Negative Feedback

Alan

As in telephony & audio amplifiers. I found this to be an interesting read (although page 4 is a little technical for me). It's a 1998 article from Stereophile, written by Martin Colloms.

I have no idea (as usual!) of how sound the reasoning is in this article, indeed, a few letters are less than complimentary about Martin Collom's line of thought. However, it served to teach me - at least in a simplified way - what negative feedback is about.

As an aside, I'd be interested in what Colin thinks of the article, as it mentions the single ended class A approach and implies it has merit - I know a few on 'Chews wouldn't argue that.

Brumjam

That's a very interesting article Alan, particularly since my speakers are meant to work best with low or no feedback amplifiers, a topic which I've been trying to get my head around from a technical standpoint.

Alan

It is interesting, in as far as I think I understand it.

Funnily enough, some of the class D stuff is being combined with class A input stages, in order to give class A 'flavour' with more grunt. One interesting and home grown example is Jon, at Reality Audio. He uses a Hypex UCD module with a class A input stage -  this, I think, is more or less what Devialet are doing at the moment, and they are certainly making waves.

I think what these guys are doing is actually positive feedback (I may be wrong), where instead of (180 degree out of phase) gain being channeled back through the same device to cancel out distortion artifacts, the class A device is being used to dominate the class D character. Class D has a relatively benign character, so is easily swamped by a class A input. So in this case, its all about amplifying a characteristic, rather than cancelling one out. I suppose all amps do this internally, from one gain stage to the next.

More akin to a pre/power amp situation really.

RobHolt

The thing with negative feedback is that it has several strands to it's effect and this can complicate assessment of no/low/moderate/high feedback listening.

On the one hand it suppresses distortion which is universally considered as beneficial, but it also changes impedance and that isn't always positive from a subjective POV.

Certainly in power amplifiers, increasing NFB will lower output impedance but not everyone likes the resulting sound. Output impedance serves to modify the frequency response of the loudspeaker system, mainly at the points were the load is reactive. For non technical, most reactive is where we see the greatest rate of change on the impedance plot, so around the deep peaks and nulls.

It also influences electrical damping and this can be critical with many vintage loudspeakers, designed for use on amplifiers with higher output impedance than is the norm today. The effect at LF can be quite substantial with some of these systems.

So changing the NFB is changing several factors at once.

PAC

Interesting subject and article.  Simply put (and technically speaking), low amounts of NF are a good thing, especially with modern loudspeakers and especially where valve amplifiers are concerned.  Zero feedback triode amps for example have next to no damping control over loudspeakers and corresponding speaker response is anything but flat.  There's no excuse for any designer these days to eliminate feedback as it makes for poor fidelity with many modern loudspeakers. 

A little feedback helps the amp to remain stable and control the load.  result is better fidelity, control and flatter frequency response.  Too much is almost as bad as too little.  Many amps use no more than a small handful of dB of feedback.  My own 869 EAR amp uses a little feedback to great effect. As mentioned above, some vintage amps were designed for higher impedance outputs and speaker loads could be 16 or even 32 Ohms.