Keithley 2700 weird jump in measurement values

[petemate]

Hi,

As some of you may know, I have successfully repaired a Keithley 2000 DMM. I also had a broken 2700, which I have now also "repaired"(That was a simple question of replacing some of those JFET switches). I thought I wanted to test the two against each other, so I produced these measurements(please note that the 2700 is plotted on the right Y axis):



As one can see, the 2000 works fine(I expected this drift due to warm-up time), but the 2700 is making some weird jumps here.

Here is a zoomed-in plot of some of the jumps:



The Keithley 2700 jumps about 170uV a single reading, then going back to the "base level" again. These jumps becomes more frequent, until it finally settles on the new level. This looks almost like some ADC LSB-shift behavior. But that can't be, since it has much more resolution than this and is able to distinguish the fine changes in the "base level" just fine. Does anybody have an idea about what could cause this behavior?

To repair the 2000, I temporarily took parts from the 2700(e.g. the TF-245 resistor array). But I made sure to put those back. And even though it would void the calibration, I don't see why it should cause these jumps..

Any ideas?

Thank you for your time.

[Kleinstein]

The jumps are rather large. 170 µV corresponds to 17 ppm of 10 V nominal FS or some 12 ppm of of the 14 V reference. So this would be something like 2 µs of the full reference active at 200 ms integration time.  This a little to much to attribute this to the slow slope part (that was broken on the K2000). So it could be something like 1 bit (not the LSB, but more like somewhere in the middle) of the result not properly transferred (e.g. aged opto-coupler).

I very much doubt it, but it could also be a problem with the reference or an related OP. So does a similar effect also happen in resistance mode ?
Another point to check could be a signal slowly changing over a larger range (e.g. a few mV in 10 V range,  or a few ohms in 10 K range - something like a PTC slowly heating or cooling), to see of the odd jump pattern repeats.

A first point to check would be the supply for the ADC part, especially the 5 V for the CPLD / gate driving logic. Also check the reference voltages, in case there is a bad solder junction / short around the TF245.

[petemate]

Hi Kleinstein,

I just had another go at it. I got to ~650 readings, then the 2700 locked up and showed the exact same symptoms as the 2000 did back before it was repaired.. Initial sampling, but "lock-up" after some time. I am fearing that I have the same problem with the 2700 as with the 2000. But we will see. Anyway, there is no reason to pursue the 2700 offset/jump issue until I get the unit fixed. I'll post again if I need repair assistance Thanks for your time!