[sdiy] Mixer levels

[Tom May]

"jhaible" <jhaible at debitel.net> writes:

> > If you combine 2 signals from the same VCO, two different waves I mean,
> > they can combine in phase with each other or out of phase, was we've
> > discussed before. In some cases, you actually don't add to the voltage
> > swing at all.
> >
> > For instance, if you combine a short pulse, with a sine and they're out
> > of phase you will have the sine intact, but with a rectangluar notch on
> > one side. That notch willgo down to zero, but it won't add anything at
> > all. Does this sound different than if that was a huge spike sticking
> > off of the sine? I don't know. But one way uses up a bunch of headroom,
> > probably clipping the mixer and the other wave uses the same amount of
> > headroom as the sine alone.
> 
> In linear systems, this makes sense. In nonlinear systems, it will make a
> huge
> difference if the spike on top of the sine is rounded by soft clipping or
> if the spike that's "shielded" inside the sine is unaltered.
> If you're thinking "HiFi" (linear, headroom, all that stuff), two such cases
> might be equivalent, or the case with higher amplitude will get additional
> harmonics from distortion (when you're exceeding your headroom).
> If you're thinking nonlinear, the clipped (steady) signal of a oscillator
> wave
> might have *less* harmonic content than the unaltered version. That's the
> joy
> of analogue synthesis (and quite different from what people have incorrectly
> called "subtractive synthesis").
> 
> A pulse with modulated pulse width is very similar to two saw waves
> which are added in a linear system (with one saw inverted, if memory
> serves):

Yes inverted, combining the inverted and non-inverted saw samples on
the SY77 is a trick for getting "PWM".

> There's a constant movement and change in your sound.
> If you're running the two saw waves thru a nonlinear system that clips
> (or rounds) its peaks, the beating (and timbre movement) will stop
> for a while, when the addition of both signals goes above the clipping
> threshold. (If you cannot get higher, you won't have any movement from
> going higher, either.)
> This is similar to a PWM with a *clipped* sine LFO for modulation
> source.
> I'm not trying to show equivalence between PWM and true beating here.
> I want to point out that nonlinearity can transform simple amplitude
> effects into something very different: a perceived change of *modulation*.

I know what you're saying, but your example doesn't seem to work.
When you add the two saws you get a pulse wave with two levels.
Clipping won't really affect it since a clipped pulse is still a
pulse.  The pulse width and timbre still change if the saws are
slightly different frequency.

This topic does bring to mind how cutting frequencies with an EQ can
actually boost the instantaneous peak level in some cases, depending
on phase, possibly exceeding headroom and causing distortion.

Tom.