All very odd. One of the crowbars is doing what it should do and heating
up a couple of degrees in the process.
Randomly about 1 in 5 times at switch on, this crowbar conducts and
drops about .9V as a rail voltage is going on that speaker line.
Oddly , at 50% mains, whatever is wrong , will drop out after 1 or 2
seconds returning mains draw to normal.
One of the output devices failing internally and correcting/"breaker"
going-o/c from current passing ? With the heatsinks off and manually
switching off the amp, IR thermometer should show which, the next time
of powering up.
There are bad design/assembly errors in this amp.
Firstly what does not at the moment seem to be the problem, TO220
pre-drivers , the legs have shoulders so can normal assembly take
advantage of thru-board anchoring of the pins , when soldered. But there
is 3mm or so of freeboard under the shoulders and 2 of the 4 such
devices are waggling loose , flexing the pcb traces with any
movement/vibration of amp.
The 4 h/s screws per h/s have no spring washers under the heads, so the
polyester of the pcb compresses with age and heat (1998 make), one of
the screws was not very tight, I suspect the precursor problem.
..