How to find local peak of sinc function ? peak detector setup?

GNU Radio
1

Hello,
I need again your help for this problem:

  • I want to locate the peaks (local maximum) of the input signal, which
    is
    sinc-function pulse (the 2 gray points in picture sincpuls.png, not the
    neighbour peaks).
    How can I use the gr_peak_detector in grc to do this ? which values I
    should
    give threshold_factor_rise, threshold_factor_fall, lookahead, alpha ?
  • If you look at my grc-picture (peakfalse.png) you will see that the
    peaks
    are not at the right places. They happen before the true peaks of the
    sine-function. How comes this ? ( My value for threshold_factor_rise,
    threshold_factor_fall, lookahead, alpha : 10,10,100,0.001 )

Thanks in advance.
grc007

http://old.nabble.com/file/p28919877/sincpuls.png

http://old.nabble.com/file/p28919877/peakfalse.png

View this message in context:
http://old.nabble.com/how-to-find-local-peak-of-sinc-function---peak-detector-setup---tp28919877p28919877.html
Sent from the GnuRadio mailing list archive at Nabble.com.

2

schrieb grc007 am 2010-06-17 23:28:

Hello,
I need again your help for this problem:

  • I want to locate the peaks (local maximum) of the input signal, which is
    sinc-function pulse (the 2 gray points in picture sincpuls.png, not the
    neighbour peaks).
    How can I use the gr_peak_detector in grc to do this ?

If you know the exact shape of the signal you want to detect, you may
get better results with doing a correlation with the desired signal
first. Then you can use a threshold or a more sophisticated peak
detector.

  • If you look at my grc-picture (peakfalse.png) you will see that the peaks
    are not at the right places. They happen before the true peaks of the
    sine-function.

You did not show the complete flow graph in your pictures, but chances
are good that its not early, but late.

Both the FFT sink and the peak detector need some time for processing
reulting in some delay so you must not expect their output to be synced.
You need to find out somehow both delays, or at least the difference of
the delay. Try you flowgraph with other imput signals, like ramp.

Patrick

–
Engineers motto: cheap, good, fast: choose any two
Patrick S.
Student of Telemati_cs_, Techn. University Graz, Austria