Always wondered what the state-of-the-art is in oscillator drift compensation; I bet
there's more to it than estimating a second derivative of one LO's phase against the other
and throwing Kalman at it until it doesn't move anymore, or whether there's better models...
but you're right. If this is more than "a one-shot measurement immediately after
calibration", you'll need to share a reference clock; even an extremely nice frequency
error of $2\cdot 10^{-8}$ between the two receivers would, at 1 GHz carrier, mean that the
phase of one rotates away from the other by $40 \pi$ every second, so you have like 10 ms
until you lose any sensible idea of direction. I don't know how well the calibration
routine is able to compensate such small frequency errors, it might actually be OK for
longer, or it might not.
Best regards,
Marcus
On 01.09.23 22:48, Marcus D. Leech wrote:
On 01/09/2023 16:39, Marcus Müller wrote:
Ah, nice! yeah, then this would work – but! any oscillator has drift, so two local
oscillators will not stay on the same phase for long; if your first couple measurements
are correct, but suddenly your directions become very wrong, that's what I'd investigate.
Cheers,
Marcus
Looks like the BB60C supports an external 10MHz reference clock, which would be utterly
mandatory for anything requiring
mutual phase-coherence, whether that's *sufficient* depends very very much on the
internal architecture of the BB60C.
On 01.09.23 19:16, Michael Berman wrote:
The gr-aoa module has a calibration phase where you connect a source with a splitter
and roughly equally lengthened cables to the two sources and it cross-correlates the
two inputs to determine an initial phase difference. Then, while the script is still
running, you disconnect the calibration source and connect the antenna's, again with
roughly equally lengthened cables to run the MUSIC algorithm.
Thank you very much,
Michael Berman
On Fri, Sep 1, 2023 at 11:00 AM Marcus Müller <marcus.muel...@ettus.com
<mailto:marcus.muel...@ettus.com>> wrote:
If you don't know the relative phase of your two receiver chains, how are you
going to
know the direction of a signal?
On 01.09.23 17:37, Michael Berman wrote:
Marcus,
Thanks for the reply! I apologize, the GR package is gr-aoa, not gr-music,
and can
be found here (https://github.com/MarcinWachowiak/gr-aoa
<https://github.com/MarcinWachowiak/gr-aoa>). The NOAA broadcast is a NBFM
signal. I am using 2 Signal Hound receivers (https://signalhound.com/products/bb60c/
<https://signalhound.com/products/bb60c/>) with 2 antennas spaced on a
beam. Do I
need to synchronize the 2 receivers with an external clock, or should they
be fine
free running independently?
Thank you very much,
Michael Berman
On Fri, Sep 1, 2023 at 9:25 AM Marcus Müller <mmuel...@gnuradio.org
<mailto:mmuel...@gnuradio.org>> wrote:
Not familiar with the details of NOAA signalling, but isn't the carrier
of an FM
signal
*the FM signal*?
For a DoA estimate, you'd correlate the different receive chains with
each other
to get a
phase; so, as long as the signals do have some bandwidth that makes the
problem
less
ambiguous, it'd work with any signal. I'm sadly not familiar with
gr-music (and
can't find
it on cgran.org <http://cgran.org>), but MUSIC works as long as the
signals at
the different receive antennas
are correlated and noise is not. You do not have to preprocess your FM
signal!
Best,
Marcus
On 01.09.23 17:09, Michael Berman wrote:
> Does anybody know if there is a way to recover a carrier of an FM
signal to
use for an
> Angle of Arrival calculation? I am using GNURadio and gr-music and I
am
trying to use the
> NOAA Weather Radio signals.
>
> Thank you very much,
>
> Michael Berman