Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[alfs]

first: happy new year !

from your experience, whats soundwise the difference, running a d-amp at about 200, 300 or 500 khz ?

and whats the effect of a higher-order loop or noise shaper to the sound?

regards
alfred

[bruno putzeys]

Open loop distortion from a class D amp is directly proportional to  the switching frequency and to the timing error (e.g. dead time related). Loop bandwidth also scales linearly with the switching frequency.
The surprising result of this is that a class D amp with a first order loop will have the same distortion regardless of the switching frequency, because loop gain improves by exactly the same factor as the open loop distortion.
Starting with a second order loop, loop gain wins the contest, and increasing the switching frequency will produce a net benefit in distortion.
Control loops taking feedback past the output filter are higher order by definition, because the output filter adds at least two poles. However, because these are complex, the improvement in loop gain is modest (10 to 20dB). That said, even loops with one real and two complex poles are very useful and are successfully used in commercial products.
A downside to higher order loops is that they have the annoying property of making the modulation non-linear. Amplifiers with normally designed higher order loops will exhibit a marked increase in distortion at high modulation indexes (see AES preprint 6693 for a mathematical analysis and a practical solution and preprint 6694 for an intuitive analysis and a mathematically perfect but possibly less practical solution).

In the end there is no clear-cut correlation between "loop order" and "sound". Much more important is actual behaviour of the closed loop. Choosing the right feedback variable (speaker voltage) has a more profound impact on the sound than the order of the loop.

[Sahib]

Hi Bruno,

I have been reading some great books on audio amplifier design by well known designer-writers in the audio circle. While one strongly prefers class A, the other prefers class B as the best design alternative in terms of sound re-production in power amps. Although the books are over 6 years old and class D has advanced in this time period, what is your preference and why? I realise that even this may qualify as subjective but I hope you do not mind.

Regards,

Cemal

[bruno putzeys]

I prefer to be agnostic about the kind of circuit one uses to arrive at a certain sonic outcome. People who try to prove that one technology is inherently better (bar power efficiency) than another usually do so because they have to for commercial reasons, or because of ideology (=insufficient understanding to look at the problem from different angles). I don't think this is the case with the books you're referring to.

What they do mean is that attaining a certain performance level is not at all equally easy or even practical with each thinkable technology. Building a good quality class A amp is rather simpler than building an equally good class D amp. I wouldn't recommend anyone planning to design a power amplifier from scratch to head into class D for musical reasons alone. Especially if you're in the "no sound, maximum transparency" game, class D is tough to crack. It's doable, but easy it ain't.

OTOH, many class D's seem to have an euphonic edge to them (one that is usually liked by the designer and by audiophile reviewers, but much less so by pro users). I've yet to meet a beginning designer of class D amps who doesn't claim his amp to be "the first truly audiophile" class D.

To keep a long story short - for a given amount of research effort (not actual design effort, that depends on one's level of experience and hence past research effort) you'll get the best results (least colouration) with class A, followed by B, and lastly D.

The reason why all my current designs are class D is simply because I've gotten the hang of it, and so far all my customers have been happy so they're good enough apparently.