Negative feedback - How much?

[Zippy]

I think I've seen negative feedback resistor values from 10K on up to open/none.

I'm adding a variable feedback resistance in one of my current builds with a value of 250K. In the interest of making it "idiot proof", I'll have a fixed lower limit of 10K on the circuit board and put the 250K pot on the back panel next to the output jacks.

How does one select a negative feedback resistance value in a new design? What's "too little", what's "too much"?

Are there theoretical considerations or is it all "feel"?

Thanks.

[diagrammatiks]

check out aiken's article here.

http://aikenamps.com/

under tech info advanced designing for global negative feedback.

there are some theoretical maximums but usually between 3-6db of feedback for a guitar amp.

[Zippy]

That's one example of where Aiken calculates the heck of a situation but leaves me wanting a lil' more. Yes, too much feedback (too low resistance) results in instability.

I see examples of "Marshall does this" and "Fender does this" but what I don't see is what target values/ratios are desirable and why?

Aiken: " Whether or not to use feedback is a subjective matter, left to the taste of the designer."

That takes it all back to my first question...

[Gaz]

Are you using a 4.7k tail resistor like most Marshalls? What tap do you plan on taking the feedback from with you hypothetical 10k resistot/250k pot combo? How many watts is the amp? These all influence you choices.

Here's a good link to mess with: http://ampbooks.com/home/amplifier-calc ... /feedback/

[tubeswell]

Negative feedback is what you want it to be. Too much will cause the amp to start oscillating madly. Too little is like... nothing. Everything in between is fair game.

[guitarmike2107]

How does too much negative feedback cause an amp to oscillate wildly?

I have done a similar set up, and made the feedback variable, by increasing the feedback you can turn a very overdriven power section into a clean amp, that’s with lots off feedback.
You can hopefully hear the effect in this MP3 where I varied the amount of NFB on a distorting SE amp - http://www.mediafire.com/file/jvd5wzzyu ... est-01.mp3

I think the amount of feedback you use is dependent on what type of amp you are building or more so what playing style you are after, i.e. for heavy metal type stuff I would go with more NFB as it keeps things in the power section tight and less sensitive, for big breathy sparkly shimmery chords I would keep it as lower.

From memory I think the amp in this video has Variable negative feedback and he does adjust it later in the video
http://www.youtube.com/watch?v=QGErfbbJ ... er&list=UL

[surfsup]

Here's an article I found interesting just googling around:

http://www.dmitrynizh.com/ecc99-6c33c-s ... _feed_back

[tubeswell]

How does too much negative feedback cause an amp to oscillate wildly?

When I say 'too much' feedback, I don't mean enough to make the amp sound good, I mean so much that the feedback signal is as strong as, or stronger than, the input signal.

(Have you got Merlin Blencowe's 1st book handy? - PP 211 - 214 cover causes of NFB instability.) In a nutshell, in a global feedback circuit when the phase shift is around 180 degrees and the loop gain is unity or greater, then the amp will achieve infinite gain and will oscillate, no matter what amp it is.

[guitarmike2107]

I will need to look it up tonight, but yes I can understand that 180 deg phase shift would cause it, your negative feedback is turned into positive feedback!

Kheunell has also written some good stuff on feedback comparisons between the Bassman; JTM45 and JCM800, there is a summary on his website

The thing I have come to realise about NFB is it can make or break a great amp but there is no magic number.

[Cliff Schecht]

How does too much negative feedback cause an amp to oscillate wildly?

When I say 'too much' feedback, I don't mean enough to make the amp sound good, I mean so much that the feedback signal is as strong as, or stronger than, the input signal.

(Have you got Merlin Blencowe's 1st book handy? - PP 211 - 214 cover causes of NFB instability.) In a nutshell, in a global feedback circuit when the phase shift is around 180 degrees and the loop gain is unity or greater, then the amp will achieve infinite gain and will oscillate, no matter what amp it is.

The phase shift required for an oscillator is over 360 degrees. If you run a 180 degree phase shift into an inverter its essentially the same as achieving 360 degrees and should cause oscillation.

[tubeswell]

How does too much negative feedback cause an amp to oscillate wildly?

When I say 'too much' feedback, I don't mean enough to make the amp sound good, I mean so much that the feedback signal is as strong as, or stronger than, the input signal.

(Have you got Merlin Blencowe's 1st book handy? - PP 211 - 214 cover causes of NFB instability.) In a nutshell, in a global feedback circuit when the phase shift is around 180 degrees and the loop gain is unity or greater, then the amp will achieve infinite gain and will oscillate, no matter what amp it is.

The phase shift required for an oscillator is over 360 degrees. If you run a 180 degree phase shift into an inverter its essentially the same as achieving 360 degrees and should cause oscillation.

A 180 degree out-of-phase sinusoidal signal is the 'opposite' to a 360 degree out-of-phase sinusoidal signal. (e.g.: phase inverter stages put the opposing signals 180 degrees out of phase to each other). NFB works precisely because it feeds a 180 degree out of phase feedback signal into the signal path. (e.g. to take the inverting side* if the PI in a bassman, the signal is inverted 180 Deg through the LTP inverting triode, then inverted 180 Deg again through the output stage on that side, and lastly inverted 180 Deg (again) in the OT secondary (ending up 180 deg out of phase overall with the initial phase at the LTP cathode).

