[USRP-users] Sequence Errors on USRP N200

Josh Blum josh at ettus.com
Sun Jun 24 19:08:01 EDT 2012

On 06/24/2012 03:41 PM, Jahshan Bhatti wrote:
>> On 06/23/2012 09:59 PM, Jahshan Bhatti wrote:
>>> Hi all,
>>> I'm writing a program to delay and playback received samples continuously
>>> with the USRP N200. Everything seems to be working fine, except for the
>>> occasional sequence error. Why does this happen? Is the network layer
>>> dropping or reordering packets?
> Regarding my delay and playback app, how well does the USRP obey the
> time_spec commands? I tried to delay the samples by one second, and it
> seems like the actual delay is off by around 4 ms (and the delay is not
> consistent). I need within 50 nanosecond precision for my application.

On tx, the timespec is converted to an absolute tick count. When the
USRP gets a packet with a timestamp, it waits waits until that time in
the packet matches its current time. All subsequent data is
backpressured the timestamp'd packet in the front.

Since N200 has a 100 MHz clock, the timespec can specify a clock cycle,
which is on the order of 10 nanoseconds.


More information about the USRP-users mailing list