Decoding FSK modulated signal

GNU Radio
1

Hey,

I’ve some home automation equipment here using the sub-ghz band that I
managed to capture with a RTL dongle. From the data sheet and the
register settings of the TI CC1100 transceiver chip that is used inside
the device I figured out the RF parameters used, i.e. 102kHz channel
bandwidth, 2-FSK modulation, 19.2k baudrate, syncbyte etc.

Attached is a screenshot of a GRC waveform I used to capture the
preamble of a single packet. Channel 2 is simply the output of a
quadrature demod block connected to a scope GUI block. I’d like to
decode that into a sequence of bits to run preamble detection etc.

I’ve seen a few OOT modules that do similar stuff but was wondering if
there is a GR’ish way to this, maybe a set of generic blocks that could
be used for that?

Thanks
Andre

[1] http://i61.tinypic.com/35mjm08.png

2

On Mon, Apr 14, 2014 at 01:25:48PM +0200, Andre P. wrote:

Attached is a screenshot of a GRC waveform I used to capture the
preamble of a single packet. Channel 2 is simply the output of a
quadrature demod block connected to a scope GUI block. I’d like to
decode that into a sequence of bits to run preamble detection etc.

I’ve seen a few OOT modules that do similar stuff but was wondering if
there is a GR’ish way to this, maybe a set of generic blocks that
could be used for that?

It looks like you are on the right track! For the next steps, I
recommend a low pass filter (with perhaps 20 kHz cut-off) followed by
M&M clock recovery and binary slicer. Check out the recent “ASK
demodulation help” thread for more details. Try setting the decimation
of the LPF to something that gives you 2 to 5 samples per symbol.

Your graph looks like the preamble is at about 1.92 kbps, not 19.2 kbps,
so I suspect that you have an error somewhere in your sample rate
configuration.

Mike

3

On 04/14/2014 05:40 PM, Michael O. wrote:

It looks like you are on the right track! For the next steps, I
recommend a low pass filter (with perhaps 20 kHz cut-off) followed by
M&M clock recovery and binary slicer. Check out the recent “ASK
demodulation help” thread for more details. Try setting the decimation
of the LPF to something that gives you 2 to 5 samples per symbol.

Hi Mike,

thanks! That’s exactly what I was looking for. I think I have to play
around with the synchronizer settings but the results are already quite
promising.

I’ll keep you updated.

Thanks again
Andre