact cleaner fixed it.

Ok opened the shielding as per your instructions and viola I found another burnt resistor! the resistor seems to be at the input of an analog devices op amp that has been set up as a buffer, I remove the resistor and replaced with my decade resistance box and all the error messages are gone! also I found out why I was getting higher readings than expected: tweaking the value of the resistor with my decade resistance box, I saw that for larger resistances I got a DCV reading much larger than expected so it's probably because this resistor got burnt and its value increased.

 so now all I want is someone to kindly send me a photo of that resistors value under the shielding :)

 
The following users thanked this post: technogreek

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Here are a couple of pics. My board looks a little different.

It looks like "01B", "018", or most likely "013". It was not clear even under a magnifier.

A meter shows it to be 1K Ohms.

« Last Edit: September 05, 2016, 04:14:46 pm by FlyingHacker »
--73
 
The following users thanked this post: adilmalik

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Wider view...
--73
 
The following users thanked this post: adilmalik

Offline adilmalikTopic starter

  • Contributor
  • Posts: 32
  • Country: gb
Here are a couple of pics. Mine board looks a little different.

It looks like "01B", "018", or most likely "013". It was not clear even under a magnifier.

A meter shows it to be 1K Ohms.

Thanks! I believe the resistor is not soo critical as long as its a stable one. I will need to re calibrate the meter tho :/

PS: i found the date code, mine is dated Oct 2011.
 

Offline adilmalikTopic starter

  • Contributor
  • Posts: 32
  • Country: gb
Wider view...

Interesting, you seem to have a different opamp than me. I have a analog devices
 

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Which op amp? Which reference IC do you have? Mine is the VRE310J
--73
 

Offline adilmalikTopic starter

  • Contributor
  • Posts: 32
  • Country: gb
Which op amp? Which reference IC do you have? Mine is the VRE310J

Its alive! And seems to be within calibration too! I checked it with another 6.5 Meter i have and they agree with each other! All the errors have gone too  :-+ :-+ :-+ :-+

The opamp the resistor i asked you is connected to. I have an AD822 and you seem to have OP282. All of your opamps seem to be different to mine.

I have the same voltage reference.
 

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Nice job!  :-+

And great deal on your meter...
--73
 
The following users thanked this post: adilmalik

Offline bendzo

  • Newbie
  • Posts: 6
  • Country: sk
Hi,
sorry for stealing the thread, but I'm not sure if I have a real problem...
I bought Keithley 2100 from eBay as faulty, it shows overload for DC voltage measurement at 10V range and lower, if nothing is connected to the meter,
but it shows 0 for 100V and 1000V, as it should, and it shows reasonable values (very close to actual value, I will try to check precision) for 10V and lower ranges if connected to lab power supply.
All self test are OK..
After switching from 100V to 10V range, it's clearly visible how is reading dropping very fast (~ 1 second) from 0 to negative values and eventually reaching overload.
I reckon that behavior for lower ranges is not normal?
Thanks for any suggestions or advises!
Jacob
 
The following users thanked this post: technogreek

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
The behavior is about normal:
The ranges up to 10 V are very high input impedance. For these ranges it is normal that an open input is not reading 0 or a value very close to it. Even a few 10's pA of bias current can be enough to slowly drive the input to overload.

The bias current might be a little on the high side if the voltage changes rather fast. To test one could have a 10 M resistor across the input and take that reading to get a measurement of the bias current. It should be below about 100 pA, thus less than 1 mV with the 10 M resistor.
 

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
The high impedance input behavior is normal, but yours does seem to go very fast. I would check the power rails. The electrolytic capacitors in these are known to go bad. Mine was kind enough to alert me to this with an error code.
--73
 

Offline Bashstreet

  • Frequent Contributor
  • **
  • Posts: 298
  • Country: gb
Yeah not bad idea to replace those caps before they puke up if they are known (rubbish)

Kind of depends if your planning to use it a lot if so i would do recap when you order some components.
 
The following users thanked this post: technogreek

Offline FlyingHacker

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
  • You're Doing it Wrong
Yeah not bad idea to replace those caps before they puke up if they are known (rubbish)

Kind of depends if your planning to use it a lot if so i would do recap when you order some components.

