Re: [Discuss-gnuradio] multiple sources synchronization ?
Hi Jean-Michel! There's nothing in your system that would make both RTL dongles and the soundcard start sampling at the same time, so naturally there's a large time offset between these. You will need to time align these signals first, before you can use the sound card signal to determine in which state (reference or unknown signal) your switch is; this is at least as hard as the problem of aligning the RTL dongles themselves. To be honest, I'd rather write an estimator that tells me, only from the signal, whether each RTL dongle is observing the reference or the signal. How do you frequency-synchronize both dongles? Have you modified them to use a common oscillator, or do you also plan to do that based on the observation of the 50MHz tone? Best regards, Marcus On 22.07.2015 09:51, jean-michel.fri...@femto-st.fr wrote: Possibly a stupid question, but might help me better understand how the gnuradio scheduler works: my objective is to make a low cost phase-referenced radiofrequency interferometer using two DVB-T dongles. Since I have observed that the PLL inside each dongle induces slow phase drift, I want to use an external RF switch to monitor a known (50 MHz) reference oscillator feeding both dongles, and then monitor the unknown signal. Current switching rate is about 50 Hz triggered by an external generator (which also synchronizes other events of the experiment, not relevant to this post). My idea for synchronizing the post-processing of phase extraction was to record on the one hand the two DVB-T dongle data flow (this I know works), and on the other hand the sound card microphone connected to the switch trigger signal. This process is summarized in the grc flowchart at http://jmfriedt.sequanux.org/damien_grc.png However, the sound card output shows a result completely out of sync with the phase measurements. I understand that the sound card and DVB-T dongles do not share the same clock sources, but considering the huge decimation factor (48*32 kHz for the DVB-T, 48 kHz for the sound card, and a phase output recorded at about 0.5 to 5 kHz, not shown on this grc chart), I would have expected the trigger signal to be more or less synchronized with the DVB-T outputs, which is not at all the case. Is the gnuradio scheduler unable to interleave two data sources as different as a sound card and the USB data flow from the two DVB-T dongles ? Is there a way I might tune my flowchart to achieve the expected result ? Thanks, JM ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] multiple sources synchronization ?
There's nothing in your system that would make both RTL dongles and the soundcard start sampling at the same time, so naturally there's a large time offset between these. You will need to time align these signals first, before you can use the sound card signal to determine in which state (reference or unknown signal) your switch is; this is at least as hard as the problem of aligning the RTL dongles themselves. To be honest, I'd rather write an estimator that tells me, only from the signal, whether each RTL dongle is observing the reference or the signal. right, this is what I ended up doing yesterday: since the reference and measured signals exhibit different power, I am recording the phase difference as well as the magnitude from the dongle connected to the switch, and use the magnitude information to differentiate the two states of the switch. How do you frequency-synchronize both dongles? Have you modified them to use a common oscillator, or do you also plan to do that based on the observation of the 50MHz tone? Yes I removed the quartz from one dongle and am using the 28 MHz output from the other dongle to clock the former. I believe that the slow phase slip is not important for passive radar applications since the cross-correlation reinitializes the phase, but for my interferometric measurement I need the long term phase drift information. Thanks for the answer, JM Best regards, Marcus On 22.07.2015 09:51, jean-michel.fri...@femto-st.fr wrote: Possibly a stupid question, but might help me better understand how the gnuradio scheduler works: my objective is to make a low cost phase-referenced radiofrequency interferometer using two DVB-T dongles. Since I have observed that the PLL inside each dongle induces slow phase drift, I want to use an external RF switch to monitor a known (50 MHz) reference oscillator feeding both dongles, and then monitor the unknown signal. Current switching rate is about 50 Hz triggered by an external generator (which also synchronizes other events of the experiment, not relevant to this post). My idea for synchronizing the post-processing of phase extraction was to record on the one hand the two DVB-T dongle data flow (this I know works), and on the other hand the sound card microphone connected to the switch trigger signal. This process is summarized in the grc flowchart at http://jmfriedt.sequanux.org/damien_grc.png However, the sound card output shows a result completely out of sync with the phase measurements. I understand that the sound card and DVB-T dongles do not share the same clock sources, but considering the huge decimation factor (48*32 kHz for the DVB-T, 48 kHz for the sound card, and a phase output recorded at about 0.5 to 5 kHz, not shown on this grc chart), I would have expected the trigger signal to be more or less synchronized with the DVB-T outputs, which is not at all the case. Is the gnuradio scheduler unable to interleave two data sources as different as a sound card and the USB data flow from the two DVB-T dongles ? Is there a way I might tune my flowchart to achieve the expected result ? Thanks, JM ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio -- JM Friedt, FEMTO-ST Time Frequency/SENSeOR, 26 rue de l'Epitaphe, 25000 Besancon, France ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
[Discuss-gnuradio] multiple sources synchronization ?
