Also regarding: “ VLF propagation graphs https://dl.dropboxusercontent.com/u/19882028/VLF/fig_02_25a.png ”
and the degree to which natural noise (and therefore indirectly SNR) correlates with Figure 2-25 (from link above); attached is a reasonable plot of natural noise, and a NASA plot of the lightning spectrum at (arguably) the near/far field boundary; the spectrum of lightning is relatively flat (+/-3.5dB) from 1 kHz to 10 kHz.
Comparing the Natural Noise Colorado Winter 1200-1600 (local time) plot with: Fig. 2-25 3000 n.m., 2000 n.m., or even 1000 n.m*.:
Figure 2-25 appears to show a far greater difference between 4kHz and 10kHz signal strength than the natural noise plot shows between 4 kHz and 10 kHz, even after subtracting 3.5dB for lightning at 10kHz (per NASA spectrum), and (unreasonably*) using 1000 n.m. as the signal-strength-weighted nominal distance of global lightning from Colorado at 3dB/1000km, 1200-1600 local time in winter.
* (one might expect the signal-strength-weighted average distance of Colorado winter daytime natural noise sources to be more than 2000 nautical miles; at 3dB/1000km I would guess closer to 3000 nautical miles for signal-strength-weighted average distance of Colorado winter daytime natural noise sources)
Figure 2-25 (analytical method) shows signal strength at 3 kHz as 16dB lower than at 7kHz, for TX-RX separation 1000km.
Numerical simulations show signal strength at 3 kHz as roughly 20dB lower than at 7kHz, for TX-RX separation 1000km.
Natural noise data seems to suggest less than a 4dB difference between 3 kHz and 7 kHz signal strength (compare to 16-20dB for analytical and numerical methods above, and compare also with a 20dB difference in Figure 2-25 using 2000 n.m. as the signal-strength-weighted nominal distance of global lightning from Colorado at 3dB/1000km, 1200-1600 local time in winter).
A 35dB discrepancy between natural noise data and Figure 2-25 is seen for 3kHz vs. 10kHz.
Which raises an interesting question that seems to relate to your experiments:
Given that discrepancies between the natural noise plot and the above numerical/analytical solutions seem to be in the 16dB to 35dB range, and given that experimental validation of the analytical (Figure 2-25) and numerical (LWPC, FDTD etc) solutions is comparatively thin in the 3kHz to 10kHz range, which of the two following categories might be in substantial error: (1) natural noise extrapolation; or (2) numerical/analytical methods? (or both?)
I wonder if your experiments in the 2.97 kHz to 6.97 kHz range might eventually reveal the answer to questions about the validity of computational and analytical methods in that range.
Thanks Paul (and Renato and Wolf!), very well!
The carrier on 5170.001250 Hz is still on the air and will run until 18 UTC.
Since it appears that you and Jacek are the only ones trying to receive my EbNaut, i'll stay at 16K25A, just to use the better code gain.
And since the last ~ 24 hour experiment was running so well, let's try 48 hours! Maybe it leads to a 30 0 30 0 phase pattern:
f = 5170.000000 Hz
Start time: 07.Jan.2017 20:00:00 UTC
Symbol length: 64 s
Duration: 45h, 30m, 40s
Antenna current: ~ 225 mA
The first time i used your calculator (http://abelian.org/ebnaut/calc.php?sndb=-63&snbws=2500&snmps=&code=16K25&sp=64&crc=16&nc=20&submit=Calculate ) to chosse the number of characters and the symbol length BEFORE the transmission :-)
With your given RAM, how many characters can you decode in 16K25A? And how long does the decode process take then?
These 2 day long transmissions mostly failed on 6.47 kHz, or gave poor results. Stacked single day transmissions were a better choice. For a 50 or 75 character message on 5170 Hz we may have to use the same technique.
I'm often thinking about the old VLF propagation graphs https://dl.dropboxusercontent.com/u/19882028/VLF/fig_02_25a.png (what was the original paper where it comes from?) which make more and more sense to me! On 5170 Hz we already see a real advantage of lower QRN relative to 8270 Hz or 6470 Hz. According to the graphs, the optimum frequency should be arround 4 kHz because the QRN from far away is attenuated much more whereas the poor propagation on that frequency is not so much expressed for 'short' (1000 km) distances. And BTW, 4 is a very nice number, isn't it!? Sooner or later someone has to do something near 4 kHz! I would be curious to see how this band (e.g. 4270 Hz or 70 km!) behaves. I can imagine that it is the best choice, even in summer or especially in summer!
When looking on the todays 'wideband' window (the upper one on http://www.iup.uni-heidelberg.de/schaefer_vlf/DK7FC_VLF_Grabber2.html) we can see that we are already diving below the QRN :-)
Am 07.01.2017 05:09, schrieb Paul Nicholson:
Decoded '73 DK7FC' from Cumiana (Renato Romero, vlf15, 504.6km)
with constant ref phase, Eb/N0 = 0.6, S/N 16.16 dB in 11.8 uHz,
-67dB in 2.5kHz.
Very strong at Bielefeld (Wolf Buescher, vlf6, 303.8km)
Eb/N0 11.6dB, 27.17 dB in 11.8 uHz, -56.1dB in 2.5kHz,
constant reference phase.
Here, improved my decode to 3.9dB when I remembered to use the
-a option which normalises the amplitude by the average noise.
I am not seeing much day/night phase shift at any site. Some
measurements on the carrier will be the next job.