class-d behavior at clipping

[raintalk]

What's the behavior of class-d at the onset of clipping all the way up to full rail pegging clipping?

Not just with a sine wave but with full spectrum music? That is the higher frequencies are riding on the lower frequencies. As described at:
http://www.st-andrews.ac.uk/~jcgl/Sc...ing/page3.html

It would seem that at these extremes the pre-filter feedback would be more stable?

I can imagine that these lower powered class-d amplifiers do clip quite a bit in real life use and no one is the wiser up until it hits some extreme.

Just curious.

[poynton]

Do class -D amps pwm "clip" ?

[BWRX]

Of course class d PWM amps clip. As you approach clipping the THD goes up and is easy to detect, very much like a class ab amplifier.

[raintalk]

Except class-d has an output filter, and it only has a couple of output devices. And they are normally just switched on and off, so other than current they aren't stressed.

Maybe the window of low THD to an unusable level is larger with class-d?

Maybe it just doesn't matter. If it sounds bad turn it down.

[Pano]

I can only speak to Tripath and TI chips, but they certainly clip, and they clip hard.

Of course it's not really the output devices that are clipping, as the are simply on or off (idealy). So it must be the comparitor circuit that's clipping - or maybe the input circuit, but I would bet on the comparitor.

The Tripath chips have a small region of "soft clipping" but it's a narrow window. Push beyond that and the crests of a sine wave just get chopped off. It does not sound nice. You can see it happening to musical peaks, too. The clips show up as bright spots at the top of waveform when seen on an o'scope.

Some of the very fast clipping you won't even notice, but push it too far and it sounds harsh. Perhaps a class-d amp could/should be built with some form of peak limit to avoid the clipping.

[BWRX]

Originally posted by raintalk
If it sounds bad turn it down.

That applies to almost all things related to audio

[phase_accurate]

Some higher-order feedback loops can behave worse than just chopping off the tops and bottoms. They can somehow "stick to the rails" and go back to normal by crossing some instability zone which may add some additional non-harmonic components.

Regards

Charles

[Ouroboros]

UCD-style self-oscillating amplifiers exhibit a sharp increase in distortion as the modulation index goes above (approximately) 85%. Well designed ones clip cleanly, although I have had prototypes exhibit some 'raggedness' on the output just before clipping.