Some help troubleshooting a Keithley 196 with bad ohms reading(s) (SOLVED)

• Some help troubleshooting a Keithley 196 with bad ohms reading(s) (SOLVED)
  Posted by staze on 28 Mar, 2020 20:04
• Hi All,
  
  I've got a Keithley 196 that isn't reading ohms right. They never settle (even when shorting the terminals together), and often go negative. I assume there's some leakage in the current source path, but I'm having a heck of a time figuring out what I should be looking at in the Schematic.
  
  Anyone have any places to look? The manual talks a bit about how the ohms path works, but it's mainly "Protection on the ohms ranges is accomplisehd by RT1, Q9 and Q16. For input voltage applied to the  terminals, Q9 and Q16 clamp the voltage to the reference resistors to a safe limit. RT1 limits the current to Q9 and Q16."
  
  The equivalent circuit shown just shows the reference resistors, which I assume isn't the issue, but rather some leakage through a FET or BJT. The second image on 5-4 may be the answer...
  
  I've also attached hi-res schematics. Any help would be greatly appreciated!
  
  Also, hope everyone is staying healthy! I'll admit, I'm working on this because I'm stuck at home... nevermind I have other meters in perfect working condition. Just figured this would be a good opportunity to fix this one.

  ---

  

  196 Schematic.pdf
• #1 Reply
  Posted by JKKDev on 29 Mar, 2020 06:55
• Disclaimer: I have never used this unit and have only looked at some pictures of it.
  
  It appears it offers a 4 wire sensing for resistance. Are you sure it isn't required to use 4 leads? Have you tried it with 4 leads? That seems to often be the problem
• #2 Reply
  Posted by staze on 29 Mar, 2020 18:30
• It doesn't need all 4 leads, it's only optional. I have another 196 that works fine with 2 leads. =)
• #3 Reply
  Posted by staze on 18 Mar, 2021 04:37
• Still working on this and finding some interesting things.
  
  One, I'd repaired this ages ago and accidentally subbed in a 680pf poly cap for a 68pf polystyrene. Found someone selling PS caps, have since replaced. No change.
  
  Found some solder joints on the main header between the analog and digital board. repaired. No change.
  
  Found that tapping on parts of the board seems to influence the reading. At this point, it behaves like it's charging something. If I measure a dead short, it slowly drops to 0ohm, then past 0ohm to negative 78ohms or there abouts, and then is stable. If I unhook the short, it climes to overload (this is in auto). All ranges seem equally "off". It does behave better after it's warm, but not hugely so.
  
  I can't find any other cold joints, nor find definitively where on the board tapping results in the change. tapping hard results in large bounces. tapping softly around doesn't result in much. The only thing that tapping softly on makes any difference is one of the relays... =/ boy I hope it's not them. Tapping hard anywhere results in swings in readings.
• #4 Reply
  Posted by coppercone2 on 18 Mar, 2021 07:30
• the relays are easy to test, desolder them and put it on veroboard and do tests within the specs of the part
  
  you can also do the mux but you might need to use a electrometer by connecting it to loads and measuring the leakage with different inputs
• #5 Reply
  Posted by staze on 20 Mar, 2021 00:24
• Ugh, yeah, I'll see about pulling the relay in question. Interestingly it's K1, which is the sense hi relay. =/ The throughholes on this board are not great, I'm always super concerned about them completely delaminating.
  
  As far as the MUX, I'm honestly not entirely sure there is one. There's some logic switching inputs, but I think the MUX itself is just a series of JFETS.
  
  Here are some higher res schematics (they're slightly too big to attach to post): https://www.dropbox.com/s/tuktvt76i6mdv0d/Keithley%20196%20Hi-Res%20Schematics.pdf?dl=0
  
  Will keep poking, and get that relay off. Hope it's not that, they're not cheap (well, the modern equivalent).
• #6 Reply
  Posted by staze on 20 Mar, 2021 23:31
• Yup, K2 relay is bad. Not entirely sure how that makes things as wonky as they are... but removed and attached to PSU, 5v across coil (20mA), and resistance across relay sometimes is 10's of ohms, sometimes is 10's of K, and sometimes is over 1M. Tested K1, and it's good. Coil resistance for both is about 250R. 
  
  Relay is a 3500-0056, which are obsolete, but also don't overly match up with Coto's 3500 datasheet here: https://media.cotorelay.com/wp-content/uploads/2014/09/15101227/3500_series_reed_relay_datasheet.pdf
  
  I've contacted COTO asking for datasheet for this old guy, but, it LOOKS like it's just a 3500, 5v, something. Judging by existing datasheet, it's somewhere between 0.5uV and 1uV EMF rating. And unknown if it has EMF shield.
  
  So, am I looking for like a 3501-05-510? Not seeing those available anywhere. Looks like Mouser has the 3501-05-511, in stock for $30 *cough*, but I don't have a spot for a shielding pin, so would have to just cut it off (or drill through the board). Maybe actually the 3540-05-910? Has higher EMF rating, but not sure that matters. Until I can figure out what the 3500-0056 is, hard to do apples to apples.
  
  Thanks!
• #7 Reply
  Posted by coppercone2 on 20 Mar, 2021 23:58
• Yea, try to take it apart now and see what the contacts look like. I have taken apart relays that look great inside but read 10meg, while the most dirty bastard switches have read sub 0.1 ohm, where I thought the contact was made out of a motor brush (turned out to be brass).
  
