[HBR] Output stage is power oscillator

[Ian Wilson]

The problem turned out to be that even with the input stage "disconnected"
it can see enough of the output to be able to have greater than unity gain
around the loop.

I had a small ceramic capacitor connected to the triode grid, not connected
to anything but sticking up 1/2" (at most) above the chassis. That turns out
to be enough. Connecting a length of screened lead to the grid instead
(with the remote end connected to the volume control) and the problem
went away.

As a side effect of trying to get rid of this I now have better bypassing
around the product detector stage :)

Once things were working, I made the trip to the garage so I am now
running the correct triode-pentode (6CX8 instead of 6AW8A). There
is a little more volume and it sounds a little more balanced. I will stop
here before getting into "transparent highs", etc.

73, ian K3IMW