[USRP-users] Still problems with USRP time from GPSDO

Josh Blum josh at ettus.com
Mon Apr 29 14:37:44 EDT 2013



On 04/29/2013 07:10 AM, Sivan Toledo wrote:
> I think that my previous message was wrong, but I still have problems with
> the GPSDO times from the USRP. Here is a printout from a UHD program that
> queries the time in 4 different ways:
> 
> usrp->get_time_now(): 1367237083:0.517574
> usrp->get_time_last_pps():  1367237082:1.000000
> metadata time stamp of recent RX samples:   1367237083:0.510384
> utime() on Linux:   1367237084:0.517167
> 
> As you can see there is a difference of a whole second between the time on
> the PC, which is derived from NTP using ntpd and should be accurate to at
> least 10ms or so, and the times on the USRP.
> 
> The USRP is an N200 with a GPSDO fitted.
> 
> We've experienced errors by whole seconds in two different units. Is this a
> bug in the firmware or image?
> 

Hey,

Sorry for the trouble.

Can I ask what version of UHD you are running? A few weeks ago I fixed a
bug on the master branch related to GPSDO time, it sounds roughly like
you describe. If you are running a stale version of master, that could
potentially be the cause. However, if this is code from maint or a
release then Im not sure.

Can you send the output of
<install prefix>/share/uhd/utils/query_gpsdo_sensors

I think that will be revealing.

-josh




More information about the USRP-users mailing list