[USRP-users] Start of Burst/End of Burst question

Josh Blum josh at ettus.com
Wed Sep 11 15:07:53 EDT 2013



On 09/11/2013 06:47 AM, Wallace, Frank L CIV NSWCDD, Q41 wrote:
> In the meta data, what do the "start of burst" and "end of burst"
> flags tell the underlying system/radio?
> 
> I am trying to send out a pulse train.  Currently, I do a "start of
> burst = true", and "end of burst = true" and "has time spec = true"
> and schedule each pulse at it's own time and send() command.  Should
> "start of burst = true" only for the first pulse and 'end of burst=
> true" only for the last pulse?
> 

A burst is a contiguous stream of samples. So burst flags are really
about sample continuity. Basically, end of burst tells the DSP chain
that no more samples will be expected, you can go into idle after
processing my samples. That start of burst is actually not processed by
the FPGA. Start is instead inferred by samples arriving into the DSP
when the DSP is currently idling.

So, the first call to send() for this contiguous stream should have
start set, the last call should have end set. If the entire burst is
done with one call, then set both start and end.

> Each pulse is about 10K samples which is way over the packetsize.


Thats ok. The driver fragments the buffer into packetsized transfers to
send out. In this case, the start flag only gets applied to the first
fragment packet, the end flag the last, and time is actually
interpolated onto each fragment.

-josh

> 
> Thanks, Frank



> 
> 
> 
> _______________________________________________ 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