[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Random behavior?

From: Tom Rondeau
Subject: Re: [Discuss-gnuradio] Random behavior?
Date: Thu, 11 Nov 2010 10:05:35 -0500

On Mon, Nov 8, 2010 at 6:01 PM, Jakub Moskal <address@hidden> wrote:
> Hi,
> We are experimenting with two USRP1's (each with RFX2400) and noticed
> some strange behavior. We use the
> /usr/local/share/gnuradio/examples/digital/benchmark_rx[tx].py. We run
> it once and the number of corrupted/lost packets is rather small, say
> 0-2%. Without modifying any default values of parameters, or touching
> the USRP's, we start it again and the number goes up to, say 40%. We
> run it again and the number of lost packets is about 90% and those
> that make it through are mostly corrupted. It doesn't always occur in
> this pattern, sometimes we get good ratio for a few runs of the
> benchmark and then it goes down again. Sometimes it no packets are
> received at all. All within the same parameters and physical positions
> of the radios. We tried different values of Tx amplitude and Rx gain,
> but radios keep behaving in this way. We attached the FFT sink on the
> receiver side and see that the signal is coming in, but packets are
> still mostly lost after a few runs, sometimes even after the first
> run.

I had never noticed that behavior before. Usually, if it's in a
position to get most of the packets most of the time, it's pretty
stable. My guess is that there's some loop that's not quite getting
the signal and sometimes locking and sometimes not.

Two things to try. See how far off frequency you are and adjust either
the transmitter or receiver to compensate for any gross frequency
offsets. The loops should take care of the rest if you're near it.

The second thing is to use benchmark_tx/rx2.py. These are the newer
versions of the digital mods/demods and should work better.

> What could be the reason for this behavior? What is the optimal setup
> for two USRP's -- in terms of physical distance and values of
> parameters?
> Thanks,
> Jakub

Unfortunately, you're just going to have to experiment with the
optimal setup for you. When developing, I probably had the 2 USRPs
spaced by about 6 - 10 feet in an office. But without real AGC on the
front-end, there's not much you can do by play with the power and gain


reply via email to

[Prev in Thread] Current Thread [Next in Thread]