ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Advanced feedback loop questions..

11/4/1999 4:12 AM
David Manson	Advanced feedback loop questions.. I have a few questions about negative feedback.. I know some basics about it.. it basically reshapes the sound by cancelling it out,, high frequency neg feedback will take away the highs,, low freq neg feedback will etc etc.. Can anyone explain some more about negative feedback in tube amps - maybe a little about what changing from 4-8-16 ohm taps do,, I also wondered what the negative feedback signal sounds like.. It comes from the very end of the amp circuit so i'm guessing that it is just a low voltage copy of the amps actual sound.. is the signal always on or is it only in effect when the guitar is played? I wondered does the signal phase have to be inverted? or is it just the same signal with an opposite polarity? anyone?

11/4/1999 4:12 AM

David Manson

Advanced feedback loop questions..

I have a few questions about negative feedback..
I know some basics about it.. it basically reshapes
the sound by cancelling it out,, high frequency
neg feedback will take away the highs,, low freq
neg feedback will etc etc.. Can anyone explain
some more about negative feedback in tube amps -
maybe a little about what changing from 4-8-16
ohm taps do,, I also wondered what the negative
feedback signal sounds like.. It comes from the
very end of the amp circuit so i'm guessing that
it is just a low voltage copy of the amps actual
sound.. is the signal always on or is it only in
effect when the guitar is played? I wondered does
the signal phase have to be inverted? or is it
just the same signal with an opposite polarity?
anyone?

11/4/1999 5:29 AM
Randall Aiken	David, you are basically correct in assuming that the negative feedback signal is a "low voltage copy" of the output signal, it is simply the output signal run through an attenuation network and fed back to a prior stage. However, there is a difference between open and closed loop "look" of the feedback signal. When the loop in an amp is opened (at the output of the attenuator), the feedback signal will look exactly like an attenuated copy of the output, complete with whatever distortion and uneven frequency response might be there. When the loop is closed, though, the feedback point will look as close to the input signal as the loop can make it, because the loop acts to remove any differences between the input and the open-loop output signal, forcing the closed-loop output signal to match the input. This "look" at the feedback summing point will change depending on how the feedback is applied. For instance, in an inverting configuration (with the non-invering input at "ground" potential), the feedback and input are summed into the same node, called a "virtual ground", and if you look at this point closed loop, you'll see almost no signal at all. In a non-inverting configuration, you'll see a signal that looks like the input. The feedback is always "there" in a tube guitar amp, but, since the amp has a transformer in the forward path, there is no feedback at DC or very low frequencies, so in effect, there is no feedback when there is no signal, since that would represent a DC signal. The signal phase has to be inverted relative to the input of the feedback stage for negative feedback, otherwise you have positive voltage feedback, which will increase gain, and too much will result in oscillation, as many people have found when installing a new output transformer with the primary wires reversed. It will howl uncontrollably when the volume is turned up. For a bit more on the subject, go to http://www.aikenamps.com and read the paper on the tech pages entitled "What is negative feedback?". It explains the stuff about taking the feedback from different taps. Randall Aiken

11/4/1999 5:29 AM

Randall Aiken

David, you are basically correct in assuming that the negative feedback signal is a "low voltage copy" of the output signal, it is simply the output signal run through an attenuation network and fed back to a prior stage.

However, there is a difference between open and closed loop "look" of the feedback signal. When the loop in an amp is opened (at the output of the attenuator), the feedback signal will look exactly like an attenuated copy of the output, complete with whatever distortion and uneven frequency response might be there. When the loop is closed, though, the feedback point will look as close to the input signal as the loop can make it, because the loop acts to remove any differences between the input and the open-loop output signal, forcing the closed-loop output signal to match the input.

This "look" at the feedback summing point will change depending on how the feedback is applied. For instance, in an inverting configuration (with the non-invering input at "ground" potential), the feedback and input are summed into the same node, called a "virtual ground", and if you look at this point closed loop, you'll see almost no signal at all. In a non-inverting configuration, you'll see a signal that looks like the input.

