It looks like a debate to me.

I stand by the intent of my original statement:

Amps seem to sound more sensitive to impedance issues to me than the math otherwise suggests.

@erik_squires I was not contesting this, FWIW.

There are far far too many other factors at play including open-loop bandwidth, compensation network, or just to make it simple, transfer function including load to simplify to an amount of feedback to equate to a particular sound.

Your simplification suggests a marginally unstable system as opposed to an over-damped system.
@heaudio123

Correct.


In order to present a nutshell explanation of how feedback is problematic (and also how to use it) the explanation has to be simplified and as much as possible use layman’s terms.


Open-loop bandwidth of course plays directly into gain bandwidth product, and compensation networks inside the feedback network itself and also in the actual signal path are of course important in any proper amplifier design. But the essence of my point is simply this:
If an amplifier has enough feedback, it won’t be bright and harsh, whether tube or solid state. Historically we’ve not seen really such an amplifier- that is why tubes are still around!


IMO most amplifier designers don’t really know how to design a proper feedback loop (as you know its far more than just a simple resistor). Meanwhile the test and measurement folks seem to realize there’s some dirt to sweep under the carpet which is why you rarely see THD analysis with the fundamental much more than 100Hz, with harmonics in the ear’s most sensitive region simply not displayed.

I will argue that we are also more susceptible to damping factor and speaker/cable impedance than we think we are, but I, in a very lage part, agree with Ralphs point, that the distortion profile is probably a large, large factor in likeability.

About that- amplifiers are far and away out in front of speakers in terms of development. One problem here is that no speaker needs more than about 20:1 for a damping ratio (not 'factor' mind you- 'ratio') and many solid state amps have far more than that, resulting in a coloration many audiophiles like called 'tight bass', something that does not seem to exist in the real world.

Roger M RM-9 has 3 feedback settings so you can dial into what your ear brian likes in a 3D world....

Not to put too fine a point on it but it can be seen as obvious that none of the settings would be correct. The RM-9 simply didn't have the gain or bandwidth to support enough feedback to really do the job properly. So any of the settings would have been a compromise.


We used to put similar amounts of feedback in our amps back 30-40 years ago, only to find that it did terrible things to the soundstage and tonality.


You can look at the deliterious effects of feedback on a Bell curve. A small amount of feedback does little damage but has little effect too- no more than about 2-3 dB. 12-20 dB is in the peak of the Danger Zone; so you will experience increasing brightness and harshness from about 4dB up to that point; above 20dB the harshness and brightness heads back to 'normal' but since very few amps have in excess of 35dB essentially 99 44/100s of all amps that run feedback will have a coloration of brightness and will be harsher than real life.

If you've ever wondered why two amps can measure flat on the bench but one is bright and the other isn't, this is why.

Since I was not invited (for good reasons as you will see) to the seminar I'll just make this post-


It should not be a debate. If you have a tube amp that 'measures poorly' yet seems to sound just fine (as has often been seen in the pages of Stereophile) and an amplifier that measures just fine and is really 'neutral' but not particularly musical then you have two amplifiers that sit at the opposite spectrum of the same problem, which is distortion and what to do about it.

In this regard with **all** amplifiers the sound of them is all about how they distort.


In this regard each has chosen a different path. On the one hand, a tube amp that 'measures poorly' is probably not running any feedback to suppress distortion, and one might want to know why not? The answer is that feedback adds distortion of its own, and what it adds is highly audible in the form of higher ordered harmonics (at low levels) and IMD. The human ear is keenly sensitive to both types! OTOH, the ear **isn't sensitive to the lower orders (2nd, 3rd and 4th), which is why that tube amp measured so poorly as it probably has much more of these in exchange for keeping the higher ordered stuff inaudible.

The solid state amp employs feedback as it has to to prevent it latching to the power supply rails (which would cause it to have a lot in common with a rock) and to allow it to be linear. So it inherently is much lower distortion, but now most of that distortion is of the form to which the ear is keenly sensitive, and comes off as brightness and harshness, since the ear converts all forms of distortion into tonality.

Again this should not be a debate! The common problem here is that neither amplifier has enough gain bandwidth product and so is a set of compromises. Tube amps, unless OTL (Output TransformerLess) usually have poor gain bandwidth product. Solid state amps usually do much better in this regard. But neither has enough- and this is what that looks like: without enough gain, you can't apply enough feedback so that the application of feedback allows the amp to correct not just for simple distortion but also the distortion caused by the application of feedback! Did you get that? Its sounds recursive because it is. Feedback can correct for itself if **enough** is applied. That value seems to be a minimum of about 35dB and 40dB is better.


But that means that the amp has to have a lot of gain so that once you blow 35dB or more away you still have enough to work with the preamp and speaker, so the open loop (no feedback) value should be at least 60dB!


Now we come to the other bit of that gain bandwidth product thing: bandwidth. Sure you can make a tube or solid state amp with that much gain, but when you run feedback around it there is an enormous possibility that it will oscillate. The reason is that when there is gain there is also phase shift caused by limited bandwidth- the more gain you add, the more the bandwidth is compromised. The phase shift can thus cause negative feedback to become positive feedback at some high frequency- and then it oscillates. A related idea in amplifier design is that of 'phase margin' which is to say that if by the application of feedback the amp oscillates, it has insufficient phase margin. OTOH phase shift causes the feedback to become positive rather than negative. 


So one way to deal with this is avoid feedback. You avoid brightness and harshness because the lower ordered harmonics mask the higher ordered harmonics. But the amp will measure poorly unless extreme steps are taken to suppress it by other means.


The other way is to stare down the gun barrels and add the feedback- now it 'measures' well but it has a brighter and harsher presentation than real life.

IOW **neither** amp is right! So this shouldn't be a debate. Instead we should be looking at the measurement standards- its obvious that since the ear is insensitive to lower ordered harmonics that they should not have the same value in the measurement as higher orders; IOW the various harmonics should be weighted. But in addition, the 2nd and 3rd in particular can mask the presence of the higher orders, so those harmonics should be in attendance in order to allow for something that sounds like music instead of electronics. But our current measurement regime takes little of that into account. In a nutshell, in this regard one of the most important aspects of human hearing (the fact that the ear senses higher ordered harmonics to gauge sound pressure) is ignored in order for the spec sheets to look 'nice'.


So this should not be a debate so much as a seminar on informing the public what is really going on. Sadly, it will probably just be a debate.