New concept of a power amp

[Wavebourn]

Thinking about power amp without coupling capacitors, with local feedbacks around output tubes, I come to some interesting solution.
Everything is DC coupled, input capacitances of MOSFETs are bootstrapped by feedback from anodes of output tubes, and their bias depends on stability of current sources in the negative rail of power. When tubes start, or some of them dies, everything else is fine. It is safe amp, not self-destructive.

Looks like some nice new construction.

Was it patented by somebody before, or not yet?

The schematic is a bit incomplete: I could not draw more because free version of Rimu Schematic software limits number of elements on the drawing.

[SemperFi]

Very interesting indeed. The feedback from plate to source makes some kind of 'super anode follower'? Some similarities with my DC coupled current mirror amps, that have been working faultless for a decade now, ie no stability problems even tho direct coupled.

[kenpeter]

two voltage followrs fighting for control.
I dont see where they forgive each other to agree a quiescent current?

[revintage]

Hey A,

Interesting idea. Still it doesn´t make sense to me.

Iq for the left MOSFET is set by R1. Then you have a current-source where current is set by R13 at the bottom that also should set Ig.

If I for some reason should be right, one might substitute the CCS with a gyrator?

Ooops Kenpeter, our posts crossed;-)!

EDIT: Maybe something like this as SE?

[smoking-amp]

I think the feedback amount for this setup will be very low due to the Mosfet source impedance being so low. Basically becomes current thru the feedback resistor versus the delta-Vgate*MosfetGm. But then the Mosfet forward gain (CCS loaded drain) increases the loop gain enormously to compensate. Still, the bias point will not be stable against Mosfet temperature drift due to the huge gain at the Drain. Revintage's gyrator looks to be necessary.

I have modeled a somewhat similar circuit that uses the P channel Mosfet similarly, but instead of input signal driving its gate, I made the bottom drain-loading-CCS a driving pentode. I thought of it more along the line of a modulated CCS load (the P Mosfet) for the driving pentode. The pentode current then has to equal the modulated CCS current, with large loop gain available (CCS versus pentode current) to make it so. An additional stabilizing feedback (not temp. sensitive) went back to the driving pentode screen grid.

[revintage]

Have simmed and by substituting A´s R5(470k) with 200k distortion seems to be OK.
The problem should be the input capacitance of the MOSFET.

[smoking-amp]

The feedback to the Mosfet source is tracking the gate input signal, so should increase the input Z (so lowering Cinput). Crss from drain to gate would be important though. P channel parts also have the worst capacitance specs. Another issue is the high impedance that is driving the output grid, but feedback controls that (at least for tube turn on, but not for turn off).

[Wavebourn]

Originally Posted by revintage

The problem should be the input capacitance of the MOSFET.

That's why it should be better to use P-type tube than MOSFET.

[revintage]

That's why it should be better to use P-type tube than MOSFET.

But seriously, with a decent driver like E88CC, Fu will be in the ballpark of 75kHz with IRF9610.

The feedback to the Mosfet source is tracking the gate input signal, so should increase the input Z (so lowering Cinput). Crss from drain to gate would be important though. P channel parts also have the worst capacitance specs. Another issue is the high impedance that is driving the output grid, but feedback controls that (at least for tube turn on, but not for turn off).

Whatever causes it, hf-response sucks.

[smoking-amp]

6L6P ?