Keithley 196 Repair

Knowing no end of Multimeter repair, I decided to pick up a Keithley 196 I found via eBay, then managed to purchase it privately through PayPal for a song (~$50).

The unit arrived a bit worse for wear, but with some Superglue and JB-Weld, I was pretty much able to repair all the case damage. The power switch also needed some percussive maintenance to get it to work reliably. Visual inspection showed only that one of the two Polystyrene Capacitors (68pf 500V) had been melted at some point. After poking it with a probe, the cap started making a very bad buzzing noise, so I replaced it temporarily with a Silver Mica cap of the same rating.

The digital board and power rails all checked out with the analog board unplugged. Once the analog board was hooked up, however, the -15V rail started to oscillate at about 120hz (see image). The real bitch was that the schematics included in the online copy of the manual are of terrible quality, and when printed are nearly unreadable. Thankfully, a very helpful individual on the eevblog forums had access to an actual original printed copy of the Keithley 196 manual, and scanned the schematic and layout pages (PDF of that here) ((Special thanks to lowimpedance on EEVblog forums)).

Ironically, before I got those schematics, I was watching some teardown/repair videos on youtube, and saw one on a repair of an Agilent E4421A Signal Generator where the issue ended up being a failed voltage regulator ((well, it was really a darlington transistor, but it was being used as a voltage regulator)) where with no load, the regulator would output 9V, but if you loaded it, it would only output a fraction of that. So on a hunch, I unplugged the analog board, and hooked in a 16ohm power resistor (40W), which means it should draw just about an amp from the regulator that should be able to handle an amp of output. The resulting output was pretty much the same. So, thinking I found it, I went to the local electronics shop, and picked up a new LM7915 (negative 15V regulator), and popped that in… with the same result. Damn. But, with that, there was really only one other option: the primary filter capacitors(s). The analog rail ones all looked to be circa 1986, while the massive 10,000µF digital one seems to have been replaced in the last few years. The ±15V filter caps were both 680µF 35V, while the 5V analog was a 1500µF 25V cap.

Knowing it rarely worked, a got out the Fluke, and tested the capacitance on the negative rail filter cap: 7nF. Hmm. Positive rail cap tested as 600µF (still good). I pulled the negative cap, and tested it again: same result, 7nF (about 1/100,000th what it should be). So, I ordered some new caps off Digi-Key (local shop would A. Want too much, and B. Probably not have that specific cap). I only had 25V caps in that size. Also figured I’d recap both rails, and the 5V rail while I was at it, since I was doing it, and knowing those caps must be getting pretty dry. I also ordered, via Mouser, a Polypropylene 68pF 2kV cap to replace the failed Polystyrene one. Polystyrene’s are nearly impossible to get in small quantities anymore ((I did find a place to buy them, but they had a $20 minimum order for something that costs $1)). Also ordered some spares incase the ones in my Keithley 199 go out, or the other one in this 196 dies (unlikely that PS caps fail without being heat damaged).

The parts arrived, and recapping the power rails was a success. Shockingly, the Polypropylene capacitor was exactly the right size to fit where the Polystyrene cap was (which is lucky since the leads on the cap were only about 1cm long). I also swapped back in the original LM7915 since it was working, and may mean less chance of having to recalibrate the unit. The unit powered up, and after 2 hours warmup, I was shocked to see that the unit was only 200µV off my Keithley 199 that I self calibrated from my bench PSU (which wasn’t calibrated either). I hope to get the unit calibrated at some point, but at this point, the cost is a bit prohibitive.

All told, I spent about $10 to fix the unit, and I now have a 6.5 digit bench meter that seems to work quite well. Highly suggested if you see a cheap one on eBay. They’re all through hole parts, and all seem to be pretty generic parts except a few of the digital chips… but those are unlikely to fail if they’re treated well enough. Good luck, should you pick one up. This one is going to stay on my bench right below it’s younger brother (Keithley 199)