[TangerineSDR] First Amateur VLF Transmission Detected at the W2NAF-KC3EEY VLF Receiver!

Wed Jan 18 12:56:44 EST 2023

A new first for the W2NAF-KC3EEY VLF receiver! Ameteur transmissions from
VLF Radio Amateur DL3JMM in Germany have been detected at the receiver and
messages have been decoded! DL3JMM's QTH is over 6500km away from
Springbrook and consists of both a land and sea propagation path. The
message was encoded using EbNaut, 8k19a polynomial coding, 16-bit crc, 5
characters message length with a 30 second symbol period at a frequency of
8270.03Hz. This information is important to configure the decoder. The
carrier's spectral peak can be seen on this spectrum plot integrated over
the transmission period. It was generated on Paul's server from the
uplinked vorbis encoded audio stream of VLF spectrum data.
[image: Screen Shot 2023-01-15 at 1.49.00 AM.png]

It also was detected at a VLF receiver at Forest, Virginia as well:
[image: Screen Shot 2023-01-15 at 1.50.03 AM.png]

I then pulled the spectrum data from the raw downsampled and hum filtered
data and integrated over the length of the transmission. To do this, here
is the vlfrx-tools signal processing chain. First, the spectrum is fed
through a 3 kHz wide brick wall filter with a center frequency of 8270 kHz.
Then, it's fed through a sferic blanker to blank out sferics based on a
setting that produced the lowest noise floor. Finally, it's fed into the
narroband spectrum analyzer with 53.07uHz resolution.

vtfilter -h bp,f=8270,w=3000
2023-01-13_04-15-00_18900s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-18840s_high_power.vt
| vtblank -a12 -d0 -t100 -v | vtnspec -v -r53.07e-6 -w0.01 -f8270.03 >
"2023-01-13_04-15-00_18900s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-18840s_high_power.dat"

It produced a narrowband spectrum with a carrier peak of 5.622e-08:
vtnspec: amplitudes 0 m/p/r: 1.203e-08 5.622e-08 4.672 p-m 6.61 sigma

The output data was plotted in gnuplot, also showing the carrier's peak:
[image:
2023-01-13_04-15-00_18900s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-18840s_high_power_a12.png]

To see if I could get a decode, I ran it through the ebnaut decoder using
the following vlfrx-tools processing chain. After the sferic blanker, the
spectrum is fed into a multiplicative mixer at the carrier frequency to
produce baseband IQ data. The output of the mixer is then downsampled and
converted to ascii data, then sent to the ebnaut decoder with the encoding
parameters of the message. -c2 tells the encoder to use both processor
cores.

vtfilter -h bp,f=8270.03,w=3000
2023-01-13_04-15-00_18900s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-18840s_high_power.vt
| vtblank -a12 -d0 -t100 -v | vtmult -f 8270.03 | vtresample -r 240 |
vtresample -r 1 | vtraw -oa | ebnaut -d -N5 -p 8K19A -S30 -k16 -r1 -c2 -v
-L20000 -PS -T60

But this did not result in a successful decode. A few more dB was needed.
Thankfully, DL3JMM did more transmissions in his transmission campaign at
the following times, he wrote:

The next transmission already started at 18:00 UTC because 18 is modulo of
6h. So they begin
next broadcasts probably:

today (01.14):

- 18:00

tomorrow (01.15):

- 00:00
- 06:00
- 12:00
- 18:00

Monday (01.16):

- 00:00
- 06:00

Here is a spectrogram showing the transmissions. Between transmissions,
only an unmodulated carrier.
[image: image.png]
You can see the first transmission on the 13th, one on the 14th, four on
the 15th, and the beginning of the one on the 16th.

