On Thu, Feb 04, 1999 at 09:04:09PM +0100, Morgan Olsson wrote:
> >1) Anybody got a pointer to where to learn about ethernet protocol and
> >bus?
> 
> OK,  IĞll clarify myself:  I am not interested in the higher layer
> protocols, as we have some odd requirements.   What we really are looking
> for are the chips, controllers etc, and possibly the lowest error detection
> and correection etc layer where included  (like CAN).
> 
> We could roll our own, but if there already exist good chips, why bother.
> And for established standards there are also nice debugging tools :)
> 
> Especially I am looking for Ethernet specification.
> 
> Who owns Ethernet?  (I think i shold look there)

Ethernet is a trademark of Xerox, having been developed
by them in conjunction with DEC and Intel. It is an
implementation of an even older network protocol idea,
called Carrier Sense, Multiple Access with Collision
Detection (CSMA/CD). Ethernet, as originally specified by
those companies, was a baseband, broadcast network carried
over a thick, 50 Ohm Coaxial cable, which almost without
exception had a yellow PVC jacket. It was also usually
marked, by hand with a felt-tip pen, every -- I think it
was -- 2.5m. One attached stations to these cables (at the
marks to minimize destructive reflections) either through
terminating the ends with huge, nasty solder-on connectors
and running the ends into two sides of a tranceiver, or
cutting away a little circle of jacket and shield on one
side and clamping on a tap that drove a probe down into
the dielectric to touch the center conductor; these were
"vampire taps". But enough of that.

Eventually, the IEEE set up a working group on Local &
Metropolitan Area Networks; this was (is) the 802 committee.
The 802.1 standard covers common definitions for all
LAN/MAN technologies, and inclues things like bridges.

802.3 is the standard that defines the CSMA/CD access
method. 802.4 is the token-passing bus, which is sort of
like an Ethernet bus but with a token-passing protocol
imposed on it; it was mostly used for manufacturing
networks. 802.5 is the token passing ring.  802.6 is the
Distributed Queue Dual Bus access method; these access
method standards go up to 802.12 or so by now.

All these standards may be purchased from the IEEE,
check out http://standards.ieee.org/catalog/index.html.
Some documents are less than $50 or so, some cost
hundreds of dollars.

There are subtle differences in framing between Ethernet
(TM) and 802.3 networks, but (a) one hardly ever sees
real Ethernets anymore, and (b) most "ethernet" MAC chips
can handle either framing.

One Ethernet chip you might take a look at is the Crystal 8900
(see http://www.cirrus.com/products/overviews/cs8900a.html).
I've seen it used in some embedded designs (e.g. some
JumpTec [http://www.jump.de] PC-104 CPU boards that I've
used have them), and it seems to be very small and consume
little power.  Cirrus Logic seems, from what I can tell,
to be pretty forthcoming with the technical information
one would need to make them work (others can correct
me if I'm wrong about this). I note, for example,
that they have a Linux driver -- in source code form --
for the chip available on their website. This made it
possible for me to get Linux running on a Jump board,
and I sent them a big thank-you for that.

I suppose that one hurdle you might have to deal with
is that most Ethernet chips will probably have an
on-chip ISA or PCI interface. If you can straightforwardly
glue the ISA interface into a PIC, it might be pretty
easy, as you suggest.  But I'm no expert on this kind of
stuff and could be totally wrong about this.

>From what you say you are doing -- looking for some sort
of moderately fast protocol with a pre-existing and easily
tapped knowledge base and easily obtained, cheap silicon,
one book (of many, no doubt) that you might find helpful is
the Principals of Computer Networking, by D. Russell. It is
an older book, first published in 1989, but I have never
found another book that does as good a job of explaining
networking technology from a historical and evolutionary
perspective; after reading that book, a long time ago, I
found that networking made a great deal more sense to me,
and I found it much easier to see similarities in protocols
that before that seemed all completely different to me.
It is written well enough to count as recreational reading,
I highly recommend it.

http://www.amazon.com/exec/obidos/ASIN/0521339928/qid%3D918175230/002-9133948-5405425

Hope this helps,

--Bob

-- 
============================================================
Bob Drzyzgula                             It's not a problem
bob@drzyzgula.org                until something bad happens
============================================================