Why do audiophiles shun feedback in amplifiers?

I have found amps with less or no feedback sound loose and diffuse with less dynamics... So why this myth perpetuated by audiophiles and even many manufacturers?
Dracule1 (Reviews | Threads | Answers | This Thread)

You did not hear the amp sound loose, you heard the amp paired with a specific pair of speakers sound loose. With a different pair of speakers you may have heard something different.

Dracule1, feedback is a very good think if you know how to use it. It improves practically everything - reduces distortions (THD, IMD), lowers output impedance*, widens bandwidth. The only problem is that amplifier presents delay to signal. When you feed this delayed signal back to input stages it doesn't subtract properly - especially when signal is changing fast. It causes overshoot after transition since gain for tiny moment was higher (feedback was late = open). It is called TIM (Transient Intermodulation) This overshoot in time domain is equal to extra odd harmonics in frequency domain. Since odd harmonics carry loudness clues we're very sensitive to them and hear it as bright unpleasant sound. In extreme case when amplifier has very deep global feedback fast transitions might even choke output transistors that stay that way for a moment (charge trapped at the junction) resulting in tiny gaps that are inaudible since brain "fills the gaps" but makes us tired. It sounds unlikely but they designed SS amps like that in 70s before TIM was discovered.

Every amplifier has some form of feedback but it is better if it is local (around one stage) in few places than one global going from the back to front. Also sane designer would make amplifier as fast as possible to reduce delay and would limit bandwidth at the input. I would also be very careful with amount of feedback - tiny overshoot might become 10x higher with 20dB deeper feedback. Reducing harmonic distortions, by the feedback, below 0.1% might not be audible but brings risk of TIM. Also reduction in output impedance might not be necessarily a good thing if your speaker is already over-damped. Class AB amplifiers often have a lot of feedback to linearize output transistors and keep gain constant when both transistors are conducting (gm doubling).

Preamps most likely have stages with local feedbacks. Bandwidth is also easier to obtain with practically no gain and no power stage.
------------------------
* - Some time ago I made example of feedback reducing output impedance:
http://forum.audiogon.com/cgi-bin/fr.pl?aamps&1315855364&openflup&21&4

Kijanki explained it pretty well as I understand it.

The argument against negative feedback is based on it being an imperfect process that introduces time delays and artificial harmonics accordingly as explained and that this inherent negative can outweigh advantages of applying negative feedback otherwise.

Audiophiles hate anything that is not perfect so being inherently impossible to implement perfectly, negative feedback is a common target of ire.

You might say that the use of global negative feedback causes the amplifier to violate one of the fundamental rules of human hearing/perception: how we determine how loud a sound is. We do that through analyzing the odd ordered harmonics rather than processing fundamental tones.

So if the amplifier has trace amounts of the 5th, 7th and 9th harmonics added, it will not only sound louder than real music of the same volume, but bright/less relaxed as well.

This is one reason why two amps can measure perfectly flat with the same bandwidth on the bench, but one will sound bright and the other won't.

As Kijanki pointed out, global feedback reduces 'output impedance' (I put the term in quotes because it is a definition that is only used that way in audio!). So amps without feedback will have a higher output impedance. If coupled with a speaker that demands a lower output impedance of the amp, tonal aberrations may result.

So you can see that a more ideal combination might be such an amplifier with speakers designed to work with higher output impedances. Then you get proper tonality coupled with no violation of human hearing rules.

For more information see: http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php