AF 447 Search to resume (part2)

[JD-EE]

EDLB, you'd want to fire up the crystal oscillator nearly continuously to avoid the chirp and initial self heating effects. And it doesn't really help you that much if the RC oscillator is moderately stable with time over the 10 ms pulse. As long as you have the 10 ms pulse you are stuck with a 100 Hz or wider FFT window.

Actual window size needed varies with shape of the actual shaped FFT filter. Gaussian is best if you know the frequency. But you can't really get that with an FFT very well. So you have to compromise and accept enhanced filter ringing.

(That's an interesting subject to me since designing proof of concept military
Hedy Lamar radios. er that's frequency hoppers. She designed the concept with some help on details from George Antheil. To make a fast hopper work right you need VERY good filter ringing characteristics or other compromises. We'd chosen the Gaussian filter approach.)

With the crystal oscillator you can use count downs to transmit X cycles out of Y total cycles easily enough. Then, since it is counter controlled, you know what the signal looks like for optimal decoding. Alas, you don't know WHEN the signal looks like that. So you have to record and post process or use a boatload of GPUs. Then you have to sit very still (kill Doppler effects) for several seconds while a few pings could add up. Then you move and sit another 10 seconds. That sounds like a lot of effort for a few dB of non-coherent signal addition. At 50ppm a 37.5kHz crystal is roughly .05 Hz off. Um, you might get decent correlation gain. edit: Oops - misuse of calculator and didn't trust my intuition. It's closer to 1.9Hz so coherent addition may not happen depending on the period precise frequency at which you presume the crystal is operating. You might be better off trading transmission time for power and use processing gain to make up the difference and distinguish the signal from the general noise characteristics such as whale song.

I've a question for auv-ee if he's read this far. What are the noise power characteristics vs frequency for 100Hz through 100kHz. (I understand 100kHz is basically down to thermal noise levels.) And I'd LOVE to know some transducer sensitivity characteristics. Alas, I don't know where to look for this data.

{^_^}