Here is something I wrote a few years back:
Now a contest with serial numbers, for instance, CQ WPX RTTY.
>
> K4GMH 599 328 328 (K4GMH)
>
> Some people say send serial number once. Others say twice. Still others
> say three times because if you get it two out of three times, you’ll know
> it’s right. I normally send serial number twice, and below I show how this
> works well. But the real answer is it depends. If you are loud and running,
> people get more than one chance to figure out your serial number, and once
> may be good enough. Caution—this may not work if the running station is
> SO2R where serial numbers might not be consecutive. If conditions are
> particularly poor, three times might be called for. It depends, which is
> why having flexible macros is a plus.
>
> Which brings us to the question of when to ask for a repeat. When do you
> have enough to log a contact? It depends. You don’t have to copy a number
> twice to be confident that you have it right.
>
> If I print 599 005 W6SX, the 599 and W6SX bracketing the 005 tell me it’s
> probably good. I’ll log it unless something else tells me to question it.
>
> One “something” is that you can often tell by ear if an exchange printed
> correctly. Or, perhaps more importantly, you can often tell by ear if an
> exchange is corrupted.
>
> If I print 599 005 0%&*, I’ll probably log 005.
>
> If I print 599 005^*A, I’ll ask for a repeat.
>
> There are lots of scenarios you can build. Contest radiosport is risk
> evaluation. Do I interrupt a 100-per-hour run to be 100% sure of an
> exchange? Or do I evaluate as above and take a chance I’ll lose a low-point
> QSO to log checking? Does my calculus change if it’s a new multiplier I’m
> not likely to find later? Of course.
>
Contest Exuberantly,
Hank, W6SX
On Sun, Jan 10, 2016 at 11:51 AM, Kok Chen <rtty@w7ay.net> wrote:
>
> On Jan 10, 2016, at 11:28 AM, G3YYD wrote:
>
> > If the checksum character is in error?
>
> Send "AGN?" of course :-).
>
> You can even automate the "AGN?", but with poor propagation, the back and
> forth could end up sounding like Packet Radio :-).
>
> Sending the serial number twice is a form error detection, that is why we
> use it today, but it is rather inefficient. And the response to an error
> is similarly, either "I don't need that mult," or sending "AGN?"
>
> Sending the exchange 3 times allow a single error to be corrected, but,
> like repeating something twice, is also inefficient. Even the rudimentary
> Hamming code is better for correcting single errors.
>
> The "3 copies" methodology has even been used in hardware in spacecrafts.
> Avizienis at JPL called it "Triple Modular Redundancy." Nowadays, in the
> same vein as error correction codes like the Hamming code (the grand pappy
> of them all), Arithmetic Codes are much more efficient hardware
> implementation than triple modular redundancy.
>
> Happy New Year 2016, David!
>
> 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
|