Bias issue

[63Deluxe]

I’m working on an old Garnet G90TR amp with 2 6V6’s. I set the bias for the tubes and all seemed well. However as I increase the volume to where the amp begins to break up, the bias on one tube goes more negative - the other tube bias V remains constant. As the bias supply for both tubes is a common source, I’m puzzled as to how one could go more negative. If I had a leaky coupling cap, it should trend more positive ie less negative. Anybody have any thoughts on this?

[EL34]

Posted in wrong board.... moving

[Merlin]

Posted in wrong board.... moving

It had a cathodyne PI. The cathodyne is super balanced below clipping, but when it starts to overdrive the power tubes it can become super unbalanced (see the bottom of this page https://valvewizard.co.uk/cathodyne.html)
So you're probably seeing the effect of bias shift produced by very unbalanced drive signals.

[63Deluxe]

Thanks, that’s a very helpful observation.
This amp was occasionally blowing main power fuse, so I was assuming this was due to one power tube doing all the work as the other tube had been almost turned off due to the more negative bias V.
In regards to fixing this issue, do you have any further advice as to how to move forward? Grid-stopper on PI input? Or is this just a byproduct of this type of PI?

[Merlin]

Increasing the grid stoppers on the power tubes to 10k should help (and on the PI too), although watch out for signs of instability in the global feedback if you do.

[63Deluxe]

Thank you for your help Merlin, much appreciated.
So nice to have a forum of knowledgeable fellow tube-amp lovers to exchange info and ideas.

[pdf64]

I’m working on an old Garnet G90TR amp with 2 6V6’s. I set the bias for the tubes and all seemed well. However as I increase the volume to where the amp begins to break up, the bias on one tube goes more negative - the other tube bias V remains constant. As the bias supply for both tubes is a common source, I’m puzzled as to how one could go more negative. ...

There's a diode effect between control grid and cathode.
When the control grid is reverse biased with respect to its cathode, it's pretty much an open circuit, ie no current flows into or out of the grid.
However as the voltage differential gets close to 0, the grid draws current - hey presto, a diode :)
That diode results in a rectification effect on large signals, as the top of their upper lobes gets chopped off. The causes the momentary bias voltage to be bigger.

Aoken describes the process https://www.aikenamps.com/index.php/what-is-blocking-distortion