The feedback is always "there" in a tube guitar amp, but, since the amp has a transformer in the forward path, there is no feedback at DC or very low frequencies, so in effect, there is no feedback when there is no signal, since that would represent a DC signal.

The signal phase has to be inverted relative to the input of the feedback stage for negative feedback, otherwise you have positive voltage feedback, which will *increase* gain, and too much will result in oscillation, as many people have found when installing a new output transformer with the primary wires reversed. It will howl uncontrollably when the volume is turned up.

For a bit more on the subject, go to http://www.aikenamps.com and read the paper on the tech pages entitled "What is negative feedback?". It explains the stuff about taking the feedback from different taps.

Randall Aiken

11/4/1999 6:52 AM
David Manson	I've read it before.. Its what kinda got me interested in negative feedback in the first place.. How exactly does negative feedback extend the frequency response? if you wanted to change your negative to a positive feedback loop all you wouldhave to do in invert the signal again right? (put an inverting gain stage on that purple feedback wire?) - also.. can you create more feedback loops (positive or negative) in your preamp fairly easily? oh.. when you said the loop is open you meant when the feedback wire disconnected right? thanks..

11/4/1999 6:52 AM

David Manson

I've read it before.. Its what kinda got me
interested in negative feedback in the first place..
How exactly does negative feedback extend the
frequency response?

if you wanted to change your negative to a positive
feedback loop all you wouldhave to do in invert the
signal again right? (put an inverting gain stage
on that purple feedback wire?) - also.. can you
create more feedback loops (positive or negative)
in your preamp fairly easily?

oh.. when you said the loop is open you
meant when the feedback wire disconnected right?

thanks..

11/4/1999 7:47 AM
Randall Aiken	Negative feedback trades off gain for "flatness". An amplifier has a bandwidth that is rolled off at the upper and lower ends. If you reduce the midband gain (by applying global negative feedback), you widen the frequency response because you have lowered the "top" of the frequency response curve, so the "middle part" becomes wider. That is a simplistic way of looking at it, but it gets the point across. Another way of looking at it is that feedback tries to make the output match the input. If the response in the forward path rolls off on the top and bottom end, the feedback will tend to increase the response at the ends to make it flat. Once again, the max "correction" that can be applied is equal to the amount of negative feedback. Trust me, you don't want to change your global negative feedback loop to a global positive feedback loop. You will have nothing but a loud squeal. You can create local feedback loops in your amp, but you have to know what you are doing, because it changes other things, like input and output impedances of the stages, etc., and you have to know whether or not to DC or AC couple them, depending on where you put them. A standard common-cathode stage can be converted to a local feedback stage simply by removing the cathode bypass cap. The unbypassed cathode resistor generates local negative current feedback, increasing the output impedance of the stage, and lowering the gain. Yes, disconnecting the feedback wire will make the amp run "open-loop", meaning no negative feedback. You will find it necessary to attenuate the drive signal to the phase inverter (or between the phase inverter and output tubes) if you want the gain to remain the same. Randall Aiken

11/4/1999 7:47 AM

Randall Aiken

Negative feedback trades off gain for "flatness". An amplifier has a bandwidth that is rolled off at the upper and lower ends. If you reduce the midband gain (by applying global negative feedback), you widen the frequency response because you have lowered the "top" of the frequency response curve, so the "middle part" becomes wider. That is a simplistic way of looking at it, but it gets the point across. Another way of looking at it is that feedback tries to make the output match the input. If the response in the forward path rolls off on the top and bottom end, the feedback will tend to increase the response at the ends to make it flat. Once again, the max "correction" that can be applied is equal to the amount of negative feedback.

Trust me, you don't want to change your global negative feedback loop to a global positive feedback loop. You will have nothing but a loud squeal.

