Have you sat there watching the meter to see if the offset jumps up when
the
relay opens? (I'm sure you have better things to do with your time!) Some
DMMs have a trending feature that keeps track of the highest voltage.
Arfa's suggestions are a good place to start.
Right now, I want to figure out what is going on and I have the time
at this moment to do that.
I measured the power and, on the left channel, the offset fluctuates
between a peak of 11.3 mV and 8.3 mV. Is this significant?
Unfortunately, I missed the meter reading right at the moment the
protection relay tripped.
It's hard to say whether or not that is significant. In itself, in terms of
absolute voltage, it is not significant. Imbalances up to about 20mV are not
uncommon, and are nothing for concern. However, the fact that it is
fluctuating may indicate that some component is 'hovering' around its
breaking down point, or that one of the rails is not stable. It is not
really usual for the voltage, whatever its actual value, to fluctuate. Does
your DMM have a peak hold facility as suggested elsewhere? This will tell
you if the midpoint shifts substantially at the time of failure. However,
note that if the relay circuit is 'quick', it might be *too* quick for the
DMM to see the voltage increase, before the relay disconnects it, so really,
you need to be taking any measurements actually at the amplifier output,
prior to the relay.
Does this amp have a set of discrete component output stages, or are they
STK hybrids ?
Monitor the supply rails as well - particularly those to the output stages -
and see if they are moving by a similar amount as the output offset. In DC
coupled amps, a very tiny shift in conditions back in a preamp stage, can
upset the DC conditions so much in the output stages, as to cause a relay
trip-out, which can make them very frustrating to troubleshoot. Use the butt
end of a reasonable sized screwdriver to have a bit of a bash around, to see
if you can provoke the problem at will. If you can, then swap to a biro pen
to pin down the actual area, and possibly, component.
You can also try a hairdryer, and a can of freezer, both sides of the board.
Bad joints which don't readily show up with disturbance testing, are often
surprisingly sensitive to heat and cold. Does the problem seem to get worse
or better the longer the amp is on ? Is it worse when the amp is 'doing some
work' ? All possibly significant factors in arriving at a diagnosis.
Arfa