Thank you for all the insights and hints. As we compile UHD from source, we can definitely change some low-level code as you proposed. Indeed, I already see some sort of RTT bias clustering with the two B210, though I have not yet enough measurement runs to make a very clear statement. Also, with two B210, any of both, and either Tx or Rx chain could influence the RTT bias, so the clustering is already a mix of two "hidden" system states.
I will test the following setup next, to more clearly identify the clustering: one X310 with CBX front-end and one B210, referenced with the same atomic clock signal. The X310 is so stable that we can use it as reference. I'll keep you updated on this. All the best, Emanuel -----Ursprüngliche Nachricht----- Von: Marcus D. Leech <patchvonbr...@gmail.com> Gesendet: Donnerstag, 10. Februar 2022 02:56 An: Ian Buckley <ian.buck...@gmail.com>; Staudinger, Emanuel <emanuel.staudin...@dlr.de> Cc: Rob Kossler via USRP-users <usrp-users@lists.ettus.com>; Julian Arnold <jul...@elitecoding.org> Betreff: Re: [USRP-users] B200 group delay variations On 2022-02-09 18:36, Ian Buckley wrote: > Quick thoughts. Since you are RTT here, the Tx and Rx RF sections of AD9361 > are driven by different RF fractional-n synthesizers, they might lock > run-to-run with different phase relationships to their shared common > reference. If you see clustering at certain (similarly spaced) time bins in > your result that might be a hint towards this being involved. I hadn't thought about that, but, yes, that *could* be a source of the variability. I'm not sure whether the AD9361 supports integer-N tuning mode, but for sure UHD doesn't expose it in the driver... It occurred to me that as RF synthesizers end up being used almost exclusively in DSP-based systems, there's little reason for fractional-N tuning, as far as I can tell--it was invented to give very-small step sizes when you're using a larger reference clock into the PLL. But with these RF chains being backed a lot more by flexible DSP architectures, where you can effectively achieve sub-1Hz tuning step size, the synthesizers don't really need the very fine step size. But that's just me waxing philosophical... _______________________________________________ USRP-users mailing list -- usrp-users@lists.ettus.com To unsubscribe send an email to usrp-users-le...@lists.ettus.com