You can create local feedback loops in your amp, but you have to know what you are doing, because it changes other things, like input and output impedances of the stages, etc., and you have to know whether or not to DC or AC couple them, depending on where you put them. A standard common-cathode stage can be converted to a local feedback stage simply by removing the cathode bypass cap. The unbypassed cathode resistor generates local negative current feedback, increasing the output impedance of the stage, and lowering the gain.

Yes, disconnecting the feedback wire will make the amp run "open-loop", meaning no negative feedback. You will find it necessary to attenuate the drive signal to the phase inverter (or between the phase inverter and output tubes) if you want the gain to remain the same.

Randall Aiken

11/4/1999 10:12 AM
anonymous	So you could just create another negative feedback loop by copying the old one.. first you would need 2 gain stages to invert the signal.. and just feed the AC signal from the plate of the second stage to the cathode of the first stage and make sure to knock down the voltage some - maybe with a good sized resistor... sound like that would work? or would you need a cap to turn it to DC..

11/4/1999 2:13 PM
Ken Gilbert	I don't mean to rain on anybody's parade, but this is exactly why I DON'T use global, or "loop" NFB in my amps--it's too complicated! Volumes of books have been written on NFB since it was discovered, and it turns into quite a region of the unknown. Trust me, there's a lot of shit going on when you close the loop, and you can't expect Randall to give you a detailed explanation of what's going to happen and why. I suggest you all start playing around with LOCAL NFB, around ONE stage. It's very easy--just take your average gain stage and add a single small resistor from output (after the coupling cap) back to the grid. Voila, NFB loop. Now, to get fancy and start doing something with that feedback, use two resistors in series, and this time throw a cap to ground at the junction. What happened? Why? FIGURE IT OUT! Then try placing a small cap across one of those resistors, and then across both of them. Try bypassing one, and then using a cap to ground as well. By localizing the NFB over a single stage, and avoiding the OPT and its hideous phase characteristics, you are assured you won't have massive oscillations... this is of course unless you add about 3 series caps from output to input and turn it into a LFO. Just take a look at any old vibrato/tremolo circuit. That's a specific type of NFB, with enough phase shifts added in by the caps to create positive feedback. By adjusting the resistance to ground, you change the time constant of the charge/discharge characteristics of those caps, and you adjust the rate at which it oscillates. Believe me, with NFB, you want to start butt-assed simple. KG

11/4/1999 2:13 PM

Ken Gilbert

I don't mean to rain on anybody's parade, but this is exactly why I DON'T use global, or "loop" NFB in my amps--it's too complicated! Volumes of books have been written on NFB since it was discovered, and it turns into quite a region of the unknown. Trust me, there's a lot of shit going on when you close the loop, and you can't expect Randall to give you a detailed explanation of what's going to happen and why.

I suggest you all start playing around with LOCAL NFB, around ONE stage. It's very easy--just take your average gain stage and add a single small resistor from output (after the coupling cap) back to the grid. Voila, NFB loop.

Now, to get fancy and start doing something with that feedback, use two resistors in series, and this time throw a cap to ground at the junction. What happened? Why? FIGURE IT OUT!

Then try placing a small cap across one of those resistors, and then across both of them. Try bypassing one, and then using a cap to ground as well.

By localizing the NFB over a single stage, and avoiding the OPT and its hideous phase characteristics, you are assured you won't have massive oscillations... this is of course unless you add about 3 series caps from output to input and turn it into a LFO. Just take a look at any old vibrato/tremolo circuit. That's a specific type of NFB, with enough phase shifts added in by the caps to create positive feedback. By adjusting the resistance to ground, you change the time constant of the charge/discharge characteristics of those caps, and you adjust the rate at which it oscillates.

Believe me, with NFB, you want to start butt-assed simple.

KG

11/4/1999 8:08 PM
David Manson	Any good books on negative feedback? TUT has a little about it but not much on creating/ modifying the loop... Weber's books neither..

Page 1 of 5

Next>

Last Page>>