  For some reason I think that having a shield if there was none originally won't hurt the guy much, just see if its connected directly through ground or through a resistance if the shield is tied to one of the legs. You could just bend the pin under neath and solder a wire to it and solder the wire to some ground, it will still do something. It's not exactly a VHF RF circuit.
  
  its ok, its just an old keithley
  
  and give radwell a call, with the virus they might be hurting for business and honor the 5 euro price on the old relay thats taking up precious shelving space, and having a few of these would not be bad parts to have in a electronics lab in the parts pin for making more advanced circuits.
  
  https://www.radwell.co.uk/en-GB/Buy/COTO/COTO/3500-0056/?redirect=true
  
• #8 Reply
  Posted by staze on 21 Mar, 2021 00:05
• I'm in the states, so that's probably gonna be brutal for shipping. The relay is completed potted, so it's not really something to take apart non-destructively. But once I get a replacement, I'll probably do that... might be interesting.
• #9 Reply
  Posted by coppercone2 on 21 Mar, 2021 00:49
• I don't think you will have luck cleaning it because its likely plated with something exotic inside
  
  I have been interested in stuff like that but if you look up plating, they usually have 2 kinds, jewlery and engineering plating. The engineering plating seems to be hard to get information about, has minimum huge order sizes, nasty chemicals ,etc. I think repair would be the most expensive option for something like that, but I can be wrong.
• #10 Reply
  Posted by staze on 21 Mar, 2021 01:04
• Oh no. Definitely not taking apart to try to repair. More just curious it’s failure mode. I am not entirely sure ohms ever really worked on this thing, so it may be there’s a nice glob of partially melted reed on the end.
  
  There’s a NOS seller of coto stuff here in the states that has some similar models. But until I get data sheet for the failed one I have no idea proper replacement to get.
  
  I should be able to confirm that’s the failure by just manually shorting the trace in ohms mode. Make sure I get good values at that point...
• #11 Reply
  Posted by staze on 21 Mar, 2021 16:58
• Did confirm the shorting the contacts manually in ohms mode resolved the issue.
  
  Now just have to get datasheet for bad relay (waiting on COTO) and find a replacement.
• #12 Reply
  Posted by srb1954 on 21 Mar, 2021 23:04
• Quote from: coppercone2 on 20 Mar, 2021 23:58

  and give radwell a call, with the virus they might be hurting for business and honor the 5 euro price on the old relay thats taking up precious shelving space, and having a few of these would not be bad parts to have in a electronics lab in the parts pin for making more advanced circuits.
  
  https://www.radwell.co.uk/en-GB/Buy/COTO/COTO/3500-0056/?redirect=true
  

  I obtained a 3500-0056 relay from Radwell a few years ago to repair my 199. Seems to work OK although there are other problems with my 199 so I don't use it very often.
  
  I never found any information on this relay and Radwell seem to be the only people in the world who hold any stock of this relay.
• #13 Reply
  Posted by staze on 21 Mar, 2021 23:06
• Contacted stateside radwell asking about stock. Their site just says "call for stock". Which I haaaaaate. So, shot them an email via contact form.
  
  kinda hoping COTO gets back to me with old datasheet, but not holding my breath.
• #14 Reply
  Posted by coppercone2 on 22 Mar, 2021 04:32
• no one is responding to my emails, normally the quote people are not as bad as this... I have a few out standing requests. I never really experienced this before, usually they are happy to give me a inflated  number. 2 weeks for some IC's without reply. Maybe I just stumbled upon some shitty operators.
  
  I have a keithley 199 also but last I checked it worked, however its been dormant pending the setup of the RF test bench over 5 years . I do like that meter.
  
• #15 Reply
  Posted by staze on 22 Mar, 2021 04:51
• May be related to the huge dry spell of ICs right now. Also COVID. Guessing some shops don’t have as many people onsite.
  
  But yeah. Makes it difficult.
  
  I have a 199 that I also quite like. These old Keithley’s are relatively nice for how much you can usually get them for.
• #16 Reply
  Posted by coppercone2 on 22 Mar, 2021 05:40
• I think they look better because I am tired of all the white stuff
  
  KEysight went to black so maybe keithley can go back to brown, they made it look good
  
  I keep thinking someone put a pile of paper down when I see a white multimeter, perhaps its a ploy to sell it to buisnesses.
  
  maybe it confuses financial people that they are buying really expensive litigation grade paper?
• #17 Reply
  Posted by staze on 22 Mar, 2021 22:19
• COTO and Radwell got back to me!
  
  COTO said it was a specific Keithley part, and recommended replacing with a 3501-05-511 (cutting the shielding pin). Radwell said they don't have any in stock.
  
  I'm not even going to bother contacting Tek/Keithley for replacement part pricing as I'm sure it's gigantic.
  
  I've purchased a 3501-05-511 ($30!) and once it's here I'll see if I should just cut the pin, or if I can either drill the board or bend the pin over and then tie to signal ground (or should I tie to chassis ground?)
  
  I've also had the bad one attached to my FG cycling the relay at about 10Hz and putting 20v of 1Mhz AC through the contacts. It's helped, but it's still not a dead short (it's gone from inconsistent to seemingly stable 0.5R). Not sure if I'll bother trying to revive anymore.
  
  Just glad I found the issue.
• #18 Reply
  Posted by staze on 16 Apr, 2021 03:51
• Finally got replacement COTO relay. Had to bend over shielding pin since it happened to be right where a trace was (so I couldn't drill the board). Got it installed, and things seem back to normal. =)
  
  Thanks for the help all!
• #19 Reply
  Posted by coppercone2 on 16 Apr, 2021 05:03
• if you want to really connect it you can get a dremel or file and cut a small notch to bend the pin under neath the part even if you want it to sit flush for mechanical reasons, or leave it elevated and put silicone on it.
• #20 Reply
  Posted by staze on 20 Apr, 2021 01:14
• Bending pin over wasn't an issue. I haven't connected it to anything. Not sure what ground I should tie to, or if I should.
  
  Anyway, working now.
  
  Thanks!