Keithley 617 Electrometer

[Laval]

Hello everyone, I just acquired a Keithley 617 Electrometer. I got it very cheap from the Physics lab at University, they were going to throw it to the garbage. The unit seems in fairly good condition besides being a bit off and requiring some calibration. There is a couple of things I'm concerned about though.

1. Dave made a teardown video of this device and there are some differences in the Analog front-end as can be seen on the picture. The first are the relays. Mine is equipped with COTO read relays instead of the rectangular relays which just means that this is probably either an older or more recent revision. My concern is with the input dual JFET which doesn't have the clip with the guard wire. On the component it is written 9446-TM LS. I wasn't able to find anything about it and the manual, at least the one I have, only refers to this component as Q308. I'm actually concerned the unit may have been tempered with.

2. The device is slightly out of spec on most ranges and needs calibration. It's not a big deal but something funny happens when measuring resistors between 20k and 30k. I connected the input on a GenRad 0.01% decade resistor box and when I go from 10k to 20k, the display starts flashing displaying 8.52K. I can hear a relay clicking continuously. This continues when I go to 30K the only difference being it then displays 16.16K. Once I get to 40k, everything goes back to normal. From there, if I go back to 30K, everything is just fine and if I continue to 20K, everything is still fine. Non of this happens if I disable auto ranging. It's like the auto ranging algorithm cannot decide which range to use and enters a loop. Has anyone seen that kind of behaviour before ? Maybe it will be fine when calibrated ...

For more reference, the Analog board has the following number: 617-162-02L and the logic board has the following number: 617-102-026. Also, the date code on the Nichicon electrolytic capacitors are all H9446 which means they have been installed in 1994.

[Smith]

I've worked on multiple 617's. None of them ever needed adjustment. One of them needed to be cleaned though. It was off by a bit, and after a IPA cleaning session everything was in spec again.  Never had to replace a cap either, so they should be good.

The single range acting up can be a EEPROM thing. My 199 from the same age had multiple issues on multiple ranges. The EEPROm had a data retention of 10 years, and over 30 years later some bits where lost.  After a quick calibration everything was working fine.

If you want to calibrate, make sure it is clean first (and well dried!) and warmed up properly.

I cant help you with the clip on the guard wire. I don't own a 617 myself.

[Kosmic]

You can find a lot of info about the 617 in this thread: https://www.eevblog.com/forum/testgear/fun-with-low-leakagebias-current-femtompere-electrometer-keithley-617/

about #1: I have one of the later revision built in 1993 (analog board revision L) and it didn't had the guard on the can of the JFET. So doesn't necessarily mean it's a modified unit

about #2: I don't remember anyone describing the symptoms you mentioned.

[Laval]

I've worked on multiple 617's. None of them ever needed adjustment. One of them needed to be cleaned though. It was off by a bit, and after a IPA cleaning session everything was in spec again.  Never had to replace a cap either, so they should be good.

The single range acting up can be a EEPROM thing. My 199 from the same age had multiple issues on multiple ranges. The EEPROm had a data retention of 10 years, and over 30 years later some bits where lost.  After a quick calibration everything was working fine.

If you want to calibrate, make sure it is clean first (and well dried!) and warmed up properly.

I cant help you with the clip on the guard wire. I don't own a 617 myself.

A software bug, due to program corruption, would indeed make sense since the problematic behaviour doesn't manifest itself with auto range off (or when ranging down) so it's unlikely a problem with the circuit itself. Unfortunately, my programmer cannot read the EPROM, I'm due for new one anyway 

About the cleaning, do I just clean the PCB (on both sides) with IPA and lint free swabs ? I just don't want to mess the sensitive board.

You can find a lot of info about the 617 in this thread: https://www.eevblog.com/forum/testgear/fun-with-low-leakagebias-current-femtompere-electrometer-keithley-617/

Yeah, I found that thread yesterday shortly after posting. I should have looked more thoroughly. It contains a wealth of information on that device. Thank you for confirming that the revision L board doesn't need the guard clip and is using a different JFET.

