> new message starts 2016-04-03 08:00:00

The signal continues to improve!  This one decoded here
'WX' with Eb/N0 = +10.7 dB

> I prepared the EbNaut Sender for period 30 and chars 3

OK that takes an extra 2.5 dB, no problem here.

I notice the .wav file start time of Stefan's decode
of the 2016-04-02 07:30 message - the file starts at
2016-04-02 08:44:55.11 so therefore it misses the
first 2095 seconds.   (Stefan is using an offset
of -2088 seconds, maybe that should be -2095).

Due to all the missing samples, 53 symbols are lost
which is why the BER was 42.9% (it counts erasures as
errors) giving an unrealistic Eb/N0 of -2.9 dB.  The
carrier strength corresponded to Eb/N0 = 3.8 dB so with
correct recording time it should have been an easy decode
and would do 3 chars with 30 second symbols with a dB or
so to spare.

--
Paul Nicholson
--