[TriLUG] OT: Home Depot and Cat 5

Rob Lockhart rlockhar at trilug.org
Wed Oct 6 00:19:28 EDT 2004


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Aaron S. Joyner wrote:

| Okay, here's where my stickler-side comes out.  There is sound
| engineering behind the 568B color code for data wiring, and it's
| usually labeled right there on every jack, patch panel, etc that
| you're likely to use (not RJ-45 ends of course, but if you're
| making those you should know what it is :p ).  In short, it's very
| important to follow the color code.  The simple reasoning is that
| (for 10/100 Ethernet) data is carried on pins 1 and 2, and 3 and 6.
|  Pins 1 and 2 are one circuit, pins 3 and 6 are the other.  One is
| used for Transmit, the other for Receive - which one is which of
| course depends on your perspective, and if you're talking as an end
|  point or a mid point (think device vs switch).  The 568B color
| code ensures that the orange pair (orange, and white with an orange
|  stripe) is used for pins 1 and 2, and the blue pair (blue, and
| white with a blue stripe) is used for pins 3 and 6.  Why is this
| important, you ask?  Well basic electrical engineering will point
| out that two wires, twisted together, will produce less inductance
| in other near-by conductors.  In other words, you don't get "cross
| talk" between the wires, and the signal is more clean.  That's the
| reason the wires are twisted so tightly in Cat-V cable, it's to
| help ensure there's no interference between the two very sensitive
| "BIG ANTENNAS" you've essentially attached to your Ethernet
| devices.

I think what you're looking for is called crosstalk.  It is
reminescent of the old days when you would actually hear conversations
on the line from adjacent wires.  This phenomenon that can be
effectively reduced by several orders magnitude if you twist the
wires.  What this does is nearly cancel out the circularly magnetic
field around the wires (due to each wire in a twisted pair carrying
the exact opposite polarity), and the near-field effects are
significantly reduced because of the near cancelling.  I say near
cancelling, because not all of the signal is cancelled out, due to one
wire in a twisted pair being closer to another wire in a twisted pair,
and thus a slight uncancelled amount of influence (Lenz' law regarding
changing magnetic field inducing an EMF - voltage - on adjacent
wires).  Note that if you have the twist-per-inch significantly high
and nearly identical from one pair to another, the exact opposite EMF
will be induced on the twists aligning up of the previously further
apart wires.  This will somewhat cancel, but being that the pairs
themselves are twisted slightly inside the cable bundle, this ends up
having some very small, but non-zero contribution of voltage on the
adjacent wires.

This may be what you mean regarding less inductance by twisting the
wires?  Maybe what you meant is mutual inductance, which is directly
related to this EMF field-cancelling phenomenon, between the donor and
acceptor of the field.

Regarding the twists, this has a couple of consequences, one of which
I believe is to ensure characteristic impedance at higher frequencies
(otherwise, the impedance would not be 100 ohms, and thus you'd get
reflections and your insertion losses would be higher).  I need to
think about the twists per inch in regards to the frequency response
and higher frequency EMI immunity.

Regards,
~  -Rob

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFBY3JQcGRHIMeR5bcRAhIQAJ9HfXs06VUuF9p52ayV+a2rTe9VYQCeJ5DX
ESynPNLKDZ6LlUJncjW29sU=
=7UTZ
-----END PGP SIGNATURE-----




More information about the TriLUG mailing list