Op Wed, 22 Nov 2006 12:22:38 +0000, schreef Mark D. Zacharias:

Suspect the bias diodes. It isn't uncommon to get a bad internal
connection in one. Replace them, even if you have to fabricate your
own from discrete parts, don't believe a meter.

To check the diodes, it helps to have some current flowing ~100 ma and
then apply heat and cool to them. The voltage change should be smooth
with temperature.


Ditto the diodes. Also, there has been a number of fake, counterfeit
transistors sold in recent years. If your MJ series did not come directly
from Motorola, Onsemi, or B&D Enterprises they could be fakes. You can
Google the topic of counterfeit transistors for an eye-opener. I've had
exactly this happen. New parts with NO other circuit problems, fail within
minutes when playing into a load.

I have found a source for more suitable power transistors (the recommended
2N3773/2N6609), and I hope to receive a set of these in de mail today.
Also, I'll check the diode combo again for unusual current/temperature
drift, although I couldn't find anything wrong during previous
measuerements.

The trouble is that the amp seems perfectly fine when operating at 90%
of the normal mains voltage. Only when it's hooked up to the full 230
volts AC, trouble hits almost instantly.

But first I think I'll whip up a sort of electronic fuse to completely
detach the end stage from the power supply as soon as the voltage of NPN
emitter relative to the PNP emitter exceeds a few volts (i.e. the bias
current exceeds 5 amps). This way, I should be able to provoke the
failure without having to spend $15 in transistors and 15 minutes worth of
work every single time. This little tool is something I've been planning
to build for some time anyway, and this is as good an occasion as any to
build it.

Thanks for your reply and suggestions (and the others too, of course);
I'll let you know if I find the cause.

Richard Rasker

--
Linetec Translation and Technology Services

http://www.linetec.nl/