Possibly a stupid question, but might help me better understand how the gnuradio scheduler works: my objective is to make a low cost phase-referenced radiofrequency interferometer using two DVB-T dongles. Since I have observed that the PLL inside each dongle induces slow phase drift, I want to use an external RF switch to monitor a known (50 MHz) reference oscillator feeding both dongles, and then monitor the unknown signal. Current switching rate is about 50 Hz triggered by an external generator (which also synchronizes other events of the experiment, not relevant to this post). My idea for synchronizing the post-processing of phase extraction was to record on the one hand the two DVB-T dongle data flow (this I know works), and on the other hand the sound card microphone connected to the switch trigger signal. This process is summarized in the grc flowchart at http://jmfriedt.sequanux.org/damien_grc.png However, the sound card output shows a result completely out of sync with the phase measurements. I understand that the sound card and DVB-T dongles do not share the same clock sources, but considering the huge decimation factor (48*32 kHz for the DVB-T, 48 kHz for the sound card, and a phase output recorded at about 0.5 to 5 kHz, not shown on this grc chart), I would have expected the trigger signal to be more or less synchronized with the DVB-T outputs, which is not at all the case. Is the gnuradio scheduler unable to interleave two data sources as different as a sound card and the USB data flow from the two DVB-T dongles ? Is there a way I might tune my flowchart to achieve the expected result ? Thanks, JM -- JM Friedt, FEMTO-ST Time Frequency/SENSeOR, 26 rue de l'Epitaphe, 25000 Besancon, France ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] multiple sources synchronization ?
The use the signal itself method is something that I've used in the past for Dicke-switched type systems. The idea was originally described by Ken Tapping of the DRAO observatory. The gr-ra_blocks OOT module that I designed includes an Oblivious Slicer block that will produce a difference value based on a block of samples, using oblivious slicing. https://github.com/patchvonbraun/gr-ra_blocks The technique works well for well-differentiated signals (large difference between reference and sky), but not otherwise. On 2015-07-22 04:01, Marcus Müller wrote: Hi Jean-Michel! There's nothing in your system that would make both RTL dongles and the soundcard start sampling at the same time, so naturally there's a large time offset between these. You will need to time align these signals first, before you can use the sound card signal to determine in which state (reference or unknown signal) your switch is; this is at least as hard as the problem of aligning the RTL dongles themselves. To be honest, I'd rather write an estimator that tells me, only from the signal, whether each RTL dongle is observing the reference or the signal. How do you frequency-synchronize both dongles? Have you modified them to use a common oscillator, or do you also plan to do that based on the observation of the 50MHz tone? Best regards, Marcus On 22.07.2015 09:51, jean-michel.fri...@femto-st.fr wrote: Possibly a stupid question, but might help me better understand how the gnuradio scheduler works: my objective is to make a low cost phase-referenced radiofrequency interferometer using two DVB-T dongles. Since I have observed that the PLL inside each dongle induces slow phase drift, I want to use an external RF switch to monitor a known (50 MHz) reference oscillator feeding both dongles, and then monitor the unknown signal. Current switching rate is about 50 Hz triggered by an external generator (which also synchronizes other events of the experiment, not relevant to this post). My idea for synchronizing the post-processing of phase extraction was to record on the one hand the two DVB-T dongle data flow (this I know works), and on the other hand the sound card microphone connected to the switch trigger signal. This process is summarized in the grc flowchart at http://jmfriedt.sequanux.org/damien_grc.png [1] However, the sound card output shows a result completely out of sync with the phase measurements. I understand that the sound card and DVB-T dongles do not share the same clock sources, but considering the huge decimation factor (48*32 kHz for the DVB-T, 48 kHz for the sound card, and a phase output recorded at about 0.5 to 5 kHz, not shown on this grc chart), I would have expected the trigger signal to be more or less synchronized with the DVB-T outputs, which is not at all the case. Is the gnuradio scheduler unable to interleave two data sources as different as a sound card and the USB data flow from the two DVB-T dongles ? Is there a way I might tune my flowchart to achieve the expected result ? Thanks, JM ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio [2] Links: -- [1] http://jmfriedt.sequanux.org/damien_grc.png [2] https://lists.gnu.org/mailman/listinfo/discuss-gnuradio ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio