Re: [Discuss-gnuradio] WWVB demodulator

[Marcus D. Leech]

On 12/26/2010 09:54 PM, Michael Ossmann wrote:
> I started to years ago but never got very far.  It looked like
> starting with goertzel was going to work.  I think I planned to dump
> goertzel output into M&M clock recovery with a 10 Hz symbol rate and
> then count the number of .1 second chunks per pulse.
>
> mossmann
I'm not too worried about computational efficiency, so I hadn't 
considered Goertzel.  My current approach
  (that I've just developed this evening :-) ), is to assume a 192KHz 
sound card input (which is congruent with
  the approach I'm currently using for VLF signal processing for 
ionospheric measurements), bandpass filter at
  60KHz (+/- 100Hz), then detect the result and data slice into 1s and 
0s at 10Hz--the data slicing is driven by
  computing a longer-term signal level average (since WWVB basically 
uses ON/OFF (well ON, and 10dB down)
  keying.

Having a variable that tracks the medium-term average allows the 
receiver to track fading and tweak the data slicer
  appropriately.

Haven't decided how I'll turn the PWM into actual 1s and 0s yet.




--
Marcus Leech
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org