Harmonics question

[Ethan Robert Duni]

>Well, the strength of the odd harmonics in the subtracted function would
>have to be exactly the same as in the initial function, and it would have
>to be phase-correct too.  
-well, i wasn't thinking so much of ELIMINATING the odd harmonics as just
changing the ratio.. yeah, phase coherency would be paramount (since the
idea is built on destructive phase addition).. 

>If you have the means to do this, then you
>probably also have the means to just strengthen the even harmonics in the
>initial signal.
-well, i was thinking of something relatively simple like running the signal
through a comparator to turn it into a square wave, then subtracting that
from the signal.. seems like that would be pretty easy to do by using an
op-amp as the comparator and powering it symmetrically.  then feed that into
a differential amplifier with the original signal.  of course, this would
require an op-amp for the comparator that would swing to symmetrical rail
voltages.  figure make a symmetrical resistor network to control the +-
comparator voltages so you can adjust the amount of signal fed into the
differential amp (so you don't cancel the fundamental completely).  probably
still need another amp on the end to make up for signal loss.. but 3 op-amps
and a fistfull of resistors doesn't seem horribly complex... or am i missing
something? anyone?

>There's another surefire way to increase the even
>harmonics: clipping.  As the amount of clip increases, the signal becomes
>more similar to a square wave.
-i'm pretty sure a square wave contains only ODD harmonics?  though that's
only a perfect square wave, which you wouldn't be getting.. 

>In a related vein, awhile back I was involved in a discussion about
>generating PWM on a sampler.  The method used was playing back a saw-up and
>saw-down wave at the same frequency, with a LFO modulating the pitch of one
>of the saws by a very small amount (around .25 Hz).  The result is a pulse
>wave with varying DC offset.  The problem was getting the LFO speed and
>depth set such that the pulse width didn't get too close to zero.
-sounds like fun; i'll have to fire up this here akai.. 
Ethan Duni