Are Class D Amps scaring the hell out of the BIG Boys ?

[DaveVoorhis]

With pure digital sources, the analog sampling stage can be eliminated and replaced with computational conversion of the source data stream directly to the Class D pulse-width modulated (PWM) output drive. The signal path is otherwise purely digital and thus distortion-free; the only potential source of analog distortion is the single, simple output stage.

I am not familiar with anyone actually being able to implement this- my understanding is that PWM is mathematically intractable in such applications. Do you have any citations?

A quick search brings up this http://www.aes.org/e-lib/browse.cfm?elib=5601 from 1991 (emphasis is mine):

Previous work has shown that oversampled noise shaping with pulse-width modulation gives a practical DAC with the potential for driving a class-D output stage to give a truly all-digital power amplifier. In this paper, we examine the effects of a new additional processing step which converts the uniformly sampled PWM into naturally sampled PWM to give essentially distortion-free performance for all signal levels and frequencies. Results from a simulation and from a hardware demonstrator are compared with those of the earlier scheme.
​

Here's a direct digital amp: https://hifimediy.com/ddx320-digital-amplifier

PCM to PWM conversion: https://pdfs.semanticscholar.org/7e00/0a6696965721a0d6d432d70fa363b9129131.pdf

A simple isomorphism between (say) PCM and PWM may not exist, but that doesn't preclude algorithmic or heuristic solutions. Admittedly, the way I wrote point 3 in my post suggested it was a completely solved problem; I should have added or relocated a "potentially" -- otherwise amply peppered elsewhere in my post -- to make that clearer. It is an R&D work in progress, like much of Class D.

 

[wyn palmer]

A quick search brings up this http://www.aes.org/e-lib/browse.cfm?elib=5601 from 1991 (emphasis is mine):

Previous work has shown that oversampled noise shaping with pulse-width modulation gives a practical DAC with the potential for driving a class-D output stage to give a truly all-digital power amplifier. In this paper, we examine the effects of a new additional processing step which converts the uniformly sampled PWM into naturally sampled PWM to give essentially distortion-free performance for all signal levels and frequencies. Results from a simulation and from a hardware demonstrator are compared with those of the earlier scheme.
​

Here's a direct digital amp: https://hifimediy.com/ddx320-digital-amplifier

PCM to PWM conversion: https://pdfs.semanticscholar.org/7e00/0a6696965721a0d6d432d70fa363b9129131.pdf

A simple isomorphism between (say) PCM and PWM may not exist, but that doesn't preclude algorithmic or heuristic solutions. Admittedly, the way I wrote point 3 in my post suggested it was a completely solved problem; I should have added or relocated a "potentially" -- otherwise amply peppered elsewhere in my post -- to make that clearer. It is an R&D work in progress, like much of Class D.

Thanks Dave.
The 1991 paper I was aware of at the time. I remember not being impressed, but I cannot remember why.
I do not have free access to AES papers any more, and I note that although the work is almost 30 years old there has not been any subsequent direct work that I can tell and no significant citations that I can find, even using Researchgate.

Clearly, it is potentially possible for an amp to be constructed essentially as a high power output DAC, however I believe the issue has always been that of managing the feedback and there have been a number of papers dedicated to this over the years- certainly the IEEE lists many, but as far as I'm aware achieving performance comparable to say, the Ncore amps has been elusive. The process is certainly not distortion free and much effort has been dedicated to reducing the distortion.
I'm aware of PCM to PWM and PWM to PCM conversion. I regularly convert formats in digital audio mixing and processing. There is some loss of fidelity at each step, alas.
I can't , as yet, find much info on the DDX320, certainly not a complete set of specs. For all I know it might even be running the output stage open loop...

 

[JoeESP9]

British electrical is terrible, and only surpassed in crappyness by Italian cars.

Sell before 50K. Does not matter if it's a Fiat or a Ferrari.




Edit: edits are en vogue in this thread.

I had an Alfa Romeo Giulia Spider from 71 to 73. If it even looked like rain it wouldn't start. A Mallory Voltmaster spark coil solved the problem.

 

[wyn palmer]

I had an Alfa Romeo Giulia Spider from 71 to 73. If it even looked like rain it wouldn't start. A Mallory Voltmaster spark coil solved the problem.

I sold my Ferrari 348 at 14k miles because it had all sorts of problems. I swore never to buy another Italian car. I also had a Lotus that had all kinds of electrical problems, but as I built it from a kit I really only had myself to blame. I've still never bought another British car.

 

[DaveVoorhis]

Thanks Dave.
The 1991 paper I was aware of at the time. I remember not being impressed, but I cannot remember why.
I do not have free access to AES papers any more, and I note that although the work is almost 30 years old there has not been any subsequent direct work that I can tell and no significant citations that I can find, even using Researchgate.

Clearly, it is potentially possible for an amp to be constructed essentially as a high power output DAC, however I believe the issue has always been that of managing the feedback and there have been a number of papers dedicated to this over the years- certainly the IEEE lists many, but as far as I'm aware achieving performance comparable to say, the Ncore amps has been elusive. The process is certainly not distortion free and much effort has been dedicated to reducing the distortion.
I'm aware of PCM to PWM and PWM to PCM conversion. I regularly convert formats in digital audio mixing and processing. There is some loss of fidelity at each step, alas.
I can't , as yet, find much info on the DDX320, certainly not a complete set of specs. For all I know it might even be running the output stage open loop...

I imagine much of the current work is probably industrial rather than academic, which keeps the academic publication rate pretty lean. It might not be considered academically interesting at the moment, either.

I did a bit more googling and found this NAD advert- er, whitepaper on their "DAC that amplifies": https://nadelectronics.com/wp-content/uploads/2017/03/DirectDigital_WhitePaper.pdf

And this: https://www.theabsolutesound.com/articles/nuforce-dda-100-integrated-amplifier-1/

 

[Frank Sol]

After Nine Years...

Ok... A little tongue in cheek & that I'm presently using a Class D amp in my main system

However

Have you noticed how the top end mags are warming-up to the class D amps? Writing good reviews and have stop complaining about being able to 'hear' switching or not switching of class D amps

It has to be a shock to the guys making 100lb+ amps = $10K+ when a 10lb $500 amp is in the same ballpark SQ wise

If you include chip amps I was close to being right

 

[hifitommy]

the big boys are already embracing them. i would love a Bel Canto, but wouldn't cry if someone gave me a Spectron which was the progeny of John Ulrick formerly of Infinity. another biggie making them is Rowland Research.

 

[AudioGMan]

For mids and highs it’s always class A/B for me. Class D is only for my car subs.

 

[Jason Lafler]

I enjoy my little collection of mini amplifiers. Visitors get intrigued. " That's your stereo?"