Potentiometers causing voltage rails short?!

[echoplex]

Hello there,

I'm currently prototyping/building my own synthesiser which use ±15V on the supply rails.
The front-panel board which hosts all potentiometers will sometimes short the +15V rail to GND when all potentiometers are soldered in.
I simply can't find the problem and it makes no sense.. sometimes I will remove some pots again and the short is gone.
Then I solder the same potentiometer back in and it's not shorting. Makes no sense.
What I found out so far: When I select the 100k potentiometers which are directly connected to the supply rails, to a higher value, let's say 110K, the problem will not occur.
Has anybody any clue what it could be? I've never experienced anything like that before.
Thanks

 

[JohnRoberts]

put 100 ohm resistors in series with the +15v feed to the pots. When one of the resistors releases it's magic smoke, that will be the bad pot.

JR

 

[David Kulka]

Probably not a bad pot but a mechanical short, something touching something, that shouldn't be.

 

[scott2000]

I had a burr mess with me once...

 

[echoplex]

This is happening on two prototypes by now.. so definitely not a bad pot but something wrong wither mechanically or with the PCBs ..
But I don't really suspect the PCBs since I've built a third one which is working..

 

[echoplex]

before the voltage hit the rails there is one 0.1uF capacitor for each rail.
I found out that somehow the pots will affect the rails sometimes and mess with the power supply.
should I try and increase the capacity for those capacitors? to let's say 0.22 uF?

 

[Khron]

Some well-lit and well-focused photos may or may not help...

 

[JohnRoberts]

before the voltage hit the rails there is one 0.1uF capacitor for each rail.
I found out that somehow the pots will affect the rails sometimes and mess with the power supply.
should I try and increase the capacity for those capacitors? to let's say 0.22 uF?

That is unlikely to make a difference.

Look for a dead short somewhere... A bad solder connection or something like that.

JR

 

[mfratus2001]

A "short" is provable. It would be measurable even when the unit is off. For dead shorts, I use my beeper function of the DVM and assemble it and determine WHEN it shorts.