Hi Florian, If you pass the arg "--ref external" to tx_waveforms, does it resolve this frequency offset?
https://github.com/EttusResearch/uhd/blob/master/host/examples/tx_waveforms.cpp#L62 Regards, Nate Temple On Thu, Dec 6, 2018 at 12:22 AM Florian Kaltenberger via USRP-users < usrp-users@lists.ettus.com> wrote: > Hi Marcus, > > I have measured this with a spectrum analyzer simply by setting markers to > the peak of the sinusoid (one marker per measured USRP) and then taking the > delta. > > Could it be that the USRP is ignoring the external reference when used > alone? Remember, I am doing the test with one USRP at a time, as the test > using multiple USRP simultaneously does not work. > > Florian. > On 06/12/2018 00:29, Marcus Müller wrote: > > oh! That means 341 ppb frequency error, which *really* shouldn't be > happening. > > I'd like to get some statistics of that error, how are you measuring > it? > > Best regards, > Marcus > > On Wed, 2018-12-05 at 12:55 +0100, Florian Kaltenberger wrote: > > Sorry typo. I did use a frequency of 3.51GHz. > > > On 5 Dec 2018, at 12:54, Marcus Müller <marcus.muel...@ettus.com> > <marcus.muel...@ettus.com> > wrote: > > Hi Florian, > > trying to get my head to understand the order of problems here: > Could you try to use a higher frequency (say, --freq 2e9 instead of > 3.5e6)? > I'd thing 3.51 MHz is out of range for the N310, anyway? > > Best regards, > Marcus > > On Wed, 2018-12-05 at 11:49 +0100, Florian Kaltenberger via USRP- > users > wrote: > > So I can confirm that there is a frequency offset between the two > USRP N310s when using only an octoclock (10MHz + PPS) to > synchronize. > I have measured with the tx_waveforms program > ./tx_waveforms --args > "addr=192.168.x.2,time_src=external,clock_src=external,master_clo > ck_r > ate=122.88e6" --rate 122.88e6 --freq 3.51e6 --wave-type SINE -- > wave- > freq 10e6 --gain 100 > on the first USRP N310 (x=10) and then on the other (x=20). There > is > an offset of 1200Hz between the peaks of the sinusoids between > the > two measurements. > Using an external LO didn't change anything either. Unless I need > to > provide any other arguments in that case? > I also tried to do a test where I use both USRPs simultaneously > ./tx_waveforms --args > "addr0=192.168.10.2,addr1=192.168.20.2,time_src=external,clock_sr > c=ex > ternal,master_clock_rate=122.88e6" --rate 122.88e6 --freq 3.51e6 > -- > wave-type SINE --wave-freq 10e6 --gain 100--channels "0,4" > but unfortunately that does not work at all at my testbench > (program > hangs and no signal transmitted). > My UHD version is 3.13.0.2 (UHD_3.13.0.HEAD-0-g0ddc19e5) > Any help appreciated. > Thanks! > Florian. > > > On 04/12/2018 21:29, Florian Kaltenberger via USRP-users wrote: > Hi Marcus and Robin, > thanks for your answers, this is helpful information. I should > add, > that I actually tried the synchronization with an octoclock > (10MHz > + PPS), but it did not give me the expected results, i.e., I > saw > some residual frequency offsets. But maybe I screwed up at some > point. Let me do some more measurements and get back to you on > this. > Florian. > > > On 04/12/2018 18:57, Marcus D. Leech via USRP-users wrote: > On 12/04/2018 10:14 AM, Florian Kaltenberger via USRP-users > wrote: > > Hi there, > I just discovered that in addition to the external 10MHz > reference in, the USRP N310 also has external local > oscilator > inputs, one for each daughterboard and each TX/RX. So does > that > mean that in order to synchronize multiple N310 in > frequency, > phase, and time, it is no longer sufficient to use an > octoclock > to provide a 10MHz reference and PPS? If so, at what > frequency > do you have to drive the external LOs and at what power? > Florian. > > > In addition to what Robin posted, I'll observe that the > external > LO port is an *additional feature* of this device. > > You should still be able to use the external 10MHz and 1PPS > ports > the same way you would with a B210 or E310, since the AD9371 > front-end chip is similar to the AD9361 chip used in the > B210 > and E310. > > The thing about synchronizing multiple independent PLL > synthesizers, though, compared to a strictly-shared-LO, is > that > the former will > experience both phase ambiguities, and have a higher mutual > phase-noise than the latter, which is why you might decide to > choose > the latter. > > > > > _______________________________________________ > USRP-users mailing > listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > -- > Follow us on Google+, LinkedIn, or Twitter! > > > _______________________________________________ > USRP-users mailing > listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > -- > Follow us on Google+, LinkedIn, or Twitter! > _______________________________________________ > USRP-users mailing > listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > -- > Follow us on Google+ <https://plus.google.com/+OpenairinterfaceOrg>, > LinkedIn <https://www.linkedin.com/company/openairinterface>, or Twitter > <https://twitter.com/osalliance5g>! > _______________________________________________ > 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
