Nixie DMM repair updateOK. So... now that the meter kinda "works", from cold, I thought OK let's start today's debugging session by measuring the zero offset from cold, and log some data, because we are TEA here eh.
So I fire it up for the first time today and... "free running" again

I immediately tap the back of the back plane PCB and wiggle that #2 Nixie board... within 5 seconds the meter decides to obey and behave !

So yeah.. I really need to suck all those solder joints on the back plane, and redo them all... might redo them with silver solder, for I gather it's more "flexible" , less prone to cracking, than regular solder. Anything that can help to keep those back plane joints from cracking twice a day... is good to take.
So I logged some data, see graph below. Yes, I am taking this seriously, see !

As I felt before, it's like an exponential. But now I have actual data to prove it !

It starts very high, 1,000mV offset, then in only 30 seconds it's already plummeted down to 500 or so. Then it slows down a lot and I switch data logging to one minute rather than 10 seconds.
After about 30 minutes it kinda settles for about 65/70mV, but I kept logging data anyway, in case it drops lower than that in the longer term... it did not, 60mV was like " its " value, it was its thing... so be it. Still wayyyy too high to fall within the compensation range of the Zero trim pot, though.
The more data you have, the better your chances of embracing the bigger picture and discern patterns, and not confuse a micro-event for an overall, guiding pattern. So it helps get a proper diagnosis quicker, and not follow red herring like I have been doing for the past week with thermal theories

Yes, I am glad I logged that boring data for so long, patiently, sitting at the bench looking at my smartphone/clock app, because while it was stagnating at 60mV or so, at 37 minutes something interesting, weird, sudden, unexpected happened.... Out of the blue, instead of reading 60mV, it read 2 or 4mV !

Just a spurious reading, a glitch, then back to its cruising level of 60mV. But... a few seconds later it did it AGAIN... and again... with every passing minute, it spent more and more time displaying near zero, and less and less time displaying 60mV.
After an hour or so, zero had become the norm, and 60 had become the rare/spurious reading, things had turned 180° !

So, I was able to use the trim pot to finely adjust it to spot on zero. 2 hours later it was still at zero, holding it just fine.. and I would still get the random 60mV reading. Well actually after a while it started climbing a bit and settled for 150mV or something, whatever, its living its life.
SO !!!
Having this sudden change to zero.. same symptoms as last night ! Must be a dodgy joint somewhere, not a leaky transistor any more !
So I will redo all the joints on that comparator board. I mean all the joints in the critical area, the front end of the two comparator section : S1 to S4, and all their associated satellite components.
There is still a thermal effect looking pattern here, though, isn't it... but I guess it must be happening within the dodgy joint itself, rather than in a component. Must be happening within this microscopic world that is the interface between the bottom of the solder blob, and the top of the copper pad on the PCB. I guess...

So... lots of wicking and solder sucking action for me soon ! Hoping that will fix it, cross your fingers !!!

-group-therapy-thread/?action=dlattach;attach=1462177;image)