Keithley 6485/6487 teardown

[plesa]

I found on my cell phone few internal photos of Keithley 6487.
It is quite similar to the 6485 ( without +/- 500V voltage source - bottom board and second transformer).
The front end contains LMC662, LMC6081. Voltage reference is LM399H.
Sorry for poor quality.

[razberik]

Hello, do you remember please whether those white EE2 relays are for range switching ?
Thank you

[plesa]

Hello, do you remember please whether those white EE2 relays are for range switching ?
Thank you

Yes, based on clicking it is used for range switching. And electrometers 6514 and 6517 are using similar relays to Meder HI series.

[razberik]

These HI Meder looks good. It is comfortable to deal with Meder ... in my native language, they manufacture some series in Prague.

Today I got a good chance to do some brief teardown. I attach pics.
It surprise me that they dont use better relays, but literally jelly-bean relays.

[razberik]

Today I was given a task to test resistivity of some materials. I was climbing with voltage from few volts to few hundreds of volts. Then suddenly relays started to clicking and OV.RFLOW:A showed on display. Totally bricked. Changing ranges does nothing.
Exactly the same behavior like this:
https://forum.tek.com/viewtopic.php?f=41&t=103860

Any tip what to do ?

[plesa]

Replace comparator U49, it should be LM339. If it is not U49, try to find LM339 in this area.
What did you tried to measure?
This meter should survive about 1-2kV on input. So you cannot damage it by measuring insulation resistance with integrated voltage source.

If you are using external power supply or higher voltage than 500V consider some protection circuit made by gas discharge tube, resistor and 2x antiparalel low leakage diodes (1N3595, DPAD5, DPAD1 or JFET).
In attachment you have minimal values of protecting resistor, it also depends on used range (lowest auto range can be selected in menu).
For measurement you described use manual range rather than autorange.

[razberik]

I was measuring insulation resistance with internal power supply.
Yes, I had AUTO setting. I learned it the hard way.

Well it depends on what boss say. Send it to repair or let me try repair it ... We have this machine for years.

[plesa]

I was measuring insulation resistance with internal power supply.
Yes, I had AUTO setting. I learned it the hard way.

Well it depends on what boss say. Send it to repair or let me try repair it ... We have this machine for years.

Keithley does not repair these meters or the repair is corresponding the new meter price. Try to replace comparator metioned above.
What level of current you needs to measure?

[razberik]

Hard to answer, when it was working, I was obtaining results in Teraohms. The guy who ordered this measurement told me these results seem to be correct.

Where and how did you figure out that LM339 use to fail ? Is it some common failure for this pAmeter Keithley topology ?

[plesa]

Hard to answer, when it was working, I was obtaining results in Teraohms. The guy who ordered this measurement told me these results seem to be correct.

Where and how did you figure out that LM339 use to fail ? Is it some common failure for this pAmeter Keithley topology ?

Teraohms it means that you was measuring  current <500pA. If you want to measure it right way, place the resistor between DUT and meter according to table above with GDT and protection diodes.
Also try to find  metal enclosure for your DUT and ground it. Also monitor humidity level, this can affect your test results as well.

I have some experience and it is common failure, same like DGxx switches is Keithley DMM ( 2001/2002). There are bunch of threads related to these repairs
Also If you are interested in this topic read
http://www.tek.com/sites/tek.com/files/media/document/resources/LowLevelHandbook_7Ed.pdf

[TiN]

Thanks for photos, I might have to add some soon  .

[razberik]

My boss agreed with trying the repair. I replaced the LM393, but no result.
It seems to me that instrument is switching ranges correctly. I mean - if I set manually some nano/micro range and then I press to AUTO, it switches to the highest range 20mA.
If I connect DMM to analog output, I receive som random voltage with nothing on input. If I short the input, the analog output changes, but I dont receive correct number either.

Any tip what to do next ?

I attach little bit more detailed photo. There are some DG411 switches. Shall they be wrong ?

TiN - I have some hi-res photos, but not very good angles and ambient lighting.

[TiN]

Feel free to upload all or any pics to my ftp, I'll reshare. There are not many photos of there online. Also if you have more photos of bottom voltage source board, would be cool to see. Access info is in my sub link .

Did you check obvious, input protection parts?
If I be you, I'd check voltage source operation. Set meter to get small output, like 1V or 5V and test if it's present.
Or opposite, source small stable signal to source and check where it disappears on circuit, assuming problem is somewhere between input connector and ADC chip
Check input voltages on front end opamps (DIP8 fella's). Be careful not to contaminate things, to prevent leakage (using unpowdered gloves helps, also protect you a bit from high voltage shock from FE supply).

[razberik]

I was wrong. There is no DC voltage at the Analog Output. There is 2VAC for odd ranges and 6VAC for even ranges. It is 100Hz frequency, looks like two combined 50Hz. WTF ?
Everything is disconnected. When Zerocheck is active, zero at output.
I was wandering about the input and I simply dont understand the purpose of the LMC662. Is this the input ? And what is the LMC6081 for ?

