Keithley 2001M Resistor low range out of specs

[Hawaka]

Hello,

I recently purchased a Keithley 2001M. I did a full recap (except for C101 on the daughter board on the digital PCB) and change the inlet filter. The unit was turning on before, so no surprise that it still worked after. I believe that this unit has very low hours as it is in pristine condition and the PCB's looked brand new. I took it to work to do a quick check on every measurement and every range. All seemed to work. Last calibration date was in 2013, so I decided to send it to a calibration lab.

Got a call today, they were unable to fully calibrate it. The problem is by the three low resistor range ( 20, 200, 2k), the measure is off by -1.5%, -0.14%, -0.14%. Does anybody has an idea what it could be?

What is really strange is that by every range on DCV, ACV, DCI, ACI, frequency and the 6 high resistor range the measurement error is less than 55% of the allowable deviation. They also tried to downgrade the firmware but it didn't change anything.

[Kleinstein]

The resistance ranges use the current sources and chances are the defect is somewhere there. It looks like the self test does not include a check of the test current, like in an ACAL function. It looks like the HW would even allow this. 

The deviation is quite large, so more like a defect and not so much a "normal" drift in the resistors or reference. Getting a problem with 3 ranges is a bit odd.
The 200 ohm am 2 K range are supposed to use the same current and thus the same error.  The highest test current could be a bit problematic, as the MC14052 switches are rather high resistance to be used with some 9 mA. Getting nearly the wanted 9 mA makes it odd that the 0.9 mA also fails.

Besides a hardware problem, there is also a slight chance that this could be from a faulty adjustment with slightly too high reference resistors used. It could  make sense to check the 20 ohm range test current and also check if the current is effected from a DUT resistor from 0 - 20 ohms. With a HW problem there is a chance the current may not be stable anymore. Looking at the dirft of the resistor reading could also make sense.

[Hawaka]

Got some news from the Cal Lab. The current source is in spec, R355/356/357/358 and R365 are good. They also replaced U325 and U332, but the same errors are still observed...

[Wallace Gasiewicz]

This may be off track.  Are there any "Clamping Diodes" on the inputs of your unit.  I do not have the schematic.   
I am repairing an HP3456 and found some leaking. If the "diodes" are removed the unit works fine. but protection is lost.  They are used in at least four places.  Maybe Keithley also uses these. Some "clamp" between Pos 15 V and Neg 15 Volt rails.
This can affect some ranges more than others. It is very hard to find low leakage diodes and some in my 3456 were actually just FETs that had Source and Drain tied together. 
I am evaluating suitable replacement FETs at present. 
Let me know if I can help. 

Wally

[Kleinstein]

The protection is quite different from the HP meters. The HP meters like 34401, 3456,3458 higher voltage PNP transistors (the newer meter with several in series) for protection. The K2001 uses back to back  higher voltage MOSFETs with photovoltaic gate drive as an extra current limit.
The ohms current source in the K2001 is a bit strange with quite a lot of parts involved.

Diode leakage would be mainly an issue with the higher resistance ranges. The ones that fail here are the lower ranges.
It is still odd that 2 ranges fail just a little.

[Wallace Gasiewicz]

Yes, Kleinstein, you are correct. 
One diode failed on the protection resistance input and that caused the high resistance ranges to be low. CR 201. This is fixed..   
The Fets used as diodes that failed are on the Voltage input and affect the low voltage readings. The ones suspect are Q106 and 107.  Probably Q106.
I believe this also affects the low resistance though.