Dear Alberto, LF Group
At 14:30 16/01/2002 +0100, you wrote:
So, what would all you judge more important, sideband independence,
or bandwidth used ? Your vote will decide the final format for Jason.
Provided the software is able to cope with receiving the signal as either
USB or LSB (for example. by having a user option to change the sign of the
measured frequency differential), I don't think having to select the
sideband would be any problem - I think reduced BW would be useful; it
would make it easier to avoid QRM like Loran and make the signal slightly
easier to generate.
Also, from your previous e-mail:
This [eliminating the bit which defines the nibble as 1st or 2nd in the
character] would gain one bit each nibble, increasing the alphabet from 64
to 256
symbols. For keyboard-to-keyboard communication, I think 64 symbols
are enough.
I was thinking more that this would mean 2 x 3 bits for each character
instead of 2 x 4, reducing the number of tones required by 1/2, and so
further halving the bandwidth. Or the "spare" bit could be used to include
some sort of error correction, perhaps.
And, regarding the 16 tone encoding scheme:
Hmmm, suppose I have just sent tone 13. Next, I have to send a delta
of 16 (the deltas range from 1 to 16). 13 + 16 mod 16 = 13, so I would
end up sending the same tone, with a delta of 0. Or maybe I misunderstood
what you said ?
That would be no problem for 16 codes - you only need to re-define possible
values of delta as 0 to 15 instead of 1 to 16. If it is important that two
successive tones never have the same frequency, ie. there can be no delta =
0, you would need 17 tones - then in the example it would be 13+16 mod 17 =
12; perhaps this is something like what Stuart is suggesting?
Cheers, Jim Moritz
73 de M0BMU
|