Message-ID: <002f01c17db2$4c388df0$0400000a@parissn2>
From: "Stewart Nelson" <sn@scgroup.com>
To: rsgb_lf_group@blacksheep.org
References: <003601c17c07$4c58fc40$587e883e@g3aqc> <3.0.1.16.20011204101542.2bff5384@pb623250.kuleuven.be>
Subject: LF: Re: Improving extreme weak signals
Date: Wed, 5 Dec 2001 18:28:45 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Hi Rik and all,

The current DFCW format is really a ternary signal.  A space (interval
of no signal) must be present between characters, because Morse is
a variable length code.  One must distinguish the presence or absence
of a tone, and the amount of energy present at the other frequency does
not help that distinction.  For example, your call would be sent as
"--- -. --... -.-- -.. " ; QSB at position: 2,9,10,11,15,16, or 20
results in valid Morse (and incorrect copy).  Errors in other positions
would be detected, because invalid codes result, but it would be
quite hard to use that information to enhance a "for human decode"
display.  If the Tx were capable of transmitting three frequencies,
and the third assigned to the space function, one could compare the
energies in each of the three bands, and choose the strongest as the
most likely sent.  That would give most of your 3 dB advantage, and
yet be compatible with current practice.

However, if you have the ability to send multiple frequencies, a much
larger gain is obtained by using a different frequency for each letter
or digit.  This gains at least 6 dB over the system above.  If the tones
were spaced 0.1 Hz apart, then your existing 4 Hz screen could show an
alphabet of up to 40 symbols.  The software could display corresponding
letters and digits on a scale, but the operator would still be deciding
which characters were sent.  It would also work with existing software,
by looking up the received frequency in a character table.  This scheme
has been described and tested by many hams, under names such as PUA43,
Piccolo, AFK, and PGP-1.

If the message length and start time are agreed upon in advance, then
systems like PUA43 offer a huge advantage, because you know what part
of the message is being sent at any time.  If you get partial copy,
you know which character(s) were received, and you fill in a piece of
the puzzle.  After several repeats, if all characters were received
at least once, the puzzle is solved.  Such a system would also permit
advanced software to combine the received energy from multiple message
repeats, forming a composite with better S/N.

The variable length nature of Morse thwarts the above idea, but perhaps
a similar code using say, four elements per character could be used.
This code would always start with a "." or "-", but could have a space
as the second, third or fourth element.  Letters with 4 dots and/or
dashes would be sent without any following space.  Letters with 3, would
have a space at the end.  One or two dots would pad out shorter letters.
There are 14 valid (per above rules) codes left over; these would be
assigned to word space, digits, and 3 punctuation.  For example if
"- .-" were assigned to "7", then your call would be sent as
"--- -. .- .--.---.. " which is two elements shorter than standard Morse,
but the main idea here is not to be faster.  It is that all characters
take exactly 8 minutes to send; if you receive "-.--" 24 minutes after
the start, you know that "Y" is the fourth character.

One could also develop a code which is a compromise between Morse and
PUA43.  For example, if seven frequencies were used, one could use two
tones, sent sequentially, to represent one character.  That would be
twice a fast as the above, more robust (no space), would fit in a 1 Hz
bandwidth, and could easily be decoded by eye.

Let me know if you think that any of these are promising.

73,

Stewart KK7KA