Stefan and Paul,
I think that a sferic spectrum of the sort that Paul made (including the
well-considered details) is generally more accurate than an FDTD solution* for
use in signal-strength prediction, but I think that aspects of the FDTD
solution (and perhaps two other characteristics) could be merged with the
sferic-spectrum to narrow the link-budget error margin for the 2kHz-4kHz
100km-1000km operating space.
1) Using the roughly-path-specific 50-sferic spectrum to extrapolate from a
successful test frequency to an estimate or prediction of signal strength at a
previously unsuccessful test frequency is a very good approach
2) Keeping in mind that the sharp peaks and nulls that FDTD shows are realistic
for narrowband sources, but are probably not at the correct frequencies and of
the correct peak/null amplitudes for the ionospheric and ground conditions for
the day/time of the test, may help with interpretation of test results. Sferics
would tend to fill in (and also shift) the sharp peaks and nulls to some extent
(representative number for this smoothing next week), so using the sferic
spectrum as the first order assumption and keeping in mind that sharp
peaks/nulls somewhat similar to those in a path-specific FDTD plot may also be
involved, might assist interpretation of test results, and planning.
3) An adjustment (perhaps) of 3db - 5dB for the general non-flatness of the
stroke (i.e. pre-sferic) spectrum between 1kHz and 5kHz might help [see
attached, Figure 2 (H)]
4) A simple first-order adjustment for envelope shape and peak frequency, as a
function of individual or average stroke amplitude [see for instance (g) and
(h) in Figure 2 of
http://nvlpubs.nist.gov/nistpubs/jres/63D/jresv63Dn2p199_A1b.pdf] could help.
I'll try a rough version of the above for the Cumiana or Todmorden path.
*FDTD models should eventually improve given guidance from experimental data,
and FDTD analysis could be improved with better real-time parameter estimates;
but for the time being the finer texture of FDTD's (all-mode) spectral plots
can be applied loosely to sferic or LWPC path characterizations as an error
window.
Meanwhile (at 0600) I see something interesting on Romero's grabber
http://www.webalice.it/rromero/live_cumiana/last-LFtest_2970.jpg
73,
Jim AA5BW
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of [email protected]
Sent: Saturday, February 4, 2017 10:55 PM
To: [email protected]
Subject: RE: ULF: EbNaut over 3 wavelengths on ULF
Stefan,
Yes, good point, I'll replot with constant antenna current instead of constant
ERP.
73, Jim
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of DK7FC
Sent: Saturday, February 4, 2017 1:32 PM
To: [email protected]
Subject: Re: ULF: EbNaut over 3 wavelengths on ULF
Jim,
i assume your plots refer to a constant ERP. Could you scale the plot to a
constant antenna current? A range of 0.1...10 kHz is fine. That could be
helpful to better understand what could be possible on which band...
Most VLF TX systems are current or even voltage limited....
73, Stefan
Am 03.02.2017 13:51, schrieb [email protected]:
> Stefan and Paul,
>
> Attached is a 2nd-pass FDTD plot (E-field only); theoretically more accurate
> with some small corrections and refinements.
>
> The attached plot and the two FDTD plots sent yesterday represent
> daytime propagation over a conductive ground similar to sea water, for
> comparison with the NAVALEX plot (daytime, sea water)
>
> The NAVELEX plot suggests: ~ 3dB more signal at 2.9kHz than at 5.17 kHz (880
> km)
> The FDTD plot suggests: far less signal at 2.9 kHz than at 5.17
> kHz (880 km)
>
> A benefit of FDTD analysis is inclusion of all modes, including evanescent.
>
> Given very limited experimental validation of propagation computational tools
> between 2kHz and 4kHz at distances between 100 km and 1000 km, your ULF tests
> may show strengths and weaknesses in the computational tools.
>
> 73,
>
> Jim AA5BW
>
>
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> [email protected]
> Sent: Thursday, February 2, 2017 3:27 PM
> To: [email protected]
> Subject: RE: ULF: EbNaut over 3 wavelengths on ULF
>
> Stefan and Paul,
>
> I've been unable to find any empirical validation of LWPC fidelity below 5kHz.
>
> Attached is a preliminary pass (E and B field amplitude vs frequency and
> distance) with FDTD, showing some deep nulls at frequencies and distances of
> interest.
>
> (compare with NAVELEX 0101 113 plot annotated by Stefan
> https://dl.dropboxusercontent.com/u/19882028/VLF/fig_02_25a.png ) The
> NAVALEX plot is based on an algebraic approximation [details shown in
> the NAVELEX document, summary of those details (and link to document),
> in this thread ~ 02:17, January 8, 2017]
>
> Empirical validation of FDTD in the 2kHz to 4kHz range is lacking but FDTD
> can provide good accuracy at high resolution (in f, d, E and B) given
> accurate h' and beta values.
> The h' and beta values used for the attached FDTD plots are nominal daytime
> values; no telling how different the result might be for actual h' and beta
> during your recent tests.
> On the other hand, the depth and sharpness of the nulls shown in the FDTD
> plots might well be relevant.
>
> I don’t know the attachment file size limit, so in case the limit is 100 kB
> or less, I will send "FDTD Plot 2 of 2 (B field)" in a message immediately
> following this message.
>
> 73,
>
> Jim AA5BW
>
>
>
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Paul
> Nicholson
> Sent: Sunday, January 29, 2017 12:26 PM
> To: [email protected]
> Subject: Re: ULF: EbNaut over 3 wavelengths on ULF
>
>
> Jim wrote:
>
> > Excellent diagnostic method and implementation.
>
> The method has some weaknesses Jim but in the absence of any artificial
> signals it's the best I can do. I must repeat the exercise when we next get
> a thunderstorm nearer to Heidelberg. Looking at the spectrum of the sferics
> will reveal if we have some ground/sky wave cancellation at that frequency
> and range.
>
> Stefan wrote:
>
> > From the known decoded EbNaut messages, what was the SNR at> Renato or
> on your side?
>
> Nothing detectable here or at Cumiana so I can't answer the question of how
> many days need stacking. Until there is some glimpse of a signal, it is
> completely unknown. I can only estimate a lower limit.
>
> With your estimated ERP I think something should be detectable in one day or
> two. But stacking two days of your recent carrier shows nothing. Even with
> my more conservative ERP estimate I would expect to see something of your
> signal.
>
> Between 2016-06-26 13:00 and 2016-07-31 08:00 you transmitted carrier with
> 30mA antenna current. I received and stacked over 34 days of carrier (about
> 15dB stacking gain) with no sign of signal. You estimated 100nW ERP.
>
> Now you can do 150mA current - about 14dB gain over July's signal. Plus,
> improvements here to filtering and blanking may add 2 or 3dB more.
>
> We can in two days exceed the stacked signal for the whole of July. Also
> with your higher ERP I can see the signal each day at Bielefeld to check the
> phase.
>
> I think we should try some daytime carrier repeats. Already we know
> there is nothing after 2 days so we will need at least
> 4 days to get a hint of signal and then at least 10 days to collect enough
> signal for some measurements. It is surely certain that we will eventually
> pick up the signal after enough days.
>
> I'm thinking about shutting down my workstation to install an extra scratch
> disk which will speed up the signal search.
> It's been busy for 137 days so no chance to close it down.
> There are still 154 jobs running though.
>
> --
> Paul Nicholson
> --
>
170204 Spectra of Sferics 1km-200km (pg 2 of 2) ''Sferics'' Volland v02....png
Description: PNG image
|
