Is it possible to design a Class D composite amplifier ?

[phase_accurate]

There are two loops: The main linear feedback loop of the class-A voltage source and the on(s)e from the switching current source. The linear feedback loop is dominatiing the behaviour.

Here you can see the pasic principle:
https://www.pes-publications.ee.eth...29_JWK_1996_Basic_Considerations_00500444.pdf

I once made simulations of the same principle but using delta-sigma modulation instead of PWM.

But it is still not clear whether the OP was talking of such a topology or if he meant to put another feedback loop around an existing "normal" class-D amp.

Regards

Charles

 

[5th element]

For what it's worth nCore and Purifi amps are essentially composite classD anyway.

Stability is obviously an issue and these amplifiers are only conditionally stable. They contain stability/clipping detection circuits that deactivate the additional composite loop should the amplifier start to oscillate. From what I understand they are only induced into instability when clipping anyway. So they simply bypass the added feedback when the amp clips.

 

[phase_accurate]

Well, higher-order loops are at the same time easy and difficult. While it can quite easily be done for linear amps, the nature of a switching amp is p...ing on the parade. If you manage to design a stable loop it will definitley improove PSSR, load stability etc. But due to the nature of the beast it will not automatically improve THD and IMD - it can actually worsen if done wrong. This is quite another league to achieve. And the methods differ between natural sampling PWM amps and self-oscillating ones.

Regards

Charles

 

[KaffiMann]

But it is still not clear whether the OP was talking of such a topology or if he meant to put another feedback loop around an existing "normal" class-D amp.

I suspect it would be the desire to have an amplifier with specs like the Devialet or the Neurochrome amplifiers only in class D and/or perhaps some mixed in class H.

Many Class D designs struggle to get much more than 108dB THD+N at the very best, I can understand the "want" to have good performance but that would also put demands on the rest of the signal chain, not many DA converters or other sources that can achieve 130dB THD+N in real life use.
Still... A good goal should be breaking the 110dB barrier for low power designs, that might enable people to use compression drivers sans pad-down resistors.

 

[phase_accurate]

For low power even an old-fashioned class-A amp wouldn't waste a lot of energy.

Regards

Charles

 

[KaffiMann]

Indeed, but let's say you'd want at least 25-50w before distortion sets in, it's still a demanding task to find DIY designs that have that kind of noise floor. Not many pro designs either.

 

[phase_accurate]

You are right and I am also looking for a good amp in this power range for exactly that purpose. But it will not be a class-D amp. And yes, I am also using resistors at the moment. It is nice when the sounds fade into silence instead of noise.

Regards
Charles

 

[KaffiMann]

it will not be a class-D amp

I think some of the more recent developments in Class H seem promising, we'll see what happens.