LTP NFB question

[Chuck H]

What's strange to me though is that I'm not seeing much (if any) gain difference in the scope shots between NFB and no NFB.?.

[catalin gramada]

Is there any NFB signal there?

Here you go. Without and with nfb applied

[catalin gramada]

What's strange to me though is that I'm not seeing much (if any) gain difference in the scope shots between NFB and no NFB.?.

Both pics near clipping at 30Vpk 4 ohm tap. See the one is 1.12vpk input, the other is 2.5vpk probes to OT output and PI input. There is 7 db nfb around and very good balance.
The PI is biased at 1.4V around so with 2.5vpk at input no clipping may occur.. I will put a guitar on it tomorrow to see how it sound and tweak the bias a little bit maybe

[Chuck H]

Looking pretty good

The only other design considerations I have are:

What is the circuit for your -100V supply? If, for example, you have more than -100V (or would that be less than -100V ) and are using a divider with a cap across the shunt serving double duty for AC bypass and ripple rejection, and the whole circuit is elevated by the feedback divider shunt resistor then there is still going to be DC elevation at the PI cathode from the 100V divider shunt being in series with the 56k "tail". This would change ALL the PI voltages. Potentially messing up your direct coupled plan. Also, with the cathode voltage circuit elevated by the feedback shunt you are inviting more power supply noise injected at the feedback node. Though not much with a 10R shunt. Just a thought.

[catalin gramada]

Looking pretty good

The only other design considerations I have are:

What is the circuit for your -100V supply? If, for example, you have more than -100V (or would that be less than -100V ) and are using a divider with a cap across the shunt serving double duty for AC bypass and ripple rejection, and the whole circuit is elevated by the feedback divider shunt resistor then there is still going to be DC elevation at the PI cathode from the 100V divider shunt being in series with the 56k "tail". This would change ALL the PI voltages. Potentially messing up your direct coupled plan. Also, with the cathode voltage circuit elevated by the feedback shunt you are inviting more power supply noise injected at the feedback node. Though not much with a 10R shunt. Just a thought.

The negative supply provided by a spare 45V winding on the PT I used. I did a doubler and fixed to 100Vdc due to some capacitors I have on hand. 470uF was used in the end and 100cps ripple is less than 0.5mV.
The voltage vary mainly by wall voltage from 92 to 100v (that is what I recorded today) .Still the bias run into 1.2-1.5 V range making the plate to vary from 175 to 190V. Still the next stage is a 12au7 CF buffer with 380v on the plate, it have a lot of excursion with a centered bias at 175 or 190 V so no a concern from this point. The voltage on negative supply start at power on so when standby is on have no ramping delay to endanger the circuit. Some protection with neon was also provided for CF. please. I will draw a schematic later, nothing fixed just some sketch and ideas to experiment like this nfb circuit we're talking about. Thanks for helping! Seems to be done.

[Chuck H]

See the one is 1.12vpk input, the other is 2.5vpk

I was looking at these two shots. The first is with no NFB and the next is from "It works" (with nfb)



I had not realized you were adjusting the input signal.

[Helmholtz]

One potential problem (as I see it) will be ensuring there is almost zero loading or through circuit impedance for the cathode voltage source.

In any case the negative supply will have to supply the tail current.

[Chuck H]

In any case the negative supply will have to supply the tail current.

Right. And being a 45V supply is the only description we can't know if it's up to more than twice the current (voltage doubler was used) used by the PI. The sine wave scope shots certainly look alright. May as well test it well into clipping before plugging a guitar in.

What's interesting about CG's amps is that he almost always tries to idealize circuits like he's building reference amps. But he's making guitar amps. As far as I can tell his amps aren't using anything like typical guitar amp designs. Which is really cool.

[catalin gramada]

Right. And being a 45V supply is the only description we can't know if it's up to more than twice the current (voltage doubler was used) used by the PI. The sine wave scope shots certainly look alright. May as well test it well into clipping before plugging a guitar in.

What's interesting about CG's amps is that he almost always tries to idealize circuits like he's building reference amps. But he's making guitar amps. As far as I can tell his amps aren't using anything like typical guitar amp designs. Which is really cool.

Hey. Is almost the same as the classic one.The real change from a functional point is grounding inverting stage directly to the ground and not to the virtual ground meant no "positive" (read in phase) nfb at the input . Of course may get rid of negative supply using 500V supply in the classic circuit. But the plates voltages + voltage drop over tail will not allow to bias the next stage in dc configuration please.

[Helmholtz]

Another thought: The effect that PI output balance gets worse with NFB applied may (at least partly) be due to the NFB compensating some asymmetry of the following circuit including the PT.

Also don't forget to look after consequences of the Miller effect.
With a 1M resistor and a Miller capacitance of maybe 70pF, the NFB will drop above around 2kHz.
Testing the output with a 400Hz square wave should show overshoot.

I always thought that the feeding the NFB to the tail as well should improve PI balance and your results nicely prove that.

[catalin gramada]

Hey. Tested with same sine input at 400hz and 10khz. Is pretty consistent.



I tied also with square at 400 hz but is no relevant. For some reasons my square input signal is extremly noisy as see. Still it shows some ringing, which is pretty normal as have no output of compensation but also don't see any severe overshoot or so. I have to make clear the bench first to proper measure.

[catalin gramada]

Late: There is some degree of hf phase shifting but don't know ithere s a reason to mess with the pole as time no sign of severe hf osscilation shows ? Otherwise looks pretty acceptable to me...

[catalin gramada]

This is the whole circuit. The output is a Hammond 1.9k (1650R?) drived by 6 Russian 6P3S (5881 kind of) at 480V for clean 120Wrms output

[catalin gramada]

This is the best I may record.The ringing could be from my signal generator or my crappy probes because I see it also on the input.

[Helmholtz]

This is the whole circuit.

Might have lost track with different versions, but with your last schematic the NFB signal doesn't pass a 1M resistor, so no problems with Miller capacitance.