Richard,

Doppler DF systems like mazerom is thinking about work by rotating a
virtual antenna in a circle -- by successively selecting one antenna at
a time from a set of antennas around a circle. (He finally got around
to saying that in the other thread...) The effect is to ADD an FM
modulation to whatever signal you're receiving. If the signal happens
to be a continuous carrier, then you end up receiving a signal with
just your antenna's FM modulation. If the transmitted signal was FM to
begin with, your FM adds in. If it was an AM signal with carrier,
you're still good to go. But if the signal looks like random
noise--does FM modulated random noise not still look like random noise?

If you have reasonably clean FM to demodulate, then you just need to
calibrate out the (audio) phase shifts of your detector, including
whatever filtering you use to extract just the modulation your antenna
put on the signal. Then you compare the phase of that detected
modulation with the phase of the antenna effective rotation to get
bearing. This is similar, in a sense, to how VOR works to give a
_receiver_ in an airplane the bearing that receiver is from the VOR
transmitter: the VOR effectively rotates a directional antenna very
quickly so that the receiver sees an AM signal: max envelope when the
directional transmission "points at" that receiver. But the VOR
transmitter ALSO is FM modulated. So all receivers tuned to that VOR
get the same FM phase reference, but the phase of the AM is different
at each receiver depending on its bearing relative to the transmitter.
In the Doppler DF, the reference phase is the antenna rotation, which
you are controlling, and then the FM phase relative to that depends on
the bearing of the arriving signal. In your receiver system, you'll
want to use a very narrow filter to reject other modulation, and the
trick then is to make the phase shift in that filter be very constant
at the frequency of interest. A switched-capacitor
(commutating-capacitor) filter will do that, if the commutation is
driven synchronously with the antenna switching. But you can also do
it with a digital filter which samples at a rate that's synchronous
with the antenna switching.

There's an article in QST -- for May, 1978 if I'm remembering correctly
-- that describes it, along with plans for constructing such a beast
for 2 meters. Anyone want mine? I tried to give it away a couple
years ago, but no takers. Worked OK last time I used it.

Cheers,
Tom