On Jan 14, 2012, at 2:35 PM, Jerry Flanders wrote:
> The only way I can be _sure_ I have it right is to see it 3 times. If
> two out of three agree, I probably have it.
Unless the error rate is very high, repeating twice (CA CA) is good enough for
error *detection* and repeating three times is good enough for error
*correction.*
IMHO, a repeat of 2 is a good compromise. In the moderately rare event of a
print where the two don't agree, the receiving station can ask for a full
repetition. Doing it three times is mostly a waste.
Of course, repeating is a really primitive and inefficient method compared to
true error correction; and will not withstand a deep QSB of more than 1 second.
With a deep QSB, you are likely to lose frame synch, so even a repeat of three
times may not print correctly anyway. Even Amtor/SITOR-B (same FSK emission as
used on RTTY) has a very decent error correction scheme for pretty long QSB; at
the cost of longer decoding latencies.
For those who are fond of sending 599 twice, sending "599 CA 599 CA" is of
course better than "599 599 CA CA" due to the behavior of RTTY under Rayleigh
fading. Not that I ever advocate sending the 599 twice, or even worse,
pretentiously calling it "RSQ." :-) I'm just trying to demonstrate the likely
behavior of Baudot decoding under HF propagation.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|