[sdiy] Continuously variable waveshaping (was Behringer Neutron)

[Tom Wiltshire]

> On 8 Apr 2018, at 06:52, Andre Majorel <aym-htnys at teaser.fr> wrote:
> 
> On 2018-04-07 21:39 -0400, Ben Bradley wrote:
> 
>> I've been wondering what the details are on these "waveshaper" things,
>> whether they're just mixing/panning between different waveshapes (say,
>> sawtooth and square), or if something nonlinear is going on or being
>> added.
>> 
>> I tossed together a four-stage wavefolder (talk about nonlinearity!)
>> not long ago, and was amazed at how a sine wave input made an output
>> that sounded a lot like a filter sweep, simply by varying the
>> amplitude of the input.
>> 
>> It was basically the bottom four stages in this schematic, and I
>> didn't even use the two diodes in the feedback path of the last stage:
>> [200-character-long Google URL]
> 
> This one ?
> 
> http://www.cgs.synth.net/modules/pic/r_schem_cgs29_wave_multiplier.gif
> 
> If so, it's part of
> 
> http://www.cgs.synth.net/modules/cgs29_wave_multiplier_v11.html
> 
> Side note about the "grinder" : Ken says the LM358 is more
> effective than the TL072 or 4558 for IC1 because slew rate is
> critical in this position. So what are those slew rates ?
> 
>  TL072  13 V/µs typical (data sheet)
>  4558    1 V/µs typical (data sheet)
>  LM358 0.3 V/µs typical (http://www.ti.com/product/lm358)
> 
> Amusing that the slowest op amp would be the best at creating
> high frequency hash.

It’s probably best *because* it’s slowest. There are plenty of examples in the stompbox world where people prefer the rubbish old slow op-amps over more modern “hi fi” options because the limited slew rate reduces the “edginess” of the high frequencies. Effectively, you’re adding a bit of roll-off.

For a wave folder, that might be true too - you want the wave folded, but lots of sharp spikes is just making high frequency hash, as you say. A little bit of rounding of some of those points might reduce the hash and still leave the interesting harmonic evolution in place.

Tom