Re: [Discuss-gnuradio] USRP internally generated signal at 1088 MHz

[Matt Ettus]

Eric Cottrell wrote:
> Hello,
>
> I got the DBSRX board back working but noticed if I run the USRP_FFT program 
> a 20 db above the noise signal appears at 1088 MHz.  It is there if the 
> antenna is disconnected and moves up and down with the gain.  It will 
> decrease and go away as the gain is moved from about 26 down to zero but then 
> so does the desired signals.  The 1088 MHz signal goes away if I set the 
> decimation to 16.  I changed the VCO divider from x4 to x2 with no change.  
> It seems to be a signal generated on the USRP board that is getting into the 
> frontend of the DBSRX.
>
> I noticed it before the board broke but did not think anything of it until I 
> realized that it is increasing the noise level and decreasing performance.
>
> I have the board in the RX B slot.  I will try it over on the RX A Slot 
> tonight.
>
>   

You are seeing is harmonics of the clock reference being sent to the
daughterboard.  There are two lowpass filters on the board, so you can
reduce this significantly by changing their time constants.  They
consist of R193+C261,  and R195+C262.  You can slow the time constants
down by increasing the C values.  I wouldn't increase the R's too much,
or you won't be able to drive the input.  Changing both C261 and C262 to
4700pF should lower the time constant enough to significantly reduce the
spurs.

Matt