[sdiy] random thought about sine shapers

[cheater cheater]

On Mon, Apr 19, 2010 at 18:48, David G. Dixon <dixon at interchange.ubc.ca> wrote:
>> Hmm, I'm not so sure it's the exact one I'm thinking of, because the
>> one I am thinking off starts with 0% PW when the saw is output, and as
>> we increase the setting, the slope of the (rising) saw to the left of
>> the (negative slope) discontinuity has a 1% PW pulse there on top of
>> it; then as we go further the saw is slowly faded out and on the other
>> hand the PW gets higher until we reach 100%. I guess you could do
>> this: feed the saw S to the pulse comparator, call the output pulse
>> signal P. Have a voltage divider which divides evenly between the
>> pulse comparator and 'no comparator'. Then those two signals get mixed
>> together. I think it would/could work well this way. Note this allows
>> you to have a sort of 'cs 80 style' saw there.
>
> What you're describing is best accomplished with a cross-fader, I think,
> unless I'm misunderstanding you.

you need the voltage of the sawtooth wave divided *before* it goes to
the shaper. Depending on how much of it goes into the shaper, the PW
will be higher, whereas the 'dry saw' in the mix will be less and
less. The peak-to-peak voltage of the PW part in the mix stays the
same all the time, but because its pulse width is lower, it sort of
'fades in' with the pulsewidth increase, while the saw 'fades out'
because of the voltage divider. It's a very simple circuit but it
makes for a very fun sound, especially when you put it through
non-linearities.

>> I might be wrong, but I believe an overdriven OTA sine shaper might be
>> nicer in sound than an overdriven transistor sine shaper. Very long
>> since I tried them, though, but the difference in sound is certainly
>> there.
>
> I've done both, and I see very little difference.  Simulation suggests that
> one can achieve slightly better THD specs with a transistor-based one, but
> we're talking about the difference between 0.5% and 0.7% or so.  IMHO, the
> quality of the sine depends much more on the quality of the incoming
> triangle than on the actual shaper itself.

Is better THD always better?

:-)

D.