[USRP-users] X310 + 2 CBX daughtercards strange behaviour

boeglen herve herve.boeglen at univ-poitiers.fr
Sat Sep 20 03:35:19 EDT 2014



I know pretty well the throughput limit of the 1Gbit
interface, that's why I use a 10 Gbit one :-) It's really impressive
since it holds the 100Ms/s using both channels. My machine runs a Core
I7 2600k + 12 Gbyte RAM on Ubuntu 14.04 64 bit (3.13 kernel, MTU set to
9000!). I use the latest firmware with GNURadio 3.7.5. 

Concerning my
problem, RF 0 channel (first CBX) has a noise floor around -100 dB on
the spectum scope whereas RF 1 (second CBX) goes up to -60 dB. However
my settings are the same for both channels. 

When mentioning the
bandwidth, It was to describe the behaviour of the signal on RF 1 (wider
noise bandwidth). 

Another noticeable difference between the two
channels is that you get a -55dB spike at 0Hz on RF 1. When performing
the Wifi test, everytime a packet enters, RF 1 channel's noise floor
increases around -60dB (over the 50MHz bandwidth). Really weird...

Best regards 


On Fri, 19 Sep 2014 17:16:20 -0400, Marcus D.
Leech wrote: 

>> Hi Markus and Mike, Thank you for your answers. It
does not change anything whether you specify the subdev or not. Markus:
yes everything is wired properly. Concerning the last email, I used a
BladeRF as a generator (0dBm + 30dB attenuator). Here is another test
with Wifi and as you can see from the two spectrum scopes (upper is RF
1, lower RF 0), RF 1 reception bandwith seems to be larger than for RF 0
(however it's set to 20MHz for both channels). I have a second X310 with
the same daugthercards and I get the same behaviour... Is there a way to
use just RF 1? I hope that going to sleep is going to help me to find an
idea to solve the problem :-) Best regards Hervé
> The bandwidth
setting doesn't do anything on the CBX card--it's only 
> used to set
the analog bandwidth between the cards baseband/IF output, and
> the
> The bandwidth delivered to the host depends, in this case, on
> specified sample rate.
> What type of host interface are you
using in this case? From the 
> spectrum plots, it looks like you're
asking for two channels of 25Msps 
> each, which
> will not "fit" over
1GiGe at 16-bit samples.
> Do you get any 'O' being output while you
run this flow-graph? What 
> type of computer?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20140920/74ccf705/attachment-0002.html>

More information about the USRP-users mailing list