Bi-pulse convertor (ON-topic)

[Ian Fritz]

Terry et al. --

Thanks for your interest in the DPWG (double-pulse waveform generator). This
was originally one of a series of circuits designed to produce waveforms
with qualitatively different harmonic content from that available with the
usual saw/tri/pul set of waveshapers. For this circuit, the initial goal was
to produce signals with a high harmonic content (analagous to the usual
pulse waveform) but containing only odd harmonics. A SHAPE control to mix
this signal with a single pulse waveform was easily added to provide a
versatile set of waveforms.

There have been three versions of this waveshaper. The first is the original
experimental unit described in EN #72 and recently redrawn by Terry. This is
a reasonably simple unit that may be quickly breadboarded to get an idea of
what these waveforms sound like. I suggest decreasing the 560k resistor at
the PWM In control to 100k.

In the second design, which I actually built into my modular, the SHAPE
control pot is replaced with a voltage controlled mixer based on a pair of
3080's with complementary drive currents. This provides voltage control of
the relative amount of even and odd harmonics. This version uses bifet op
amps (LF351) throughout. There were some minor issues with symmetry of
clipping, OTA offsets, etc., but these don't seem too crucial to me. The
circuit uses ac coupling at the OTA inputs. This version of the DPWG also
has a control (manual) to mix some of the original signal (usually tri or
sin) with the output, to provide an even wider range of timbres.

The third version of the DPWG was designed to work with the SSM 2033 VCO
chip, and is driven by the 0 - 7.5 V sawtooth waveform produced by that
device. (The design could easily be modified with different saw amplitudes
and/or power supply voltages, but I haven't done so yet, except on paper).
This version of the DPWG operates somewhat differently from the other two. A
pair of window comparitors produce the two pulses. The widths of the two
pulses are the same (PW) and are set by a control-voltage level. The first
pulse is positioned at the start of the saw ramp and the position of the
second pulse (PP) is determined by another control-voltage level. The second
pulse can have either the same or the opposite polarity as compared to the
first pulse. Four control voltages drive the unit: PW, PWM, PP and PPM
(initial width and position and modulation of each). If the pulses have
opposite polarity and the second pulse is in the middle of the cycle, then
only odd harmonics are generated. Changing PP then adds even harmonics,
similarily to the SHAPE control of the expirimental circuit.

I have always liked these waveshapers -- even without sweeps -- and I use
them all the time in my recordings. Definitely worth your time to experiment
with!

  Ian


----- Original Message -----
From: "KA4HJH" <ka4hjh at gte.net>
To: <synth-diy at node12b53.a2000.nl>
Sent: Thursday, December 28, 2000 2:19 AM
Subject: Bi-pulse convertor (ON-topic)


> Here's a little project for everybody (and I keep thinking I already saw
> this in ElectroNotes or somewhere but...). This evening I got to fool
> around with the software synth Reaktor and I really like the harmonic
sweep
> you get with the "bi-pulse" oscillator. This is basically a pulse wave
> where the positive and negative halves are generated separately and work
in
> opposition. The top varies from 50% to 1% while the bottom varies from 50%
> to 99%, and vice versa.
>
> The waveforms (converting from a ramp) look something like this:
>
> 0% CV
>
> ----------              +5
>           |
>           |
>           |              G
>           |
>           |
>            ----------   -5
>
> 50% CV
>
> -----                   +5
>      |
>      |
>       ---------          G
>                |
>                |
>                 -----   -5
>
> 75% CV
>
> ---                     +5
>    |
>    |
>     -------------        G
>                  |
>                  |
>                   ---   -5
>
> Sounds like PWM but distinctly different. Should also work with a triangle
> input. Obviously this is a simple quantization problem, which has been
> discussed before. Anyone have a schematic?
>
>
> --
> Terry Bowman, KA4HJH
> "The Mac Doctor"
>