UK Vintage Radio Repair and Restoration Discussion Forum - View Single Post

Leon Crampin · 12th Nov 2005, 4:30 pm

A most interesting analysis of a well known problem. Metal migration is not by any means a new phenomenon - silver can be very difficult in this respect and can cause havoc in thick film resistor modules and ceramic capacitors. Alloys are generally less troublesome although brass often loses the zinc.

It's fairly obvious to see now why these devices generally short collector to case, as there is a maximum voltage gradient between these two points if the case is positive. So, why not a rescue job? The case connection has to be at earth potential from an RF point of view to be useful as a screen, but why not connect it to the most negative point in the circuit, rather than positive? This may have the effect of making the whiskers grow back.

Accordingly, I have today set up an experiment, using a Bush TR130, which has 3 AF117s and is a very accessible to work on. Starting with a working set, I moved the two screen connections for the IF amplifiers to V- (decoupled). There was no detectable change in performance, but care must obviously be taken to ensure that the transistor cases do not touch the IF cans. I then thought about a "rescue" job, and fitted an AF117 with a short collector to case to the mixer stage. In this case, moving the can connection to V- had no effect, of course. Fitting a 100k resistor in the lead restored normal operation of the set (much as it would if you just snipped the lead), but maintains the bias across the whiskers.

I intend to keep this set permanently powered up to see if the short eventually clears - easily checked by shorting the 100k resistor. I'll let you know how this develops, but it may take a while.

Possibly, to preserve the AFxxx devices for posterity, the case polarity with respect to the other transistor electrodes should be reversed. We'll see.

12th Nov 2005, 4:30 pm	#26
Leon Crampin Octode Join Date: Nov 2005 Location: Surrey, UK. Posts: 1,873	Re: Whiskers in germanium transistors etc A most interesting analysis of a well known problem. Metal migration is not by any means a new phenomenon - silver can be very difficult in this respect and can cause havoc in thick film resistor modules and ceramic capacitors. Alloys are generally less troublesome although brass often loses the zinc. It's fairly obvious to see now why these devices generally short collector to case, as there is a maximum voltage gradient between these two points if the case is positive. So, why not a rescue job? The case connection has to be at earth potential from an RF point of view to be useful as a screen, but why not connect it to the most negative point in the circuit, rather than positive? This may have the effect of making the whiskers grow back. Accordingly, I have today set up an experiment, using a Bush TR130, which has 3 AF117s and is a very accessible to work on. Starting with a working set, I moved the two screen connections for the IF amplifiers to V- (decoupled). There was no detectable change in performance, but care must obviously be taken to ensure that the transistor cases do not touch the IF cans. I then thought about a "rescue" job, and fitted an AF117 with a short collector to case to the mixer stage. In this case, moving the can connection to V- had no effect, of course. Fitting a 100k resistor in the lead restored normal operation of the set (much as it would if you just snipped the lead), but maintains the bias across the whiskers. I intend to keep this set permanently powered up to see if the short eventually clears - easily checked by shorting the 100k resistor. I'll let you know how this develops, but it may take a while. Possibly, to preserve the AFxxx devices for posterity, the case polarity with respect to the other transistor electrodes should be reversed. We'll see.