USRP Software Radio

cancel
Showing results for 
Search instead for 
Did you mean: 

USRP-2921 spur without any input

Hi,

When using USRP in my experiment,I found a very strange spur at the center frequency of spectrum.I couldn`t figure out the cause.Could you help me out?

spur.png

0 Kudos
Message 1 of 4
(6,766 Views)
The USRP has greater than 80dB of SFDR (spurious free dynamic range). Being a direct conversion receiver, the peak you see is likely a combination of LO leakage and DC offset in the baseband IQ signal. This is expected and will be very small as it is in your graph, below -80dB.
0 Kudos
Message 2 of 4
(6,762 Views)

Hi,

I found something interesting,could you help me out?

1.When I set IQ rate to 25MS/sec,I got 25MHz span of spectrum.However,the power of the spectrum on the far left 2.5M and the far right one decreased around 5dB..I wanna figure out the cause.

2.I can`t reconfigure signal(even use niUSRP property) in a while loop when i wanna change the carrier frequency in USRP-2921,but it works in USRP-2920.I am wondering why....(I used niUSRP EX Rx Continuous Async Reconfig on the Fly.vi)

0 Kudos
Message 3 of 4
(6,752 Views)

Great questions.

1. The pass band of filters inside the onboard chip in the analog section of the NI USRP-2921 are visible for very wide spans.  It's flat for short spans, but as you can see the curvature on a very wide span such as 25MHz. Only about 80% of the bandwidth is usable due to filter edge effects, so your real-time bandwidth is closer to 20 MHz.  The 2921 does include adjustable baseband filters for Recieve at fixed values of 15M, 19M, 28M, 36MHz that can be accessed from the niUSRP property node titled Bandwidth.  (the curvature you are seeing could also be caused by the antenna, but this can easily be ruled out by simply removing the antenna.

 

2. The USRP is always sampling at 100 MS/s on I and Q giving a theoretical realtime bandidth of 80MHz.  Because of the onboard analog filters reduced this to about 40MHz and down-sampling done digitally mathch the data rate to that of gigabyte ethernet, making 25MHz is the widest possible bandwidth for 16bit samples.  The property node you are using to change frequency on the fly allows you to move around within the 40MHz window without re-tuning the LO (local oscillator).  In order to move outside of that 40 MHz bandwidth you would need to stop the Rx task, retune the LO center frequency, and restart the task.

Message 4 of 4
(6,749 Views)