Joe,
It's а very interesting result! I think the only internal reason of such slow
sine-wave fluctuations is a interference of the mains frequency. During the
power-on Floating Logic CPU IC801 is autodetect the mains frequency (50, 60, 400
Hz) and setup appropriate output lines (400Hz, /60Hz) and ADC clock divider for
proper PLL phase synchronization. It may take more than 5-10 minutes, during
which the filter's capacitor C806 voltage and V-T (ADC) clock changes slowly. After all,
clock oscillator outputs 5.24 MHz which must varies with the mains frequency.
If phase synchronization is failed due to some reason, then effective Common
Mode Rejection is largely reduced.
First of all you can try to adjust initial PLL frequency in accordance with the
Solartron 7081 Maintenance Manual, 9.8 (page 9.5). It's very simple.
The next steps, I think, are isolation of the input leads and Input Amplifier
infuence. To do this, I recommend: 1) disabling probes and short the input with a
copper wire (photo). 2) repeat testing after the execution of a command "MODE,
TEST 0". This command disables the input amplifier and ADC shorts.

With best regards,
Mickle T.

MK, Andreas, Mickle:

Thanks for the replies.  Now, additional details and more questions.

Yes, the air conditioner was operating during the collection period with the
times of collection shown on the graph reflecting Central Daylight Savings
Time in NW FL.  It was a slightly cooler day than usual but not cool enough
to allow the A/C to remain off for roughly an hour and a half which would
have started about 10:30 AM local time.  Therefore the question arises of
how to 'shield' a sensitive DMM from transient air currents in the workshop.
I have thought about that but I am not sure how to go about it without
having a box that would progressively heat up as a result of the meter being
turned on and in the box.

There were 13,057 total data points collected over 41,369 seconds with the
front panel set to '7 x 9' or roughly a 3.2 sec collection interval, per the
front panel labeling.  Using the data above, the actual collection interval
was about 3.17 sec.  Therefore, each 2000 data points (vertical line on the
graph) represents about 1 hour and 45 minutes.

What I do have that would, perhaps, interfere with the collection is a whole
house stand-by generator system.  It runs automatically about the time the
data 'smoothed' out but only for about 20 minutes.  It comes up to speed,
automatically switches the house to the generator then, at the end of the 20
minutes, automatically switches back to the mains.  The 'switch' is never
completely 'in-phase' but the controller does try to 'sync' the switch.  The
computer I used to collect the data is powered by a UPS but the meter was
not.  If there was a power 'glitch', the meter would have turned off then
back on as the power was restored.  However, it would not have gone back
into the collection mode selected.  Therefore, there was no power
interruption.  However, clearly, there was a 20 minute period of time when
the power frequency was not 60.0 Hz.  It was very likely 60.1 Hz which is
what the generator is set to in order to facilitate synchronization with the
mains when switching the load.

Whether this is sufficient to 'disrupt' an 'oscillation' of the 7081 is
unknown but an interesting question.

I clearly need to read the manual about the issue that Andreas and Mickle
raise and probably repeat the experiment.

Mickle's picture of the 'short' is another possibility but there is an issue
with the front panel connector on my meter that I think has to do with using
wires in the past.  It is a bit 'flaky' and I have tried to 'adjust' it
using a small 'pin' to lift the contacts and bend them slightly to make them
more secure.  I will probably use a new connector insert, and solder the
pins together and use that as the 'short', thus saving wear and tear on the
connector.

One other question is raised by the picture.  Per the Operating Manual Part
1, paragraph 10.1, the two terminal input lead shows the red wire connected
to pins 4 and 5 (see paragraph 2.6) and the black wire connected to pins 1,
2, and 3.  This is how my 'home made' test leads are constructed.  However,
Mickle's picture shows the 'short' connecting only pins 2 and 4.  My test
leads, shorted, connect all 5 pins together.  Is that an issue?  Would this
explain any of the data collected?  Did I construct my test leads correctly?

Thanks again to all who are contributing to my education.

Joe