Stefan wrote:

> f = 8270.100000 Hz
> Start time: 05.May.2017   05:30:00 UTC
> Symbol period: 10 s
> Characters: 100
> CRC bits: 0
> Coding 8K19A

Decoded after a bit of work

 "THIS IS A NEW LONGEST EBNAUT MESSAGE ON VLF.
  881 KM ON THE 36 KM BAND.
  JOIN THE VLF ACTIVITIES! 73!"

list rank: 0
constant reference phase
carrier phase: 113.5
carrier Eb/N0: 0.2 dB
carrier S/N: 27.81 dB in 20.2 uHz,
            -19.14 dB in 1Hz,
            -53.11 dB in 2.5kHz

I started with daytime settings for sferic blanking and got a
close but false decode.  Reducing the blanker threshold almost
to night setting produced the correct decode.

I'm noticing that the probability of false decodes with long
messages is much higher than I expected.   It is something to
do with the message being many times the constraint length of
the convolutional code. This I will have to study to understand
properly.  There are 30 bits of redundancy for the outer code
to use but we don't seem to get the full benefit of it. Some
trials with 200 character messages are showing this clearly.

--
Paul Nicholson
--