Twin Reverb AB763: Fender fiberboard leakage and conductivity

[22uf]

How many millivolts would be too much? I know that's a "how long is a piece of string" thing. But I am guessing all these old boards have some degree of conductivity by now.

I'm seeing a peak of 20mv in the bias board at various spots near the diode eyelets and 10-15mv or so around the plates on the main board.

My power transformer was running at 60c so I tore it down and pulled out all the gunge (a lot of it) and sanded down the loose laminations to a mirror finish and re-assembled.



Its now pretty steady at 45-55c (30c ambient). Its also now drawing 0.6-0.8amps where as before it was drawing 1.0 amp pretty solidy (probably because I changed some leaky coupling caps for M150's). I'm otherwise really happy with how it plays .. just worried that I can't really kick this out the door to my buddy and call it done with a conductive board. Or can I? I'm not very experienced with these (or generally).

It doesn't seem to be causing any issues, the noise floor is great since I swapped out the treble caps for micas and replaced the RCA jacks and the tone stack caps etc.

[sluckey]

I would not be concerned about 20mV leakage, especially if the amp has no ill symptoms.

[22uf]

Cheers Sluckey!

I'll send it on its way then. 

[pdf64]

…sanded down the loose laminations to a mirror finish and re-assembled…

Transformer laminations are supposed to be insulated from each other, by being etched to form an oxide layer. As I understand it, that keeps the eddy currents contained and limited.
That’s why laminations are used, rather than steel blocks.
The point being that it wouldn’t seem to be beneficial to polish them off to a shiny, bare metal finish.

[Willabe]

Each transformer lamination is painted to isolate it from the lams next to it. If you remove that paint, the lams short out.

Very bad thing to do to a transformer.

[PRR]

The 20mV can be compared to the smallest important voltages in the circuit. Typically grid-cathode bias. 0.020V is "nothing" compared to typical 1V biases.