OK ! I worked all evening and morning on the Rochar DVM.
I am lucky, this one is not causing too much trouble, I was able to revive it and make good progress, almost there.
1) Gave it a good clean, looks much more pleasing up close in the flesh now. Looks nice and not a bio-hazard any more
2) Fixed the dead digit : was simply the tube itself that was dead (shit happens...) ... luckily I had a spare one... thanks to a generous TEA member who sent it to me some time ago. His identity shall remain preserved, but you know who you are, so thank you very much !
3) Looked under the chassis at the Vref board. It's got the same two big yellow dipped WW resistors with that special tempco, that I had trouble with, with my other meter... some may remember it ?
They were open-circuit because of moisture getting in somehow, and corroding the leads from the inside. Didn't know how they were constructed of course, so I sadly ruined them when trying to crack them open. So any repair was out of the question.
Anyway, in this meter same problem. One of the two yellow resistors measured open-circuit. Indeed you could see green stuff leaking were one of the leads traverses the yellow "jacket", for lack of a better term. This resistor was the one that's setting the charging current for ramp generator that does the voltage/time conversion. So I thought... open-circuit resistor means zero current charging the timing cap, hence no ramp, hence of course the instrument is unable to measure anything... so since this was identified as a show stopper, I decided to start the repair process from this angle, and go from there. There was no point troubleshooting anything until this problem was fixed.
So I tried to fix this resistor, to see if I could do better than last time. Yes indeed. I managed to crack it open without damaging the coils inside.
Noticed that on one end, the wire had come off the "cup" of the terminal. So I unwound the wire a bit, to get some slack, soldered the resistor back to the board, then tried to solder the wire to the board / terminal strip as well. I managed to do it after half a dozen attempts... the wire is so thin that it kept breaking when I worked it with the iron tip. Also, my tweezers are too big and crap quality to reliably grab that tiny wire. Not to mention I could barely see the wire with my naked eyes to begin with. The opti-visor came to the rescue but it was only good when inspecting the resistor up close... but working distance with this optivisor is too crap to let me do any soldering work, so had to solder pretty much blind, not really seeing that wire... I took a pic up close, with a white sheet of paper to help see the wire.
Anyway, once soldered, the instrument as I hoped, was now alive ! It can take measurements !
4) ... but only in single-shot mode... the trigger circuitry had a problem. If I trigger manually, using the push-button, it works. But if I set the trigger to "AUTO" mode, it doesn't measure anything, no go.
Luckily I fixed it quickly, was easy : the trigger circuitry has x3 5uF axial electrolytic caps to get the oscillator/flip-flop of the AUTO mode going, as well as the reset circuitry. I tested them with the chinese component tester... all 3 were "Unknown or Damaged part" woops.....replaced all 3 with "brand new" 40 year old NOS Nichicon 4.7uF caps I had in my stock, (radial sadly, of course) which somehow measured just fine on the chinese component tester..... AUTO mode now works fine !
5) OK so at that point the meter basically works, I am overjoyed.
I then tested it with my chinese Vref that can output 2.5 / 5.0 / 7.5 / 10Volts.
On the "native" / 5V range, it didn't need much in term of zero and full scale calibration.
Then I tested the 50 and 500V ranges and they are slightly off (a few LSDs, still reasonable), so might play with calibrating those ranges once all else is all and good.
6) All these first steps went quite quickly... hadn't even yet taken the time to check the power supply... so I now did that. All seems good, slightly excessive ripple on some rails so might replace the filter caps later on, but for now the ripple is still plenty acceptable, I think, given that the meter... works.
the 200V supply for the display tubes is not fluctuating any more, so I guess the caps have now reformed. Of course they must be on their way out, and will die soon I guess.... and unsurpringly this particular rails is the one having by far the most ripple, at 10V or so (didn't scope it, but the DMM says 3.1V, so make that times 3 or so to get actual peak to peak value).
7) So at that point I basically had a nice working meter, so I could now start testing it more thoroughly to shake more bugs. Bugs I found :
7.1) On the 5V / most sensitive range, 1mv resolution, with the inputs floating, it displays zero at first but then it steadily, slowly increases, increased.... A cap must be charging somewhere... but where ? Not overly concerned about that in the short term, given that as soon as you short the inputs it displays a solid "0000", so clearly whatever is causing the drift when floating, is not capable of travestying an actual measurement when you want to make one.
EDIT : checking the manual, it says the 5V range is super high-impedance, greater than a Tera ohm. So I guess it doesn't take much parasitic capacitance anywhere in the front-end, to cause the symptom I am seeing ? So chances are, it's not a fault, it's just normal behaviour...
7.2) Now for an actual problem : at first everything was fine when I was measuring the 10V output of my reference... meter would display "10.00", eh. However after a while, it started spuriously displaying "12.00" instead... and as time went on, it's now pretty much et on 12V rather than 10.
Trying letting the meter cool down a bit, but when I powered it up again, it was set straight away to 12V rather than 10. So maybe not a heat related issue then... rather something that went wrong and is now permanently wrong... which in a way is a good thing since solid faults are of course so much easier to trouble shoot than random or spurious ones...
Anyway, I did some testing and it looks like the problem lies in the counter stage/Section of the display board n°2 (from the left) : when you trigger a measurement, all the counters / digits gets reset. This works fine. However where this particular board is having is a problem, is when it wants to "reset" when it overflows :you reset it, then it can count just fine from 0 to 9, BUT if it needs to overflow and roll back to 0... it rolls back to '2' instead. So I guess I need to look at the schematic of the counter of that board, and try to understand what could be going wrong.
So overall I am quite happy. it's in good nick, inside out, and basically works, almost, just fine, after only a day working on it.
So since it's going quite well, I will keep working on it in the coming days. Try first to fix that counting issue, then replace a cap or two, then do some more restoration work. Need to fix the usual problem of the orange colour filter falling off thanks to the foam holding it in place, turning to dust.
Also noticed, buried, hidden behind the front panel by the input jack, an electrolytic cap in series with a resistor, wired across the negative jack and earth/chassis jack. Cap seals do'nt look great and the resistor is clearly cooked.