Hi Piotr, A couple of comments - regarding setting the dboard_clock_rate, is this needed? I don't know what the value should be for the CBX so my suggestion would be to eliminate this argument so that UHD uses default settings (unless you know this is a good value) - regarding the "master_clock_rate" and the "rate", your command line shows an incompatibility of these 2 parameters. The rate should be an integer division (preferably some multiple of 2 or even better 4) of the master_clock_rate - so perhaps 25e6 or 33.3e6. Alternatively, it was previously possible to set the master_clock_rate to 184.32 MHz so that you could get a rate of 30.72 MHz, but I'm not sure if this is still supported. Of course, this all depends on your sample rate assumption in your waveform file.
I'm not sure that either of these will help, but wanted to pass it along. Rob On Thu, Mar 5, 2020 at 5:32 AM Piotr Gawłowicz via USRP-users < [email protected]> wrote: > Hi, > > Recently, I have started playing with the Matlab WLAN toolbox. So far, I > have managed to generate a waveform of 802.11n frame, transmit it using > USPR x310 with UBX-160 daughterboard and successfully receive the frame > using commercial-off-the-shelf (COTS) hardware - Intel Wireless-AC 9260 > NIC. The setup also works perfectly when using USRP B205mini. > > However, when I transmit the waveform using CBX-120 daughterboard (in the > same x310 USRP), is not received by the COTS Intel NIC. I want to play with > MIMO and I need to use CBX-120 daughterboard as we have 4 of them and only > one UBX-160. I tested all of our 4 CBX-120 daughterboards with the same > result. > > I use exactly the same command in case of CBX-120 and UBX-160 (both in > same x310 USRP so I change subdev parameter), ie.: > ./tx_samples_from_file > --args="type=x300,addr=192.168.10.2,master_clock_rate=200e6,dboard_clock_rate=50e6" > --subdev A:0 --type=short --rate=30e6 --freq 2412e6 --gain=25 --repeat > --delay=1 --file 80211n_waveform.dat > > I have noticed that Connection Type is different in the case of CBX-120 > and UBX-160, i.e. IQ vs QI, and generated waveforms accordingly in MATLAB, > but without success. > > I connected USRP RF outputs to WiFi NIC using RF cable (with 45dB > attenuation), but still, the same effect, i.e. UBX-160 works, CBX-120 does > not work. > In the case of UBX-160, the RSSI of the received frame equals -55dBm (as > reported by NIC), so it is still a pretty good link. > > After running the calibration procedure (as described here > https://files.ettus.com/manual/page_calibration.html), the CBX-120 > started working somehow, i.e. some frames are received but not really > reliably (only few out of tens transmitted frames are received). In the > case of UBX-160, all frames are received correctly (of course over the > cable). > The frame is generated with MCS 0 (BPSK, and code rate 1/2), so it is very > robust and can be received with SNR of 3dB. > > I have tested also a couple of the UHD versions and hence FPGA images, but > with the same result. > Can someone explain what is the difference between CBX-120 and UBX-160 > daughterboards that cause the described behavior? Should I set some more > parameters to overwrite default ones in the case of CBX-120 to achieve > better results? > > Best Regards, > Piotr Gawlowicz > _______________________________________________ > USRP-users mailing list > [email protected] > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >
_______________________________________________ USRP-users mailing list [email protected] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
