Hi,

I played around with the subdev specs. They are all alike. I found the 
bug. For some reason GRC generated flowgraphs lack a call to
`self.uhd_usrp_source_0.set_gain(gain, 0)` whenever the gain variable is 
changed. This was not a huge problem for X310s because they perform 
quite nicely with RXgain=10 in my setup. But for the N310s, I needed to 
set RXgain=40 before I generate the flowgraph because the gain callback 
is missing.
I will update GR and open a new issue if the bug persists.

Thanks for all your help!

Johannes



On 22.10.19 17:38, Robin Coxe wrote:
> The subdev spec mappings for the N310 are in the getting started guide 
> on the Ettus KB.
> 
> 
>           UHD 3.13.x.x+
> 
>   * RF0 = A:0
>   * RF1 = A:1
>   * RF2 = B:0
>   * RF3 = B:1
> 
> 
> Have you tried simple sanity checks such as inputting a single tone into 
> each N310 RX chain and progressively increasing the gain and plotting 
> with ascii FFT?   (We did this analysis extensively during V&V testing, 
> as indicated on the N310 datasheet, so unless your system is configured 
> incorrectly or has a HW problem, you shouldn't see what you're seeing.)  
> If you think there's a HW problem with your unit, I'd suggest contacting 
> NI for an RMA.
> 
> -Robin
> 
> 
> On Tue, Oct 22, 2019 at 3:16 AM Johannes Demel <de...@ant.uni-bremen.de 
> <mailto:de...@ant.uni-bremen.de>> wrote:
> 
>     Hi Robin and Marcus,
> 
>     Thanks for your hints.
> 
>     UHD 3.14.1.1 work perfectly fine with my X310s. Only the N310s seem to
>     have problems. I use the default HG FPGA image that ships with 3.14.1.1.
> 
>     For the moment, I use `VERT2450` antennas. The ones you can order from
>     Ettus. I transmit/receive @2.484GHz. I mounted antennas on all antenna
>     ports. So for the N310s that's a total of 8 antennas per device.
>     So far I did not specify a `Subdev Spec` in GRC.
> 
>     I tried a few more options. Also, I unmounted the antennas on the RX2
>     ports on the N310s and made sure the `TX/RX` antenna port is configured.
> 
>     Here are my observations:
>     1. X310 -> X310 everything works fine TXgain=20, RXgain=10.
>     2. N310 -> X310 everything works, though I need to raise TXgain=38 a
>     bit.
>     3. X310 -> N310 works but it seems like sensitivity is still bad. I
>     need
>     to raise TXgain=70. RXgain seems to not make a difference at all.
>     4. N310 -> N310 bad performance but transmission is possible. Need to
>     raise gains to almost maximum for bad performance.
> 
>     This is the N310 setup code generated by GRC. I tried different `Subdev
>     Spec`s for both TX and RX. I obtained the best results with `B:0`
>     for TX
>     and RX. Though, `best` is still way worse than X310 performance.
>     uhd.usrp_source(
>           ",".join(("addr=X.X.X.214", "")),
>           uhd.stream_args(
>               cpu_format="fc32",
>               args='',
>               channels=[],
>           ),
>     )
>     self.uhd_usrp_source_0.set_subdev_spec('B:0', 0)
>     self.uhd_usrp_source_0.set_center_freq(tfreq, 0)
>     self.uhd_usrp_source_0.set_gain(rgain, 0)
>     self.uhd_usrp_source_0.set_antenna('TX/RX', 0)
>     self.uhd_usrp_source_0.set_samp_rate(samp_rate)
> 
>     Since Option 2 N310 -> X310 does yield good results, I assume that the
>     N310 RX chain or configuration is the problem. Though, I have no idea
>     how to debug that and which knobs to turn.
>     I tried to observe the spectrum with an N310 via `uhd_fft` from
>     `gr-uhd/examples/grc` and it feels like the N310 just doesn't case
>     about
>     RXgain settings.
> 
>     Johannes
> 
> 
>     On 18.10.19 20:42, Marcus D Leech wrote:
>      > What antenna are you specifying and which port are you rurally
>     plugged in to.
>      >
>      > Are you using offset tuning?  That may be necessary for Narrow
>     signals near the tuned frequency, due to DC offset removal.
> 
> 
> 
>     On 18.10.19 16:24, Robin Coxe wrote:
>      > Hi Johannes.  That low response from the N310 sounds fishy.  Have
>     you
>      > tried another channel besides 0 on your N310?  It might be that:
>     1) the
>      > TX and RX cables are connected backwards (note that the order on the
>      > panel is funky due to layout constraints with the AD9371 RF IC on
>     the
>      > daughter cars. 2) signals are actually on the adjacent channel—
>      > double-check the UHD subdev mappings listed in the Getting
>     Started Guide
>      > because if memory serves they changed at some point— or 3) or the
>     SMA
>      > connector center pin connections might be cracked.
>      >
>      > Also, I’m not sure how compatible GR3.8 is with UHD 3.14.1.
>        Someone
>      > who still works on Ettus products mentioned at GRCon that you
>     have to
>      > use master-next branches, but those use the new RFNoC, which is
>     still
>      > largely uncharted territory with the N310 as far as I can tell.
>      >
>      > -Robin
>      >
>      >
>      >
>     ------------------------------------------------------------------------
>      > *From:* USRP-users <usrp-users-boun...@lists.ettus.com
>     <mailto:usrp-users-boun...@lists.ettus.com>> on behalf of
>      > Johannes Demel via USRP-users <usrp-users@lists.ettus.com
>     <mailto:usrp-users@lists.ettus.com>>
>      > *Sent:* Friday, October 18, 2019 4:06 AM
>      > *To:* usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>
>      > *Subject:* [USRP-users] N310 sensitivity
>      > Hi all,
>      >
>      > I figured out how to use our new N310s.
>      >
>      > I ran into the next issue. Let me describe this one briefly.
>      >
>      > With our X310s I set TXgain=20 and RX_gain=10. Both devices are 1-2m
>      > apart. I observe a really nice RX constellation with gr-gfdm and
>      > XFDMSync with bursts every 1ms (burst length ~50us). With our N310s I
>      > set RXgain=60, TXgain=70 and still, the constellation is very noisy.
>      > My assumption is that some part of the RX chain is not configured
>      > correctly. But I don't know what the problem is in particular.
>      > Also, I used an X310 as transmitter and an N310 as receiver. But the
>      > result is the same, the RX constellation is really noisy.
>      >
>      > Another observation is, that my RX sample stream in a GR time
>     sink does
>      > show quantization artifacts. i.e. I can see discrete steps which I
>      > assume are due to low RX sensitivity. The RX value amplitude is
>     around
>      > 0.0005 for the samples that I get of a USRP source. I assume this
>     should
>      > be a higher value. With the X310s it was more like 0.1.
>      >
>      > Do I need to take special care when I only use 1 antenna port?
>      > Is there another AGC setting that I need to configure for N310s?
>      >
>      > Johannes
>      >
>      > Software
>      > UHD: 3.14.1.1
>      > GR: 3.8
>      > _______________________________________________
>      > USRP-users mailing list
>      > USRP-users@lists.ettus.com <mailto: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

Reply via email to