Amplifier Feedback Good or Bad?

[gene]

Feedback is an ubiquitous and unavoidable basic technique that keeps the world running. It is only controversial in consumer audio, and then only specifically in power amplifiers. Some audiophiles believe too much negative feedback will diminish sound quality in amplifiers while others believe no feedback is even better. Is this an audiophile myth that needs debunking or is there merit to this claim? I worked with Bruno Putzeys to better answer this question. This article is a summary of that work with a very long Q&A session to come as well as a Youtube livestream.

Read: Is Too Much Negative Feedback Bad in Amplifiers?

About Bruno Putzeys
Bruno Putzeys (b. 1973). Trained as an electrical engineer at the Belgian National Radio and Film Academy. While his sights were always firmly fastened on sigma-delta AD/DA conversion and class D amplfication, his attic experiments still involved a lot of valves, vinyl and big heatsinks. He learned the ropes of R&D during a 10-year stint at Philips. His time at Hypex saw a wide adoption of his amplifier circuits into various high-end audio products while work at Grimm Audio turned into a first foray into active loudspeakers. He currently divides his time between Kii Audio, a manufacturer of complete active speaker systems and Purifi, an R&D outfit specializing in amplifiers and loudspeaker drive units.

 

[Verdinut]

Well, for some designs it is good but too much is too much, right?

 

[M Code]

IMHO...
When an amplifier has excessive negative feedback it become less stable under dynamic program conditions when pushed hard....

Just my $0.02...

 

[gene]

IMHO...
When an amplifier has excessive negative feedback it become less stable under dynamic program conditions when pushed hard....

Just my $0.02...

Myth

 

[jeffca]

Not being an amp designer or even an electrical engineer, my expertise about this is next to nil, but, after reading a lot about amps and their designs over the years, I learned one thing: there are very few universal rules when it comes to amp designs... especially today.

Amplifier feedback done wrong, like anything done poorly, sucks. In audio and physics, there is no free lunch. But, then again, what happens when you discover something new?

Negative feedback was overdone by amps in the late 70's just to get low THD specs. That had some severe drawbacks in those types of amps. Then Bruno Putzeys came up with his stuff and found a way to use a ton of negative feedback in his class D amps to fantastic effect.

THX and John Siau of Benchmark came up with with the achromatic (AAA) AHB-2 amp design that uses parts of class A, B and D design for what seems to be a new topology. It uses a feedforward correction scheme and it's one of, if not the, best amplifiers ever. It also uses a switching power supply which, common wisdom dictates, is supposed to be for cheap, mediocre equipment. The PSU, though, is so good that I was told by their head of sales that hooking the amp up directly to the wall is suggested and that, in the off chance of a catastrophic voltage spike such as lightning, the amp should be fine.

The best thing about all of what I mentioned is that it's been independently verified (not the AHB-2 lightning strike, sorry). This isn't supposition, hype or snake oil.

So, my thought is this: I don't care about the technology as long as it's proven to work.

By the way, love the photo of the old Marantz receiver at the top of the article.

 

[Gmoney]

Didn’t Dr. Harman Publish an article about negative feedback? Phase lock? Don’t Quote me on that Gene. I’m just a little peon who enjoys reading your post. Gene, I’m not answering your Survey till you make me. cause all 3 look like Trick questions no matter which one I pick I‘m sure i’ll get it wrong.

 

[Gmoney]

I picked #1. seems I read somewhere that less is best. Come on guys Genes giving us a free Seminar on amp Theory and application
For those of us that would be Me and everyone else who needs to get their learn on!

 

[M Code]

Myth

Says who?

Just my $0.02. .

 

[PENG]

Says who?

Just my $0.02. .

Imo if such a claim was made as sort of a general statement then I would agree with Gene that its myth, that a lot of readers might have taken as fact by the power of hearsay.

 

[M Code]

My comments are based upon my personal experience, we have worked with some of the leading audio amplifier design engineers and managed projects for certain well-known brands for amplifiers up to 500W per channel. Note that this is for Class A-B analog amplifiers, the respective engineers design the amplifier to have great sonic performance and minimum distortion. First they design the amplifier circuit without feedback, then apply minimum amount of feedback to meet the brands' marketing distortion specs. Note that many of these products have been highly reviewed, received multiple credits for outstanding sonics and achieved major sales objectives....
So yes I may be opinionated but it works for me and satisfies my clients....

Just my $0.02...

