For posterity here is a solution I arrived at.


I was ignoring the md.start_of_burst and md.end_of_burst fields. In order to 
send many contiguous transmit requests without receiving erroneous late errors, 
send the first transmit request in a chain with md.has_time_spec true, 
md.start_of_burst true. Send middle packets in a chain with md.has_time_spec 
false and SOB/EOB false. Send the final packet with has_time_spec false, and 
EOB true.



From: Richard Joseph Muri
Sent: Tuesday, June 2, 2020 1:59 PM
To: '[email protected]' <[email protected]>
Subject: Sending multiple contiguous transmit requests



Hello,



I've been having an issue with the USRP where I've found if I have a continuous 
waveform segmented over multiple packets, it is difficult to transmit it on the 
USRP. The scenario is as follows:



I am using a USRP x310 with UHD version:

[INFO] [UHD] linux; GNU C++ version 4.8.5 20150623 (Red Hat 4.8.5-36); 
Boost_105800; UHD_3.13.0.3-24-g00019176



Packet 1 has 100 samples. I do a send starting at time 0 with 100 samples. 
Packet 2 has 50 samples. I do a send starting at time 100 with 50 samples. 
Packet 2 is marked late and doesn't send properly.



Packet 1 has 100 samples. I do a send starting at time 0 with 100 samples. 
Packet 2 has 50 samples. I do a send starting at time 101 with 50 samples. 
Packet 2 sends properly, but now I have a sample delay inserted into the middle 
of my waveform.



Packet 1 has 100 samples. I do a send starting at time 0 with 99 samples. 
Packet 2 has 50 samples. I do a send starting at time 101 with 50 samples. 
Packet 2 sends properly, but now I have a blank sample inserted into the middle 
of my waveform.



The obvious solution is to reassemble my fragmented packets before making a 
tx_streamer->send() call. I'm wondering if there is a way to replicate the 
behavior of a single large transmit request using multiple smaller transmit 
requests?



Thank you,

Richard

_______________________________________________
USRP-users mailing list
[email protected]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Reply via email to