Why there can never be too much negative feedback - and how the myths around it arose.

[antcollinet]

While looking around for info relating to a post here, I found the following articles (By Bruno Putzeys of Hypex fame) on the EDN site.

Negative feedback in audio amplifiers: Why there is no such thing as too much​

https://www.edn.com/negative-feedback-in-audio-amplifiers-why-there-is-no-such-thing-as-too-much/

A detailed and technical discussion of how negative feedback works and how important it is for audio fidelity (when properly implemented)


Part 2

https://www.edn.com/negative-feedback-in-audio-amplifiers-why-there-is-no-such-thing-as-too-much-part-2/

Where he covers the audiophile controversy surrounding NFB, and outlines his thinking about how it arose. Particularly useful to counter some of the NFB misinformation floating around out there.


I found them incredibly useful as a retired (very much non audio) electronic engineer - though I'd need to spend more time to refresh my knowledge of the control theory maths to fully get to grips with it all.

If, like me, the detailed maths side needs more work to follow, there are plenty of words to explain what is going on - so you can still follow the reasoning. Though if this is the case I'd strongly advise against anyone of the "NFB is evil" mindset attempting to counter the information with unfounded verbiage.


(The articles are not new - it is possible they've been linked here before. Apologies if so)

 

[restorer-john]

The articles are not new - it is possible they've been linked here before. Apologies if so

Pretty sure they've been linked more than a few times...

There's some other self serving 'article'- the 'f-word' or words to that effect. (f for feedback)

 

[Newman]

Is 100% too much?

 

[restorer-john]

Is 100% too much?

Nah, unity is so sweet.

 

[charleski]

All this blather about NFB being bad dates back to the '70s, when it was discovered that opamps of the time with limited open-loop bandwidth would behave badly under too much global feedback: the old transient intermodulation bogeyman. It was part-and-parcel of the 'Japanese amps sound bad because they have low distortion' myth.

http://hifisonix.com/wordpress/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf

One moment, please...

The solution was rapidly found and has been a standard part of amplifier design ever since. But that hasn't stopped audiophiles banging on about it and there are plenty of unscrupulous manufacturers happy to cater to their magical thinking.

For one perspective of the history involved you can read John Curl's article here:

Wayback Machine

but be prepared to roll your eyes at the interpersonal bitching, which is nothing new when it comes to audio engineering.

 

[restorer-john]

It was part-and-parcel of the 'Japanese amps sound bad because they have low distortion' myth.

Absolute myth, I agree. It's a pity certain Class D designers decided to have a go at engineers from the past.

 

[antcollinet]

All this blather about NFB being bad dates back to the '70s, when it was discovered that opamps of the time with limited open-loop bandwidth would behave badly under too much global feedback: the old transient intermodulation bogeyman. It was part-and-parcel of the 'Japanese amps sound bad because they have low distortion' myth.

http://hifisonix.com/wordpress/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf

One moment, please...

The solution was rapidly found and has been a standard part of amplifier design ever since. But that hasn't stopped audiophiles banging on about it and there are plenty of unscrupulous manufacturers happy to cater to their magical thinking.

For one perspective of the history involved you can read John Curl's article here:

Wayback Machine

but be prepared to roll your eyes at the interpersonal bitching, which is nothing new when it comes to audio engineering.

Thanks for those links - I'll have read later.

 

[fpitas]

but be prepared to roll your eyes at the interpersonal bitching, which is nothing new when it comes to audio engineering.

The egos are often large and tender thanks for the links.

 

[fpitas]

All this blather about NFB being bad dates back to the '70s, when it was discovered that opamps of the time with limited open-loop bandwidth would behave badly under too much global feedback: the old transient intermodulation bogeyman. It was part-and-parcel of the 'Japanese amps sound bad because they have low distortion' myth.

http://hifisonix.com/wordpress/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf

One moment, please...

The solution was rapidly found and has been a standard part of amplifier design ever since. But that hasn't stopped audiophiles banging on about it and there are plenty of unscrupulous manufacturers happy to cater to their magical thinking.

For one perspective of the history involved you can read John Curl's article here:

Wayback Machine

but be prepared to roll your eyes at the interpersonal bitching, which is nothing new when it comes to audio engineering.

It's especially convenient for the tubers, since tube amps typically have very modest amounts of NF (or none, these days )

 

[charleski]

A lot of the early problems with NFB were linked to a desire to save money by using opamps, as the LM741 came out in 1968 and generated a lot of interest. Unfortunately the 741 is awful for audio - noisy with low bandwidth and slew rate. The NE5532 launched ten years later with ten times the bandwidth and low noise and remains a perfectly good choice to this day. It also meant that all the early issues with large amounts of NFB just went away.

 

[Tom C]

I can remember being taught in tech school (1979-1981) that NFB reduces distortion, but there’s a limit to how much you can use. That if you use too much, it exaggerates sibilants. Spoken “s” sounds too essy.

 

[antcollinet]

I can remember being taught in tech school (1979-1981) that NFB reduces distortion, but there’s a limit to how much you can use. That if you use too much, it exaggerates sibilants. Spoken “s” sounds too essy.

If you look at my second link above this might be explained. Negative feedback up to about 15-20dB creates 3rd and higher harmonic distortion. This could well lead to sibilants. From 20dB onwards, all distortion reduces. So "low to moderate" negative feedback can create problems, which are eliminated by higher levels.

 

[fpitas]

I can remember being taught in tech school (1979-1981) that NFB reduces distortion, but there’s a limit to how much you can use. That if you use too much, it exaggerates sibilants. Spoken “s” sounds too essy.

That's...odd.

 

[Tom C]

Well, I was a tech taught by techs. No engineers involved.