David Bridgen said:
I don't see it Nigel.
With a given supply rail the peak output voltage will be, ignoring output device constraints, the same value as the rail, and the peak power calculated from that.
With the same rail, the r.m.s. voltage, from which the power is calculated, will be 0.707 of that value. This is true irrespective of the class of operation.
Powers supply sag shouldn't be a problem in a properly designed amplifier and as far as this discussion is concerned is a red herring.
By the way, what do you mean by "limits before clipping can be reached."
It's vitally important that a digital audio system, either power amplifier or a CD recorder doesn't clip - for example a 16 bit CD recorder would suddenly change from 65535 to zero, as the 16 bit value overflows. With a switching amplifier (class D) it's not quite the same, but if the output stage clips it leaves the relevent transistors turned ON for much longer than the switching time, which could well spell disaster?.
In order to prevent this happening the amplifier includes a 'limiter', if the output is getting too large it will clip the input, keeping the output stage below the clipping point.
400W RMS sinewave into 4 ohms requires 113V p-p, which presumably the amplifier will deliver? - so it's a perfectly accurate 400W RMS amplifier. If you try to drive the amplifier harder, the limiter will kick in, preventing the output exceeding 113V p-p.
If you now drive the amplifier with music, fairly quiet, with loud peaks, and a peak tries to exceed 113V, the limiter will kick in again limiting the output to the same 113V - except now it's limiting on a short peak, rather than a continuous sinewave.
This isn't unique to the Behringer amplifier, it's likely to be common to any similar class D PA amplifier - certainly the others I know about all do the same.