Tried to decode *** messages from the overnight carrier Oct
10th/11th 18:00 to 06:00 and the best result was 2 chars 50
second symbols 16K25A with 18 bit CRC at +1.0 dB, BER 42.8%,
constant reference phase.

It was an averagely noisy night.

Looks like we'll have to use daytime signal and combine
multiple days.  Probably a method we will need for 2970 Hz.

> A message is sent several times and the sequences are
> added until a decode appears.

If added coherently.  Or send with longer symbols, splitting
the sending across multiple days.

Well I can test those two methods here with simulated signal and
noise.

--
Paul Nicholson
--