Re: [Discuss-gnuradio] Synchronising USRP2's in GRC.

[Douglas Geiger]

 For the old(er) USRP2 driver (i.e. in the current released version of
GNURadio) - the C++ block supports the functionality to synchronize
the clock, and they are available in python as well, but I don't
believe the GRC block exposes this functionality. It would be
relatively trivial to add a drop-down in the block configuration to
select a clock reference. Setting the clock time is a little trickier,
but is do-able in GRC. All the modifications need to be done in the
usrp2_sink_xxxx.xml and usrp2_source_xxxx.xml files (

 I've been taking a look at doing this with the new UHD driver (i.e.
for the next GNURadio release). The low-level UHD driver supports
synchronizing the LO to the reference clock, and setting the time
according to the PPS (or just according to when you tell it to via the
set_time_now()), however the gr-uhd block does not yet.  Adding this
functionality is relatively trivial for the C++ code, but I've run
into a problem with SWIG-izing the calls (for use with GRC) due to the
UHD's use of boost::uint32_t in the time_spec_t structure. As I'm not
a SWIG expert, any advice on how to deal with this would be
appreciated. I do expect to be able to share the patch when I get it
working.

 The particular problem I run into is that time_spec_t is defined to
want a boost::uint32_t in the constructor (for the number of seconds),
along with a double (for fractional seconds) - but SWIG apparently
doesn't recognize boost::uint32_t is a typedef for an integer type, so
while I don't get compile-time errors, by the time I fire up python to
test out the function call (set_type_next_pps(const uhd::time_spec_t
&time_spec) in this case) - and try to construct a time_spec_t object
to pass, python throws an error as it doesn't believe I'm passing the
correct type.

 Doug

On Mon, Jun 21, 2010 at 7:02 AM, Zohair <address@hidden> wrote:
>
> Hi everyone,
>
> I need to use several USRP2's in the same GRC programme (MIMO). Is there a
> way to sychronise them to the same clock in GRC itself?
>
> Best,
>
> Zoh
> --

-- 
Doug Geiger
address@hidden