AD) I am getting the exact AC signal from AD829 video amplifier output.

[plesa]

OK, so it seems that other components has been damaged. Another components in the chain is LMC6081. This will also explain the weird analog output readout.
During zerocheck is input shorted, so you can check input impedance. Analog output is direct output from input stage.
Are you sure that on input was only voltage 500V?
I have few 6485/6487 for service and all of them has damaged comparator only. Never input stage blown. On their input were voltage between 1-5kV. This meter do not have any protection on input. And all input protection needs to be external.

For theory check chapter 4 in manual
http://www.tunl.duke.edu/documents/public/electronics/Keithley/keithley_485_picoammeter_manual.pdf
There is also schematic but 6485 is similar, but different.

[razberik]

Hello. My boss gave me more time on it. I made detailed pictures of it and they are being uploaded to xdevs. Check it out TiN ;-)

I was lurking around the preamp section. I was trying to reverse the schematic. There is quite small amount of components around, however I was doing it in work with board installed, so I was unable to track the traces under relays precisely. So I have some garbage schematic, attached. What is interesting for me is using positive inputs of opamps, but hey, they did it the same way in Kei 485. Anyway, no feedback loop is closed now in my schematic.

What I have so far found, that AD829 video amplifier is outputing rail-to-rail (+/-15V) swing in 100Hz frequency. There is no corresponding input, however I dont know the feedback network of it.
If this was some non-precise circuit I would definitely give it a try to desolder it, but I have to think twice at this delicate instrument. Any tip ? I have strong belief that this AD829 is the driver of the feedback loop.

AD) plesa - yes, I am absolutely sure that 500V was not exceeded. It happened while sourcing about 300V from internal supply and suddenly relays were clicking like crazy and then OV.FLOW.

[TiN]

Thanks for photos, nice to know that same voltage source board used in 2501 too.
Did you check inputs on AD829, what voltage is there?

Here are all photos from razberik, moved at permanent location:

[Kleinstein]

Seeing a 100 Hz signal makes me think about the supply: could be a capacitor problem and thus too much ripple. At least check the supply voltages with a scope.

[razberik]

Here you go. The preamp schematic.



It was hard to track some tracks. Some of them I didnt find out.
I wonder how the U25 LMC662 is supposed to work. 

Need some time to rearrange the schematic into more readable form.

[plesa]

Did you check the power supply for AD829 and also check the output of LMC6081 by scope or DMM.
If you needs some measurement of functional meter I can help.

[razberik]

I replaced the AD829 and BUF634, but no change.
When AD829 is removed and I short the GND/Guard and output pin on AD829 footprint, I receive some value on display. When BUF634 is removed, it reads 0.00001nA.
Any idea ?

[razberik]

Some news.

plesa - Yes, I have checked all power supplies, these are all OK. I didnt reply because this was the absolute first thing to check. No ripple, no significant noise, just nominal voltages.

I have tracked the voltages around U25 LMC662 and U27 LMC6081. This differs between mA+uA and nA ranges. On mA+uA ranges, I receive saturated rail-rail swing at the U26 AD829 output, on nA ranges, it saturates to negative supply.

IN+ on U26 is -17mVDC all the time. I receive -11.2mVDC on fully functional Keithley 6485 I lent from my colleague. I believe this is OK.
I believe that saturation is driven by the input at IN-.

On nA ranges, the U27 LMC6081 behaves strange to me. Output is -30mV all the time. IN- is at zero. When I touch the DMM probe IN+, the DMM reads some miliVolts ramping up. Starting value is random, but it ramps towards few miliVolts. No change at the output.

I used HP 3478A for reading voltages. (When I read e.g. 10VAC, it corresponds to +/14V rail-rail swing)
All is referenced to GND/Guard. The instrument has shorted Analog Output GND with chassis.

I attach the input area schematic with voltages and waveform of the rail-rail swing at the U26 output.

[razberik]

As I indirectly promised, here is the rearranged schematic. I also found out the relay switching matrix, noted in schematic for ZCHK OFF and nanoAmps range.
K9 is operated during startup and switch off only.
What is interesting is the LMC662 circuitry. It seems that only the input FETs are used to compensate input current of LMC6081. Thus these two opamps are matched ?

[TiN]

razberik doing great job on schematics reversing, my respect!  Keep it up. 

To contribute at least something, I took few ugly photos of my latest toy, KI 6485. As expected, very very almost same  Obviously no point of having triaxial on 6487, as guard is not driven actively.

[lukier]

To contribute at least something, I took few ugly photos of my latest toy, KI 6485. As expected, very very almost same  Obviously no point of having triaxial on 6487, as guard is not driven actively.

Hmm is that the right front panel? 

I have the old 485 from the ugly-brown era and it also has just BNC, no triax. I wonder how similar is the front end circuitry, maybe I'll take some photos and compare (AFAIR no LM399 in 485).