Suomi100 satellite beacon not decoding with gr_satellites CLI

[mikaelnousiainen]

I've stored several samples from the Suomi100 satellite in 32-bit I/Q format (50k sample rate) and tried to simply check if I've got any valid data. The gr_satellites CLI won't detect any packets in any of my samples. I've attached two of the packets with strongest signals (in a ZIP file) here if you'd have any suggestions for me: suomi100-samples.zip

My first question is: what does the demodulator even expect to find in I/Q data? Should the center frequency of the transmission be at baseband - or at some other frequency - or does it matter at all? Is my SNR simply too low?

I used Airspy R2 and a hand-held Arrow II antenna for tracking/RX.

I'm running the command:

gr_satellites "Suomi 100" --iq --rawfile suomi100-test-5-50000hz-float32.iq --samp_rate 50000

[K4KDR]

Thanks for posting your IQ files! Looking at them in Inspectrum, just based on experience they don't look QUITE strong enough to decode a 9600 baud signal:

... but it never hurts to try. Sample-5 was a bit stronger than #4, and you can see a bit of symbol separation (the lower graph below). However, 9600 baud signals can be quite particular and even when you think "just maybe", they won't decode without sufficient SNR.

Not to state the obvious, but at least at first glace it appears that we need stronger RX either by an improvement in tracking, antenna gain, an LNA right at the antenna, or all three.

Just to confirm, are the settings in the .yml file still correct for this satellite?

  9k6 FSK downlink:
    frequency: 437.775e+6
    modulation: FSK
    baudrate: 9600
    deviation: 2400
    framing: AX100 ASM+Golay

[mikaelnousiainen]

@K4KDR Thanks for confirming my suspicion, I actually ended up doing exactly the same kind of experiment with the GNU Radio FSK demodulator as you - just before looking at your answer. :)

I think I'll definitely need an LNA -- and possibly a better antenna too!

AFAIK the parameters are correct, nothing should have changed with the satellite transmissions.

[mikaelnousiainen]

I can confirm I was able to decode a single packet with an LNA and a different kind of antenna. Thanks for quick answers here!

[daniestevez]

The FSK demodulator, which is what is used here, requires the signal to be at baseband (see #159). The packets shown in @K4KDR's screenshot have a frequency error that is large enough that the demodulator won't work.

[mikaelnousiainen]

Aha, thanks, that's an important fact then! Maybe something worth adding to the docs of demodulators!

[mikaelnousiainen]

I did a test and mixed the center of the signal to baseband in all my samples with the same result - the FSK demod results are not good enough. Time to figure out some potential upgrades to the RX hardware :)

[K4KDR]

I did a test and mixed the center of the signal to baseband in all my samples with the same result - the FSK demod results are not good enough. Time to figure out some potential upgrades to the RX hardware :)

Your replay method is correct. It's particularly handy to configure the 'BFO' as a variable slider which is used as the frequency of the signal that you multiply against your I/Q sample. That allows you to continuously loop a small range of packets and adjust on-the-fly to move the observed signal to zero. As has already been said, that allows the remainder of the flowgraph to do whatever is required to extract the payload.