* the signal from non-inverting side of the LTP also get inverted 180 Deg in the output tube on its side, but because that output tube is on the opposite side of a PP primary, the signal there is re-reinverted to be in the same phase in the secondary. The outcome w.r.t. feeding back the signal from the secondary is the same - i.e. the secondary is producing a signal 180 Deg out of phase with the signal at the point at which the NFB signal is being fed back into the signal path.

My understanding of the reason that 180 Deg NFB oscillates with unity open loop gain or greater, is that the fedback signal is (equal to or) stronger than the input signal. Being approximately 180 deg out of phase is the pre-condition for this to happen, but the oscillation is caused by the fed-back signal being stronger than the input signal, thereby giving the loop infinite gain.

[martin manning]

(e.g. to take the inverting side* if the PI in a bassman, the signal is inverted 180 Deg through the LTP inverting triode, then inverted 180 Deg again through the output stage on that side, and lastly inverted 180 Deg (again) in the OT secondary (ending up 180 deg out of phase overall with the initial phase at the LTP cathode).

* the signal from non-inverting side of the LTP also get inverted 180 Deg in the output tube on its side, but because that output tube is on the opposite side of a PP primary, the signal there is re-reinverted to be in the same phase in the secondary. The outcome w.r.t. feeding back the signal from the secondary is the same - i.e. the secondary is producing a signal 180 Deg out of phase with the signal at the point at which the NFB signal is being fed back into the signal path.

You're 180 degrees out here... The signal from the preamp gets inverted only twice on the inverting side, once in the PI and once by the power tube(s) on that side, so the signal at the secodary has the same phase as the preamp signal. Since the "positive" feedback signal is applied to the non-inverting side of the PI, it becomes negative feedback wrt the preamp signal.

[Ian444]

Has anyone considered what happens when the output tubes start clipping? Would it be true that for the duration of the clipped signal, the feedback would be less than what it would be if the output stage was not clipping? If true, then the total signal applied to the output tubes would increase when the output tubes start to clip, making them clip even harder. Is this the case?

[Jana]

I think there might be some misunderstanding of the terms phase shift and positive feedback and what can happen. The phase shift isn't referring to the known and constant phase shift that occurs from the grid of a tube to its plate. That is a 180 degree shift. No problem there. The other phase shift that can happen is because of capacitors and inductors (OT). I forget all the math and whatnot but the jist of it is that at certain frequencies there is a phase shift that occurs. What happens is that the negative feedback loop, in conjunctive with presence cap, inductance of the OT, and other components, will shift the phase of the signal at certain frequencies. Thus, what is supposed to be negative feedback will start to become positive. It might be at above audio frequencies so you might not hear it but it is oscillating. That's my understanding and experience of it.

[tubeswell]

(e.g. to take the inverting side* if the PI in a bassman, the signal is inverted 180 Deg through the LTP inverting triode, then inverted 180 Deg again through the output stage on that side, and lastly inverted 180 Deg (again) in the OT secondary (ending up 180 deg out of phase overall with the initial phase at the LTP cathode).

* the signal from non-inverting side of the LTP also get inverted 180 Deg in the output tube on its side, but because that output tube is on the opposite side of a PP primary, the signal there is re-reinverted to be in the same phase in the secondary. The outcome w.r.t. feeding back the signal from the secondary is the same - i.e. the secondary is producing a signal 180 Deg out of phase with the signal at the point at which the NFB signal is being fed back into the signal path.

You're 180 degrees out here... The signal from the preamp gets inverted only twice on the inverting side, once in the PI and once by the power tube(s) on that side, so the signal at the secodary has the same phase as the preamp signal. Since the "positive" feedback signal is applied to the non-inverting side of the PI, it becomes negative feedback wrt the preamp signal.

But if that was the case, then in an SE amp (like a 5F2A) the signal at the secondary would be wrong because it would be in phase with the output tube (which would be 360 phase lag behind the pre-amp driver stage's cathode), and so it would be positive feedback. The OT secondary has to be 180 out of phase with the output tube plate in order for the NFB signal to have a 'cancelling' effect on the pre-amp/LTP cathode.

Driver/LTP (inverting) + Output stage (inverting) + OT (inverting) = 180
[180 Degrees] + [180 Degrees) + [180 Degrees)

The OT secondary is 180 deg out of phase with the primary anyhow because the flux flowing around the core produces a back-EMF which opposes the primary VAC and limits the current to a steady magnetising current which lags behind the primary voltage by 90 degrees. Simultaneously the EMF generated across the secondary, is itself 90 deg out of phase with the flux in the core. So this adds up to either 180 deg out of phase, or in phase, depending on which way you hook up the secondary. But you have to hook up the secondary so it is out of phase with the output tube plate, because the output tube plate is in phase with the cathode of the driving/LT stage, and you don't want positive feedback (which is what you'll get if the oT secondary is in phase with the OT plate). That's my understanding after re-reviewing the relevant chapters of Merlin's 1st and 2nd books anyhow. I'm open to discussion or correction of my perception, but at this stage I'm not persuaded to an opposite point of view (no pun intended).