Metrahit 24s - Rotary switch doesn't do anything

[Zeitkind]

Hiho,
have a Gossen Metrawatt Metrahit 24s here with a kinda strange problem. History of this meter is unclear, looks as it was heavily used.
If I power the meter up, it is set to AC and automatic - which is the first position of the rotary switch. It measures AC voltage quite OK - at least it somewhat agrees with my Sanwa and other meters I have. I don't have a reference for AC, so well, it seems fine, no idea.
So, first basic function works and I can also select auto or manual range - both work.
Turning the rotary switch does - nothing.  The meter stays in AC settings, although rotating changes the floating voltage on the leads a bit before settling around somewhat 3mV on any the positions of the rotary switch.
The meter has no obvious damages, we cleaned the board and poked around a bit - no shorted stuff found.

Anyone any idea where to look?

[coromonadalix]

a careful teardown could show you  the problem,  i dont know if it is a dial knob   or it has the dial contacts on it / in it to do the required functions

[Zeitkind]

I already had disassembled the meter (for cleaning, it was full of plaster kind dust and such). The rotary switch looks fine, the contacts look fine albeit a bit worn, no other visually noticeable problems. Contact tracing the switch pads were not really successful and the mechanic seems fine (contact pressure and such), but before trying to dig myself a rabbit hole, I decided to ask the great oracle called eevblog.. ^^
I didn't expect problems, so stupid me already reassembled the meter (i.e. soldering back the input contacts).

Edit: Found some pictures here in eevlog, so yeah, will desolder again and inspect the other side of the PCB - I guess the encoding is done on the other side?
https://www.eevblog.com/forum/index.php?action=dlattach;topic=168338.0;attach=653892;image

[Zeitkind]

So, desoldered again. I did not really look at this side while cleaning because I didn't want to remove the display and its conductive rubber stands - which needs to be removed.
The rotary switch seems to work as a voltage divider I guess. So starting with 5V at the right, the first switching position (P1) has R1 and R(all-the-rest) which gives ~4,4V. The next position P2 has ~3,88V with R1+2, P3 3,38V with R1+2+3, P4 at 2,88V, P5 at 2,4V, P6 at 1,9V, P7 at 1,45V, P8 at 0,96V and P9 at 0,49V. So, switching seems fine on the mechanical side, but I have no idea where this voltage goes to. It ends on this side at a resistor and a small cap and a via to some caps on the other side. Should go somewhere to the CPU, but couldn't really find a connection. 

[giosif]

The center contact pad to which the rotary switch connects each of the contacts you mention, that pad goes to pin 47 of the CPU IC.
Check if you have continuity to that pin in your case.

[Zeitkind]

Well, the "GMCAD30" QFP has 52 pins. If the mark on the top is the orientation mark, then it's numbered as on the picture attached. I can visually trace the signal (it's a more or less translucent 3 or 4 layer PCB), let's call it position signal Ps, to around that area and other vias, but not to pin 47. There are two vias in front of the chip, one goes to pin 47, the other above has Ps, but it's not connected to Pin 47, there is about 1,2 MOhm between. There is no pin directly connected to Ps.

[giosif]

Sorry, I should have been more specific: what I call the "CPU IC" is the other chip, closer to the top edge of the board.
"GMCAD30" is what I call the "ADC IC", since I think that's one of its functions.

[Zeitkind]

Ah, stupid me, I thought the switch readings would be done also by the ADC. So, the QFP with 100 pins is not only the display driver but also the main CPU.
OK, pin 47 is indeed directly connected to the position signal (0,16 Ohm). Hmm..

Edit: And it's not pulled down, I can measure eg. P8 with ~0,96V on pin 47.

[Zeitkind]

Well, the signal reaches the controller, but the controller does not react. So it seems that the input of the controller is either blown or something else goes wrong, but I don't see much I could try or test now. Resistance between pin 47 and other pins varies from 20k to several meg but in-circuit. Doesn't really look dead, but well Anyone knows if this µC is obtainable? Or is it a pre-programmed custom one? If not, this meter is toast. 


Edit: It's alive!
Weird, the moment I decided to toss the meter to the box with unrepairable trash, the switching came back to life.  Absolutely no idea why or how. All I did is reassembling the thing and powering up "for the last time". Won't trust this meter much, but well, it's good enough for measuring low voltage stuff and resistance.

[giosif]

It must have sensed the impending abandonment in the drawer with "irreparables" and decided it is not that bad to be useful, after all. 

Glad to hear things worked out, in the end.
And, maybe just me, but I would probably try to bring back the problem by tapping/wiggling various components or areas of the PCB.
Or by slightly twisting/bending the PCB.
OTOH, that could be a rabbit hole and you may be better off just living with the risk (that the issue might reappear).

[shakalnokturn]

I'd give both CPU and ADC ASIC a solder reflow.