[USRP-users] IQ different offsets

Marcus D. Leech mleech at ripnet.com
Thu Mar 17 15:47:02 EDT 2016


On 03/11/2016 05:48 AM, Pol Henarejos via USRP-users wrote:
> Dear all,
>
> I am doing some tests with my two USRP X300 with two SBX120 in each, 
> and I get some DC offsets in IQ. Configuring one USRP as a transmitter 
> conveying a phasor (x(t)=cos(2pi*f*t)+1i*sin(2pi*f*t)), and another 
> USRP as a receiver, I get the following picture:
>
> http://s11.postimg.org/bz0k9wgqb/offset_iq.png
>
> As you can see, there is an offset between real and imaginary parts. I 
> have run the calibration tools and I ensure that the calibration data 
> is loaded when I set up the TX/RX scripts. I ran the following code
>
> uhd_cal_rx_iq_balance --verbose && uhd_cal_tx_iq_balance --verbose && 
> uhd_cal_tx_dc_offset --verbose && uhd_cal_rx_iq_balance --verbose 
> --subdev B:0 && uhd_cal_tx_iq_balance --verbose --subdev B:0 && 
> uhd_cal_tx_dc_offset --verbose--subdev B:0
>
> in the USRP 1 and USRP 2. These commands were ran sequentially (first 
> with one USRP and after with the other USRP) to avoid interferences.
>
> When I start TX/RX scripts, both display
>
> -- Loaded /home/phenarejos/.uhd/cal/tx_iq_cal_v0.2_XXXX.csv
> -- Loaded /home/phenarejos/.uhd/cal/tx_dc_cal_v0.2_XXXX.csv
> -- Loaded /home/phenarejos/.uhd/cal/tx_iq_cal_v0.2_YYYY.csv
> -- Loaded /home/phenarejos/.uhd/cal/tx_dc_cal_v0.2_YYYY.csv
> -- Loaded /home/phenarejos/.uhd/cal/rx_iq_cal_v0.2_XXXX.csv
> -- Loaded /home/phenarejos/.uhd/cal/rx_iq_cal_v0.2_YYYY.csv
>
> where XXXX and YYYY are the serial numbers of the daughterboards, 
> different in each script.
>
> Also, I lock the input reference clock of RX to the output reference 
> clock of TX.
>
> Is there something wrong?
>
> Thank you for your time.
>
> Regards, 
Two things to try:

Try tuning so that your tone is away from the DC area.   So, if you're 
tuned to X, have your tone be at X+10Khz or so, and see what the
   difference is.

Also, very-recent UHD versions have made some improvements to the I/Q 
calibration routines, so you might consider upgrading to the
   latest, and re-running your calibrations.







More information about the USRP-users mailing list