Spurious spikes in FFT

My HW configuration: USRP1 with DBSRX, no input signal.

Using the script to plot FFT usrp_fft.py -f 1003.25M -d 8 -g 72 I see
two
spurious spikes at 1000 MHz and 1004 MHz (the next one is at 1008 MHz).

http://www.nabble.com/file/p23854396/fft_no%2Bsignal.png

Where does these spurious signals come from? In the thread “Spurious
Spikes
Seen with usrp_fft.py”
http://www.nabble.com/Spurious-Spikes-Seen-with-usrp_fft.py-td14747991.html#a15025079
one of the explanations is the 64MHz clock, but then the spikes should
be in
64MHz intervals.

On the picture below with 0dBm CW input signal at 1015 MHz you can see
the
spurious spike is significantly strong. Reducing the gain removes the
both
signals - the CW and the spurious.

http://www.nabble.com/file/p23854396/fft_with_signal.png

Does anyone have experience how to remove/suppress the spurious signals?

View this message in context:
http://www.nabble.com/spurious-spikes-in-FFT-tp23854396p23854396.html
Sent from the GnuRadio mailing list archive at Nabble.com.