Usrp_multi and the USRP2

GNU Radio
1

Is there any code similar to usrp_multi.py to use multiple USRP2’s
already (or is it unnecessary)? As I understand it, with the USRP2, you
either use the MIMO cable, or an external reference to synchronize the
clocks - then on the host side you would still need to do the
align_on_samplenumbers/deinterleaving, etc. business that usrp_multi.py
took care of for the USRP1.
If nobody has code available to do this, should I expect any major
hurdles in trying to convert the usrp_multi.py for use with the USRP2
(besides converting the source create calls, removing the subdev calls,
etc)?
If I’m using the MIMO cable, do I need to make any additional settings
(e.g. sync_to_pps(True) or similar) to synchronize the master to slave?
http://gnuradio.org/trac/wiki/USRP2GenFAQ says the VCXO locks on to the
external reference, if one exists - which implies this is also the case
for the MIMO cable, just wanted to verify that was true.
Looking at test_mimo_tx.cc in the u2-mimo-wip branch I see it is
possible to setup one usrp2 interface, and send samples to different
channels - is that accessible from python now - or is it still code
that’s in development? If that’s the case, is it possible at all to get
synchronized samples out of two usrp2’s with a MIMO cable right now with
python, or should I be writing something in C++ to do that for me?
Perhaps
Thanks,
Doug

–
Doug G.
Research Assistant
Communications and Signal Processing Lab
Oklahoma State University
http://cspl.okstate.edu
[email protected]
[email protected]

2

On Tue, Feb 03, 2009 at 02:33:15PM -0600, Douglas G. wrote:

etc)?
python, or should I be writing something in C++ to do that for me?
Perhaps
Thanks,
Doug

Doug,

There’s currently no production code that supports using the USRP2 and
the MIMO cable. We have test code that confirms that the h/w works as
we expect, but we’re not going to sort this out until after “the great
reworking” of the host, firmware and on-the-wire spec as part of our
move toward VITA-49 (VITA Radio Transport), an emerging standard for
digital IF.

In the meanwhile, you can make this work by providing your own
star-distributed 10 MHz reference and PPS. All USRP2’s will be
coherent and will share the same timestamps. You will have to invoke
config_mimo(MC_WE_LOCK_TO_SMA) and sync_to_pps() on each USRP2.

Eric

3

Eric B. wrote:

There’s currently no production code that supports using the USRP2
and

Eric

Understood. Looking around at code that uses config_mimo (which looks
like it’s only available to C++ apps right now) - I see rx_samples.cc
uses it, but it looks like that’s fallen out of use, and is a bit crufty
now. Any reason I couldn’t modify rx_streaming_samples.cc to call
config_mimo?
Thanks,
Doug

–
Doug G.
Research Assistant
Communications and Signal Processing Lab
Oklahoma State University
http://cspl.okstate.edu
[email protected]
[email protected]