Re: [Discuss-gnuradio] Multi-rtl - making multi-channel receiver out of multiple RTL-SDR dongles

[Juha Vierinen]

In my testing, this phase rate difference seemed constant and I could simply calibrate it out by looking at the phase rate term estimated from the phase of the cross-correlated noise. The samples stay aligned for hours, so the issue is caused by the tuner or the DDC. One theory that I had was that the multi-rtl driver somehow sets up the dongles differently, but I never got around to looking at the code. 

I think this was the script I used to figure out what the phase drift was:

https://github.com/jvierine/chirpsounder/blob/master/apps/passive_radar/rnoise.py

Here's the IQ plot to prove that the cross correlated phase is stable over 6000 seconds:

http://kaira.sgo.fi/2013/09/16-dual-channel-coherent-digital.html

juha

On Wed, May 25, 2016 at 12:14 PM, <address@hidden> wrote:

From our own experience with dual-dongle measurements, the phase drift
seems to be strongly related to R820T(2) temperature. We reduced significantly
the phase drift by gluing a large heat sink common to both chips on both
dongles, without completely removing this effect (we aim at measurements lasting
multiple hours). At the moment the only option we could think of (mail to this
mailing list dated 28 May 2015) is switching to a reference clock to calibrate
for the phase difference between the local oscillators, but the actual cause
of the drift remains a mistery (probably due to the implementation of the PLL,
but I cannot understand why Phase Locked Loops would drift in Phase !).

JM

> It wouldn't be as simple as it was for me as a developer and as it
> (hopefully) is for the end user without your hardware mod.
>
> Can you say something more about the residual center frequency
> difference? Where might it come from? I prepared little test of
> coherency between the receivers (multi-rtl/examples/test_multirtl.m).
> Among all the figures that it shows there is a plot of relative phase
> offset of signals coming from the receivers. In fact I have seen linear
> phase change on that plot - that corresponds to some central frequency
> offset. If I know what is the source of this offset maybe I will be able
> to find some way to fix it in software.
>
> --
> Piotr
>
> W dniu 25.05.2016 o 08:24, Juha Vierinen pisze:
> > This is awesome! I'll definitely try this out soon. I use one off
> > python scripts to find the sample offset and the small residual center
> > frequency difference. This simplifies the process significantly.
> >
> > This should make it much easier to implement a passive radar block, or
> > an interferometry block.
> >
> > juha
> >
> > On Tue, May 24, 2016 at 4:03 PM, Piotr Krysik <address@hidden
> > <mailto:address@hidden>> wrote:
> >
> >     Hi all,
> >
> >     I want to announce new GNU Radio related project prepared by me -
> >     Multi-rtl:
> >     https://github.com/ptrkrysik/multi-rtl
> >
> >     It is a Gnu Radio block that combines multiple RTL-SDR receivers into
> >     one multi-channel receiver.
> >
> >     Only hardware modification to RTL-SDR dongles required is connecting
> >     them to a common clock source (i.e. one of the dongles' oscillator as
> >     Juha Verinen showed once). Each channel can work on a different
> >     central
> >     frequency.
> >
> >     Everyone who wants to know how it was achieved is invited to read my
> >     github page:
> >
> >     https://ptrkrysik.github.io
> >
> >     Best Regards,
> >     Piotr Krysik
> >
> >     _______________________________________________
> >     Discuss-gnuradio mailing list
> >     address@hidden <mailto:address@hidden>
> >     https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
> >
> >
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

--
JM Friedt, FEMTO-ST Time & Frequency/SENSeOR, 26 rue de l'Epitaphe, 25000 Besancon, France