Some phasemodulation thoughts

[Martin Czech]

:::The task would be to add the output of a sawtooth VCO together with the
:::modulation signal. 
:::
:::To allow for deeper phasemodulation one could "wrap" the phase signal with
:::a circuit similar to the serge waveshaper. The folding should occur at
:::+/-(3/2)pi. Let pi=1 (volts :-) and set up a chain of full wave rectifiers
:::to fold the modulation voltage at +/-1.5V +/-3V +/-4.5V ...
:::The amplitude of the saw would have to be 1V (2Vpp).
:::glitches. But they could be compensated like in Jorgen Bergfors' sineshaper.

snipped...

Hello Rene, I just had the same idea on the weekend. Ayear ago I asked the
list how to do analog phase modulation, no answer. Now, it is very simple:

How would you do this in the digital domain:

You have a sine table and a phase pointer to this table.  The phase
pointer gets an additive component,, this is the modulations signal.
The final phase pointer result is modulo, it wraps.

Analog, this would mean:

the sine table is replaced by a triangle network with a sineshaper at the
output.  Such a triangle network is simply a piecewise linear function (PWL),
it can be done with precision half way rectifiers or transistors as
emitter followers.  Such networks are nothing special, severall "morphing"
or interpolation circuits which were proposed on the list use them.

For every 180 DEG of phase modulation one HWR is needed, at the end one
needs two summer stages to add up all HWR signals with appropriate sign.

The precision HWR gives more dc accuracy and lower drift, but is slow.
The emitter follower HWR is faster, but needs another junction
(complementary emitter follower) for 1st order temp compensation.


Ok, the phase pointer is ... simply a sawtooth. Preferably with fast
discharge slope.  The modulation input is just added to the sawtooth.

Over the weekend I did a gnuplot simulation, and it worked imedeately.

In reality we will have some artefacts, however:

1. the sine shaper (ota or diode network) is of course nonideal

2 . the single segments of the PWL may not match exactly, but this could
be trimmed

3. this is the biggest: the saw discharge will need time, and the opamps
will also need time, thus we get large glitches in the output signal.
Sometimes they will look like in a saw->tri shaper, sometimes the will
wiggle once more. These glitches will add partials at Fc+-Fm, therefore
they could be masked by PM sideband frequencys. There are ways to
make these glitches a bit smaller.

4. this is PM, so at low modulation speed the percieved frequency change
will be small. This circuit can not do these fantastic analog frequency
sweeps, of wide frequency modulation, from sub audio to 10kHz, or so,
if you know what I mean. So, analog frequncy modulation is still necessary,
but maybe not the through zero variant, with all the implicit problems.


The advantages will be:

1. can be added to any reasonable existing osc.
2. no problems with frequency or phase stability
3. ratio independend harmonic contend

m.c.