The RFNoC Replay block would be a good starting point, if you want to do
this all in the FPGA.
On Thu, Jul 18, 2019 at 2:04 PM Marcus Müller via USRP-users <
usrp-users@lists.ettus.com> wrote:
> At the very benign 20 MS/s, I'd really say your first option is the way
> to go. The rest probably won't work very well du to turn on/off
> behaviour requiring you to zero pad a bit to flush your TX data chains.
>
> You can of course also write an RFNoC block to store and generate data
> in-FPGA, but really: at 20MS/s just continously stream. Even a bog-
> normal Gigabit Ethernet card has plenty enough bandwidth to do that. I
> doubt sending a sequence from RAM will occupy much CPU on your host PC.
>
> Best regards,
> Marcus
>
> On Thu, 2019-07-18 at 22:58 +0200, Cédric Hannotier via USRP-users
> wrote:
> > Dear all,
> >
> > I would like to periodically send a frame with an USRP X310 (frame
> > length: 320 samples, rate: 20 MS/s, period: 1-500 ms). However, I
> > struggle to find the best way to implement it. What I have tried so
> > far:
> >
> > 1. Append zeros to the frame to reach the expected period.
> > However,
> > this consumes too much bandwidth due to the zeros.
> >
> > 2. Use tx_streamer->send() with a tx_metadata_t.time_spec and
> > tx_streamer->recv_async_msg(). Using that, only the frame is sent,
> > saving most of the bandwidth. However, with small periods, it tends
> > to
> > print some 'L'.
> >
> > 3. Buffer a batch of send request on the USRP, then wait a
> > specific
> > time (using eg. recv_async_msg() until the returned metadata
> > contains
> > the penultimate time_spec (I expect that the time_spec returned is
> > the
> > one specified in the send metadata)) and redo. The issue is that I
> > was
> > not able to find the buffer size (is it related to the
> > tx_streamer->get_max_num_samps()?). I would like to fill the buffer
> > without overflow.
> >
> > I was hoping that I could save the frame in an USRP's memory, and
> > then
> > ask it to periodically send the frame with a specific period. Is it
> > possible?
> >
> > Here is an example of (2):
> >
> > template <typename samp_type>
> > void send_from_file(const uhd::usrp::multi_usrp::sptr &usrp,
> > uhd::tx_streamer::sptr tx_stream, const
> > std::string&
> > file,
> > const double period)
> > {
> > size_t data_size = get_file_size<samp_type>(file);
> > std::ifstream infile(file, std::ifstream::binary);
> > std::vector<samp_type> buff(data_size);
> > infile.read(reinterpret_cast<char*>(buff.data()),
> > (std::streamsize)(buff.size()*sizeof(samp_type)));
> > infile.close();
> > size_t num_tx_samps = buff.size();
> > std::cout << file << " " << buff[0] << " " << num_tx_samps <<
> > std::endl;
> >
> > uhd::tx_metadata_t md;
> > md.start_of_burst = true;
> > md.end_of_burst = true;
> > md.has_time_spec = true;
> > md.time_spec = usrp->get_time_last_pps()+5.;
> > double timeout = md.time_spec.get_real_secs();
> > uhd::async_metadata_t md_status;
> >
> > while (not stop_signal_called) {
> > tx_stream->send(&buff.front(), num_tx_samps, md);
> > if (tx_stream->recv_async_msg(md_status, timeout)) {
> > if (md_status.event_code !=
> > uhd::async_metadata_t::event_code_t::EVENT_CODE_BURST_ACK) {
> > std::cerr << "Error: " << md_status.event_code
> > << std::endl;
> > exit(EXIT_FAILURE);
> > }
> > } else {
> > std::cerr << "timeout before sent" << std::endl;
> > exit(EXIT_FAILURE);
> > }
> >
> > timeout = 0.1;
> > md.time_spec += period;
> > }
> > }
> >
> >
> >
> > Best Regards,
> > Cédric
> >
> > _______________________________________________
> > USRP-users mailing list
> > USRP-users@lists.ettus.com
> > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
>
> _______________________________________________
> USRP-users mailing list
> USRP-users@lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
