The Zero Positioning Coding;more processor power for Class D?

[The golden mean]

Today,I read the German Hi-fi magazine Stereoplay(#8 2006), and laid my eyes on an article about some reseach going on at (the) Institut für Theoretiísche Elektrotechnik , in Hannover.
Sorry, I think I´m unable to do a good translation of the article,
but the main thing seems to be, to transform the distortion components of a Class D amp,out of the frequenses that we can hear.This seems to be done , by "tricks" in the phase domain , a lot of math,and consequently with a lot of computer power; " not less than eight (Single) processor IC´s are being used in the test set-up".

Maybe someone more gifted than me in the German language, can contact:
Herr Professor Wolfgang Mathis, at this site:

http://www.tet.uni-hannover.de

If you can invite him or his "Mitarbeitern" (fellows) to write something about it here, I think this will be appreciated.

The process is called ZePoC, and the mag is reporting, that they await to both listen to and measure an amp based on the principle , this fall.

[classd4sure]

Sorry, I think I´m unable to do a good translation of the article,

This is exactly what a Sigma-Delta/noise shaper does. So leads me to believe what you're talking about is something like an SD converter but with the added complexity of 8 digital filters as noise shapers?

Not very interesting if that's the case.

[phase_accurate]

A presentation in German can be found here:

http://getwww.uni-paderborn.de/viva/...itenberger.ppt

The modulation used is the sum of two digitally computed signals whose "payload" is in phase but whose distortion products are out of phase. The distortion products therefore get cancelled upon summation. Basically some related principles are used that are known as the "phase method for SSB". It therefore needs a so-called Hilbert-Transformator, i.e. a thingie that generates a constant 90 phase shift throughout a wide frequency range which makes it a little impractical for some applications. Furthermore it doesn't use NFB but that doesn't necessarily have to stay that way.
It is NOT the same thing as SD modulation.

I have seen that some of these guys wrote a book (actually his dissertation) exclusively on class-d amps, something that is very rare BTW:

http://www.tet.uni-hannover.de/paper...rger_2005.html

Regards

Charles

[classd4sure]

Ok, it's a little bit more interesting now.

So I'm confused, is it cancel the distortion or push it out of band? Why resort to all this complexity if they're not even taking feedback....???? Exactly what distortion is it that they're attempting to cancel?

Thanks

[Jaka Racman]

click modulation

[classd4sure]

Excellent, thanks!

[phase_accurate]

It definitely looks interesting. But before a fully integrated version is available it isn't worth the effort IMO (apart from development work of course).
Have a look at the measurements of the practical implementation and you'll see that it certainly doesn't outperform the better ones of the current class-d amps. And the latter need less effort in terms of component count, computing power, know-how .........

Regards

Charles

[The golden mean]

The article I read is rather short, but a diagram shows the new principle to have virtually zero distortion in the audio band (with a three-tone signal) . And as a comparision, there is a diagram from a "conventional" PWM-amp , as well. No distortion components are seen at the level of -140 dB and below 50 kHz, in the new approach...

And maybe the team has taken the principle further?The block diagram published in the mag, doesn´t seem to correspond exactly with the ones provided in a link above.
It is often said, that added complexety seldom pay of, but there may be exceptions to this rule?

I hope that the mag gets a sample for evaluation. If so , I will report their findings.