search

dernasherbrezon / r2cloud

Decode satellite signals on Raspberry PI or any other 64-bit CPU.
Apache License 2.0
257 stars 30 forks source link

Generate spectrogram of weather satellite signals #34

Closed dernasherbrezon closed 6 years ago

dernasherbrezon commented 6 years ago

Will help with debugging

zeroby0 commented 4 years ago

Hi! Can you zoom in the spectrogram to the center frequency? Currently it's multiple MHz wide, so the actual signal is just a thin line.

dernasherbrezon commented 4 years ago

Do you have an example? Currently it is 240khz or 288khz wide. And you can't "zoom" because FFT was created once and all buckets are pixels. So if you want to zoom into the part of the spectrogram, you have to define smaller FFT bucket and run FFT on the original data.

zeroby0 commented 4 years ago

Here is a spectrogram for a NOAA 18 pass.

I thought the thin line in the center is the NOAA transmission, but if the current width is 240Khz, then the 40KHz NOAA signal is too thin to be the thin center line. If the width is 240Khz, then there is no need to zoom in more.

But isn't this spectrogram too symmetric? That's a bit suspicious, do you know what's going on here?

spectogram-1

dernasherbrezon commented 4 years ago

Hey, couple of things:

  1. The line right in the middle is dc offset.
  2. The actual signal is within thick lines on both sides. It doesn't have clear APT spikes like here, because of small spectrogram resolution.

  1. They are symmetric because NOAA signals are fm demodulated and stored in .wav file. They don't have imaginary part, thus symmetric after FFT.
zeroby0 commented 3 years ago

If they are symmetric, can we drop one half so we can have a better use of space? Maybe also a smaller spectral width for NOAA satellites so we can see it better?