30 years later...

[gstopp at fibermux.com]

     Four waveforms in thiry years... there's a reason for that! The 
     triangle, sine, saw, and rectangular cover almost all of the harmonic 
     content requirements - no others are really needed. The electronic 
     function generator provides these waveforms easily, so that's why they 
     are what they are.
     
     See here's the deal - if you draw a waveform, any waveform, some 
     squiggle, with curves and slopes and whatever, and define a start 
     point and an end point, and then oscillate it, it will probably sound 
     just like one of the four waveforms above or a simple combination of 
     them. If the same wave repeats over and over again, it is a static 
     sound. It can be buzzy or dull or thin or full, but it will be a 
     boring tone.
     
     To be interesting it must move. It must change in subtle ways from 
     cycle to cycle in order to be an improvement on the four basic 
     waveforms. Pulse width modulation, beating VCO's, and animation will 
     make things more sonically interesting. So will wavetable scanning. 
     Using a sequencer as a waveform "graph" will give you some nice buzzy 
     yet boring waves, but if you try this also try moving the pots while 
     you listen and you will hear the animation.
     
     Shift registers and low-resolution wavetable scanners sound very 
     gratchety and machine-like because there's a lot of higher frequency 
     "sub-waveforms" inside the main waveform. These are actually pretty 
     cool and I think I should try to experiment in that direction some 
     day! PPG owners already know this stuff. I heard a Digisound EPROM 
     scanning VCO over at Kevin Lightner's a while back, and it was pretty 
     scary.
     
     I like the "backwards" waveform idea. You'd need to modulate time 
     itself for that trick, maybe in a few years? Make it 
     voltage-controlled? :)
     
     - Gene


______________________________ Reply Separator _________________________________
Subject: Re: 30 years later...
Author:  Christopher_List at Sonymusic.Com at ccrelayout
  Date:    10/31/96 10:39 AM
     
  Let's not forget the various "pseudo-digital" waveforms that you can do 
  yourself;
  1. Stepping a sequencer at audio frequency. A four or eight step 
  sequencer, with one or two rows, dedicated to a task like this (i.e. no 
  internal clock, no up / down, etc) would be super easy to build - 
  basically just some pots, a 4017, and an op-amp or two...
     
  2. Stepping through eprom values using a VCO'd square wave and a counter. 
  This guantees the waveshape will always be the same.
     
  3. Scott Gravenhorst's "Complex Waveform Generator" that uses XOR 
  feedback logic
     
  While the PPG and ProphetVS use method 2 - as far as I know, they do it 
  digitally, not using a VCO to drive the circuit. Therefore, it's safe to 
  say that if you're looking for any of these methods in an actual 
  production synth, you won't find them.
     
  More stuff to think about;
  I always thought it would be cool to stack up four VCOs and have them 
  hardwired to be at different intervals relative to one another and have 
  volume settings for each one - so that you could have instant partials. 
  There'd only be one set of controls for all the VCO's, and there'd be one 
  set of the usual sine/tri/saw/pulse outs. The cost in MAT-03s would make 
  this one expensive VCO!
     
  As I mentioned a while ago, a great method of getting new wave shapes is 
  to use a VCO (at audio frequency) to;
  1. X-fade between two other VCOs (or a VCO and the same VCO filtered) 
  2. Control the CV on "Timbre Modulator #2" - the full wave-rectifying 
  x-fade thing.
     
  Lastly, it would be really easy to build some kind of x-fade or flip-flop 
  or counter into your VCO so that you had an output wave form that came 
  out as half square then half saw then half tri or some combination of the 
  above (which I guess is what you want - everything done at the VCO level 
  - no other circuits). Though this would color the sound, but I don't 
  think it would change it too drasticly.
     
     
  What sort of a waveform would you like to see - outside of a sample? Take 
  the Waldorf uWave - they put all kinds of different waves in there - but 
  most of them sound very similar (or the same) - until you filter them or 
  mix them - so why bother?
     
  - CList
     
  ps - what I'd like to see is a wave form that goes <backwards> 
  - so it would look like this on the scope;
  &&&&&&&&&&&&&
  - :)