PMR446 Receiver in GNU Radio Companion

GNU Radio
1

Hello,

I have build an PMR446 Receiver in GNU Radio Companion. I can detect the
channel and also demodulate it, but the quality of the receiver is not
that much good. I want to amplify my received signal, Anyone have
suggestion how to do it.

Best Regards,
SAJJAD SAFDAR
MSc Communication Engineering

2

On 12/20/2012 12:52 PM, Sajjad S. wrote:

MSc Communication Engineering

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page
You don’t say what type of USRP you’re using – is a sample rate of 256k
supported on your USRP of choice?

Also, your RF is set to 0, perhaps increase this some and see if things
improve?

3

On 20/12/12 17:52, Sajjad S. wrote:

I have build an PMR446 Receiver in GNU Radio Companion. I can detect the
channel and also demodulate it, but the quality of the receiver is not
that much good. I want to amplify my received signal, Anyone have
suggestion how to do it.

If I read your flowgraph correctly. Your sample rates are not aligned

You have 256k coming out of your USRP [See Martin’s reply on this]
You then decimate that by 5, which gives 51.2k, but your NBFM receive
is expecting 44k1.

You then feed that 44k1 out to a standard squelch that is expecting to
be fed at 32k, which you then send onto an audio sink expecting 44k1,
and a EX GUI FFT sink expecting 256k

And as Martin mentioned, you have 0 db Gain. Increase that to 20 or 25
for testing.

I’m actually surprised that it managed to detect and demodulate anything
with those values, although your basic flowgraph is fine. Do you not get
lots of underflows and overflows, as well as a slowly updating FFT ?

If it were me, I’d grab 250k of B/W off the USRP, Use a LPF instead of
a BPF, Decimate by 5 (to 50k), then use a rational resampler to get to
44k1 (or just get the rational resampler to go from 250k to 44k1), then
do the NBFM receive

Note NBFM receive block can do some resampling, but it has to have
the quadrature rate divisible by the audio rate (eg 88k2 and 44k1 are
fine, but 44k1 and 250k are not), so you might want a rational
resampler block that will do just about any two B/Ws, although takes
more CPU

Iain