Fluke 8860A

[rsd]

If in voltage auto range and with open leads connection, the measured input voltage keeps getting more and more negative. It resets at about -200mV, probably with the first range switch (the relays click). And then begins again.

It also happens in the lowest two fixed ranges (there is no relay clicking between those two). With the higher sampling rate, it is charging faster. On Manuel it charges to over negative 1V. I did not take it apart yet. Wanted to check first if that is expected from a meter that old, or if somethings to be fixed 

Schematics are attached. Any Tips?

[rsd]

My best guess would be Q6 or Q3? Since those would only cause trouble on the lower ranges. See attached picture.

[rsd]

Measured voltage across a 100k Resistor is about 0.003 to 0.000mV, so 0.030nA leakage? That seems to be fairly low. So it is probably pulsed, which makes sense in terms of the different rates for different sampling frequencies. So that does rule out the marked JFETs from my previous answer.

[Kleinstein]

It is normal for a high impedance input to drift from the bias current. 30 pA is usually considered OK, though not great. Specs are often like < 50 pA with a few meters also specifac <20 pA.
The dependence on the sampling rate suggsts that much of it is from the switching part / charge injection at the FETs. I see no extra precharge circuit or accurate compensation of the gate charge. For the given circuit I would consider the behaviour normal to good.

[rsd]

Thank you very much for the clarification.