Distorted Spectrum after applying gr.freq_xlating_fir_filter_ccc

GNU Radio
1

Hey all,

I transmit a carrier at 2.45GHz using one USRP and recieve using
another
USRP as it is required for my application.
But the spectrum i get has a frequency offset of about 7.68KHz as shown
in
the attached file.

So, I used gr.freq_xlating_fir_filter after usrp_source to
demodulate
the signal to baseband (based on the advice by Micheal Dickens,thank you
by
the way).

However, the signal spectrum I got is not exactly at zero, it is around
320Hz.
Moreover, the spectrum is too distorted as shown in the 2nd figure
attached.

So, can this distortion be removed using the parameters or is there some
other solution?

Thanks!
Bruhtesfa

*Details: *

  • The sampling rate of the original recieved signal from USRP was
    64MHz/128=500KHz, so I used decimation factor of 500 in the xlating
    filter
    such that the output baseband signal will have a sampling rate of
    500KHz/500=1KHz.
  • since the frequency offset of the signal in 1st figure is -7.68KHz, i
    used the opposite freq_offset=7680 in the xlating filter.
  • i used cut off frequency =500Hz and width of transition
    band=200Hz.
2

On Jan 19, 2009, at 5:46 AM, Bruhtesfa E. wrote:

Details:

  • The sampling rate of the original recieved signal from USRP was
    64MHz/128=500KHz, so I used decimation factor of 500 in the xlating
    filter such that the output baseband signal will have a sampling
    rate of 500KHz/500=1KHz.
  • since the frequency offset of the signal in 1st figure is
    -7.68KHz, i used the opposite freq_offset=7680 in the xlating filter.
  • i used cut off frequency =500Hz and width of transition band=200Hz.

Everything looks OK “on paper”, but I think your filtering will be
tight given the sample rate of the output of the xlating filter. It
very well might be that the offset needs to be the actual value (not
the negative of it) … I really don’t remember any longer.

My advice is to play with the xlating filter using more bandwidth:
reduce the decimation to, say, 50 (BW = 100k), and change the filter
to be wider and more relaxed (say, BW=25k, transition=15k, in order to
encompass both the original signal as well as the shifted copy of it
if the offset is not exactly correct). I think if you play with the
variables a bit using more bandwidth, you’ll figure out what’s going on.

Good luck! - MLD

3

Michael D. wrote:

On Jan 19, 2009, at 5:46 AM, Bruhtesfa E. wrote:

Details:

  • The sampling rate of the original recieved signal from USRP was
    64MHz/128=500KHz, so I used decimation factor of 500 in the xlating
    filter such that the output baseband signal will have a sampling
    rate of 500KHz/500=1KHz.
  • since the frequency offset of the signal in 1st figure is
    -7.68KHz, i used the opposite freq_offset=7680 in the xlating filter.
  • i used cut off frequency =500Hz and width of transition band=200Hz.

Everything looks OK “on paper”, but I think your filtering will be
tight given the sample rate of the output of the xlating filter. It
very well might be that the offset needs to be the actual value (not
the negative of it) … I really don’t remember any longer.

My advice is to play with the xlating filter using more bandwidth:
reduce the decimation to, say, 50 (BW = 100k), and change the filter
to be wider and more relaxed (say, BW=25k, transition=15k, in order to
encompass both the original signal as well as the shifted copy of it
if the offset is not exactly correct). I think if you play with the
variables a bit using more bandwidth, you’ll figure out what’s going on.

Good luck! - MLD

Thank you so much for all your help Michael!!!
I will try it out.

Bruhtesfa