USRP1 DBSRX phase noise

Hello,

I would like to capture signals using the USRP1 (rev 4.5) with DBSRX.
For
the first test I am using CW signal from generator at 1.501GHz, 0 dBm
(verified to be precise). To capture the signal I use the
usrp_rx_cfile.py
(usrp_rx_cfile.py -R a -d 8 -f 1.500G -N 40064 -s -g 45 usrp.dat)

I plot the FFT using Matlab (rectangular window, FFT length 65536).
At first it appears fine (there is a DC component, but OK):
http://www.nabble.com/file/p23724900/unzoomed_1.5GHz.png

Zooming in the CW peak:
http://www.nabble.com/file/p23724900/zoomed.png
There is a lot of noise, the peak is not sharp (width 2 kHz). Looks like
phase noise to me. Or is there very fast frequency drift?

Thinking the signal might be saturated, I’ve reduced the gain. The
frequency
range is smaller, but also the SNR
http://www.nabble.com/file/p23724900/zoomed_25dB%2Bgain.png

I’ve tried to use hi quality external reference clock (64MHz), following
the
instructions form USRP Clocking Notes. Unfortunately there was no
improvement.

Is there an explanation for this behavior?

Thank you.

PS: Sorry for multiple posts, if any…


View this message in context:
http://www.nabble.com/USRP1-DBSRX-phase-noise-tp23724900p23724900.html
Sent from the GnuRadio mailing list archive at Nabble.com.