Hi Folks,
Following on from the posts over the last few days on compiling
GNURadio for the Raspberry Pi, and having just taken delivery of
mine, I thought I would give it a go.
Having installed Raspian, I managed to get UHD and GNURadio compiled
and installed. This took the better part of 24 hours! Wasn’t really
expecting too much, but setup a simple flow graph as follows:
USRP Source —> UDP Sink
A remote machine with a UDP source was showing a WX GUI waterfall.It
managed 250k with no under or over flows. I added a second part to the
flow graph as follows:
TCP Source —> USRP Sink
The remote system had a null source sending data to the PI, and again
I found that I had no under or over flows, except for the odd glitch
(mainly) at the start, and if I tried to do other things with the Pi
at the same time.
I did try a sample rate of 500k, but got underflows quite severely,
especially when running “Full Duplex”. Wasn’t surprised to be honest.
While “only” 250k, This is more than wide enough for a lot of things,
especially within the Amateur Radio domain.
One reason I was so interested in this test is to determine the
feasibility of putting a URSP1 up at the mast head.
With the Pi, and a Fibre <-> Copper Gigabit media converter (for
lightning mitigation), all the RF gubbins can be up the top, and
(virtually) no co-ax loss, with all the heavy duty SDR work being
done by more powerful computer(s) on the ground
(Yes, I know the N2xx and E1xx series have LAN connectivity. They also
only have one transmit and one receive slot, and still need the media
converter for lightning mitigation)
And yes, I would love a USRP device with two transmit and receive
slots, and fibre LAN connectivity rather than copper 
I’ve only tried it with one RX and one TX chain so far. I suspect dual
RX and TX may introduce some under and overflows, even at the minimal
250k sample rate. Possibly dropping to 8 bit samples might help here,
I’ll have to experiment.
Anyhow, just thought other folks might find my results interesting.
Iain