 

[gene]

My comments are based upon my personal experience, we have worked with some of the leading audio amplifier design engineers and managed projects for certain well-known brands for amplifiers up to 500W per channel. Note that this is for Class A-B analog amplifiers, the respective engineers design the amplifier to have great sonic performance and minimum distortion. First they design the amplifier circuit without feedback, then apply minimum amount of feedback to meet the brands' marketing distortion specs. Note that many of these products have been highly reviewed, received multiple credits for outstanding sonics and achieved major sales objectives....
So yes I may be opinionated but it works for me and satisfies my clients....

Just my $0.02...

This summary article states why feedback gets a bad rap. Most early amp designers didn't really know how to do feedback properly and those amps had a lot of LF feedback but it was not effective for full bandwidth. We have a much larger editorial that goes over this coming soon.

 

[Bruno P.]

When an amplifier has excessive negative feedback it become less stable under dynamic program conditions when pushed hard....

Stability at and after overload is absolutely one of the more realistic worries. The problem is called "integrator windup" in control theory. It's probably the toughest part of doing a feedback control system, and it's certainly the most common reason why people claim they "choose not to" use lots of feedback: because they couldn't make "lots" work at all. Of course it's only a choice if one actually has both options...

By sheer luck, the classical arrangement of first putting local feedback around a power stage and then closing a first order global loop around that neatly sidesteps the problem since all stages in the circuit clip at the same time and come out of clip at the same time. So for that category of amplifiers (which is almost all respected class A/B workhorses) the problem is no longer there.

There are also ways of comprehensively solving integrator windup in fully global loops. I patented one that I found particularly suited for audio but quite a few others can be found in academic literature. Once you have integrator windup prevention down, you can make an amp with a very high order loop clip as cleanly as a first-order one. In fact you can get rid of any detectable side effects. Meaning that a thing like "excessive feedback" no longer exists. After all, it was only "excessive" while it kept the thing from working at all.

When researching the question "why does feedback have a bad reputation" I basically kept bumping into the same pattern: a technical challenge presented itself and instead of taking it on, people just turned back and started professing their dislike of feedback instead. The article I prepared for Gene goes into some of those, but I left integrator windup out because it was getting rather long anyway. So thanks M Code for bringing it up so that I could mention it here at least.

 

[Matthew J Poes]

Says who?

Just my $0.02. .

Bruno Putzeys, leading amplifier designer and expert in feedback circuits. Designer of some of the best performing amplifier circuits in existence.

 

[M Code]

BRUNO...
Many thanks for the info...
I am a major fan of Hypex amplifiers being a leader in Class D solutions...To my ears they do sound good..
Are there any audio AVRs available today that utilize your feedback design approach using Class A-B amplifiers?

Just my $0.02...

 

[Gmoney]

This summary article states why feedback gets a bad rap. Most early amp designers didn't really know how to do feedback properly and those amps had a lot of LF feedback but it was not effective for full bandwidth. We have a much larger editorial that goes over this coming soon.

Gene, can’t wait for your editorial to be posted up! I gotta feeling lots of AH members will enjoy the read I will for sure. Not to mention a lot will be schooled up by your editorial.

 

[PENG]

BRUNO...
Many thanks for the info...
I am a major fan of Hypex amplifiers being a leader in Class D solutions...To my ears they do sound good..
Are there any audio AVRs available today that utilize your feedback design approach using Class A-B amplifiers?

Just my $0.02...

I would be interested to know that too, but regardless, I am not worry about the class AB amps designed (with "lots" of negative feedback)/build by the usual suspects such as D+M, Yamaha, Outlaw, Parasound, Bryston, Anthem, ATI etc etc etc. . And I suspect even Passlab's class AB amps would no longer be deprived of the same lol.. May be I should email Mr. Pass.. I am sure you know why I mention him right?

 

[Pogre]

What a fascinating topic. I chose option 3 for the poll.

 

[Gmoney]

What a fascinating topic. I chose option 3 for the poll.

playing it safe uh Pogre? I picked #1 because I read that low negative feed back was good, better or great. You should know by now Gene aka Master Yoda loves to trip us up with loaded or Trick questions just to test knowledge of his Forum members. I know first hand cause Gene, has/likes to Pimp out some of us on his YouTube channel. Anyone remember his Article on his YouTube on where Gene goes “Freedom”. from Braveheart? I’d Volunteer to be a test subject just to be noticed on his YouTube channel, would a Honor in my book.

 

[highfigh]

I want straight wire with gain.

What- someone had to say it.....

 

[PENG]

I reluctantly voted 2 because 1 is clearly false. If no feedback is used you are going to have distortions that are likely going to exceed the threshold of audibility, period.. My only issue with 2 is, the words "best sounding" always bother me, as it is subjective by nature. If those two words are deleted, I would have voted happily instead reluctantly.