Crossline negative feedback?

[skeezbo]

Daydreaming, possible stupid question: would a large resister placed after the PI (where a cross line master volume or cut control would go) create the same positive sonic effects as standard negative feedback? Spikes in the frequency response and other aberrations would be cancelled without effecting the gain of the phase inverter and would leave the other side of the LTP open as an input for a reverb loop, etc. what do you think?

[jjasilli]

Not a stupid question, but no.  A resistor of high enough value for the circuit will appear to that circuit as a NC (No Connection); hence no effect.  Too low a value will be a short circuit.  In between. . .


A resistor shows the same impedance to all AC frequencies (at least the frequencies we are dealing with).  See:  http://hyperphysics.phy-astr.gsu.edu/hbase/electric/acres.html


So such a shunt resistor must decrease the PI voltage output, and do so for all frequencies.  But to be effective as NFB, only certain frequencies must be affected, not all of them. 


To get a resistor to function in a filter circuit, it must be used in conjunction with a reactance, such as capacitance (R-C filter) or inductance (R-L filter), as part of the filter circuit.


Note: the resistor may see such reactance in nearby tubes and/or caps.  But attenuating the output of the PI is counterproductive; and post PI seems an unlikely location for a suitable filter, if one could even be designed.

[skeezbo]

I assumed that since the resistor would be between two out of phases sources the reduction by phase cancellation would average the frequency response to some degree. Thanks.

[HotBluePlates]

I assumed that since the resistor would be between two out of phases sources the reduction by phase cancellation would average the frequency response to some degree. ...

That's a cross-line master volume. All it winds up doing is adding {Signal} to {-Signal} to reduce the net drive to the output tubes.

... would a large resister placed after the PI (where a cross line master volume or cut control would go) create the same positive sonic effects as standard negative feedback? Spikes in the frequency response and other aberrations would be cancelled ...

But how does the new cross-line negative feedback force any change of operation? In fact, how does feedback cause any change of operation the way it's normally used? [Questions to ask yourself]

Feedback reduces distortion at an amplifier's (single gain stage or entire amp) output by comparing the output signal of an active gain stage to the input signal; where there's a difference, the feedback loop takes some of that stage's gain and throws it away (which makes it negative) in the process of counteracting the change/difference between input & output.

And the "change" which is counteracted is distortion.

Along the way, some other effects happen, such as raising or lower of internal or output impedance. Which change is made, and in what direction, depends on how the feedback is derived from and applied to the amplifier.

I assumed that ... the resistor ... would average the frequency response to some degree.

"Averaging frequency response" sounds a lot like "equalization" to me. That's a lot to ask of a humble resistor, who responds to all signal frequencies essentially the same.

And since this resistor is only on the output of the phase inverter, it doesn't compare output to input. Instead, it's only comparing 2 different outputs. It has no gain of its own to "use as muscle" and throw away in order to force a change.


This is probably a good time/place to talk about feedback & how it works. Or not... You were talking about "evening out spikes in frequency response." Since that sounds like EQ, it's probably good to ask whether we should focus on the goal (EQ) or the means of reaching the goal (possible NFB implementation).

Which one do you care about?

[jjasilli]

Expanding on my Reply & Hotblue's response, a fixed resistor across the PI outputs is counterproductive, because it will attenuate the PI signal.  We could simply have fed the PI less input signal to begin with, or designed it for less gain, without adding a component afterward. 
But a pot could be useful as a MV.


Reducing overall vol is not the purpose of NFB.

[skeezbo]

Thanks for the response HBP. I am only thinking about the effect on an ac-30 or tweed deluxe circuit that can be loose on the low end or spikey on top compared to designs that do use negative feedback. So, without real negative feedback the "resonance" effect that dampens low end response is out, and a cut control might help tame the highs. Plus I need to get out my books and re-read the chapters on negative feedback design; I am in over my head about how it really works. Thanks again!

[PRR]

We usually want NFB to sense the *loudspeaker* signal. So it "sees" the OT and speaker interaction.

The usual driver/inverter has smooth frequency response far beyond the audio band (or trimmed-to-taste to reduce bass splatt and hypersonic trouble). Doesn't need "aberrations ..cancelled".

[HotBluePlates]

Thanks for the response HBP. I am only thinking about the effect on an ac-30 or tweed deluxe circuit that can be loose on the low end or spikey on top compared to designs that do use negative feedback. So, without real negative feedback the "resonance" effect that dampens low end response is out ...

Yep, and if you want to fix that, you add negative feedback around the output stage.

As PRR notes, what you hear as "loose low end" is an inability of a high-internal-impedance pentode being able to exercise firm control over a speaker cone's movement, especially at its bass resonant frequency.

When you add feedback from the speaker to the phase inverter, the loop senses the speaker's big motion and implements a counter-acting signal to reign-in speaker flap.

So NFB acts directly on the problem. Any approach without using NFB is just trying to cut bass at the point where the speaker flaps the most. But that's also a very narrow frequency range... You'll need a complex filter to notch that out, and you're still not really correcting the problem at the source.

[skeezbo]

I get it. Problems with speaker resonance are not "aberrations that can be cancelled" but ones that can be controlled by signal returning from the output. I will still re-read those chapters. Thanks again!