Hi Jim,
Thanks for your interesting contribution.
I am optimistic that we will get some data for arround 4 kHz and below
on the path to Alex.
Before the decode of that single character on 5170 Hz i had expected
that the SNR will be about 10 dB below 6470 Hz and maybe even more.
At some point on the way down to DC we, or the RX station, has to
critical ask himselfe if the sebsitivity of the system is still (nearly)
as good as possible. Missing a dB on the RX site can cause the TX to
stay on for dozens of hours longer...
73, Stefan
Am 26.10.2017 01:35, schrieb [email protected]:
Stefan,
The attached doesn't show noise vs frequency vs season/distance very well, but
examples include:
a)
Colorado data seem to suggest less reduction in noise as frequency decreases
from 9k Hz to 3 kHz, for more-distant lightning (this by comparing seasonal and
time-of-day Colorado data and knowing that Colorado and Florida are important
sources)
b)
In contrast to (a) above, comparing California 0800-1200 data (wherein even
summer noise may be nominally from 2000 km away) to Colorado Summer 0800-1200
data (wherein late-morning local lightning probably dominates lightning from
elsewhere), the distant-lightning case (California) shows a considerably
greater reduction in noise as frequency decreases from 9k Hz to 3 kHz.
Neither of the above clearly shows noise vs frequency vs season/distance, but
perhaps they suggest that many variables are involved.
Also notice that Colorado Spring 0800-1200 noise minimum is uniquely shifted to
6 kHz, with only a few dB difference from the 10 kHz peak. Maxwell and Stone
noted that as a curiosity.
Once again I'm guessing that your upcoming experiments may provide some of the
most relevant data available.
73,
Jim AA5BW
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of DK7FC
Sent: Wednesday, October 25, 2017 4:55 PM
To: [email protected]
Subject: Re: LF: ebnaut 5170 - decode nr Moscow
Hi Alex, Paul, VLF,
I found that the fascination risis with the square of the wavelength!
My congrats again to your ufb VLF RX and thank you for your effeorts. We are
very successful.
I never expected that we can transfer messages on 5170 Hz over that path with
such a rate. Now we have some important data: We can expect 6 dB in
23 uHz during an average quiet night. So now we can calculate how many days
must be expected for a longer message, like we did on 6470 Hz. I would like to
try a 12 character message soon. It takes just 2 weeks or so :-)
Paul, maybe you are interested in wav files for reprocessing. I attached the
23th and 20th...24th(weighted). The size is very small...
For the 23th i got an even better result with a time offset of 1754 seconds.
My ERP on 5170 Hz is 2 dB lower than on 6470 Hz. But the SNR we observe seems
not to differ much from that on 6470 Hz. The opposite of what all the curves
predict!!!!!!
So what will happen arround 4 kHz???!!! Will we be able to send messages over
that respectable distance? I bet!
From 4 kHz it is not so far until we reach 2.97 kHz!! That's my dream for
this season, 2 characters on 2.97 kHz over that path!
We were focused on that one graph showing VLF propagation as a function of the
frequency, for different distances (and constant ERP!). But the y axis shows field
strength and not SNR! What is when most of the noise comes from far away (>
4000 km), which is the case in winter on our path!? Then the noise will drop
stronger than the signal? This first experiment seems to be a first sign for that
assumption. So what happens on 4.47 kHz relative to 5.17 kHz? The ERP will drop by
just 1.2 dB. But the noise may fall much stronger!! We will check that too :-) And
on
2.97 kHz, relative to 5.17 kHz: Just 5 dB less ERP and maybe a much better
noise reduction!
73, Stefan
Am 25.10.2017 18:23, schrieb Alex K:
Hello VLF, Stefan
I'm glad to report about sucsessfull decode of 1-char ebnaut message
sended on 5170.1 Hz 20,22,23 and 24 October.
Message was decoded by weighted stacking, unweighted stacking and from
23 okt file only. Other files give very poor decode with many false
decodes or only false decodes.
In attacment we can see best decode from weighted stacking.
Also all ebnaut-files is available here:
http://rn3aus.narod.ru/5170.zip My VY congratulations to Stefan! Ab
ebnaut-rx V0.7a
input file sum.wav
sample rate 0.342935 per second
rx frequency 5170.1
file start time 2017-10-24 15:30:58.390 polynomial 8K19A crc size 6
number of chars 1 block size 12 symbol period 180.000 number of
symbols 240 list length 20 start offset 1754.000 freq offset 0.000000
cores 1 skipped 1755.434703 seconds to start initial reference phase
-27.5
---------------------------------------------
found =
list rank 0
reference phase 0 0 0 0
carrier S/N 12.10 dB in 23.1 uHz, -34.26 dB in 1Hz, -68.24 dB in
2.5kHz carrier Es/N0 -11.70 dB carrier Eb/N0 4.54 dB info bit period
7578.95 seconds symbol error rate 79/240 = 32.917 %
Es/N0 from symbol errors -10.0 dB
Eb/N0 from symbol errors 6.2 dB
Shannon capacity 1.9 bits/hour
Shannon efficiency 25.2 %
symbols file sum-symbols.csv
elapsed 2 seconds
|
