[ILUG] Shannon's law (was : Genesis)

[Paul Jakma]

On Tue, 7 Mar 2000, Kenn Humborg wrote:

  To properly represent an _analogue_ signal with discrete samples,
  you must sample at at least twice the highest frequency in the signal.

oops yes, of course. why would anyone want to construct a bitstream
of rate 2Y by sampling a bitstream of rate Y? :) doh!

(my brain is starting to overheat i think, must get some sleep).
    
  > also, 6.2KHz seems a very poor figure for even the shittiest copper
  > line, doesn't it?
  
  It's not really the line itself (although shitty _long_ lines are a 
  contributor).  It's the circuitry at each end of the line.  Remember that
  the 'protocol' (i.e. electrical interface) at that phone socket on your
  wall goes back about 70 years.  Specifications like 'how to tell a ring
  signal' and 'how to signal off-hook' are remarkably primitive.
  
  There are also filters at various points to limit the signal to the 300
  to 3400Hz band.  Nasty, evil, 12-pole filters...

yukk.. any reason why? (part of the telco spec i guess?)
  
  Later,
  Kenn
  
  PS.  Funny that we're talking about phones.  My phone was dead
       when I came home this evening.  Called in a fault report and
       now it's working again.  Pretty good considering the rep said
       no one would be available until Wednesday morning :-)
  
  
  

-- 
Paul Jakma	paul at clubi.ie
PGP5 key: http://www.clubi.ie/jakma/publickey.txt
-------------------------------------------
Fortune:
Alea iacta est.
	[The die is cast]
		-- Gaius Julius Caesar