Content-Transfer-Encoding: 8bit
Message-ID: <5.1.0.14.0.20011205110343.00a863e0@gemini.herts.ac.uk>
Date: Wed, 05 Dec 2001 12:56:14 +0000
To: rsgb_lf_group@blacksheep.org
From: "James Moritz" <j.r.moritz@herts.ac.uk>
Subject: Re: LF: Re. Improving extreame weak signals.
In-reply-to: <000b01c17d71$5ce0d720$7273883e@g3aqc>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dear LF Group,

At 09:41 05/12/2001 +0000, Laurie wrote:
>...but it would seem sensible to have defined time and
>frequency "slots" into which the signal could drop,so that all the available
>energy could be used rather than it be spread out in timeand frequency.

I think we have been here before ... If you chose the optimum resolution 
for the speed of QRSS you are using, you will get the best signal to noise 
ratio between "dot" and "no dot" when all the samples containing a signal 
are used for the FFT algorithm that calculates the spectrum of the signal. 
This suggests you would get the brightest dot on the screen if the start of 
the FFT is made to coincide with the start of the dot. But in fact this 
more or less happens already; suppose you have 30 s dots and have set up 
the FFT to use 30s worth of samples. Normally, the spectrogram software 
will perform an FFT at least once every few seconds, let's say 3 seconds, 
using the received signal from the previous 30s. Even if the relative 
timing of the dots and FFT is totally un-synchronised, at least one FFT 
will start within 3 seconds of the start of the dot being transmitted, and 
so be very close to optimum. The FFTs performed on data from before and 
after the optimum time will contain less signal, so the dot displayed on 
the screen will fade in and out, reaching a peak of intensity in the 
middle. The only effect of synchronising the start of the FFT with the dots 
will be to eliminate the sub-optimum FFTs, and retain the optimum one where 
the timings coincided - the effect would be the same as superimposing an 
opaque mask on the screen with slots coinciding with the timing of the 
transmitted dots. I suppose this would not be hard to do, but would it be 
an advantage? You would get rid of some "clutter" on the display, but it 
would be harder to tell when bursts of noise and so on had occured. The 
display would not actually contain any more signal information, though.

I think Rik's idea of displaying 2 tones differentially ought to work, but 
it would place quite stringent demands on frequency stability. The current 
spectrograms are not too fussy about exact frequency, so long as the drift 
is smaller than the FFT resolution during one dot period, and the signal 
stays on the screen. But if we were to compare two tones, it would require 
accuracy in the frequency shift that was smaller than the resolution of the 
FFT - a few millihertz with the longer dot lengths. Judging from last 
year's experience with Wolf, it is quite hard for this kind of accuracy to 
be set up and maintained throughout the transmit/receive system, when the 
equipment being used includes amateur-type rigs, sound cards etc.

BTW: I hope to get my DFCW modulator finished as soon as I get a bit of 
time to do it - should be a real advantage for someone whose callsign is 
mainly dashes!

Cheers, Jim Moritz
73 de M0BMU