only true if the number is sent once only;
with 599 123 123 the probability that both USOS
characters are hit is just marginal..
I vote for "no dashes"!
Andy, DF4WC
-------- Original-Nachricht --------
> Datum: Fri, 08 Feb 2013 08:45:26 -0800
> Von: Kok Chen <chen@mac.com>
> An: RTTY Reflector <rtty@contesting.com>
> CC: Bill Turner <dezrat1242@yahoo.com>
> Betreff: Re: [RTTY] WPX suggestion
> On Feb 8, 2013, at 4:19 AM, Bill Turner wrote:
>
> > Please, please, please, NO dashes. While it does speed up sending, it
> > interferes with the USOS (Un Shift On Space) function which most folks
> > agree is a much better thing to have. USOS greatly reduces requests for
> > repeats, especially when there is QRM and QRN.
>
>
> It is another myth that an extra character helps when you are transmitting
> all numerals.
>
> The extra FIGS that follows the Space actually hurt. The reason is this:
>
> if the probability of the error of a single character is Q, then the
> probability that you do *not* have to repeat an exchange of N characters (all
> N
> characters survives) is (**N is the power of N)
>
> 1 - (1-Q)**N
>
> For example, if Q is 0.1 and N is 8 (e.g., 599-123, i.e. transmitting
> <figs>599<dash>123), then the above probability is 57%.
>
> By transmitting that extra character, the probability that all characters
> are received correctly is now
>
> 1 - (1-Q)**(N+1)
>
> Or, using the same probabilities, the probability of not having to repeat
> is now 61%.
>
> The above with random errors.
>
> Lets see what happens when a dash is specifically replaced by Space that
> is followed by a FIGS, and we allow partial reception to count (i.e., the
> other person can still see the 3 digit that what was sent correctly -- it not
> safe, but I present this in case it is further argued that a FIGS somehow
> will magically transform a 599-123 into something readable again).
>
> Recall that the dash can be received wrongly, but unless it turns into
> LTRS, it will just be printed as some other FIGS character, e.g., into a dot.
> Recall that a dash is Baudot 0x03 and a LTRS is a Baudot 0x1F. The
> Hamming distance is 3. So the probability of a dash turning into a LTRS is
> very
> low. I.e., if we use the same Q above of 0.1 (10% of the characters are
> wrong), the probability of a dash turning into a LTRS is 0.3%.
>
> Lets see what happens when you send the Space followed by a FIGS. The
> Space can still get mangled and turn into something other FIGS character. But
> lets look at the FIGS that has to also be sent. Given the same Q as
> above, that probability that the FIGS can be turned into a non-FIGS character
> is
> 9.7%! When that happens, you will receive "599 xQWE" where x is
> anything but a FIGS character.
>
> What if the space is mangled? well, it is the same as if the dash is
> mangled... and guess what, you print exactly the same thing as what dash
> message. I.e., If the space is mangled into a dot, you copy 599.123, which
> you
> also copy 599.123 if the dash turned into a dot. Remember that "123" only
> prints incorrectly for the dash guy when the dash turns into a LTRS, and the
> receiver prints 599QWE.
>
> The only other possibility is if the preamble FIGS is received as
> something else (say, as a dot).
>
> The Space guy is received as ".TOO 123" and the Dash guy is received as
> ".TOO AWE" Most of us (and some software) will know what either means, and
> not need a repeated exchange, but a newcomer to RTTY will be scratching his
> head in both cases, and ask for a repeat.
>
> So, do yourself a favor. If the exchange consists of only numerals, use a
> dash in place of a space. If the exchanged is mixed, e.g., "599 NH" then
> don't use a dash. But, don't do it blindly. Dashes are not always
> better, but in the case that the original posted presented it (all numerals)
> it
> is better.
>
> 73
> Chen, W7AY
>
>
>
>
> _______________________________________________
> RTTY mailing list
> RTTY@contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|