How many modulation methods in Class D amplifier ?

[xuhaoz]

Hi, all,

I am a beginner in this forum, I have read some articles about the modulation method in class D amplifier, and I found that PWM and Sigma-Delta method are the most popular methos used in Class D amplifier.

Do you know any other modulation methods?

What's the relationship between COM and PWM?

Thank you very much.


yours
Jimmy

[phase_accurate]

How many do you like ????

Fun aside: Yes the main modulation schemes are indeed PWM and delta-sigma. Both have numerous sub variants however.

COM is a self-oscillating PWM modulator scheme by B&O.

Regards

Charles

[SpittinLLama]

Keep in mind there are two sub classes to Class D, Class AD and Class BD. AD means there are only two states for the output, typically the supply rails. BD has three states, usually the supply rails and 0V. Both PWM and delta-sigma can be done in AD or BD. To better describe modulation scheme the type and the sub classe can be used. But then there are lots of other 'things' done that are highly confidential to the companies that create them. All done to improve THD perofromance, reduce EMI, etc. Searching around on the web will pull up lots of info. Just don't be fooled by marketing classes such as Class T and Class I.

-SL

[dusfor99]

Hi,

Yes PWM and Delta Sigma. PWM is good for helping out with THD. The reason being that since the driver stage will have a rise a fall time that is not 0 so when trying to make a certain output level, say half-scale you will be driving the output stage with a 50% duty cycle but there will be a little bit extra due to the rise and fall time. Since in PWM the switching frequency is fixed, this extra amount is the same no matter what output level you are trying to create. To instead of making THD, it simple make a slight DC offset. If you were to use Delta Sigma, (meanign the output switching frequncy varies with output amplitude) the artifact of rise and fall times will cause THD. Think of it like this.
In Delta Sigma, when triyng to make a 1/2 ful scale output, the output switching frequency might be say 800kHz. However when going to 3/4 full-scale, the frequency is now only 600kHz. Since everytime yo switch, you add an extra error caused by the rise and fall time, the error you add is actaully dependent on the output amplutide you try to generate. This will cause THD.


Sorry if I bore you with some of the limited theory I know, but I thought you might be interested.


Dustin

[xuhaoz]

Hi, I really appreciate for your explaination.

BTW, if I use PWM topologies, which method do you recommend to implement on a chip ? half bridge? full bridge ? or self-oscillating(hysteresis , phase-shift controlled) ?

Also , please recommend some articles and threads , thank you very much.

Jimmy

[fredos]

You can make lot of variant. What I use is fixed PWM at 500Khz a low level and with a PLL slow down to 250Khz a hight power. This help a lot to reduce THD a low level and increase effiency at hight level, when THD is (a bit) less important. Just remember to do first switching frequency step of at least 20Khz to avoid beating between channel if you use common supply! That some kind of delta-sigma modulation without disadvantage! The more hard with this technique is to design feedback that will remain stable...Easy to do do if you think a lot about this! That what I use in my HVI amplifier..

Fredos
www.d-amp.com

[phase_accurate]

Some Self-oscillating toploogies behave like that without further effort !

Regards

Charles

[dusfor99]

Hi,

I dont have any experience with "self ocsillating designs" but I have designed a few different digital modultors. One that I did was for doing a headphone amp, and I have also designed modulators that are for high power (100W RMS or so) but really, the modulator is the same, just the ouptut stage can be scaled up for more power. The prototpyes I have designed have been able to reach 106dB of DNR . (That is take a -60dB input signal and notch it out, then measure all the rest of the noise in the 20-20kHz bandwidth and A-weight). I have found that TI has come out with a digital modulator that goes to a bit better than this, so I bought their refference design and will measure it. The modulators I have used are Sigma-Delta (a 9th order one) and PWM. I am really interested in learning the other meathods people use because I think it just plain cool.

CLD.

[Pafi]

this extra amount is the same no matter what output level you are trying to create. To instead of making THD, it simple make a slight DC offset.

This is not true. Timing error caused by fall and rise time depends on output current in a nonlinear way, no matter PWM or sigma/delta.

[dusfor99]

Hi Pafi,


Well im sold on that, I will take back that statement.


Thanks for catching me on that,


CLD