LTP NFB question

[Helmholtz]

Yes, that's a variation of uneumann's circuit using the same principle of separating the 2 NFB paths.

Regarding the amount of speaker signal injected to the tail, the bootstrapping effect should work best if the signal at the lower end of the 56k tail resistor is almost the same as the signal at the cathodes. With almost same signal at its ends the 56k looks like a very high tail impedance as seen from the cathodes and thus improves balance.

Make sure your voltage divider resistors have sufficient power rating with speaker voltages above 20V.
I.e. the 560R should be a 2W type.

[catalin gramada]

Yes. And Uneumann found a way to show those shunt appendix have a lot of sense to be used in circuit and is not an accident of design as many texts states. I wish to know to do simulations before but even so all the tests confirmed that. Definitely the solution worth to be keep it and not rejected as an accident of nfb differential design, because it works.

[Helmholtz]

Yes, it's a pity that none of the tube books I know adresses these effects.
Happy having learned something new from this and uneumann's thread:

1) Applying NFB to a LTPI increases imbalance.
2) Feeding some NFB signal to the tail bootstraps the upper tail resistor and thus improves balance (so Merlin is wrong here).

[catalin gramada]

1) Applying NFB to a LTPI increases imbalance.
2) Feeding some NFB signal to the tail bootstraps the upper tail resistor and thus improves balance (so Merlin is wrong here).

You may say it twice. Thank You for helping.

[catalin gramada]

Forgot to mention...with nfb in the grid and proper balance adjustment path in the tail the non inverted 100nF nfb input cap is no longer need it.

[catalin gramada]

As Helmholtz state, for the best balance we should find the same signal in shunt as in cathode, meant perfect common mode. I found I have to raise the series 560 ohm to 680 ohm for the best (still 3mV over the 56k tail) but outputs show slightly imbalance...I don't know how much the symmetry of tube ac voltage swing it count in respect with bias point was choosed but of course the tubes may have their own habits. But as time it show starting symmetrical clipping think is allright as it is.
Yellow-PI input
Blue-OT output
scooped at onset clipping.

[catalin gramada]

Forgot to mention...with nfb in the grid and proper balance adjustment path in the tail the non inverted 100nF nfb input cap is no longer need it.

I think this was stuping thing to say. The path contain DC component meant the V2 grid will not stay at 0V , so correct will be the previous version with the nfb injected through the 100nF cap and return 1M to the grid...I think
So this one will be

[catalin gramada]

This is a hard one for me. There are 80 mV DC on the shunt and less than 0.1 mV DC on V2 grid. Practical such amount of DC to the grid have no influence on the operation. Does it have any sense to isolate the input with a cap please ?

[Helmholtz]

I'd say if shorting the cap doesn't noticeably change plate voltage, there's no need for the cap (and the 1M grid leak).

[catalin gramada]

More clear

[catalin gramada]

I'd say if shorting the cap doesn't noticeably change plate voltage, there's no need for the cap (and the 1M grid leak).

Thanks!

[Helmholtz]

More clear

Yes and as the DCR of the 4 Ohm secondary should below 1R, the actual grid DCV is probably lower than your meter is able to correctly read.