LED bias and negative feedback

[passaloutre]

I'm building an amp loosely based on the 5E11 Vibrolux, which has a cathodyne phase inverter, fixed bias on the power tube grids, and global negative feedback from the OT secondary back to the cathode of the stage preceding the phase inverter (see photo).

For a little more gain, I decided to try the LED bias that I've read about on Merlin's site. It occurs to me now though, that the negative feedback typically relies on the resistive voltage divider formed by the NFB resistor (56k here) and the cathode bias resistor. With an LED for bias here, there is no resistive voltage divider, but my knowledge is lacking to figure out what the consequences are.

Is the NFB circuit performing similarly here to how it would with, say, a 1k5 cathode bias resistor? Or is it doing something unexpected?

Should I just keep it simple and stick to a 1k5 so this will do what I want?

[Willabe]

Play it with the mod, then disconnect the NFB wire and play it, what do you hear?

[passaloutre]

Well I can confirm the amp "works" as drawn. It sounds good, and I know I'm getting feedback because I initially had the OT plate leads reversed. I will certainly try a simple resistor bias this evening, but I'm more curious about the science and how the LED behaves vis-a-vis negative feedback. I learn a lot of theory on this site, and this circuit brought up a question I don't know the answer to, so I'd love to hear what the experts have to say.

My understanding is that a diode acts as a short if the voltage across it is greater than its forward voltage, so my intuition is that any feedback signal greater than +1.7V gets dumped to ground, while signals under that threshold are applied as negative feedback??? I'm certain that explanation is incomplete, if not entirely incorrect, so I'm hoping someone else can explain it.

[Merlin]

The LED behaves like a fully bypassed cathode, so there is no feedback at all in your configuration.

[passaloutre]

Thank you Merlin. So simple when you put it that way. So that means that a bypass capacitor (so long as it covers the frequency range of interest) would also negate the negative feedback? Because negative feedback relies on being able to change the voltage at the cathode?

Ah yes, it just clicked in my head. And a presence control is the case where the capacitor does not cover the full frequency range...