> But why is the Eb/N0 so high,

Very strange.

Only 30.5% symbol errors were counted, which is where
the Eb/N0 = +5.1 comes from.

But the carrier level: 11.2 dB in 24.0 uHz corresponds
to Eb/N0 = -1.1 dB.

The laws of physics have changed?  Or I have a software
bug or made some mistake?

Or something strange about the signal?

Symbol error map:

 http://abelian.org/vlf/tmp/160512a.gif

The night was low noise and daytime signal appears good too.
It looks like a really good signal, except that the overall
average calculates to only 11.2 dB in 1/41600 seconds.

Will investigate after I have finished work.  Interesting to
see what today's signal does - last night was low noise too.

--
Paul Nicholson
--