UPWM

[terminator12]

I have read that NPWM(natural analog PWM) is fundamentally linear whereas UPWM(digital PWM) distorts the input signal. Intuitively, are not the pulse widths linear to the input amplitude in both the above cases?

Could someone please give me intuition on this or can point me to some resource which gives intuition on this. I have referred Karsten Nielson's thesis too, but could get only a mathematical proof.

[Pafi]

Intuitively, are not the pulse widths linear to the input amplitude in both the above cases?

Not amplitude, but sampled values of the signal! And the difference is in the sampling points. As long as the input signal is DC, or very low frequency, the two methods are equivalent, but at high input freq the naturally sampled and uniformly sampled PWM takes different sample values. This can be seen if you draw them.

Basically both methods generates some problematic spectral components, but while UPWM generates harmonic distortion, NPWM generates non-harmonic components. I can't tell why.

PS: NPWM is not neccessarily analog and UPWM is not neccessarily digital. Both can be realised in both way, but the basic operation in theory is always analog! True digital "PWM" is a totally different thing (sigma-delta converters).

[terminator12]

pafi, thanks for replying. I have referred many papers. Everywhere, people tell UPWM is nonlinear and come up with techniques to approach NPWM. But they never explain intuitively why UPWM is non-linear and why NPWM is linear.

Another doubt I have is that, even though the comparator is such a non-linear block, the NPWM output spectrum is totally linear for the signal in the baseband. Can anyone please discuss their angle of viewing this property of the PWM.