Re: LF: Re: 9kHz noise level

[Andy Talbot <andy.g4jnt@googlemail.com>]

WSPR looks at the energy in 1.5Hz bins, or chunks, and effectively filters the signal to this bandwidth.   If the noise impulses ar eof significantly shorter duration, they will be smeared outas well as reduced i magnitude and probably end up appearing quite close to Gaussian.     Certainly, on 73kHz the impulsive Loran interference and natural spikes appeared completely Gaussian when filtered to the sub-Hz bandwidth we used there. 

 

However, for 9kHz signalling, I think you would do a lot better by going for coherent signalling. The path should be very stable so you shoul dbe able to retian coherency for hours. Use GPS timing and at that frequency you can get an absolutely coherent local reference and timing, so try long duration PSK, with symbol timings based on ten second intervals, or minutes.

 

Andy
www.g4jnt.com

On 26 February 2010 17:07, Roger Lapthorn <rogerlapthorn@gmail.com> wrote:

Two questions...

(1) Does anyone know how modes like WSPR behave with signals buried deep in impulsive noise like sferics? It seems that most of the natural noise at <9kHz is of this nature. Man-made noise is another matter. Perhaps there are certain modulation types that are well suited to get through for this sort of impulsive noise? I am no expert so have no idea but would value the views of people like Andy.

(2) Regarding OFCOM and NoV permits for <9kHz is it not reasonable that they should have some overview of any transmissions (radiated, conducted or inducted) which could be sources of interference? We all appreciate that the radiated energy will be tiny, but near-field signals can be a significant interference source and over a considerable (local) distance at these very low frequencies. A most basic example could be interference to telephone, inductive loop audio systems, hi-fi or even pipe tracing systems. If any interference did occur, OFCOM would have some leads for investigation. It is not as if they are charging for NoVs.

73s
Roger G3XBM

On 26 February 2010 19:51, Alexander S. Yurkov <fitec@omskcity.com> wrote:

> Subject: LF: Re: 9kHz noise level
>

> Dear LF,
>
> sometime back in 2000 I attempted to measure the daytime noise background at 9 kHz (see http://www.qru.de/vlf-theorie.html). I was using my regular LF antenna at my suburban QTH. The effective height is ~ 9 m, calibrated on LF by comparison to a small loop, and assuming that it would depend only little on frequency. Using good old Spectrogram with moderate averaging and placing the cursor between the visible impulsive spherics, I arrived at a background noise of about 5 dBuV/m referenced to 1 Hz (equivalent to 142 dB kTo).

Seems noise level depend strong on the location. Anyway it is obvious for
industrial noise.

> Empiricaly I found that some noise limiting or blanking was essential to
> maximize SNR for narrow bandwidth reception.

Theory confirm this empirical rule. Few years ago i do some a theoretical
study of optimal signal recieption in the condition of strong spherics
noise. Such a noise is substantialy nongaussian then conventional theory
assumed gaussian noise is not adequate. The study show conventional
filtering (or FFT, it is the same) is not optimal. To get optimal SNR one
should do some nonlinear processing before filtering. This processing is
very similar to limiting. There is some an article on the subject on my
web page www.qsl.net/ra9mb but unfortunely it is in russian. And it is
very theoretical:-) I wonder if such a processing really improove SNR on 9
kHz. Such a processing was realized in GLFER program as author told to me.
But i don't test it.

Regards,
Alexander/RA9MB

--

http://g3xbm-qrp.blogspot.com/
http://www.g3xbm.co.uk
http://www.youtube.com/user/G3XBM
G3XBM    GQRP 1678      ISWL G11088