[sdiy] analogue phase modulation

[Magnus Danielson]

From: harry bissell <harrybissell at copper.net>
Subject: Re: [sdiy] analogue phase modulation
Date: Wed, 10 Jan 2007 21:50:24 -0500
Message-ID: <45A5A5F0.8040003 at copper.net>

Hi Harry!

> Yeah !  iirc the archive discussion was about the sawtooth
> phase shifter and could we extend it past one cycle...

Two in a row? :-)

Actually, all one have to do is to take the conceptual idea (do the circuit
analysis properly and you will get it) and then just extend it further for a
CV that ranges multiple cycles.

One such method would be to construct a synchronous sub-tone out of N waves,
apply the sawtooth phase shifting and then unwrap to N waves again. Effectively
will the N waves to 1 wave conversion be acheived by the staircase sawtooth
wave of a N step counter and by having each step matching the amplitude of the
sawtooth the synchronous 1/N frequency sawtooth have been shaped. The sawtooth
phaseshifter mearly moves the time for the reset which is done through a double
wizardy, by adding a suitable PWM (generated from the sawtooth with the CV) the
original reset is canceled and a new reset is generated. Just adding the PWM
to the sawtooth would shift the DC value as the CV changes, but this is linear
with the CV so subtracting out the CV will cancel the DC shift. This takes two
op-amps and 4 resistors. Back to the N cycle shifter, the shifted sub-tone
now needs to be converted back to the normal frequency and this can easilly be
done by just subtracting out the staircase. Voala!

The problem with this will be the non-linearity of the sawtooth, the
unstability of amplitude (temperature and De Franco compensation) and the
matching of the incomming reset slope and staircase slope with that of the
PWM reset slope.

The staircase waveform is easilly generated by a digital divider clocked by the
incomming sawtooth reset waveform. The outputs is then combined through a
resistor network (effectively a DAC) to form the staircase waveform.

So, it should total out with 4 op-amps and a suitable divider chip. Toss in a
buffer/inverter chip to shape up the digital waveforms for luxury version.
A handfull of resistors and a few caps for slewrate limitation where needed.

Op-amps with suitable bandwidth and slew-rate should be used for optimum
performance.

> which if I'm following the discussion we are talking about
> shifting over MULTIPLE cycles, not +/- 1/2 cycle...

It seemed they only needed +/- 1/2 cycles anyway, and then it works straight
out of the box.

> otoh I'm just barely follwing :^P

Me too.

Cheers,
Magnus