[Discuss-gnuradio] Getting a -135 dBm/Hz noise spectral density in USRP
From:
Nazmul Islam
Subject:
[Discuss-gnuradio] Getting a -135 dBm/Hz noise spectral density in USRP N210
Date:
Wed, 27 Jun 2012 01:51:14 -0400
Hello,
I am trying to measure channel path loss using two USRP N210. I need to measure the noise floor so that I can find the threshold for pathloss measurement. I am using the attached flowgraph to calibrate the receiver signal strength. I tried two ways to calibrate the receiver power and both show me approximately -135 dBm/Hz noise spectral density.
1. At first I don't transmit anything and measure the power of the floating point numbers that is coming out of rx USRP block. ThenĀ I transmit a PN sequence of -5 dBm power at 17 MS/s from an USRP. I connect this USRP through a 60 dB attenuator to the receiver USRP. This input (-65 dBm) seems to double the power of the floating point numbers. I infer that the my USRP noise floor is equivalent to -65 dB at 17 MHz.
2. I also apply an external 1 MHz noise source of -75 dBm power to the receiver USRP. This leads to an increase of 3 dB power in the receiver FFT plot. This also suggests I have -75 dBm noise power at 1 MHz.
Both these experiments suggest -135 dBm/Hz noise spectral density in my rx USRP. However, this is very far from theoretical noise spectral density (approximately -170 dBm/Hz). My Rx USRP N210 has a WBX daughterboard and my Tx USRP has an SBX daughterboard. The ettus websites suggest only 6 dB noise figure (http://code.ettus.com/redmine/ettus/documents/16). If my USRP daughterboard is not faulty, I must be giving some wrong parameters. I put the RX gain as 30 dB to get the full range of WBX daughterboards (http://files.ettus.com/uhd_docs/manual/html/dboards.html). Shall I increase it more?
Thanks for your help.
Nazmul
-- Muhammad Nazmul Islam
Graduate Student Electrical & Computer Engineering Wireless Information & Networking Laboratory Rutgers, USA.