VC Waveshaper from the test bench

[gstopp at fibermux.com]

     Hi DIY,
     
     I've got a new voltage-controlled waveshaper design that really has me 
     excited.
     
     A couple weeks ago I was bringing up my full-blown ASM-1 system and I 
     was looking at stuff on the oscilloscope whilst turning trimpots. 
     Along the way I got a look at the waveform produced by two slightly 
     detuned square waves and I noticed something...
     
     (Digression - the sound of two detuned square waves is a very profound 
     sound. I like it a lot. To achieve it properly, you must either start 
     with *perfect* square waves, or use pulse waves and tweak the duty 
     cycle to *exactly* 50%, because the sound of a perfect square wave has 
     very distinctive harmonics. This sound was made famous by Keith 
     Emerson's solo at the end of "Lucky Man". I like to use three or four 
     detuned square waves for it as well, for a very "evil" sound.)
     
     Anyway on the scope I saw a pattern that reminded me of a circuit. The 
     waveform was a stepped wave with three levels - positive voltage, 
     ground, and negative voltage. The widths of the positive and negative 
     rectangular sections were constantly changing - between maximum width 
     (no ground level, square wave) and minimum width (ground level only, 
     no sound, meaning perfect cancellation).
     
     I recognized this waveform - it looked like the drawing in 
     Electronotes for the output of the "double pulse waveform 
     shaper"! (It's in the "Timbre Modulator" section.) I did some 
     scribbles on paper and discovered that the two different 
     approaches (combining two square waves or using this circuit) 
     would in fact produce the same result. I built it, and sure 
     enough, triangle in, change the width, and it sounded exactly 
     like two detuned square waves.
     
     Once I had the protoboard up and the headphones on, my mind 
     started to wander as my fingers stirred up a pile of parts on 
     the bench. I wondered what would happen if I added more 
     comparators and summed their outputs together with resistors, 
     and used a resistor ladder for the reference thresholds driven 
     by the pulse width inputs. I tried this and the sound was very 
     profound - a little between pulse width modulation, with strong 
     odd harmonics at every point, and VCO hard sync. On the scope I 
     could see that the waveform looked like a very low-resolution 
     stairstep approximation of the triangle input, with the steps 
     moving from wide to narrow as the PWM input varied. The sound 
     had strong aliasing-type harmonics, much like a bad sampler or a 
     EPROM-scanning VCO, but you could move these harmonics around 
     with the PWM input. At the extreme ends, all comparators were at 
     the same state except one, so a very nasal narrow pulse 
     resulted.
     
     Essentially this circuit became a low-resolution flash converter 
     with equal-weighted summed bits plus voltage-controlled 
     converging and diverging thresholds.
     
     I tried a sawtooth input instead - very differnt, even harmonics 
     now, yet the same profound harmonic sweeps! I was quite 
     impressed. You can sweep from silence (no pulses) through narrow 
     pulses, into a square wave, and past this back into narrow 
     pulses and then silence again. In the case of the sawtooth input 
     this means a transition from saw-like even/odd spectrum into 
     pure odd-only (square) and out again, all the time with 
     sync-like harmonic sweeps yet not so harsh as hard sync.
     
     I'll be building a few of these, to be sure, if only to see what 
     it sounds like to have two or three of these running in 
     parallel!
     
     - Gene
     gstopp at fibermux.com