[sdiy] Synthesizers...

[Simon Brouwer]

Scott Nordlund schreef:
>
> I've never seen any synth-related reference to this (guess it would be
> in electronotes if it's anywhere), but saw-derived PWM and tri-derived
> PWM do have different spectra.  PWM from a sawtooth wave results in an
> upward (or downward) sideband shift proportional to the modulation
> frequency, while PWM from a triangle wave results in a simultaneous
> upward and downward sideband shift.  This is why saw-derived PWM can
> sound "out of tune", especially at low frequencies.

Here is how I see it...

As explained in other messages, PWM simulates the beating of two sawtooth
waves of different frequency.
For example, if you use a 1 Hz triangle to fully modulate the pulse width,
then with a VCO frequency of 1000 Hz it sounds like the addition of two
sawtooth waves, one of 999 Hz and one of 1001 Hz, if you derive the pulse
signal from the triangle.

If you derive the pulse signal from the sawtooth then it sounds like one
saw of 1000 Hz and one of alternatingly 998 and 1002 Hz. So in this case
one of the saws is not detuned but the other one is twice as much out of
tune.

At a VCO frequency of 100 Hz this will become 99 + 101 Hz and 100 + 98/102
Hz respectively. So the relative detuning of the two saws will be larger
at low frequencies (also with pulse signal derived from triangle).

The semimodular I'm (slowly) building will have two dedicated PWM LFO's
that will track the pitch voltage to compensate for this.

-- 
Vriendelijke groet,

Simon Brouwer
-*- nl.openoffice.org -*- http://www.opentaal.org -*-