Subject: Re: Manchester encoding w/ Cat-3 cable
From: Henry Spencer <henry@zoo.toronto.edu>
Date: Aug 01 1996
Newsgroups: comp.dcom.cabling,comp.dcom.lans.ethernet

In article <4tnkdd$p94@charlie.lif.icnet.uk> evansr@europa.lif.icnet.uk
(Richard Evans) writes:

>: Just a nit; but if ethernet uses differential manchester and it
>: basically doubles the bit rate by potentially including a
>: transition the beginning and middle of each bit, how does it work
>: over CAT3 tp? ...
>
>Manchester doubles the bit rate by leaving out the clock signal
>during the times when it is not necessary.

Note, though, that Ethernet uses Manchester II, aka Phase Encoding.
PE sends 0 as 10 and 1 as 01 (if I recall the polarity correctly), so
it always has a transition at the middle of each bit cell, and may
or may not have transitions at cell boundaries.  With the worst-case
data, all 0s or all 1s, it has transitions at all cell boundaries and
is a 10Mhz square wave; with the best-case data, alternating 0s and 1s,
it has no transitions at cell boundaries and is a 5MHz square wave.

PE has no smarts and no tricks like leaving clocks out when unnecessary.
It's a very straightforward signalling system, and it has a lot of good
properties, but it *is* rather inefficient because of its 1->2 expansion
of the data bits.  Modern RLL (run length limited) codes use more moderate
expansions, e.g. 5->7, with encodings carefully designed to preserve at
least some of the nice properties of PE.
-- 
 ...the truly fundamental discoveries seldom       |       Henry Spencer
occur where we have decided to look.  --B. Forman  |   henry@zoo.toronto.edu