2 modules ideas

[Hairy Harry]

No voodoo is needed. What you need is the wave wrapper
proposed by Thierry Rochebois... or if you really want
a linear triangle, the variant of his circuit that I
designed.

The wave wrapper will (at certain input points) give triangle
wave outputs at 2X, 3X, 4X, 5X.... depending on the number of
series stages.

This was posted before. Since we are in "no binaries" mode...
if someone wants to host I'll send the schematics and we can all
look....

H^) harry


>From: "Grant Richter" <grichter at execpc.com>
>To: jbv <jbv.silences at wanadoo.fr>, "Synth-DIY" 
><synth-diy at node12b53.a2000.nl>
>Subject: Re: 2 modules ideas
>Date: Tue, 15 Aug 2000 11:54:34 -0500
>
> > 1) "additive" synthesis
> >
>
>The Buchla Model 148 Harmonic Generator did this.
>
>From the catalog "Generates a fundamental and its first
>nine harmonics (Harmonic Numbers 1-10 (sic))".
>
>The method used was to start with a triangle oscillator,
>followed by cascaded non-linear function generators to
>generate harmonic triangles, then convert the triangles to
>sines with the classic JFET waveshaper.
>I never have found a schematic for this, but once talked
>to a fellow at U of Cal. that had seen it.
>
>The initial triangle can be turned into a 2X triangle by
>full wave rectification followed by a gain of 2 and offset
>back to a zero center either by AC coupling or DC summing.
>This would allow you to generate 1,2,4,8 triangle wave
>frequency multiples. Once again, the triangles are then
>converted to sines.
>
>The 3X NLF must have used some of that Buchla voodoo.
>How you get a 3X nice triangle wave from a 1X triangle
>wave, is not obvious to me. But if you can do it then
>the 6 multiple just uses the FWR stage again. And 9 could
>use the x3 circuit again.
>
>5 and 7 would seem to be very tough. Perhaps a multiplier
>between lower order terms (I can't visualize it).
>
>Another method would use the phase detectors from 4046s
>with a divider chip to drive EXAR VCOs with pulse and
>sine wave outputs. (pulse for the divider and sine to
>listen to). In this case the PLL slew rate could be
>exploited for effect and a slew pot could be put on
>each harmonic. That's bound to sound interesting.
>
>The least expensive method I can think of would use a
>multisampled EPROM look up table. A 8K EPROM divided
>into 8 sections gives a sine lookup table of 1024 for
>the fundamental. You would store 8 waveforms each with
>a higher harmonic. You could use a 400 Khz 10 Bit National ADC
>(ADC1061) for the 10 LSB of the EPROM.
>You then phase lock an HC4046 at 8X the ADC one bit. That
>goes to a 4024 which scans the eight different harmonics through
>a 4051 to eight sample caps buffered by two TL074s.
>
>Chip count 9 for 9 harmonic sines at multiples of input triangle
>TL072 input buffer to ADC
>National ADC1061 A/D
>EPROM
>DAC0800 D/A
>4046
>4024
>4051
>2 X TL074
>
>This has one input for a triangle wave and eight sine outputs of the
>frequency multiples 2-9 of the input.
>
>Essentially what you are doing is what Buchla did above, but using a hybrid
>digital approach. The EPROM is also being used for sine conversion. Better
>results might be obtained by outputing triangles from this scheme and
>keeping the analog sine shapers.
>
>
>
>
>
>

________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com