Absolutely, I would replace all of them next time you are in there... especially if they are branded "Stone." Those were the ones that died on mine. They started to bulge, but luckily had not leaked yet.
--73
 

Offline Bashstreet

  • Frequent Contributor
  • **
  • Posts: 298
  • Country: gb
Yeah no point to keep known bad caps. Of course OP will make his mind as what will be best option to him some people do not bother changing caps at all before there is a pop and smoke :blah: We are all different  :-DD
« Last Edit: November 20, 2017, 04:58:11 pm by Bashstreet »
 

Offline bendzo

  • Newbie
  • Posts: 6
  • Country: sk
Thank you!

Measured bias current: (1Meg -> -6.3mA) 6.3nA (consistent with other values too)

All electrolytic caps have been replaced (original was rubicon though)

I've checked most of "obviously precise" resistors and resistors burned in adilmalik's meter, they seems to be fine... voltage reference seems to be fine.

I checked keithley 2000 schematics, 2100 is reasonably similar, but I don't know what should I check, in relation to the issue.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
6.3 nA is really a little too high for the bias current.

May guess would be one of the JFETs in the input switching went bad. With some luck one could find the culprit with local heating/cooling. Chances are high that the current would go up / down if the already bad JFETs gets a little warmer or colder. So with a well isolated heater or cold rod one could test the JFETs (likely SOT23 form factor).

One point that could be measured directly would be the bootstrapping of the over-voltage protection. This point should be rather close to the input voltage.

How is the offset in the 100 V range ? If not compensated numerical, one would expect some -63 mV. One should also so a small (e.g. 0.6 mV range) difference between open and shorted.
 
The following users thanked this post: technogreek

Offline bendzo

  • Newbie
  • Posts: 6
  • Country: sk
Hi,
thank you!
I'm attaching thermal image of board, one JFET (looks to be input switching) is really hotter, I will try to take thermal images for different ranges (this one is for 10V) to see if there is any difference.

100V range is holding 0 very nicely, there is a small ADC noise of few last digits but it's really random (fluctuating around zero).

Could you please roughly describe how to measure "bootstrapping of the over-voltage protection" and how to confirm that JFET is bad by measurement?
I don't need full manual just few points.

Thank you very much!
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
A leaky JFET should not get really warm at that current level ( a few nA).  However if a FET gets really hot from other parts, it will leak.

As the 100 V range seems to be Ok, chances are the culprit is the JFET that is turned on in the 100 V /1000V range.

An in circuit measurement is difficult for the JFETs. There is a chance to find the bad ones from local heating or cooling.

To check the bootstrapping of the overvoltage protection, one should measure the voltage at the protection diodes, where the resistor for bootrapping goes to. It should follow the input rather close (only a few mV off).

It might be a good idea to get at least a crude schematics of the input section by reverse engineering. Chances are the Keithley 2000 is similar.
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
It might be a good idea to get at least a crude schematics of the input section by reverse engineering. Chances are the Keithley 2000 is similar.
Unlikely since the 2100 is not designed and built by Keithley themselves. It is the illegitimate sibling of the 2000 family of DMMs. Perhaps it should not bear the Keithley name at all, and be called the "Snow 2100"  :-DD
 

Offline bendzo

  • Newbie
  • Posts: 6
  • Country: sk
A small update, I started to do simple reverse engineering of the input circuitry (I found later that somebody did it few days before me, https://www.eevblog.com/forum/repair/keithley-2100-repair-jfet-issue/msg1359980/#msg1359980 ) and realized that Q204 a Q203 are connected in parallel, then I finally realized (complete brain fart from my side |O ) what Kleinstein meant by testing JFETs with cold/hot rod, so I did soldering iron and freeze spray test on FETs, only Q204 reacted to cold temperatures (leakage was double??), high temperature had only very small effect on both Q204 and Q203. So I removed Q204 and leakage current dropped to 100pA! New MMBF4393Ls are already on their way with new relays (classic and solid state AQV258), is there any point in replacing other components (LT1050, OP282, other FETs,...) just in case?
Thank you!
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
It is odd to see a bad part to show more leakage when cold. The more normal way would be more leakage when hot. Anyway broken parts sometimes do odd things.
Congratulations to finding the fault.

It might be worth checking the solder joints for the over-voltage protection (e.g. sprac gaps), maybe MOVs.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf