Return to KLUBNL.PL main page

rsgb_lf_group
[Top] [All Lists]

Re: VLF: SpecLab filter optimisation tests

To: [email protected]
Subject: Re: VLF: SpecLab filter optimisation tests
From: Stefan Schäfer <[email protected]>
Date: Wed, 29 Dec 2010 14:50:41 +0100
In-reply-to: <[email protected]>
References: <[email protected]> <[email protected]>
Reply-to: [email protected]
Sender: [email protected]
User-agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
Hello Lubos,

Is the difference of the minimum of the single turn VLF loop higher than on the multi turn loop? If so, then the single turn loop would be best for getting an optimal SNR. Interesting!

My DFCW-3 window on my grabber became a DFCW-600 window for while. This is to compare the old and new SL settings. The upper window uses the new settings. I am curious about the difference. Hopefully i can get an improvement in OE3GHB's next transmission :-)

73, Stefan


Am 29.12.2010 08:21, schrieb Lubomir Bobalik:
Hello Stefan!
Many thanks for your effort to improve SL setting. It is as important as having good antenna. :-) 
I wrote your new parameters to my SL setting to try this one. Gerhard, OE3GHB, was transmiting on 8970 las day, but I didn´t see your signal. At first, I was surprised, but then I
realised, that my single-loop antenna was beamed to you. It means, that it was in minimum receiving for Gerhard. I am going to turn it at 217° and we will see. When I tested single-loop I noticed, there was difference -20dB between direct beaming and +90° beaming on the beacon in Germany. It is very much!

Dne 28.12.2010 16:39, Stefan Schäfer napsal(a):
Dear VLF,

In my 9th experiment, my signal was well received at 4X1RF in 2873 km distance. Chris has recorded the whole procedure and sent me 4.5 GB on a DVD as a registered letter (many thanks Chris!). I could successfully extract the 8970 Hz part in a 880 MB file. Then i re-processed this file several times in SpecLab, adjusting the FFT / Noiseblanker / Clipper / Filter -settings. This recording is optimal for re-processing and looking for optimal settings since the SNR is relatively high at the beginning (up to 12 dB) and becomes lower (due to rising QRN) at the end. There is a longer section on 8970.00000 Hz and later the message "CU" that is partly visible and partly lost in the QRN. So the goal was to make the whole message visible or at least as much as possible.

The final improvement is rather marginal, probably not much more than 2 dB. The pessimists will say its a pity that the SNR improvement is not more and the optimists will say its good that we obviously used almost perfect settings in the last tests ;-)

So, let me tell you about the results:
The left half of the attached picture shows the transmission as received by using the current distributed USR files/settings (4.5 mHz FFT, 60 s/pixel, 3 kHz band filter BW, 1 kHz filter slope, Clipper threshold = 6 dB above avrg., Noise blanker set to 2 ms / 9 dB, 0.05 sec.). The right part shows the best SNR optimisation i achieved. Of course this is all a subjective decision, there is no strong difference but i find that the "CU", at least the "C" can be better identified. Reducing the FFT bandwidth should actually improve the SNR but the readability decreases! There is not really an improvement between 4.5 mHz and 2.8 mHz, at least in this DFCW-600 transmission.
The SL settings of the best achievement are: 4.5 mHz FFT, 60 s/pixel, Clipper threshold = 0 dB, Filter BW = 3 kHZ, Filter slope = 1 kHz, Noiseblanker at 6 dB / 0.0003 sec ramp time, max. pulse width 0.05 sec.

Radio amateurs who are not operating in QRSS/DFCW mode may say there is no difference between the results ;-)

OK, the different is very good visible. Well done!



On 6470 Hz there was no significant improvement, even in longer FFT times. But my tests on that part are still not finished.

BTW, without using the SpecLab internal bandfilter, noiseblanker and Clipper that signal would have been TOTALLY LOST!! There wouldn't even be the faintest trace of my signal!!


Yes, using all this tools is important, it is "big gun" agains a noise.

I noticed, that Clipper red box indicace "C" continously. The message says, that "signal is 0,0dB ABOVE clipping (A); avrg=-5,9dB". Is it correct? 

As a conclusion i would say that the current SL settings are well suitable but a few dB SNR improvement could be achieved. Of course this depends on the kind of QRN/QRM and the antenna that is used. The RX stations who can receive my signal at 15 dB SNR or better will not see a significant difference but the new settings may be interesting for stations above 1000 km or below. Those who can set up the SL paramaters may try this. I will provide a new USR file for some RX stations in high distance before my next VLF experiment.

Any comments/hints are welcome.

When I was active on 137kHz, I had problems with noise and QRM as well. My best SL configuration used function "reassigned diagram". Using this function gave very good results. Would be usable on VLF as well, or is it not suitable for QRSS600? I cann´t try it, becouse I haven´t any control signal on 8970kHz.


73, Stefan/DK7FC


Have a nice last some days of 2010 year!

Lubos, OK2BVG



<Prev in Thread] Current Thread [Next in Thread>