Fluke 8520a - Multimeter 10VDC input reading goes up by itself

[The_Parp]

Good day !

I bought a Fluke 8520a from EBay.

When I turn on the device, all is fine.
All the power rails are within the specs
When running the diagnostic program, on test sometimes fail. It's a VAC voltage gap but I don't think is related to my other main problem.

The unit seems to have two different routes for VDC reading (low and high voltage, controled by a relay). When the unit is in 100VDC resolution, the voltage displayed stays at 0 (or near). When it gets to lower resolution (10VDC or lower), the displayed voltage goes up (or down) by itself. Well pass beyond the reading limit of this resolution.

If I try to read the voltage at the VDC input posts with another multimeter, the input seems to de-energize and goes back to 0, then moves again.

When a VDC source is applied (10VDC) to the input, the reading is steady and accurate at every resolution. In that case, if I lower the input by the source, the voltage reading follows even at lower resolution, but if I unplug the source, the input stays energized and voltage reading goes up or down (still only at lower resolution, 100VDC res reading is fine). When I swap between the resolutions without a source, the reading resets.

I checked the Zeroing circuit TP that should pulse at InputVDC+2V and when the input energize itself slowly, the limit is 18,9VDC or so (pulse is there, but voltage too).

Since shorting the input de-energize it, I went with a multimeter and tested many TP, some were shorting the input too. Pass the DC buffer, I wasn't able to short the input.

Does anyone have experience with this instrument or have encountered the same problem?

Here is a vid (turn of the sound, I'm on an old chair)
https://1drv.ms/v/s!Aj0t8sj3fkXvja4NTQwfrTflZvu2tQ?e=hJdi0V

I know that the manual warn for a ramp up but 18VDC? Is it normal?

My plan for now is to recap everything from the input to the end of the DC buffer. I'll keep you updated.

Thanks for your time!

[Kleinstein]

For a DMM with a high Z input it is normal that an open input is drifting quite a lot and often drifting all the way to saturation in one direction.  So the observed dirft could be normal.
The inputs are very high resistance (specs are something like > 1 G or > 10 G), but the actual resistance could well be signifcant high, like in the 100 G range or more.
On the other side there is some bias current of a few 10 pA typical (specs are often < 50 pA). With 10 pA and 10 Gohm one gets 10 V as an expected open circuit voltage and chances are the bias and / or resistance are higher and also not very stable.
The input bias slowly charge is capacitance (e.g. some 100 pF) at the input. With some Fluke meters there could be additional capacitance from an input low pass filter that can be enabled.

One can use the speed of the voltage drifting to estimate the input bias:  ad a low leakage capacitor (e.g. PP or PS type film) of some 1-10 nF arcross the input and than measure the rate the voltage changes.  Checking the input bias is a good idea as dirt and possibly ESD damage (gradual) could lead to higher than normal current.

Changing the electrolytic capacitors may be OK, but not always needed. Some of the older ones are still OK and one could check a few and only replace them all if some test bad.
The film an ceramic capacitors usually don't need to be replaces, with the exception of maybe RF suppression caps at the mains side that show cracks (e.g. the "Rifa" ones in a clear epoxy case).

[The_Parp]

Thank you very much for the answer.

I will still change tant cap since 2 died on me while performing tests. But As I understand it, there is no real problem. I will try to measure the rate  and see if there is a filter I can disable.