The trick would be to oversample. I'm not sure how you would do that with just UHD, but with GNU Radio, you would just use a rational resampler block. That is, setting the sample rate to 20 Msps and using a 2X resample ratio. Ron On 7/17/24 08:38, Brais Ares Fernández wrote: > I've just tested exactly the same setup but using a *B210*, and it > does *not *happen. > > So it definitely seems to be what Ron pointed out. > > @Marcus: yes, the signal used is sampled at 10e6 and I can see its > spectrum in Matlab and it's a clean OFDM. Actually we tried one OFDM > signal modulated by us, and other completely different one by 5GMAG > <https://github.com/5G-MAG/rt-mbms-modem> and that effect always appear. > > Then... Using N210, is there no way that I can filter out that > aliasing stuff? > > > ------------------------------------------------------------------------ > *De:* Ron Economos <w6rz@comcast.net> > *Enviado:* miércoles, 17 de julio de 2024 16:05 > *Para:* Brais Ares Fernández <bares@gradiant.org>; > usrp-users@lists.ettus.com <usrp-users@lists.ettus.com> > *Asunto:* Re: [USRP-users] Re: Replicas out of band in OFDM signal > (N210+CBX) > > Yeah, I think UHD sets it to some calculated value that can't be > overwritten. > > > BTW, here's what the spectrum looks like with a very wide analog > filter. So the default bandwidth setting on the N210/CBX is just a > little too wide to fully suppress that first alias. > > > Ron > > > On 7/17/24 06:39, Brais Ares Fernández wrote: >> Hello, Ron. Thank you for the answer. >> >> I've already tried but it seems to do nothing: >> >> $ sudo /lib/uhd/examples/tx_samples_from_file --freq 1500e6 --rate >> 10e6 --gain 20 *--bw 10e6* --type float --file ~/OFDMsignal.raw >> --args="name=sdr1" >> >> I even tried weird values like 100e3 or 100e6, but no apparent effect >> over the spectrum at all. >> ------------------------------------------------------------------------ >> *De:* Ron Economos via USRP-users <usrp-users@lists.ettus.com> >> <mailto:usrp-users@lists.ettus.com> >> *Enviado:* miércoles, 17 de julio de 2024 15:00 >> *Para:* usrp-users@lists.ettus.com >> <mailto:usrp-users@lists.ettus.com> <usrp-users@lists.ettus.com> >> <mailto:usrp-users@lists.ettus.com> >> *Asunto:* [USRP-users] Re: Replicas out of band in OFDM signal >> (N210+CBX) >> >> Those are aliases. Try setting the analog bandwidth with the --bw >> parameter. >> >> >> Ron >> >> >> On 7/17/24 03:58, Brais Ares Fernández wrote: >>> Hello, >>> >>> This might be a common effect but I don't comprehend why it happens >>> and if it is avoidable. >>> >>> I'm using a N210+CBX to transmit a 10 MSPS *OFDM signal*, with Fc = >>> 1.5 GHz. When transmitting, I can see some kind of "replicas" at >>> +-7.5 MHz from Fc, always with a 40 dBc difference. >>> >>> $ sudo /lib/uhd/examples/tx_samples_from_file --freq 1500e6 --rate >>> 10e6 --gain 20 --type float --file ~/OFDMsignal.raw --args="name=sdr1" >>> >>> >>> I've already tried changing lo_offset and executing the calibration >>> utilities for IQ imbalance, but none of that seems related to the >>> problem. >>> >>> Could anyone provide any insight? >>> >>> Regards. >>>