[sdiy] What is a quadrature oscillator good for?

[Cornutt, David K]

From: Andre Majorel [mailto:amajorel at teaser.fr]
> OK. Here's a sample that sounds like what I think you mean :
> 
>   http://www.teaser.fr/~amajorel/misc/qpsk.mp3
> 
> The carrier is a 1 kHz sine. The modulator is swept from 1 Hz up.
> The modulator does not directly appear in the output, except that
> every time it goes through a cycle, the phase of the carrier is
> increased by 90°.

Wow.  That is really interesting.  In the beginning you
can hear the pops from the discrete phase changes while the
modulation is in the sub-1Hz range, and then it starts to
sound kind of like an AM mod sweep, but not quite.
I've never heard anything quite like that before.  Thanks.

> As the frequency of the modulator goes above the carrier
> frequency, the spectrum gets pretty nasty ; the aliasing madness
> your can hear from 18" on survived x1000 oversampling.

Yes, I don't think that noise has anything to do with
the sampling rate; it isn't "aliasing" in the Nyquist sense.
I think it's a mathematical consequence of the modulation
freqency exceeding the carrier frequency; this causes the
carrier and modulation to "exchange places", sort of
(as the primary lower sideband goes past 0 Hz).  I've
seen the same basic thing in some of my AM experiments.

I was thinking of an experiment setup using a four-valued
modulator which would select one of the 0, 90, 180, or 270
phases for each step of the modulation.  I'm going to
try some Csound runs this weekend, but I was also thinking
of an analog setup, perhaps involving a quadrature oscillator
(with inverting and non-inverting outputs) and something
like a 4-in-1-out switch where the modulation selects one
of the four outputs at any given time.  It could be done,
for example, with some level comparators and two MOTM-700
routers (the four inputs into the X and Y halves of the first
one, and the second one selects either the X or Y output
of the first one).  

Another possibility, more in the analog domain, is the
related concept of quadrature amplitude modulation (QAM).
Here you have your in-phase (I) and quadrature (Q) outputs,
with each one going into a VCA.  There are two modulation
signals.  The catch here is that, to really do it the way
the radio people do it, the VCAs have to have the ability
to go "past zero" and output an inverted signal with gain
if the control voltage is negative; this way the phase
angle that results with the I and Q signals are combined
can sweep through all four quadrants.  (Right off the top
of my head, I can't think of an off-the-shelf VCA that
does that.)

I've got an old telemetry textbook, from the days of analog
telemetry systems, that describes some oddball modulation
techniques that I've never heard of in any other context.
I'll have to go read that book some more and do some studying.
Definitely time to do some experiments this weekend!