[USRP-users] Synchronized tx/rx

Martin Braun martin.braun at ettus.com
Wed Jul 5 17:40:05 EDT 2017


On 06/27/2017 05:29 AM, ROBIN TORTORA via USRP-users wrote:
> Windows 7, UHD 3.9.2, X310 with UBX daughtercard.
> 
> 
> I have an app that does time syncd tx/rx on the same daughtercard.
> 
> 
> When I schedule a timed transmit, what does the time spec actually
> represent?
> 
> 
>     Is it the time that samples start through the Tx chain?

This depends. On your UHD version, yes, it's when it starts going into
the Tx chain. On newer UHDs, it's the time it gets sent to the DAC.

> Similarly on the recv side, what does the time spec actually represent?
> 
> 
>     Is it the time samples start coming through the Rx chain?

Same answer, in reverse.

> Here is my issue:  I have "calibrated" out the delay of the loopback by
> sending a test pulse through tx chain starting the tx at time T1 while
> simultaneously starting the Rx chain at T1 also.  Lets call the measured
> delay 10 samples, i simply remove 10 samples form the beginning of the
> rx data and continue on merrily.
> 
> 
> 80% of the time, everything is completely syncd perfectly for the length
> of the Tx/Rx.
> 
> 
> 20% of the time, the rx is off by 1 sample, so the delay is actually 11
> samples rather than 10.  Everything is syncd, just off by 1 sample.
> 
> 
> This is executed with the same freq and sample rate.
> 
> 
> Is there anything organically "wrong" with my approach?

It looks sensible. What's your CPU rate (i.e. what's your
interpolation/decimation)?

-- M

> Is there anything I can do differently to eliminate the 20% of the time
> 1 sample mismatch?
> 
> 
> It seems I am straddling some time bin and wondering if a different
> approach would more stable and looking for any suggestions...
> 
> 
> Thanks,
> 
> 
> 
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
> 





More information about the USRP-users mailing list