[Lowfer] Spatial and time diversity for QRSS

[craig wasson]

John;

Maybe I'm reinventing the wheel here.  Is the summing I propose exactly the
same thing my Spectrum Lab software is doing when I tell it to do a FFT
with an input size or length of say 512?   Is it taking 512 samples 1 cycle
apart and doing some sort of summing/averaging on them?  In this case I
assume each sample is one cycle at the frequency I'm analyzing.

So for a 1000 hz audio decode of a QRSS signal a FFT length of 512 is
looking at about half of a second per bin?  Or is it not that simple.  I'm
still struggling to understand decimation.  I see now that I shouldn't have
quit taking math in college after differential equations.

I guess my idea was to take advantage of the fact that a QRSS signal is
pretty constant over the length of a dot, so I could repeat the whole FFT
process over and over for a few seconds without blurring the edges too
much.  But maybe I'd get the same effect by increasing the FFT length to
several thousand?

Andy - thanks for reminding me about "stacker.  I had looked at the
"stacker" program a long time ago, but had forgotten about it.  It does do
what I was talking about with time diversity summing at the presentation
layer - which eliminates all the issues with phasing.  I originally got
frustrated with the two-step process of first stitching then stacking -
maybe I'll suggest the developer combine those and have it produce a more
real-time argo-like display.

The spatial diversity I will look into more.  I have an SDR-MK1 receiver
that has two independent tuners running off of a common clock - so the
output of the two should be in phase.  I can combine the signals from two
different antennas of two different types and located in different parts of
my lot.  At LW the signals should stay in phase even if the antennas are a
few hundred feet apart and the signal is in line with the two antennas.  I
should just need to run the two outputs through a mixer and into Spectrum
Lab, although from what John said the best I can expect with two phased
receivers would be a 3db improvement.  Maybe 3db will make a difference?

I use the computers doing all of this signal processing to help me heat the
house in winter, so I'm getting ready to try some new ideas.

Craig - N6IO



On Thu, Nov 5, 2015 at 12:12 AM, JD <listread at lwca.org> wrote:

> Hi Craig. White noise differs from other random noise mainly in its
> spectral composition.  Summing two random sources of equal level doubles
> the noise power (3 dB), whereas summing two coherent sources in phase
> doubles the signal voltage (6 dB).  Thus, if you can achieve phase
> coherence of two copies of the signal, and the noise accompanying each is
> random, then you get a net 3 dB S/N advantage with each doubling of
> sources.  If the noise is not truly random in the two copies (ie, powerline
> buzz) or the signal cannot be lined up in precise phase match, the
> advantage will be less.
>
> I'm not sure about your time diversity ideas where the same signal is
> overlaid on itself with time delay. That will null certain specific
> frequencies each time, according to the delay period, but it will also
> introduce smear in the signal rise and fall times, and has the need (as you
> noted) to keep the signal carrier in phase with itself or it too will
> suffer cancellation part of the time along with the noise.
>
> The third time diversity option I'm testing is to fold in 2 or more copies
>>>> of the same QRSS signal from adjacent iterations of the full call being
>>>> sent.
>>>>
>>>
> Doing this with adjacent iterations of the call at the display stage can
> be quite effective.  As you noted, NDBfinder was a good example, but it
> really wasn't designed for QRSS.  I'm not sure if anyone has been working
> on an improved version.
>
> I wish I were enough of a software guru to take a long-term set of FFT
> data from a spectrum program and try overlaying the bins periodically to
> display a summed result.  It'd take some user input to line up the
> instances, though, unless the repetition rate were known very precisely;
> and of course, if the transmitted signal or the receiver drifts in
> frequency during the capture it wouldn't work at all because adjacent
> iterations wouldn't be depositing energy in the same set of bins.
>
> John
> ______________________________________________________________
> Lowfer mailing list
> Home: http://mailman.qth.net/mailman/listinfo/lowfer
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Lowfer at mailman.qth.net
> Post must be less than 50KB total for message plus attachment!
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>