Rob / Marcus, The initial-5GHz LO was my problem. I was not providing it at startup. Thank you guys for the info. The mirroring is resolved!
-Scott On Thu, Nov 3, 2022 at 8:48 AM Marcus D. Leech <patchvonbr...@gmail.com> wrote: > On 03/11/2022 09:36, Rob Kossler wrote: > > When running N310 with external LO, the device will automatically try to > run a calibration at device startup (actually in the software make N310 > device) where it expects the external LO to be at 5 GHz. If you supply an > LO at a different frequency, this calibration will be ruined. Is it > possible that you are not supplying a 5 GHz LO at startup? Note that you > can change it after startup to your desired freq of operation. > > And, responding to another discussion point about QEC correction, even if > you disable the "tracking cal" you can keep the "init cal" in place. I > assume this means that an initial calibration is performed and then the > calibrated gain/phase settings are left untouched afterwards. So, > disabling the tracking cal should not mean that you are not performing any > correction - just not updated dynamically. > Rob > > Thanks, Rob. I had, once again, forgotten about the initial-5GHz > requirement. > > The complete list of possible "init" and "tracking" CAL options is here: > > https://files.ettus.com/manual/page_usrp_n3xx.html#n3xx_mg_calibrations > > > On Wed, Nov 2, 2022 at 4:28 PM Marcus D. Leech <patchvonbr...@gmail.com> > wrote: > >> On 02/11/2022 15:53, Kenneth Burchfield wrote: >> >> Marcus, >> I agree. Is there a way to tune the sideband suppression so that I can >> attenuate more of the mirrored signal? >> With the internal lo, everything seems to just work. Are there extra >> configuration steps that I need to take to adjust for the external lo? >> >> I'm trying to find that out now. >> >> I suspect that spending an afternoon with the AD9371 datasheet might >> provide some clues--but as to whether those are >> actually exposed in UHD is another question. >> >> With the internal LO, the chip has internal algorithms for optimizing >> phase and amplitude balance. Those algorithms aren't >> "in play" when you use an external LO, apparently. >> >> One thing that *could* be done, but is tedious, is to adjust the phase >> and amplitude balance of the baseband signal to >> compensate for the imbalance of the mixer. This is basically what the >> internal algorithms in the chip are doing, as >> far as I can tell. >> >> >> >> Thanks, >> Scott >> >> On Wed, Nov 2, 2022 at 2:41 PM Marcus D. Leech <patchvonbr...@gmail.com> >> wrote: >> >>> On 02/11/2022 15:27, Kenneth Burchfield wrote: >>> >>> Marcus, >>> >>> The sidebands do not change with the LO level. >>> >>> I am basically running the tx_waveforms example from UHD 4.2.0.0, and I am >>> adding the argument tx_lo_source=external. Attached is a picture of the >>> spectrum. I am generating a SINE with wave-freq input 2e6. The center spike >>> is some LO leakage. The spike at -2e6 is the mirroring I am seeing. >>> >>> Thanks, >>> Scott >>> >>> The LO suppression looks to be about 55dBc, which is not horrific. >>> >>> The sideband suppression appears to be only about 30dBc, which is not >>> very good. >>> >>> >>> >>> On 02/11/2022 12:09, Scott Burchfield wrote: >>> >>> Hi, >>> >>> >>> Are there any updates on this problem. I normally try not to reply to old >>> threads, but I am seeing the same the same spectral mirroring Rob described >>> earlier. My setup is also the same as Rob’s.I have tracked some newer >>> threads that mentioned disabling TX_QEC_INIT, but I am still seeing the >>> mirroring after disabling QEC. >>> >>> https://www.mail-archive.com/usrp-users@lists.ettus.com/msg07489.html >>> >>> Thanks, >>> >>> Scott >>> >>> >>> It seems to me that if you use an external LO, that LO has to go through the >>> phase-splitter on the mixer, and that phase-split is subject to >>> errors, so turning off QEC may not be that desirable--although I think >>> the mixer on the N310 (using an AD9371) >>> >>> is a 2XLO, so the phase error should be very very small. >>> >>> Do the unwanted sidebands change with LO level? >>> >>> >>> >> _______________________________________________ >> USRP-users mailing list -- usrp-users@lists.ettus.com >> To unsubscribe send an email to usrp-users-le...@lists.ettus.com >> > >
_______________________________________________ USRP-users mailing list -- usrp-users@lists.ettus.com To unsubscribe send an email to usrp-users-le...@lists.ettus.com
