Hi Marcus, thanks for the reply. So if I don't want this function (precisely timed bursts) but i only want bursty transmission, I should only use the length tag, right?
My understanding of bursty transmission is that the USRP does not wait for the buffers to fill up: when it sees the length tag, it waits for appropriate number of samples (value of length tag) to arrive and transmits this burst immediately (sends it to DAC). Furthermore, this mode of transmission will eliminate long delays (order of 5-10 seconds depending on the sampling rate) that I see when i change some aspect of my tx-ed signal to appear on the RX side. The way I explain these long delays is that the USRP TX buffer should fill up before actual transmission occurs. Is my understanding correct? thanks again, Achilleas On Fri, May 7, 2021 at 11:19 AM Marcus D Leech <[email protected]> wrote: > I can answer at least one of your questions. > > The tx_time tag is used for TDMA type applications where the burst > *timing* is critical. > > You send the burst a little in advance with the time tag, and the USRP > won’t commence transmission until that time. > > Sent from my iPhone > > On May 7, 2021, at 10:19 AM, Achilleas Anastasopoulos <[email protected]> > wrote: > > > I posted the following on discuss gnuradio list but I am also posting here > in case this is more appropriate. > --------------------------------------------------------- > Hi all, > > I am reading from here: > > https://wiki.gnuradio.org/index.php/USRP_Sink > > how to do burt transmission with the usrp_sink. > My questions have to do with the second option: > > ---- > Using tagged streams (See Tagged Stream Blocks > <https://wiki.gnuradio.org/index.php/Tagged_Stream_Blocks>). To use this > capability, you must specify which string the usrp_sink block should be > looking for to denote the length of the next PDU to be transmitted. Set the > "tsb_tag_name" parameter in the usrp_sink to whatever string your radio > application uses to denote your PDU length. A commonly used string for this > purpose is simply "tx_pkt_len". If using Tagged Streams for timed bursts, > you must include your "tx_pkt_len" tag and a "tx_time" tag on the first > sample of a tx burst. If your first "tx_pkt_len" tag has an offset of 0, > and your packet length is 1000 items, your next "tx_pkt_len" and "tx_time" > tags must appear with an offset of 1000. TX bursts should not overlap, and > there should not be gaps in samples between bursts. > ---- > > Q1: why do we need two tags to make this happen? I would think that > "tx_pkt_len" tag would be sufficient: when a tag like this is found by the > usrp sink then it waits for that many samples to come in and then it > transmits them in burst mode. Then it waits for the next such tag and so on. > What is the meaning/use of the second tag "tx_time"? > > Q2: Does the "tx_time" tag have to be exactly that name (as opposed to the > "tx_pkt_len" tag which can be user defined?) > > Q3: The text above is talking about PDU's but my understanding is that a > PDU is a special type of a PMT, while the USRP sink input has to be a > tagged stream, not a PMT/PDU. Should the input be a PDU (to a message > port?) or should it be a tagged stream? > > thanks > Achilleas > > _______________________________________________ > USRP-users mailing list -- [email protected] > To unsubscribe send an email to [email protected] > >
_______________________________________________ USRP-users mailing list -- [email protected] To unsubscribe send an email to [email protected]
