[volt-nuts] Fluke 845A mystery

[Will]

My friend asked help with his Fluke 845A. It is late colour, very
clean inside and outside. The higher ranges work fine and the readings
are accurate (for example when the needle is on the 1V scale mark
multimeter reads 1.002V). So the chopper seems to be working.

The problem is instability on the lowest ranges. Noise level is ok,
around 200nVpp on the 1uV range, but zeroing is extremely difficult.
Even a tiny adjustment results a huge change with delay.

Amplifier DC voltages were as expected, but we found out that
adjusting the ZERO potentiometer affects the +/-15V power supply
rails. The overall supply voltage changes or the ground shifts or
both. For example the negative supply can vary from -11 to -18V. Any
change in supply/ground changes the zero setting and we have a chain
reaction.

All the electrolytics measured like new. The +/-15V switching power
supply works fine when we replaced the rest of the circuit with load
resistors. Because of the global feedback it is very difficult to
divide the circuit into parts and check each of them individually.

Any help much appreciated.