Content-Transfer-Encoding: 8bit
Message-ID: <200012211940.OAA11457405@indyweb.cgocable.ca>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Date: Thu, 21 Dec 2000 14:38:48 -0500
To: rsgb_lf_group@blacksheep.org,  bpsk@qth.net
From: "Bill de Carle" <bill1@cgocable.ca>
Subject: LF: AFRICAM V2.1
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Greetings to everyone.

Just uploaded latest AFRICAM (V2.1) to my web page at:

http://cafe.rapidus.net/bill1/bbs.htm

AFRICAM 2.1 has some improvements, some bug fixes and some additions.

Under poor SNR conditions it is difficult to accurately measure the
phase (frequency) of a received signal.  The older versions used to
measure frequency by measuring the phase in the middle of each bit
time relative to a reference frequency (fcar) - then by noting how
much the phase drifts upwards or downwards over time, the offset
relative to that reference frequency was obtained.  That technique
works fine when the actual frequency being received is very close
to the entered fcar value (nominally 800.0 Hz) - but it becomes
increasingly difficult to get a good reading in noise as the received
frequency moves away from its nominal value.  That's because the
energy response is way down the skirt of the Integrate & Dump filter
except right at the nominal frequency.  The latest version takes
this into account by measuring the incoming frequency at a reference
frequency equal to the best guess we have obtained so far.  As the
reference freq gets closer to the actual received frequency the measure-
ment improves.

This means you will likely have more success using the AUTOFREQ system
than before.  If you can hear the signal at all, I recommend you try it.
If there is no audio trace of the signal you'll probably still be better
off to set the rig right on frequency as best you can and keep your
fingers crossed, hi!

The SLOG command line parameter wasn't working properly.  It works now.
In case you don't know what that did, it writes the measured phase value
at the center of each received bit to disk - format is an integer, which
represents the measured angle in 1/10 degrees 0..3599.  That data can
subsequently be plotted using PLOTPHAS.COM - you can see how noisy the
phase determinations are, which gives a good idea of what bit error rate
to expect.

The SHOWPH command (shows reference phase (modulo 180) on screen - now
averages over each frame - tends to be less jittery.  Probably more useful
for the laser ranging folks.

Added support for ET3 - that's a mode similar to the other ET modes but
more robust in noise - it uses a 32-bit frame.  Each symbol differs from
all others in the set in at least 13 ways.

Added support for non-differential encoding for all the ET modes.
This is still experimental - you shouldn't use it unless you know the
transmitter is sending non-differentially encoded BPSK.  The advantage
is it's more robust in QRM - disadvantage is there is a 2-way ambiguity
in the reference phase value which at this point has to be resolved by
trying both possibilities (the FLIP button does that).

At present the only way to generate non-differential ET is with a keyer.
I've upgraded MATIC to MATIC V4.6 - it's in the KEYER.ZIP file on my web
page if anyone wants to try it.  It now supports the new ET3 codeset with
both both differential and non-differential encoding options available.

I am currently transmitting a short test message on 6776.8 Khz.  That's
a license-free band available in Canada.  It's just below the 40-Metre ham
band, so propagation ought to be similar.  The parameters are: ET3, diff,
MS10, run-length 31.  ERP is less than 5 milliwatts!  If you want to try
copying the message, invoke V2.1 of AFRICAM with "SB ET3 MS10".  If you can
hear the signal, set TC=10, TCF=5 and set AUTOFREQ ON.   If you can't hear
the signal, use the default settings.

Would appreciate signal reports,
Bill VE2IQ