Keithley 191 Repair Help

[John3354]

First post here.  Noob here setting up an electronics lab for tinkering and I have been picking up a fair amount of kit.  The most recent purchase is a Keithley 191 multimeter that I picked up off eBay for $20. 

The seller stated that it only flashed random digits when turned on, but I bought it with the thinking that it could just be noise or at most it could be a switch issue like Dave had in his Keithley 177 repair video.  No such luck so far.  I plugged in a set of leads and shorted them and the digits are still pretty random.  Too much of a swing for just random noise.  So I lubricated the switches hoping that would solve the issue.  No luck there either.  Same issue.

Next I started checking voltages.  I followed the manual here:  https://doc.xdevs.com/doc/Keithley/191/191%20(A%20-%201979)(30183%20Instruction).pdf

+5 and +15 look good, within a tenth or two on voltage.  No issue there.  But -15 fluctuates between -12.75 and -13.  As the manual says it needs to be -15V +-10% that is well out of spec.  What is the next step?  Should I pull the regulator out of the board and test it on the bench?  Or, is there a cap I should be looking at?

Sorry for the basic questions.  I am a bit of a noob, but I thought this would be a good way to learn.  I hope I have not gotten in over my head.

[jkf1000]

That voltage definately sounds low, could be the regulators or more likely something pulling the voltage due to loading. Dont forget to check also the voltages before the regulators as you will need a decent overhead as the old regulaters required a good 2 to 3 volts higher to operate reliably. Try isolating circuits one at a time and then recheck your voltages, also check for hotspots.. Karl

[lowimpedance]

Also while in the same area check the PS filter caps which may have dried out enough to allow excessive ripple which would lead to regulator dropout and thus fluctuations of the rail.

[John3354]

Dont forget to check also the voltages before the regulators as you will need a decent overhead as the old regulaters required a good 2 to 3 volts higher to operate reliably.

Getting roughly -17V at the input, but with some fluctuation, which means that:

Also while in the same area check the PS filter caps which may have dried out enough to allow excessive ripple which would lead to regulator dropout and thus fluctuations of the rail.

...is probably a winner.  Also, it seems like someone may have already had a go at it because as I look at it two of the caps are newer models than I would have expected.  Only two of the original caps appear to still be there.  What is left are 10,000uf and a 470uf caps.  I have 470uf caps in stock, but will have to order the 10,000uf ones.  At 37 years old they are probably both past their "Best-by" dates.  I am seeing a fair amount of fluctuation on the multimeter on the supply to the -15V voltage regulator.

I pulled the 10,000uf cap before and tested it with my cheapy component tester and it seemed to be OK, but I do not know how much I should trust that.  Any recommendations for a good LCR meter that will not break the bank?

I am also going to go back over the schematics and verify that the proper spec caps are in place.  The person who came before me may have been a bit  .  The #1 rule of troubleshooting is to always assume that the person who worked on it last is an idiot, even if that person was you.

[jkf1000]

The filter caps are always going to be a prime candidate for failures. Also keep your eyes open for any old tantalum capacitors. They can fail short circuit without warning, and have caused further damage to neighbouring components on some of the equipment I have repaired. There will probably be a plethora of suggestions as far as what LCR to choose, I personally use the DER EE DE5000, ridiculous value for money thats gives a lot of bang for the buck. It is the ESR that you need to be confident in when checking electroltics. A lot of people suggest a shotgun approach with old electrolytics. Basically just changing everyone they see. Personally I prefer to find the faulty ones first to get the equipment up and running and then determine whether to continue changing the others as preventative maintenance. Karl

[cncjerry]

Replace the caps and regulator, don't bother testing it either unless you have the equipment.  You can buy a lot of caps and regulators for the cost of a decent LCR meter.  There are cheapos on eBay that work well, little PIC and a display that will test everything.

I have a bunch of Keithley meters. A couple have a relay right after the input connectors and they are known to fail.  Actually think I have two 191s, is that the meter with the optional A/C voltage and it has 4w ohms?  Can't remember, but look for a relay. If so, pop the cap off carefully and run some paper through the contacts or at most, some 2k or higher sandpaper.  Random numbers could be a floating input.  Does it show OL in ohms mode?  That will tell a lot, otherwise you probably have a filter cap problem.  Check the rails on AC mode.

[cncjerry]

I haven't looked in the cabinet but looked at a picture on the web, and I am pretty sure I have two 191s, one with and the other without the A/C feature.  I don't understand why they would have had A/C as an option, but if you switch it to A/C and get a display of 1EEEE then it is missing that feature.  Let me know if you need me to check anything for you.  Send me a PM as I'll be traveling the next two weeks and might not logon to EEVBLOG.

