Loading ...
Sorry, an error occurred while loading the content.

1963Re: [SpectrumLabUsers] NWC signal spectrogram [3 Attachments]

Expand Messages
  • wolf_dl4yhf
    May 2, 2014
    • 0 Attachment
      Hello Edgar,

      you wrote:
      > I expect that the signal composition will be different, but can not
      seem to get the spectrum colours to be as smooth as those in the above

      Difficult to say how the 'AWESOME' software avoids the steps in the
      spectrogram colours when zooming in so deeply.
      Math dictates the tradeoff between spectral and time resolution - it's
      impossible to achieve a true time resolution of "fractions of a bit
      time" and -at the same time- a frequency resolution of "fractions of the
      frequency shift" (if MSK is considered FSK with a modulation index of 0.5).

      Possible ways they do it:

      A - individually interpolate the colour for each pixel in the
      spectrogram, even if those pixels fall inside the same FFT frequency bin

      B - use a radically different FFT windowing function (which is almost
      zero within most of the FFT window) and a much longer FFT. The net
      result would be similar as A : spectral smearing between neighbouring
      FFT bins.
      Since there is no such FFT windowing function hard-coded in SL (yet),
      the only way to achive both high resolution in the time- and frequency
      domain is to use the 'reassigned spectrogram' as already suggested here
      some time ago.

      All the best,
      Wolf .
    • Show all 3 messages in this topic