Fluke 8845A Not holding calibration

[EddyInChicago]

Hi All,

I have a Fluke 8845A that is going out calibration. The following is what I was told by the person that calibrated it and it's history.

"Hey there Ed.
The Fluke 8845A sn: 9407014 has had a history of losing the calibration corrections made on the DC Volts function.
2008 our first year to calibrate the item, we received the unit being out (high) by 0.03% on all DC ranges.
We would go through the soft calibration and take our outgoing data and we can see the unit has come back into tolerance.
The next year we would receive it again and it would now be reading 0.04% high on DV functions only. Again we would go through the soft cal and get great results on our outgoing data.
Third cal cycle the DMM showed errors not only in DCV, but all other functions being 0.2 % high.
Then it was sent off to Fluke for Repair/Calibration, (Certificate # 2441853-9407014:1309112306)."

Has anyone had this problem before (or something similar)?
What components can be causing this? (Where should I look)?

Any help is appreciated. I also have the calibration data for this DMM. If that helps, I can attach them.

Thank you,

Ed
Seems the Firmware did not cure it long term.
When I took the data this morning, it nearly mirrored the 2008 data sheet for as far as I got.

[Kleinstein]

I don't know the details of the 8845. A more or less continuous drift of the scale factors would point to an analog problem, not a problem with the digital / software side. In theory a failure of the calibration memory can cause trouble but this would be more like a sudden total failure and is usually caught by some check sums to give an error message.

The analog parts they may show drift would be the voltage reference and depending on the details of the DMM also some resistors in the ADC could effect the calibration. Some DMMs do a self check or automatic calibration on the ADC gain factor and other don't. Not sure how the 8845 handles this. For the prime range (e.g. 10 V) there should be no other stage with gain.
Other ranges have the input divider or additional resistors for the gain.
Besides drift in the resistors it could also be board leakage due to contamination. For the reference there may be an indirect effect from the supplies, though not very likely.
One could check the input current / bias current. This is relatively low effort via a high resistor or drift rate with a capacitor at the input. The input bias is not directly related to a gain error, but contamination would likely also effect the bias.
Another point may be a visual inspection, e.g. looking for leaking capacitors (this is a known problem with some Keithley meters).
Checking for reference drift takes time.