Those meters are a lot of fun.  I pull one out now and then for a little nostalgia. Either that meter or the 178a had the relay problem keeping the voltage floating and the numbers went everywhere because of the high impedance input.  There was some reason that it also did it on the ohms setting too, something to do with switching the current source in and then using it as a volt meter to read the resulting voltage from the fixed current source.  It's been a few years...

[John3354]

Is there anyway to test caps without an LCR meter?  I have a few multimeters including a UT-61E.

When I test the tantalums with continuity check, one of the tantalums shows ~50ohms resistance while in circuit.  Does that tell me anything, or would I have to pull it from the circuit to be able to tell?

Unfortunately, I do not have any tantalums but I will get an assortment on order.  It will take a while to make it to the US on the slow boat from China though. 

[RayRay]

The UT-61E can also test capacitance FYI, so you could definitely use it to measure the capacitance.
For best results, you'd need to test the capacitors out of the circuit (if you measure capacitors that are connected in parallel, you wont get an accurate reading). Set that aside though, in order to fully test a capacitor, you need to measure both the capacitance and ESR.
Your UNI-T can only measure the capacitance, which is better than nothing of course, but it wont tell you for 100% whether the capacitors are good or not (as in some cases, a capacitor can have good capacitance, but bad ESR levels, which will make it function poorly!)
You can get the MESR-100 (ESR meter) from a variety of places (such as fasttech/dealextreme/banggood, ebay or aliexpress sellers), it's reasonably affordable ($55) and provides a great value for money.

[John3354]

I finally sat down and traced the circuit path back to a 470uf 50V cap that was on that rail, pulled it and my cheapy component tester did not even recognize it as a cap, my multimeter was also confused.  So I went scrounging and found an old UPS with an equivalent spec cap in it, desoldered it and put it in the 191 and now all is right with the world.  Voltage readings are stable and everything seems to be working properly.

Thanks fellas.  Much appreciated.  Now, to go bid on some more old Keithley stuff.  Ever since Dave's videos I have thought this vintage stuff was pretty cool, plus it is more than sufficient for my needs.

Needless to say, I will go ahead and order a full replacement of caps just to make sure.  There is no sense in not swapping them all out while I am at it.

[jkf1000]

Congratulations on a succsessfull repair. As with most repairs, after the initial visual check over, testing voltages will usually point you in the right direction. Always a nice feeling to use equipment you have brought back from the brink and I am all for keeping viable equipment out of the landfill. Time now to look for your next project.. Karl

[John3354]

Congratulations on a succsessfull repair.

Thanks.  Much appreciated.

One thing though, when I used the electrical contact cleaner I think I got some on the plastic front bezel.  It is not as clear as I remember.  Or...maybe I did not pay much attention as I was concentrating on getting the thing working?  I tend to get tunnel vision and only concentrate on the most pressing issues first.  Are there any suggestions for cleaning up vintage plastics?

[jkf1000]

The first rule with plastics is caution. Whatever treatment product you choose, always test on an area you wont normally see to check if there is a reaction. There are polishing products on the market but I have never used any of them.
 You can restore the finish on clear plastics such as display units using very common chemicals. What I use is a progressive step depending on how bad the finish is. You can start with something as rough as 2000 grit wet and dry paper if the finish is really bad and work to finer from there. I use various products. For medium scratches and discolouration a kitchen cream cleaner works well. Here in the UK we have a product called Cif which has a slightly gritty consistancy which works really well for me. I then clean the surface and polish again with common toothpaste, then after another clean the final polish is with normal automotive polish. Just do it in steps. I have had fantastic results on VU meters which looked like new when finished.. Karl

[cncjerry]

I suggest care with the abrasive cleaners or they will take the lettering off the front.

The 191 is a decent meter for the $20 you paid.  4W ohms, etc.

Keep an eye out for a Keithley 195 or 196 as they are a notch up.  The 196 especially.

[John3354]

Update:  The meter is still running.  Checking it against my Fluke 45 and UT-61E and other meters here show it to be within the least significant digit on all ranges that I am able to test.  It seems to be in calibration, or at least good enough for my needs.

I still have not had time to clean up the front panel.  To do so, I will have to desolder the input jacks.  So, I am wondering if it would be worthwhile to replace the input jacks with some that would accept shrouded plugs so that I can use conventional multimeter leads?  Any thoughts?

Also, today I just missed a 197A for $34 that likely only needed a fuse.  I got busy with work and let the auction slip by.  Oh well, on to the next one.  I think I like test equipment more than projects.