Paul was able to achieve a decode at Forest, VA for the midnight
transmission on the 16th, however, I still needed a few more dBs for that
transmission and recommended I stack both midnight transmissions. Stacking
feeds both transmissions through an additive mixer and aligns timestamps
and sampling. This allows the signal power from both transmissions to
add and will increase the likelihood of a decode. First, I converted both
midnight transmissions to baseband IQ data using the following vlfrx-tools
signal processing chains:

vtfilter -h bp,f=8270,w=3000
2023-01-14_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1.vt |
vtblank -a12 -d0 -t200 -v | vtmult -f 8270.03 | vtresample -r 240 |
vtresample -r 1 >
2023-01-14_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt

vtfilter -h bp,f=8270,w=3000
2023-01-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1.vt |
vtblank -a12 -d0 -t200 -v | vtmult -f 8270.03 | vtresample -r 240 |
vtresample -r 1 >
2023-01-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt

Then, I fed both into the stacking script:

bash stack2
2023-01-14_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt
2023-01-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt
>
2023-01-14-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband_stacked.vt
2023-01-14_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt
1673740200.000000 3.914e+05 1.532e+11
2023-01-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband.vt
1673826600.000000 5.056e+05 2.556e+11

Finally, I converted the stacked transmissions to ascii and fed it into the
decoder. I GOT A SUCCESSFUL DECODE! The message is in brackets with an
Eb/N0 of 2.5dB!!

vtcat -S600
2023-01-14-15_23-50_16000s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-16000s_1_r1_baseband_stacked.vt
| vtraw -oa | ebnaut -dp8K19A -r1 -S30 -N5 -c2 -PU -L50000 -v
initial reference phase -48.3 amplitude 2.244e+01
phase   0    0    0    0    0
found rank 0 ber 3.5156e-01 Eb/N0 1.2 M -8.639339600e+02 ph 0 0,0,0,0
[BERND]
carrier phase: -11.9 deg
carrier amplitude: 5.749e-02
carrier Eb/N0: 2.5 dB
carrier Es/N0: -10.05 dB
carrier S/N: 17.04 dB in 65.1 uHz, -24.83 dB in 1Hz, -58.81 dB in 2.5kHz
elapsed 2
phase   1  180  180  180  180
phase   2   30   30   30   30
phase   3 -150 -150 -150 -150
phase   4  -30  -30  -30  -30
phase   5  150  150  150  150
phase   6   60   60   60   60
phase   7 -120 -120 -120 -120
phase   8  -60  -60  -60  -60
phase   9  120  120  120  120
phase  10   90   90   90   90
phase  11  -90  -90  -90  -90
vstack: max height 15

I also wanted to mention the CPU usage during the decode. EbNaut uses
powerful error correction and processing for message decodes. CPU usage for
the decoder process was 130%, meaning one core was completely utilized and
the other was 30% utilized!! Paul mentioned that this is the only amateur
radio mode that uses more power to decode the message than it does to
transmit the message!

Congratulations to Paul and Bernd DL3JMM on a successful transmission
campaign and a first for the W2NAF-KC3EEY VLF Receiver!!

More about the EbNaut mode can be found here <http://abelian.org/ebnaut/>.

Jonathan
KC3EEY
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tapr.org/pipermail/tangerinesdr_lists.tapr.org/attachments/20230118/5fb8e2c0/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 2023-01-13_04-15-00_18900s_DL3JMM_8270.03_8K19A-crc16-30s-5chars-18840s_high_power_a12.png
Type: image/png
Size: 6764 bytes
Desc: not available
URL: <http://lists.tapr.org/pipermail/tangerinesdr_lists.tapr.org/attachments/20230118/5fb8e2c0/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screen Shot 2023-01-15 at 1.49.00 AM.png
Type: image/png
Size: 69761 bytes
Desc: not available
URL: <http://lists.tapr.org/pipermail/tangerinesdr_lists.tapr.org/attachments/20230118/5fb8e2c0/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 825825 bytes
Desc: not available
URL: <http://lists.tapr.org/pipermail/tangerinesdr_lists.tapr.org/attachments/20230118/5fb8e2c0/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screen Shot 2023-01-15 at 1.50.03 AM.png
Type: image/png
Size: 51156 bytes
Desc: not available
URL: <http://lists.tapr.org/pipermail/tangerinesdr_lists.tapr.org/attachments/20230118/5fb8e2c0/attachment-0007.png>