Mueta amp, DIY questions

[jwagnerhki]

Hi,

I'm about to design a 1kW peak @ 8ohm hifi subwoofer/general power amp. I had previosly built a ~100W classic triangle wave + comparator setup which was ok but had way too much audible hiss.

Anyway, after looking through the huge bunch of all-digital PCM=>PWM controllers (Texas Instruments etc) with no feedback and quite mysterious THD+N specs, I found Tripath & Pulsus and others, and finally the Mueta site.

I'm curious if anyone here has tried the Mueta approach? A self resonant hysteresis controller setup with voltage and current feedback from the load:

http://www.mueta.com/AES.pdf

The approach looks very interesting indeed, esp. with the relaxed requirements on timing and mosfet switching accuracy and on the filter components and all...

But - does it work or the output sound as good as advertised? I.e. has someone already built something like it? Like, anyone seen reference designs? Or heard other comments on the setup?

About the mueta amp design: what sets the lower bound for the switching frequency? The highest bound seems to be set by the circuit delays. But for the lowest bound, when the output voltage has reached the input reference i.e. zero voltage error - the switching freq then falls very low. But how low exactly? How to prevent it from falling too close to the audio range? Or do I have a fundamental misconception here...

many thanks,
cheers,

- Jan

[IVX]

Hi Jan,
main point of the mueta approach it's a additional current feedback loop from filter cap. This loop works to correcting distortion at high frequency of sound range, and low active for subwoofer range. Without current feedback mueta amp seems ordinary hysteresis controller with added linear voltage feedback loop probably. I'm did exactly so in my amploiD design. see http://www.diyaudio.com/forums/showt...85#post227985.
Best regards.

[Jaka Racman]

Hi,

I did some simulation regarding Mueta amplifier and i have mixed feelings. The man feature of this amplifier is not high frequency feedback but filter capacitor current feedforward. While capacitor current feedforward works OK and is stable, the amplifier is rather unstable when one implements global feedback.

You can get idea about topology from page 41 of this lecture .http://www.aes.org/sections/la/Meeti...0D-2-25-03.pdf . Actually frequency is determined by current ripple hysteresis through filter capacitor and it does not vary more than in any other self oscillating class D design.

Anotrher links you might want to consider is US patent no. 6 552 606 and data shet of forthcoming mueta chip http://www.mueta.com/App_sheet_DL.pdf .

As far as I am concerned, there are some differences between patent claims and AES paper and the actual implementation in the data sheet. It seems as the authors are still learning. I still think that self oscillating design proposed by Subwo1 holds greater potential.

Best regards, Jaka Racman

[IVX]

Hello Jaka Racman
You are right, regarding mueta infos, some differences is really exist, but the author is dead if he any more does not learning. . However four opamps at signal path + mueta chip it's not so good for sound IMHO. Why you still think that self oscillating design proposed by Subwo1 holds greater potential? Thx for CLASS D-2-25-03.pdf it's a really nice stuff.
Best regards.