Wandering reading on vintage DMM HP3470 (HP34750A + HP34702A)

[Swainster]

Hi Guys,

I seem to have caught something from my 34401A repair so now I have another patient, but this one has a more subtle problem than being burnt out. I bought this HP 3470 "measurement system" from the Bay and initial impressions where that it has survived it's near half century life remarkable well. It shows a few dints and scratches, and was probably dropped on it's arse at one point, but was very clean and still operational. A few tests showed that its reading were pretty close - probably good to 3 digits, but this is a 5 1/2 digit, 200000 count instrument so...

Actually the first problem experienced was that after warming it up for a couple of hours I started doing some comparisons with another multimeter, whereupon the display got stuck. Power cycling didn't help and I figured that this was probably the longest it had been turned on for a while, and something had probably popped. Luckily reseating all the DIP ICs got it chooching again so it was probably just a bit of oxidation. Since then it's been behaving itself except that the reading slowly oscillate by about +-20 counts. This even happens with a shorting plug in place so seems to be internal to the device somewhere. The oscillation looks to be regular and "sine-like" with a period of something like 10 seconds (I didn't measure it).

One thing to note is that this is a US option meter (option 060), and I live in a 50Hz country (option 050). I managed to find the correct HP part to convert to option 050 but it will be a while before it arrives. Could this issue be solely due to power line aliasing? I don't have an AC bench supply which can generate 60Hz mains so cant really do much to confirm this, however +-20 counts seems to be quite a lot of noise.

When I had the PCBAs out (only one screw required, which was missing anyway), I did a quick check on some of the electrolytics in the power supply. Only one seems to be showing signs of aging, which is pretty good going for nearly 5 decades. Not really sure if it's a aluminium or tantalum electrolytic as its a HP part number. Interestingly the part is marked with 20V, but the parts list specifies 30V and the actual measured voltage is well over 30V. I guess HP knew what their own parts were capable of. So could this be the cause of the wandering reading? Naturally I'll be changing this one out. BTW, the component date codes were from '72-'73, indicating this is an early model. Unfortunately the original serial number has been lost and replaced with a very authentic looking serial number tag but with a nonsense serial number stamped on it.

And finally on a general note, to give this meter another few decades of life, what (if any) parts should I change as a matter of course? Or should i just keep my sticky fingers out of it as much as possible (if it ain't broke, don't fix it). For instance, I measure a few of the tantalums on the board and they all seem spot on.

[Swainster]

Still trying to troubleshoot this problem.  I temporarily swapped out the electrolytics at the transformer output but although the rail looks much healthier on the 'scope, the wandering reading is still there. I've found that it is a bit more random than I first though, and can go up to few hundred counts on some ranges. One interesting observation is that the noise pretty much disappears when I ground the main board e.g. with a 'scope ground clip.