Feedback Question/Clarification

[fizzard]

I have seen a few posts in various threads that criticize various amplifiers because of their high open loop gain. The reason for this criticism is the high open loop gain requires a large amount of negative feedback, and negative feedback is bad.

According to my understanding the closed loop gain is determined by the input and feedback impedance values. For sufficiently large values of open loop gain, the closed loop gain is independent of the open loop gain. So a further increase in open loop gain won't change the amount of negative feedback required.

Can I get some clarification please?

[Cauhtemoc]

I have seen a few posts in various threads that criticize various amplifiers because of their high open loop gain. The reason for this criticism is the high open loop gain requires a large amount of negative feedback, and negative feedback is bad.

Negative feedback is not bad, it is a fundamental part of transistor amplifier design. Read this:

http://sound.westhost.com/articles/distortion+fb.htm

According to my understanding the closed loop gain is determined by the input and feedback impedance values. For sufficiently large values of open loop gain, the closed loop gain is independent of the open loop gain. So a further increase in open loop gain won't change the amount of negative feedback required.

Feedback factor = Open loop gain - Closed loop gain

For example, an amplifier with an open loop gain of 60 dB and a closed loop gain of 20 dB will have a feedback factor of 40 dB. An increase in open loop gain by 10 dB will increase the feedback factor by 10 dB.

[jcx]

Rod Elliot gets some points right, but goes one claim too far:
http://www.diyaudio.com/forums/showt...ht=#post920583

negative feedback does "create" new distortion products in the course of greatly reducing the total error

presumably the original question relates to Otala's "phase intermodulation distortion"- Otalta showed that the phase shift of integrating feedback can change the phase relations of the intermodulation distortion products from some static nonlinearities to resemble "phase modulation" spectrum vs"amplitude modulation" spectra that are seen with constant feedback over frequecy

the topic has been discussed several times - I like simulations to illustrate points:
http://www.diyaudio.com/forums/showt...ht=#post489927

[P.Lacombe]

In few words : No, negative feedback is not bad.

Excess of negative feedback is bad ! Because its usually causes instability, specially with complex loads, and can give a dirty sound.

Nelson Pass said that negative feedback is like a credit card... handle with care !

[john curl]

No feedback is the best feedback!javascript:smilie('')

[jcx]

Originally posted by P.Lacombe
In few words : No, negative feedback is not bad.

Excess of negative feedback is bad ! Because its usually causes instability, specially with complex loads, and can give a dirty sound.

Nelson Pass said that negative feedback is like a credit card... handle with care !

again I believe that opinion, while common, is easily demonstrated to be wrong

negative feedback does cause the increase of higher harmonics with low order distortion amplifying circuits, curious cancellations of particular distortion harmonics are possible with some device' distortion characteriscs and a specific (to that device, bias condition, load) small excess loop gain

some "low feedback" designers imply that they tune these cancellations to minimize the harmonic distortion component they claim is particularly dissonant such as the 7th

It is hard to believe in the practical possibility to tune the loop gain to the device' distortion characteristic to get even 40 dB reduction in one or two harmonics

It is not so hard today to get 60 dB excess loop gain over the audio frequency range, giving a greater reduction than the best cancellation for all distortion components

A high gain, low bandwidth negative feedback amplifier will have higher gains at lower frequencies - since music is commonly shown to have ~ 3KHz power bandwidth most musical signals in a "low bandwidth" amplifier will see > 20 dB more loop gain and corresponding reduction in distortion components at those frequencies compared to a low gain, high bandwidth amplifier with flat loop gain to 20KHz

If the low gain, high bandwidth amplifier has restricted excess loop gain to 20-30 dB then the high gain amplifier’s errors at all audio frequencies will be hundreds to thousands of times less – you wouldn’t think it would be hard for the low feedback crowd to show the PIM/PMD or other supposed high gain “dynamic” errors sound hundreds of times worse than the large distortion components in their amps

[implicit in the above is the assumption that the power output stage distortion is dominant and you compare high/low gain amps with the same output stage performance]

[lumanauw]

[implicit in the above is the assumption that the power output stage distortion is dominant and you compare high/low gain amps with the same output stage performance]

I agree with JCX. The bad rap of negative feedback often comes from an amp with global feedback and bad output stage. The bad output stage wrapped in global feedback contributes more bad things than the feedback itself. (when a good output stage is used in the same global feedback amp).

[GRollins]

High feedback tends to create a false "hi-fi spectacular" sound quality. Some people describe it as "detail" but real guitars don't have that much finger sound on the strings and human voices don't have that sort of sheen to them. For me the deal killer is that it causes deterioration of the image. Real images don't sound that way.
I've built any number of circuits over the years and found that the less feedback I used the better they sounded.
People who claim otherwise rarely seem to get out and hear live, unamplified music so they lack a referent, but they sure do yap loudly about how perfect negative feedback makes things. I used to...until I started listening to classical and jazz. That changed my perspective, which previously had been based solely on rock. Once I started noticing that things sounded wrong at home on my stereo I started asking questions. The answers--and sometimes the lack of same--led me to experiment with differing levels of feedback.
Note that if negative feedback were truly the answer, we'd have reached audio nirvana in the late '70s when designers were routinely using grotesquely high levels of feedback. That the concept was an abject failure is demonstrated by the relative paucity of classic solid state designs from that era. And note that the few that you can name were not on the bleeding edge of the feedback wars. (There's always some joker who comes in and lists every bloody thing that was in production at the time...as though mere production constitutes a "classic." Jeez...)
You might also want to note that damping factor (which is always second or third on the list of so-called benefits of negative feedback) is vastly overrated. Why? Because tube amps, which are notoriously possessed of low damping factors, can sometimes outpace solid state amps in the bass.

Grey

[fizzard]

Originally posted by Cauhtemoc
Feedback factor = Open loop gain - Closed loop gain

For example, an amplifier with an open loop gain of 60 dB and a closed loop gain of 20 dB will have a feedback factor of 40 dB. An increase in open loop gain by 10 dB will increase the feedback factor by 10 dB.

That pretty much answers my question. I was scratching my head about how reducing the open loop gain would let you reduce feedback.

[Mooly]

GRollins, what can I say! I totally agree with all your observations and have come to similar conclusions.The majority of designs are all based on the same circuit topology, and while some have very impressive "on paper" specifications they are uninvolving to listen to.With my last design I opted for a different approach, single ended input stage(distortion mainly 2nd harmonic & rises at 6db/ octave compared to 18db/octave for long tailed pair), low open loop gain, mosfet output using lateral fets, and a low damping factor- 0.22ohm.A servo allows for d.c coupling.A simple preamp, OPA604, and discrete series/shunt FET input switching, which works better than any mechanical switch and which also allows for full remote control.This is an amp where 95% of recordings sound wonderful, not just a few "demo" discs.It has opened my ears to some amazing music, string quartets, solo violin, the list goes on.It can only be described as compulsive listening.
Regards Karl