[Smith]

A software bug, due to program corruption, would indeed make sense since the problematic behaviour doesn't manifest itself with auto range off (or when ranging down) so it's unlikely a problem with the circuit itself. Unfortunately, my programmer cannot read the EEPROM, I'm due for new one anyway 

About the cleaning, do I just clean the PCB (on both sides) with IPA and lint free swabs ? I just don't want to mess the sensitive board.

You could try calibrating one channel and see if that fixes the issue on the single channel. It should overwrite the existing EEPROM calibration data, that could do the trick. If not, I recommend some XGecu programmer, I still use the TL866-Plus, and never found anything it couldn't read or write, and I never had a single issue with the software. I use one at home, and at my job. There are a few updates versions though. Should do even better, specially for the price. Pro tip: get one with the included adapters, specially the ZIF socket versions for SOIC and wide SOIC.

The sensitive area around the input and the triaxial connector are the biggest concerns for contamination. They should be cleaned, normally a good wipe with some 99% IPA on some lint free wipes or cotton swabs should do the trick. This reduced the input noise on my 6517 at about 100 fold. Don't forget your cables too, but as these units use the nowadays quite rare twin slot triaxial connector I guess you don't have that much cables to clean 

[Laval]

You could try calibrating one channel and see if that fixes the issue on the single channel. It should overwrite the existing EEPROM calibration data, that could do the trick. If not, I recommend some XGecu programmer, I still use the TL866-Plus, and never found anything it couldn't read or write, and I never had a single issue with the software. I use one at home, and at my job. There are a few updates versions though. Should do even better, specially for the price. Pro tip: get one with the included adapters, specially the ZIF socket versions for SOIC and wide SOIC.

The sensitive area around the input and the triaxial connector are the biggest concerns for contamination. They should be cleaned, normally a good wipe with some 99% IPA on some lint free wipes or cotton swabs should do the trick. This reduced the input noise on my 6517 at about 100 fold. Don't forget your cables too, but as these units use the nowadays quite rare twin slot triaxial connector I guess you don't have that much cables to clean 

I will order the new version that does T48. It comes with a bunch of adapters. I haven't received my Triax connector and cable yet. These two notches connectors are ridiculously overpriced  I also have to work on the enclosure.

Thank you for the knowledge, this is an interesting device.

[Smith]

You could convert it to 3-lug triaxial, these are quite "common", yet still expensive. There are 2 and 3 lug to BNC converters available, these can be quite handy in the not very low current measurements. This way you can use standard coax (will do for a lot of measurements), but beware as there are 2 types, grounded to the external ground, and to the guard ring. But these are also quite expensive on their own. Maybe you can find one cheap somewhere on the interweb.

[Laval]

You could convert it to 3-lug triaxial, these are quite "common", yet still expensive. There are 2 and 3 lug to BNC converters available, these can be quite handy in the not very low current measurements. This way you can use standard coax (will do for a lot of measurements), but beware as there are 2 types, grounded to the external ground, and to the guard ring. But these are also quite expensive on their own. Maybe you can find one cheap somewhere on the interweb.

Using a more standard 3-lug Trixial is a good idea since they are used by other equipment like old Keithley 236/237/238. I just need to get a good grasp of the different coax types. The connectors are still expensive though. For now I have ordered a cheap one (the blue coax). Hopefully, they will at least be real Triax coax. If not, it didn't cost much.

[Smith]

Just be sure not to get the yellow Belden 9222 (I believe) triax cable, this stuff is very thick, and performs terrible. The Keithley cable is so much better, but it is sold per inch for a reason...

[coromonadalix]

fw from xdev's

[Laval]

I couldn't fine a 1000 pf 0.1% cap for calibration of the Coulomb range so i got a variable cap (25 pf - 465 pf) that I will put in parallel with a 620 pf cap to trim to 1000 pf. Since my LCR is a 0.1% device, I should be abl;e to get to 1000 pf at 0.1% precision (should I need to calibrate the unit). The only thing I can't get atm is the 100 G resistor.

I should receive my cable and connector tomorrow, I'm not too optimistic about it since it is ridiculously cheap.

fw from xdev's

Thanks