From: utzoo!decvax!cca!hplabs!hao!menlo70!sytek!zehntel!teklabs!tekcrd!azure!michaelk@sri-unix
Newsgroups: net.ham-radio
Title: Re: Thoughts on CW
Article-I.D.: azure.1502
Posted: Sun Nov 28 20:28:18 1982
Received: Wed Dec  1 06:05:43 1982

Running CW at 0 w.p.m ideally gives you a bandwidth of zero hz (Awful
clean, stable xmtr). When you key it, you are modulating it, producing
sidebands.  If you increase the w.p.m rate, the bandwidth increases. If
your transmitter doesn't "waveshape" the keying action 
then the bandwidth increases much in the way the modulating square wave
has a bandwidth wider than it's fundamental. At 20 w.p.m. (assuming
suitable "waveshaping") the bandwidth is still under 100 Hz or
so.  By the way, bandwidth does not determine the information transfer
rate, it only limits it.  Human voices (like in SSB mode) are not
efficient in bandwidth use. In the second I might say "hello", I could
send 120 characters with my 1200 baud modem that uses the same (or
smaller) bandwidth . It would take some three seconds to send that using
CW at 20WPM, but at 1/25 the bandwidth, yielding a net 8 times better
use of bandwidth.  Further, sending English language by any mode
also cuts down the information transfer rate -- English contains much
redundancy.  An obvious example of this is your *NIX "compact" program.
You can make a text file 30% smaller WITHOUT loosing any information
content (it can be converted back, can't it?).
For more detailed, and exact numerical representations, get hold
of a fellow "into" Coherent C.W. -- They operate with VERY narrow
bandwidth, such that you can pull out copyable CW when you have
a noise to signal ratio (>1).

Mike Kersenbrock WB4IOJ
Aloha, Or.