[Lf] [Fwd: LF: RE: Synchronised QRSS]

[Andre' Kesteloot]

Talbot Andrew wrote:

> The data entering an FFT process is premultiplied by a waveform -
> usually something like a raised cosine - to remove the sudden end
> effects of arbitrary start and stop of the samples.   The effect is to
> reduce the resolution, giving an effective bin size greater than the
> actual value, but reducing sidelobes caused by frequencies sitting
> between bins so there is now no need to position signals exactly in
> bins.  Choice of window type allows trade off between resolution and
> sidelobe level.  eg the Hamming window gives an effective  bin size 1.85
> x the raw value and a sidelobe level of 48dB maximum.  The Blackman
> Harris window however, gives a bin size nearly three times the 'raw'
> value but at -86dB sidelobes.
>
> By matching FFT overlap + resolution with signalling rate (dot length),
> we already have an optimum signal detection system based on pure power
> detection alone.  Jiggling about with precise timing intervals will tell
> exactly where on the screen a dot or its absence should be, but not
> whether it is there or not.   There cannot be a gain in SNR just by
> playing about with timing.  An improvement can only come about by either
> reducing signalling rate and resolution bandwidth together, or going to
> coherent carrier recovery techniques so signals can be treated
> vectorially rather than as scalar amplitudes.   This means (B)PSK.
>
> Sorry folks, but the laws of Physics, Shannon,  Nyquist and Murphy all
> intervene !   If you want a signalling system with more performance than
> 3s SLOWCW, it will either have to be 30s SLOWCW   to gain 10dB for the
> same power or  -  shock horror  -  build / buy some extra hardware and
> go to PSK.   There are no quick fixes.
>
> Andy G4JNT
>
> > As well as synchronising the dots and dashes of QRSS with the
> > receiving FFT software, might it be possible to control the TX and
> > RX frequencies with sufficient accuracy that all the received signal
> > would end up in one FFT "bin"? This would then eliminate the
> > smearing of the frequency axis. I guess this would require TX and
> > RX frequency accuracy of about 1ppm, similar to BPSK.
> > >
>
> --
> The Information contained in this E-Mail and any subsequent correspondence
> is private and is intended solely for the intended recipient(s).
> For those other than the recipient any disclosure, copying, distribution,
> or any action taken or omitted to be taken in reliance on such information is
> prohibited and may be unlawful.