BlueWolf's Howl

« The Stats Jump | Bluewolf's Howl | One more shot »

December 03, 2005

Same or not?

Every once in a while, you'll run across something that appears on first glance to be the same. But, if you look closer, you'll find that there really is a difference. One of these things is -- cabling.

If you don't know how to properly terminate a cable, you shouldn't do it. But, that doesn't stop a lot of people...and then trouble starts. When it comes to CAT5 patch cables -- yes, it does make a difference whether or not you terminate the ends properly.

In a straight-through cable, each wire is connected 'straight-through.'

1 <---> 1
2 <---> 2
3 <---> 3
4 <---> 4
5 <---> 5
6 <---> 6
7 <---> 7
8 <---> 8

I've heard a number of people claim that in this case, the order of the wires doesn't matter. As long as the same order is applied to both ends of the cable, it'll work. And when you look at the diagram, this looks like it should be true. But, if you look a little closer, you'll see it's not.

In CAT5 cabling [unshielded, twisted pair], you'll notice that pairs are twisted together. This was not just for neatness. It has a purpose. The twisting prevents/lessens crosstalk. Crosstalk is when signals leak between wires in a cable.

Twisted pair cabling is a common form of wiring in which two conductors are wound around each other for the purposes of canceling out electromagnetic interference known as crosstalk. The number of twists per meter make up part of the specification for a given type of cable. The greater the number of twists, the more crosstalk is reduced. Twisting wires decreases interference because the loop area between the wires (which determines the magnetic coupling into the signal) is reduced as much as physically possible. The directions of current generated by a uniform coupled magnetic field is reversed for every twist, canceling each other out.

So when you're cabling ethernet, you want the ethernet signal (1/2) (3/6) to be in the same pairs. Likewise, when you're cabling a T1 patch cord, you want the signal for that to be in pairs too (1/2) (4/5). A properly terminated cable will use the order:

1 (pair 1) white-orange
2 (pair 1) orange
3 (pair 2) white-green
4 (pair 3) blue
5 (pair 3) white-blue
6 (pair 2) green
7 (pair 4) white-brown
8 (pair 4) brown

As you can see, ethernet uses pairs 1 & 2 while T1 uses pairs 1 & 3. This arrangement allows the signal to travel the length of the wire in twisted pairs. If you terminate the ends of a patch cord with the pairs next to each other (regardless of color):

1 (pair 1)
2 (pair 1)
3 (pair 2)
4 (pair 2)
5 (pair 3)
6 (pair 3)
7 (pair 4)
8 (pair 4)

As you can see, this will result in the ethernet signal using pairs 1 & 2/3 [wires 1, 2, 3 and 6] and T1 using pairs 1 & 2/3 [wires 1, 2, 4 and 5]. The tricky part is that it will 'appear' to be fine. Inexpensive cable testers won't pick up on this. They will simply send voltage down the wire and it will be detected at the other end. You will think that it is a good cable. For extremely short lengths (less than 2 ft), you may not notice any problems. The problems may quickly arise as the cable length increases. In this case, the inexpensive tester will detect the voltage, but when the cable is used, the voltage difference is enough to cause errors (or worse --> intermittent errors which you'll be chasing like a ghost in the machine).

At the Application Layer (of the OSI model), we see text and graphics. At the Physical Layer, the devices only see voltage. The machines don't really 'see' 0s and 1s. They detect increases and decreases in voltage. If a 0 gains some voltage from crosstalk, it can look like a 1 to the receiving device. If a 1 loses voltage from crosstalk, it can look like a 0 to the receiving device. This is not a happy situation.

Granted, error-correction is built into protocols like TCP/IP. And it will resend packets that happen to not transmit properly (regardless of reason). But, that's not only inefficient...it wastes bandwidth. And at some point, enough crosstalk will disable communication. You'll see errors and you'll spend a lot of time (and perhaps money) trying to fix it. You'll probably notice the errors on the receiving device (because that's where you'll look first). You'll try to replace it. Dang. It still doesn't work. Then you'll get the bright idea that perhaps you should replace the transmitting device. Dang. That still doesn't fix it. Well, you've tested the cable numerous times during all this replacing. It tested good. But now you have no other choice but to replace it too. You finally break down and buy a guaranteed patch cord of the right length. Ah-ha! Finally, it works.

Sure, a pre-made patch cord is much more expensive than the one you made. But, compare the price of that patch cord to the cost of the time you spent trouble-shooting this problem. Compare it to the cost of the devices that you may or may not be able to return (since the original devices were actually functioning properly). How did you save money in this? Well, the best way to 'save' money is to spend the time and effort necessary to learn how to do such a simple task correctly the first time. And if you don't have the time or the inclination to do so, realize it and spend the few extra bucks that it takes to purchase a pre-made cable. You'll really be saving money, time and trouble down the road.

[PS - When trouble-shooting a problem that appears to be hardware-related, always replace the least expensive device (the cable) first. ]

Posted by BlueWolf on December 3, 2005 12:06 PM