The 7075 lives! It ain't perfect, but it's not far off either.
The root cause of the problem was as expected:
It really, really helps if the digital logic actually gets data and can pass it on to the memory and counters!
It took absolutely ages for the meter to warm up to a stable internal reference voltage. I had the fire extinguisher on the bench at the ready as both the 7075 and the Thurlby supply are unknown quantities! After close on two hours, the internal reference stopped increasing in voltage and settled down to 6.999413V (varying +/- 2uV from 411 to 415).
Once I'd got a stable reference, I ran the self tests:
Test 1 (Internal Reference) Should be: 7V +/- 0.005V is: 6.999413 PASS!!
Test 2 (A-D Internal Zero) Should be: 0V +/- 0.0002V is: -0.000280 FAIL (just)
Test 3 (AC Converter) Should be: 10V AC +/- 0.01V is: 10.0002 SUPER PASS!!
Test 4 (Ohms Converter) Should be: 10.000R +/- 0.02K is: 10.003780K FAIL
So, voltage looks reasonable, but ohms is a little high. I turned to my simple references.
First I checked shorted probes on the voltage range:
Looks OK!
Then, my Fisher Electronics LT1021 reference, which was calibrated at 10.00002V and had been powered on since I powered on the meter:
That's a FAIL.
Then I plugged in my (hardly) Kelvin clips and shorted them:
Promising...
Some Vishay precision metal foil resistors - 5K00000 and 10K0000 at 0.005% tolerance:
FAIL - too high
Also FAIL - too high
Then a Vishay 23K524 at 0.1% resistor:
Pass!
And finally, a Bourns 1R0 at 0.1%
Pass!
All, in all; a bit of a mixed bag.
There's lots of cleaning to be done. Particularly all the copper springs which should connect the case panels to the meter's ground. I doubt that they make anything more than a spectacularly bad connection at the moment considering how tarnished they are.
My test leads are far from optimal - the connections into the Fischer plug are white bootlace ferrules which have a reasonably good fit once they are crimped, but they're nickel plated copper. My Kelvin clips are just simple crocodile clips with both wires connected at the screw on the longest arm.
Even so, the voltage and resistance differences from what is expected are too large to be caused by my crappy test leads. So, it's not quite a "winner, winner, chicken dinner", but I think it'll come good with some more fettling.
It is worth getting hold of the 7081 operating manual and having a look at section 3, "warm-up procedure after prolonged low-temperature storage". My 7081's stability certainly improved after self-cooking for several days.
Thanks for that tip, I'll go and study that. At least I have some time over the weekend to be with the meter while it's on. I've no idea of the condition of the mains filter and I won't be leaving it turned on without oversight for some time.
And "get a reference" is a vast topic in its own right T'aint as easy as going out and buying one, unless you have lots of cash to spare plus (for some options) space and load-bearing capacity
Yeah, I'm painfully aware of the depths of volt-nuttery I've exposed myself to by getting a 7.5 digit capable meter. It's a very deep and very expensive hole...
Even so, the voltage reference comes with a calibration sticker and the voltage was measured with a calibrated meter. The Vishay resistors should be within a reasonable distance of their original tolerance and I'm not even close on either, so there's certainly more work to be done.