I think you guys are making this overly complicated. A sharp rise in distortion suggest that the amp is "clipping" whether if it's from voltage rail saturation or not enough current. It will look the same, clipping don't have to be an audibility metric, it doesn't have to mean huge distortion number, it don't even have to be visible on a oscilloscope. It's just a rapid rise in distortion, I think that the problem here is that Amir's wants to define the metric used and his rationale for picking a "knee", which indeed suggest clipping occurs. I assume that what he wanted to express is "the beginning of steepest slope in the power delivery graph means max power is obtained" but it have to be just assumption because he doesn't care to explain and answer the questions. If it's well defined, I don't mind that, like in this example there are 3 knees, Amir chose the steepest rise starting point as max power. OK, but it started rising before that. Softer clipping from below 200W at 4 ohms, harder clipping past 350 or so, they are all technically into "clipping" but defining where it starts to matter will always be arbitrary, or basing this on "audibility" will also always be just subject to any interpretation. We do need a number, I liked
@restorer-john Idea on defining a operational spec at 250 mW and define this as when it can't match this THD spec max power is attained, I am also perfectly ok with. 0.01, 0.1, 1 numbers. As long as it's defined and it's clear. Let's not bring audibility into this, it's a power spec, it should define what power an amp can provide and you have to put a treshold somewhere and that treshold have to be clear. It's a power specification, not a fidelity specification. Me, for example. I don't hear the level distortion on this scope image. I don't know what THD percent it is. It's from a IEM wireless monitoring system. I don't know the level of distortion and don't hear it, but it's still a clipped signal. I have an ancient System one AP on a shelve but can't be bother to plug it in. It don't matter the number, it's clipped. It don't matter, trying to tie max power to an absolute "clipping point" will not work because as restorer john said, not everything will have the same slope, and it's ok to have peaks in that non linear zone, It appear that Amir don't want "any" clipping, but it will always be hard to define. The distinction should not be differentiate distortion to clipping, it is all about distortion,but one is a fidelity spec, one is a power behaviour spec, of couse the 2 are not to be confused clipping occurs when distortion rise, we are one derivative away.
View attachment 247780View attachment 247786