AW: AW: VCO idea and env

[Haible Juergen]

	>One little comment I can't suppres: The above mentioned .html
document
	>always says that triangle waves are used for scanning (am I
right?).
	>But this means that every interpolator stage appears TWICE in one
	>cycle.  This is not quite what one expects, if you are going do
"draw"
	>a PWL.  Instead a sawtooth wave should be used (what happens on the
	>"flyback" ?  glitches ?) I ask this not to annoy Juegen, but maybe
I've
	>not understand the circuit.

Hmm, yes, if you have static waveforms and complete graphical 
control in mind, scanning with a saw wave is probably the best.
Just adjust the level of the saw input signal so that it comprises
exactly the full range from 1st to last slider. Glitches depend
on your saw reset time and input level / offset adjustment
in that case.
But as someone has mentioned, static waveforms are kind of
boring. The IS comes to full bloom with varying input levels
(or offsets etc.). I've made a lot of experiments since I built it,
and most rewarding setup is with the input signal running thru a VCA
(ADSR-controlled) before it goes into the IS. Even better, use an
external monosynth (or guitar, or anything else), which already has
a certain loudness contour. As for the waveform of the input signal,
low harmonic contents gives a more dramatical effect. That's why
I use triangles most of the time. But saw waves produce some nice 
effects, too.
For clean output waveforms, a 1 VCO source is fine. If you want the
IS to become alive, you may use a 2nd VCO or a LFO, either of which
would just be *added* to the main input signal. (No extra modulation
required here; just sum the two together.) Make sure that this second
source has a much lower level than the first one. A constant, low level
VCO2, and an ADSR-enveloped high level VCO1 is a good choice.
This is, for rather "tame" waveform sweeps. If you want to go wild,
feed a polyphonic synth into the IS. (A short VCA release time is a good
idea here ..) You might want to set a range of rather "weak" and monotonous
nonlinearity to process polyphonic sounds. Then add one "dent" (non
monotonous segments) at one end, and drive your polyphonic input source
into this region from time to time with a volume pedal etc.
Opps, it always becomes a long post when I write about the IS ... sorry.

BTW, the idea with the triagle input is inspired by the Hammond Scanner.
This one is rotating, i.e. start point and end point connected together, and
this would correspond to a saw modulation. BUT Hammond has wired the
scanner to get a symmetrical (sine ? triangle?) function nevertheless.
But this is for the other IS application, filter scanning, of course.
Just to explain the history of the idea ...

Even with other than saw input signals, a graphic slider repesentation
is still useful: In general, you don't draw the waveform, but the dc
transfer
function. So I wasn't exactly precise when I spoke of the forthcoming
distortion device. Graphically controlled nonlinear distortion courve
might fit.

In the ideal case, the PWL corners are quite sharp. The cubic spline
idea on my homepage was a joke, of course. But maybe the corners
should be rounded by using a less ideal rectifier. Something like the
MS-20 triangle converter comes to mind. But I haven't got a working
circuit for that version yet. Temperature compensation is important when
you go down near the threshold voltage of diodes ...

JH.