<html>
<body>
At 10:32 PM 23/08/2006, jeff millar wrote:<br><br>
<blockquote type=cite class=cite cite><blockquote type=cite class=cite cite><pre>OK.
I can see that if all txs/rxs are fed from the same oscillator then the
fact that any modulation is possible, including phase, suggests that it
can
be done.
I'd need to understand how a constant "DC" phase shift is
represented in I/Q.
</pre><font face="Courier New, Courier"></font></blockquote>Imagine a
local oscillator (LO) mixing with a carrier (RF), the output (IF)
contains the difference between the two frequencies. If the LO and
RF happen to have the same frequency, then the IF comes out at 0 Hz, or
DC. The actual DC voltage depends on the signal amplitudes and
phase of the LO vs RF. When exactly in phase the output has the
maximum positive value, when exactly out of phase the maximum negative
value, and when at 90 degrees the output has zero value. Measuring
the output voltage provides a hint of the RF phase, but two different
phases can produce the same output (ambiguity).<br><br>
Now route the RF to two mixers and use two LOs 90 degrees out of
phase. The two IFs become "I" and "Q", each the
same IF but now they provide enough information to determine the RF phase
without ambiguity.<br><br>
In general, I and Q provide two copies of the same IF, but 90 degrees out
of phase. This enables a number of signal processing tricks such as
canceling one sideband to create SSB. It also enables "zero
IF" receivers commonly used in SDR.</blockquote><br>
I have a rudimentary understanding of QSD. My question was really a more
specific one: what is the mathematical function applied to I and Q in
software that will produce a constant phase shift on the carrier that it
modulates? I wasn't really expecting anyone to answer. When I have a
better understanding of the math (I'm working on it) then I'll be able to
answer it myself. Something doesn't smell right though - something about
frequency response of the I/Q output. Anyway, I'll get over it.<br><br>
<br>
<blockquote type=cite class=cite cite><blockquote type=cite class=cite cite><pre>I
think an outboard A/D->memory->D/A unit would have some buyers
though.
Nothing to do with SDR but it uses the knowledge gained.
</pre><font face="Courier New, Courier"></font></blockquote>What
do you mean by ADC-memory-DAC?</blockquote><br>
Sample the RF, put it into memory, wait, take it out of memory, then D to
A (eg LT2208/FPGA/DDR RAM/Some D/A). There's going to be a minimum shift
caused by the processing delay but that will cancel out if all of
the signals have the same circuit. I can't see this working at very high
frequencies due to the amount of memory required and its speed. Could be
feasible at HF though, where antenna gain means unmanageable
size.<br><br>
<blockquote type=cite class=cite cite><blockquote type=cite class=cite cite><pre>Calibration
would be an interesting exercise.
</pre><font face="Courier New, Courier"></font></blockquote>The
arrays may not need calibration. Some papers on "blind signal
separation" and "principle component analysis" discuss how
to separate signals using randomly placed, uncalibrated antenna
arrays. You don't know which direction the signals come from, just
how to separate them from each other.</blockquote><br>
Interesting. You need to know the direction for transmitting though. Also
if you are looking for a signal from a specific direction (eg a beacon)
and no others.<br><br>
Murray VK6HL</body>
<br>
</html>