[RTTY] Question for the Experts

[Bill Turner]

On Tue, 21 Aug 2001 17:01:45 +0100, Mobile Workshop wrote:

>If you remove the dash, then it often comes up as TOOAPPQAPPQ where the A is
>the more often received character, but it could be anything.

_________________________________________________________

One of my favorite RTTY issues!  Forgive me if I repeat stuff I've said
before, but I do believe it is important, especially to newcomers.

The key to the issue is a type of operation called unshift-on-space
(USOS).  USOS is used by most TNC's by default, but there is a little
quirk of contesting which can make it ineffective.

Here's a little background:

As you are probably aware, the Baudot character set uses a figures shift
character and a letters shift character to change between letters and
figures and other punctuation.  This is needed because Baudot only uses
five bits and five bits can only display 32 characters - not enough for
all the letters, numbers and punctuation.  For example the letter Q and
the number 1 are identical in Baudot - except the 1 is preceded by a
figures shift character.  This is a non-printing character of course;
you don't see anything on your screen.  It is merely an instruction to
your receiver to treat what follows as a figures-shifted character
instead of a letters-shifted one.  Likewise, when you return from
sending numbers to sending letters, a letters-shift character is sent.
This is how your receiver keeps things straight.

The problem comes about when some QRN, QSB, QRM or whatever causes a
loss of the shift character.  Baudot was originally devised for wired
communication where this is not a problem.  With radio, unfortunately it
is.

In an attempt to correct the problem, USOS was devised.  What happens in
your receiver is that when a space character is received, the receiver
resets itself to letters-shifted characters.  This works because your
USOS transmitter knows that after sending a space character, if the next
character is a number, the figures shift character has to be sent first.
The reasoning behind this is that letters are much more common in
ordinary conversation than numbers are and it works quite well for
ragchewing... but contesting is a little different.

Often in a contest exchange, what follows a space is a number, not a
letter.  The problem (follow me closely here) happens when the
transmitting station does not send a space before the number, but rather
sends a hyphen.  For example, 599-001-001.  If the ORIGINAL figures
shift character is lost, the receiver will print TOOAPPQAPPQ, the hyphen
being a figures-shifted "A".  In the same situation, but with a space in
place of a hyphen the receiver would print TOO 001 001.  The missing
ORIGINAL figures shift character causes the 599 to print as TOO, but the
subsequent (and more important) serial number prints correctly, thanks
to USOS.

The reason people use hyphens is that the transmission goes faster - a
hyphen character takes the same time interval as a space, and (here is
why it's faster) the hyphen is also on the figures shifted set so
therefore does not require sending another figures shift following the
reset to letters shift caused by USOS.  I don't blame you if you have to
re-read that a few times.  It is anything but readily apparent.

To summarize, hyphens make the transmission faster, but spaces along
with USOS make reception more accurate when QRM, QSB or QRN is present.
I can understand the folks who want a fast transmission - in a contest,
faster is better IF IT'S ACCURATE.  But if it isn't accurate, a lot of
time is wasted asking for repeats.  

Anyway, I prefer accuracy, so you will always see my exchanges with
spaces.  For the folks who prefer speed, may I suggest setting up an
F-key which repeats only the significant part of the exchange?  You'll
be using it more than you would otherwise.  :-)

If you'd like to see the actual Baudot character set, go to:

http://www.dataip.co.uk/Reference/BaudotTable.html

Hope this helps.

73, Bill W7TI