RE(2): spectrum of PWM

[Haible Juergen]

> 	>You should be able to model it using Bessel functions.
> 	>
> 	>Consider this:
> 	>
> 	>You have two sawtooth waveforms which are at exactly the same
> 	>frequency (SHARP! THE same) which are added together 1:-1. One of
> these
> 	>are phase-modulated in the relation to the other. Now, as we change
> 	>the phase modulation will the resulting pulse-width change. Now, we
> 	>can handle this using Bessel functions in the frequency plane.
> 	>I think scetching something along these lines will make you arive
> at
> 	>an exact solution.
> 	>
> 	>Is that sufficient for you?
> 
	>Excellent. I think that's the solution !

I was too fast. That's the saw-based PWM (minus DC shift) that is modelled.
Any ideas for the triangle-based PWM ?
(With the "center of gravity" of each pulse being invariant, rather than the
rising
edge as in the saw-PWM case)

JH.