[sdiy] analogue phase modulation

[Magnus Danielson]

From: Eric Brombaugh <ebrombaugh at earthlink.net>
Subject: Re: [sdiy] analogue phase modulation
Date: Wed, 10 Jan 2007 16:03:08 -0700
Message-ID: <45A570AC.8010000 at earthlink.net>

> JH. wrote:
> >> When I was doing the Modcan VCDO research I found that 314% mod index
> >> was enough (funny - that number looks familiar).
> > 
> > 314% as in 3.14 periods, or as in 3.14 (= half of one period) ?
> > 
> > Ok, must be the latter, or there would be no significance in the Pi*100% 
> > number.
> 
>  >Bing< Give that man a cigar!
> 
> > And indeed. I made some simulations with phase modulation of only +/- Pi, 
> > and it gives a very rich spectrum already.
> 
> That's exactly the way to look at it: we're modulating +/- 1/2 cycle of 
> the wave about some nominal center. As the carrier ratio increases this 
> gives some nice wrap-around/foldover effects that introduce a lot of 
> higher harmonics.
> 
> > So ...
> > If we _only_ need sin / cos function fitting for [-Pi .. Pi], two 
> > multipliers and a well-tracking
> > QVCO (or an ordinary VCO plus a dome filter instead of QVCO), then building 
> > DX7-style
> > operators completely in the analogue domain will be very easy, and I wonder 
> > why we've
> > had so much trouble about it in the past.
> > 
> > Too good to be true? At the moment, I don't see any catch ...
> 
>  From my work in the digital domain I found that the timbre really falls 
> apart if the modulation/carrier ratio isn't precisely controlled. 
> Without a controlled ratio you'll get beating effects that, while 
> interesting quickly become anharmonic and degrade into something that 
> sounds more like simple ringmod. The Cynthia ZO has some capability for 
> locking the modulation & carrier frequencies together to help with this.

OK, so if +/- pi radians is all it takes, then inserting my sawtooth phase
modulator

http://rubidium.dyndns.org/~magnus/synths/schematics/sawphaseshift.pdf

should be enought. Just do a normal sawtooth core, and sine waveshaper and
away you go.

In order to do the sine/coside function you either need that special Analog
Devices chip or you have another sine/cosine generator with a phase locked
square generator of nominally the same frequency. The phase locking is to be
a full integrating second degree PLL and the tweakability is only about +/- 10%
or so. The squarewave then samples the sine and cosine waves. The PLL is then
pulled by adding the phase CV to sum up with the output of the PLL phase
detector. Naturally, the frequency needs to be high enought for the aliasing
arthefacts to be small enought after the anti-aliasing filters. Could work, but
it is another half-complex setup.

Cheers,
Magnus