upgrade is back

[RebelMan]

By the way, it's interesting to read what a well-regarded amplifier designer says about how much power is needed to drive typical speakers to "normal" listening levels without distortion when playing music.  For example:

I reviewed the article but found nothing of which that was foreign to me.  It bears pointing out that the author was overly simplistic with some of his comments.  Since I have no desire to discuss circuit theory in this thread I will limit my response to the quote you provided and the inferences you are suggesting.

First of all speakers are not simply resistive loads.  They are complex impedance devices and their thirst for power is driven by variations of frequency and volume over time.  Since power is a product of BOTH voltage and current, clipping an amplifier can occur when EITHER one of these are in short supply.  In general, an amplifier that can output high levels of voltage but low levels of current will be weaker than an amplifier that can output low levels of voltage but high levels of current (proportionately speaking of course).  

For example, if Amplifier A can output 500 watts into an 8 ohm load and 500 watts into a 4 ohm load it will be weaker then Amplifier B that can output 250 watts into an 8 ohm load and 500 watts into a 4 ohm load.  Since impedance varies with time and the input signal, Amplifier B is better equipped (built) to follow the impedance curve of the speaker.   As the impedance of the speaker varies the output of Amplifier B will follow in proportion to the demands placed on it, we call this linear behavior.  Under the same circumstances, Amplifier A is less equipped to follow the impedance curve of the speaker.  In this case as the demand of the speaker varies the output of Amplifier A will still follow but not in proportion to the demands placed on it, we call this non-linear behavior.

Devialet amplifiers are built with stiff voltage rails and therefore are very linear designs.  They are quite capable of driving 2 ohm loads which covers the vast majority of loudspeakers people are likely to own.

For the author to say that that a 500 watt amplifier is necessary to drive a 90dB sensitive speaker (without clipping) and then exclude all the details about the setting from which he makes his conclusion is GROSSLY misleading.

The point about clipping and the surprisingly high power levels required to avoid it is fundamentally not about linearity.  Rather it's due to the difference between test signals (sinusoids) and music signals which typically have a high crest factor.  As the article points out, this can be observed quite easily by a simple experiment - and obviously Mr Sanders has done so - which leads to his conclusion which I quoted.

Anyway, that may be a moot point; I highlighted the article to point out that there are sound, objective reasons why more power can improve audio fidelity due to factors that are not obvious from a naive analysis.  Indeed Mathieu Pernot agrees, stating "infinite power is the ideal".  If you choose to disagree with him, and Mr Sanders, that's your perogative of course.

It's precisely about linearity because his whole premise is about signals clipping and under what circumstances.  He makes many assumptions and leaves the details out.  He also has an agenda to sell equipment.  A more famous Nelson Pass sees it quite differently but he also has an agenda.  Tit for tat.