🚀 go-pugleaf

In article <42qk8c$a8n@anarchy.io.com>, Jim McCoy <mccoy@io.com> wrote:
>Well, sort of.  The current phone system samples signals at a rate of 8 KHz,
>and so there is a finite limit on how much information can be sent over
>a signal such as this

Actually, it's not even "sort of".   It's just plain wrong.  A
conventional analog phone line has a bandwidth of about 3000Hz.  The
8KHz has to do with the sampling rate required to represent this analog
signal as a digital one.  (The Nyquist rate.)  It's quite possible the
analog signal passes from beginning to end without ever being converted
to a 64Kbps digital signal.  (Well, not really.  In this age of digital
switching, it's just about a certainty that the signal gets digitized
at the central office, if not sooner.)

>the Shannon limit, named for the the Bell researcher who discovered it and
>invented the field of information theory at the same time.

Yep, that's it.  For a typical "voice grade" circuit, the Shannon
Limit is 24,000 bits/second.  (This result falls out from plugging a
few values into a simple formula.)

One of the things that effects that value is the line noise.  Thus,
science says exactly what people report:  when things are very, very
good you can get 28.8K.  That is, however, pushing the envelope, and
frequently you will get less.  Sometimes much less.

I strongly recommend |Understanding Data Communications| (published
by Sams) for anybody interested in this stuff.  Earlier editions were
carried by Rat Shack.  I bought the 3rd ed. a few years back at
Bookstop.  It does an *excellent* job of explaining the technical
fundamentals of data communications, particularly modems, without
getting bogged down in esoteric engineering.
-- 
Chip Rosenthal              Ban excessive reposts: /^X-Mailer: Mozilla/h:j
Unicom Systems Development
For a good time:  http://www.unicom.com/john-hiatt/
PGP key:  http://www.unicom.com/personal/chip.html