Help in conlfict PWM Vs Sigma-Delta

[danielpasti]

Hello people. Im from brazil, so I hope you dont care about my english. But I have a great problem. Im making a first order sigma delta modulator to integrate in a chip. It is allready done, but I cant explain, mathematic, to my teacher what about the SNR between Sigma-Delta and PWM (Open loop). I know that the acuracy, for the same fs, remains the same in both cases, but the SNR changes. Is there any equation for SNR in PWM mode ? How can I argument this ? thanks people

[jwagnerhki]

There are different kinds of PWM, for example uniform PWM and naturally sampled PWM. A short googling gave http://www.energy.ece.uiuc.edu/chapm...s/SVMpaper.pdf and www.eecg.utoronto.ca/~kphang/ece1371/pwrdac.pdf although they mention SNR (mainly the latter publication) it's not really in-depth mathematical. And then there's also noise-shaping in PWM that can be used.

Probably googling for "npwm upwm snr" and "pwm noise shaping snr" should turn up PDF's with nice SNR formulas in them... ;-)

[phase_accurate]

And it furthermore depends on what you call Sigma-Delta. If you are talking about the same thing as I do then the 1st-order delta-sigma is definitely worse than the PWM modulator.

Regards

Charles

[danielpasti]

Well. The first order sigma delta Im using has a fs = 5MHz (which is ease to achieve in chips development). It rounds 70dB of SQNR. Why is it worst then a PWM modulator ?

[phase_accurate]

Because a PWM modulator has infinite resolution (theoretically !!)while not even an ideal SD modulator has infinite resolution.

Since you are sampling at 5 Ms/s an SD modulator might be easier to implement than a PWM modulator. The PWM modulator OTOH wouldn't need very high switching frequencies to achieve a dynamic range of 70 dB.
A 2nd order SD modulator wouldn't be much more complicated to implement than a first-order one BTW.

Regards

Charles

[danielpasti]

Right. I ll find some about this OTHOT. But, Im developin this one for analogic application only, in other words, for use in Hearing Aids devides, so it is an anolog input and not an PCM input. Wherever, will be this, I cant change it now. The receiver of the HA have a cut off frequency of 10KHz, so I wont need resolution for freqeuncies above this. I found in the book ("an introduction to electroacoustics amplifier and design" ) that they treat some Sigma delta modulator as a PWM modulator. For example, it has a PWM modulator with feedback, and it is mixer with de input sinal in the entrace of a integrator, so it is and SD and not a PWM. I opened the articles listed above and I didnt found any equation for SNR for an open loop PWM modulator. When it is a closed loop modulator it is easy to obtain the SNR. IN OTHER WORDS, I have to compare and First order SD modulator with an open-loop PWM modulator with equations ...thanks

[SpittinLLama]

SD modulators have what is called idle patterns or tones. It is worse with lower orders and gets much better with 3rd order and higher. You might want to check into this before going for the design completely. Are you planning on running the hearing aid without a filter on the output of the Class D?

-SL

[danielpasti]

Yes . It must be without an output filter. The receiver receive directly de PWM bitstream. As I told before, the fs=5MHz. The pwm would even be better then to this SD? The negative feedback wouldnt improve the linearity for the audio band ?

[SpittinLLama]

SD will not work very well without an output filter or a SD design with a zero state (Class BD) which then has resolution issues with low signal levels (think quantization). I would recommend you use a PWM Class BD design since it has no resolution issues and can run filterless. Both TI and National make such chips for 5V applications. National also makes SD chips, 2nd and 3rd order filterless and a 4th order filtered.

-SL

[danielpasti]

Please people. I need your help. I still havent found what Im looking for. Some of you told that the PWM is better than a first order SDM. But I didnt understand it. I want a simple relationship SNR for a open loop PWM modulator. Is this true that I need to switch 2.9GHz a simple PWM system to achieve 16BITS of resolution? Why they are switched at 400KHz? Im talking about a simple open loop PWM, because the first order SDM is not so dificult to build, and I want to compare it. Please somebody help. thanks