The recent 'TEST' messages are decoding very well at Bielefeld,
range 301.7km.   From the online recording of Wolf's vlf6
E-field receiver

            Eb/N0   S/N    Phase
 2016-11-11  5.7   -59.4   -42.5
 2016-11-12  4.3   -60.5   -36.3
 2016-11-13  8.5   -55.1   -37.2
 2016-11-14  6.7   -58.1   -34.1
 2016-11-15  5.0   -59.8   -38.1

All with constant reference phase, and the decoder is working
on a copy of the signal which has been through a stage of
vorbis compression/decompression.  No normalisation used.

The signal processing pipeline for the above goes like

 vtread -T2016-11-11_08:00,+27900 /raw1/vlf6 |
    vtcat -p |
    vtfilter -ath=5,ul=6400 -h bp,f=7900,w=3000 -h bs,f=6400,w=60 |
    vtblank -a2.0 -d0 -t1 |
    vtmult -f 6470 |
    vtresample -r240 |
    vtresample -r1 |
    vtraw -oa |
    ebnaut -dp8K19A -r1 -S60 -N4 -L15000 -c4 -PS -v

vtread extracts the raw signal from the 20 day loop of signal
from Bielefeld stored on my VLF stream server in Germany.

The vtcat -p is necessary to pad with zeros the gaps produced
by occasional interrupts of the vlf6 vorbis uplink.

vtfilter is applying an auto hum notch up to 6.4 kHz, band pass
centered 7900 +/- 1500 and a band stop 6400 +/- 30

vtmult mixes down to baseband I/Q.  Two stages of resampling because
vtresample won't do a large ratio in one step.

vtraw -oa converts to three column ASCII for the decoder input.

Now to try Cumiana vlf15 ...

--
Paul Nicholson
--