Symptom
I have a mixing console, where one 6-pin stereo pot is errartic. When I turn it all the way down I sometimes can still hear the left channel at full volume.
Explanation
The circuit is a regular voltage divider where the wiper is connected to a JRC NJM 4580 op amp inverting circuit with 2 SMD resistors and one SMD capacitor (reverse enigneered).
The only explanation I have is that the pot sometimes looses its GND connection. I connected it to the GND of the other channel, but this didn't change anything, so the error is not on the PCB.
The pot appears to be an ALPS RK12.
So the disconnection must be in the pot itself.
Mysteries
What I don't understand is the following: I can resolder the pot and the error disappears only temporarily. I can even bend the pin during resoldering and it does not change much. However, sometimes when the error is present, all I have to do is touch one pin of the pot with a screwdriver and the error disappears.
How can that be? There is no way a loose contact inside the pot can be that sensitive, or can it? It appears like the signal the screwdriver introduces makes to error go away. Also I am amazed because broken pots usually have a problem with the wiper. But a broken "bottom end" is kind of rare.
I wish I could imagine how a loose connection of the SMD op amp could cause such behavior. I am ready to buy that an SMD circuit responds to screwdriver tapping. I haven't resoldered the op amp yet.
And finally
I cannot easily replace the pot, because I need to take the whole device apart which may cause more harm than good. Also the pot is an industrial pot and not easy to get. And worst of all: I'm beginning to doubt that it is really the pot.
Am am aware that answers to my question may be speculative. But I am running out of ideas, so that'll be okay.
Best Answer
Here is what the inside of a different type of pot looks like:
Your pot will have similar construction inside to this one, though it will be hidden by the molding. There is a metal part that connects to the circuit board. In the photo it's held by a rivet or similar crimped construction to the substrate, which has a silver conductive printed portion overlapping the resistive element.
If the rivet on the 'low' (CCW) side becomes loose, the exact symptoms you describe will happen. Same if the substrate is cracked. Your type of pot may be slightly different- perhaps the lead is embedded in a conductive epoxy rather than a rivet, but bottom line is that there is a transition from the pin to the substrate to the element and somewhere it is intermittent.
If you replace the pot, the problem should go away.
But first, check that you have not missed a lifted pad on the PCB which is more likely on a single-sided PCB. A crack between the pad and the trace can be difficult to see- wiggling the pot and looking at the pad with a magnifier in good light may help. Be sure to clean any flux off of it first with proper flux remover.
Components such as pots are frequent sources of trouble, either from breakage or from damaged PCB conductors because they're connected to the outside world, and users can be rough by times.