Hello everyone In my scenario I do continuous transmissions with one X300 USRP at a rate of 184.32 MSPS (UHD 3.15). Initially I used a timestamp only for the first transmission to align data with PPS, but now I want to re-align also after some rarely occurring underflows.
My approach was to add a time_spec to each tx_streamer::send() call: Each send() transmits 10ms of data, and accordingly for each call the time_spec is incremented by 0.01s. What I now see is that eventually an underflow will happen, and afterwards all further transmissions lead to “L” events; the transmission never recovers. I also tried to set the transmission time further in the future to allow recovery when the event_code shows an underflow, but this doesn’t solve the issue. I could image that I need to work with start_of_burst / end_of_burst flags, timeout settings, etc., but tbh the documentation isn’t very clear on that topic. Further observation: I’ve added time measurement to my tx loop and noticed that also the time_spec set for the first transmission doesn’t seem to work. The first transmission gets a time_spec 2 seconds in the future, subsequent transmission increment this time_spec by 10ms. But on execution the first four transmissions take about 10ms, the fifth transmission takes about 1.5 seconds and all subsequent 10ms again. What I expect is a 2 second delay for the first transmission. So I really seem to have some serious issues with timed transmission. Any help, explanation of the mechanism or hint in the right direction is welcome. Max
_______________________________________________ USRP-users mailing list -- [email protected] To unsubscribe send an email to [email protected]
