From: utzoo!decvax!harpo!esquire!brl-bmd!wmartin
Newsgroups: net.ham-radio
Title: Re: Thoughts on CW
Article-I.D.: brl-bmd.453
Posted: Wed Nov 24 16:30:01 1982
Received: Thu Nov 25 10:06:25 1982
References: sri-unix.4329

I am somewhat confused as to the traditional concept (i.e., I read it before
somewhere more than once) of "bandwidth used = information transfer rate".
Let me postulate this: A very slow CW signal is transmitted by generating
an RF signal 100 Hz in bandwidth, switching it on and off once per second
to make a second-long dit. This on-off switching speed is increased, assuming
the transmitting and receiving equipment can handle and detect it, to ten
times per second, then 100 times per second, then 1000 times per second.
Instead of one-second dits, we now have one-millisecond dits, right? All
along, the signal has been 100 Hz wide, because the SIGNAL has never been
changed, all we have been doing is turning it on and off faster. So our
information transfer rate has multiplied by a factor of 1000, but the
bandwidth is unchanged. Couldn't this process continue, reducing the dit
length to a microsecond, and then a nanosecond, until the pulses are so
short that they cannot be distinguished from a continuous signal? The
bandwidth is still unchanged, isn't it? Or is it some magic effect of
the switching that turning something on and off at a higher rate will
somehow increase the bandwidth it occupies? If it does, is this some
effect like splatter? If so, isn't eliminating it just an engineering
problem, not a theoretical issue?

Hmmm...  Will Martin