Re: Hmmm...
>The trick is to design an audio amp that cancels out the crappiness of the op-amps in it. (Hint - negative feedback.) Then use decent power FETs for the final stage output.
That's very 1970s thinking, if you don't mind me saying so. The problem with this approach is that the finite bandwidth of the forward part of the loop introduces artifacts which don't show up with a sine wave (you'll get ludicrously low THD numbers) but rather show up as "Transient Intermodulation Distortion" or "why does my Sinclair amp sound like crap?". People didn't tumble to this for years mainly because they didn't have the testgear or circuit analysis tools to show it, they just knew that certain amplifiers sounded great (including most valve amplifiers) while others didn't. (Valve amplifiers have a more linear and much lower gain forward path.)
Its all moot these days, though. Quite apart from modern semiconductors having effectively infinite bandwidth by audio standards most modern amplifiers are class 'D' -- digital, with PWM output. We're well past the point where a commodity processor can handle all the functions of signal collection and conditioning. Its really life coming full circle -- early transistor amplifiers started out life as servo amplifiers, military project designs 'repurposed' at home by enthusiasts, with modern audio amplifiers just being repurposed, and downrated, servo drivers.