Improving extreme weak signals]

[Andre Kesteloot]

Rik Strobbe wrote:

> Hello group,
>
> First of all my congrats to the ZL6QH team and W4DEX for setting a new
> milestone. Besides the new distance record it also shows that 136kHz DX
> progation isn't limited to northern (southern) latitudes.
>
> Now another item :
> In the recent 136kHz trans-pacific tests and the 73kHz trans-atlantic tests
> the mode was DFCW with 2 minute dotlength.
> I believe that both DFCW and these long dotlengths could make it possible
> to gain a few more dB SNR :
>
> 1. Assuming we use a 2 minute dotlength where the dot starts exactly on an
> even minute the receiving software could take advantage of this
> 'synchronisation' as it would exactly know over what period the signal has
> to be integrated. This would not only allow to get rid of the 'blur' but
> also strongly reduce the FFT calculation time (that could be used for other
> SNR improving handlings). Further it might be possible that the knowledge
> of the exact 'dot period' can allow a better discrimination between
> presence/absence of a signal.
>
> 2. With the current receiving software what you see on the screen is the
> strength of a signal compared to the noise. Using DFCW with a exactly known
> shift would make it possible to show the 'differential' signal. This should
> give a 3dB gain over QRSS.
>
> Eg : Assume that we could agree to a standard shift of 0.1Hz. Further
> assume that the receiving software shows a 4Hz segment (let's say 798 to
> 802Hz). So at the 800Hz line on the screen we would not see the absolute
> strength of the 800Hz component but the difference between the 800.0 and
> 799.9Hz components. If a 'dash' is transmitted (on 800.0Hz) the
> differential signal would be positive, if a 'dot' is transmitted (on
> 799.9Hz) the differential signal would be negative and if nothing is
> transmitted the differential signal would be (close to) zero.
> If we connect these differentential values to a good color pallette a DFCW
> signal would be shown on a single line (the 'dash' frequency) where a 'dash
> could appear as one color (eg. red) and a dot as another color (eg. blue)
> while noise (no signal) would appear as the inbetween color (eg. brown).
> We would have to get used to this kind of appearance of a DFCW signal, but
> the profit would be a 3dB better SNR.
>
> I wonder if the above makes some sense (I'm not a DSP specialist) ?
>
> 73, Rik  ON7YD