[sdiy] A question about Chorus

[Ian Fritz]

Hi Bernie --

As an aside, I've been playing with some softsynths (iPad apps, 
especially), and I see that many of them have "supersaw" waves, which seems 
to be basically the same as the MPWA. Magellen actually has four waveform 
animators (which they refer to a "unison").  So possibly the MPWA could be 
the most influential of all your projects!  I still have my personal 
version working, and Cynthia and Yves have somewhat simpler analog 
implementations for sale.

Ian


At 09:55 AM 9/1/2013, Bernie Hutchins wrote:
>Hi Tom -
>
>I saw your SDIY question about "Chorus" effect with parallel delay 
>lines.   I think you are wondering (without coming even close to 
>believing!) if you might be able to get away with a single delay with a 
>complex control waveform instead of expensive multiple delays, summed, 
>with simple delay-controlling waveforms.  No such luck!
>
>I recall thinking about pretty much a similar issue with my Multi-Phase 
>Waveform Animator.  This was in an unnerving setting of being about to 
>describe the MPWA the next day at an AES show in 1980.
>
><http://electronotes.netfirms.com/AES4.PDF>http://electronotes.netfirms.com/AES4.PDF
>
>  One generally tries to anticipate questions the audience might ask 
> (never once guessing correctly) and I thought - what if I had just summed 
> the eight LFO's and fed the control sum to one sawtooth shifter?  Would 
> that be the same thing?  I think this is pretty much equivalent to your 
> question.
>
>These questions are deep, and slippery.   So we look for a simple 
>counter-example.
>
>Suppose you had just two delay lines, with the results summed.  One delay 
>is fixed.  The second delay then is made to sweep through this fixed 
>value, and assume the delay is swept slowly as would happen with a chorus. 
>Assume further that your signal is a single sinusoidal identical at the 
>input of both delays.  At the point where the sweep makes both delays the 
>exact same, you get a doubling of the signal in the summation.   But move 
>the delay forward or backward a bit, and you will not only not have a 
>doubling, but are headed for a null.  This is a classic comb 
>filter.   There is no single delay, no matter how complex the control, 
>which does this - the output always has unity gain.
>
>With a single complex control signal to a single delay, you will get only 
>very slight modifications to the spectrum as a result of a modulation of 
>phase.  Your control is very low frequency with a chorus effect.  The comb 
>filtering on the other hand greatly modifies the spectrum dynamically.  I 
>think that's the "why" here.
>
>We did learn a lot from the MPWA.  We found that it takes something like 8 
>processing devices in parallel for a good effect, but that 16 were too 
>many (just too homogeneous).  Another thing was the audio demo failure at 
>the AES show (the big ballroom of the Waldorf) to hear the effect.  In 
>that room, there was already immense reverberation which homogenized the 
>effect, as in comparison for example, to listening in a small room or with 
>headphones. Fortunately I had my "co-conspirator" Lester Ludwig there who 
>was familiar with the real effect and understood the acoustic problem at 
>the show.
>
>Good question.  Achim's explanation in terms of crest factor is also very 
>useful.
>
>Bernie
>