There is an absolute limit to signalling efficiency, determined by Shannon in 1948 of -1.64dB S/N (normalised bandwidth)   WSPR/WSJT are within about 3 - 4dB of the limit, so we'll only ever get this improvement to play with.   Of course, we can go narrower and narrower and longer and longer, but there comes a point where the ionospheric path just won't support the signalling any more

 

And that's it.  All that's left is more EIRP, directional Rx antennas, noise cancellation and...

 

For the signal in noise story, see if you can find a copies of :

 

"Communication in the Presence of Noise",  CE Shannon,  Proc.  IRE  (now IEEE)   January 1949 

which is the seminal paper that sets the  -1.64dB S/N limit

  

--- and ---

 

"Poisson, Shannon and the Radio Amateur" ,  JP Costas, Proc IRE, December 1959   which spells it all out in a delightfully readable way

 

Most human modes, morse and QRSS, and Hell all come out at a roughly similar normalised S/N, and machine modes, at the moment roughly 4 - 6dB better than the best humans ops.   As shown, coding could further improve things by another couple of dB.

 

So who's going to be the first with MFSK Turbo coding on LF.  ?

That's real Amateur Radio as far as I'm concerned.  Not faffing about struggling to dig tones out of noise or looking for lines of different coloured dots

Andy
www.g4jnt.com