Cable ends: how important is standard?

Barely. And don't ever expect one wired like that to work at more than CAT3 (10Base-T) speeds.

Those standards exist for good reason. By wiring as you describe, instead of wiring to standard, you'll end up splitting the pairs which is a Really Bad Idea.

If the cables you're dealing with are not wired to standard, then they should be re-done correctly. Period.

Good luck.

You don't have to follow the standard, so long as you use correct pairing. As how far your cable would work, is anyone's guess. Also, the problem is not inductance, it's the lack of shielding and the unbalance caused by mixing pairs that causes problems.

Inductance is the opposistion to current flow, along a conductor, not the coupling of signals between adjacent wires. There is however, inductive coupling, between conductors, along with capacitive coupling. The characteristics of the cable determine both.

A foot or two...

Fix the terminations -- the wire is paired for very good reasons.

I note that if you don't follow the 568A/B standard and just connect the pins according to matching colors, straight thru, then a cable will work. If it's short, anyway. How long would you think that a CAT5e cable need to be before the failure to follow the standard would cause so much inductance that the cable might fail? I'm wondering if a 50-meter cable might fail for that reason. A network cable tester confirms signals on each connector, but NIC cards don't recognize it. It's not wired according to standard. (Don't ask me why: it's a long story I don't fully understand myself.) Thx.

It *is* inductance; as well as the imbalance. As correctly configured, the transmit data is on one pair; the receive data on another.

When you split the pairs, you get a long linear transformer between the one wire of the {say} transmit pair is now coupled to another single wire. When that other wire is the receive pair... so instead of *many* dB of isolation between the transmit and receive pairs; you get near zero.

There is no shielding with UTP, that's what U stands for - unshielded. It's not unbalance, it's crosstalk between the transmit and receive pairs. When pairs are split, the crosstalk is, well, it's intolerable.

No, inductance is the physics effect, for lack of a better label, that causes "inductive coupling"...

But it's not worth debating definitions. The OP has a choice: punch down the cable correctly, or not expect it to work.

David Lesher wrote in news:cothnv$cm2$1 @reader1.panix.com:

inductive

Hey Guys,

Since you both seem to know a lot about this, can you explain it to me in layterms? Basically, from what I knew before I started dealing with computer networking, a wire is a wire and as long as wire 1 goes to terminal 1, wire 2 goes to terminal 2, etc. the cable will work. For instance, this is the case with phone cords.

However, I have been much disabused of this notion lately. So, basically, in terms for the common man like me, why do the pairs need to stay together?

Thanks. Ryan

Most NICS (network interface cards) have transformers that have windings that drive the line and receive the signla from the far end. The winding is like a see-saw in that there must be a difference in the voltage on the ends to have a current flow, like if both ends of the see-saw are pushed down equally, there is no movement.

An interfering signal that affects the twisted pair line affects both conductors equally so there is no difference between the conductors, and there is no current flow in the winding, and no interference gets thru the transformer. If you have one conductor 'different' i.e. longer or shorter, or if it's mixed with another pair, then there is a difference and the current flows in the transformer winding, and interference gets thru.

The RS-232 signals were each on a single conductor, relative to the ground. The data rates were slow, and the voltages were high so any interference didn't affect the signal enough to cause errors. But as data rates increased, the world needed a better system, so the RS-422 standard with balanced pairs was developed. This allowed the data rate to be increased orders of magnitude from ten thousand to a million bits per second.

But ethernet over twisted pair is still much faster than this, so the balanced pairs are even more critical to maintaining good signal to noise ratio.

The ends of the Ethernet are transformer coupled. What this means is they can differentiate between common mode and differential mode noise.

If you have 2 parallel wires, say spanning a room, both are antennas, picking up the noise from {say} the next room. I think you can see that to the most part, they pick up the SAME noise. That's common_mode input. But the next receiver stage, what the wires connect to, will ignore that and only look at the pushme-pullyou difference between wire 1 and wire 2's potential. That difference comes from whatever drives the input end of those two wires.

OK, now take those two wires and twist them together into a tightly pair. This raises the ability of the pair to differentiate between the outside, common_mode {noise} and the intended pushme-pullyou signal to the nth degree. Why? "an exercise for the student. [1] Basically, things cancel far better.

Now, you can (and do...) put another twisted pair right next to the first one and {esp. if the twist rate is different between the two pairs} almost zero signal will leak from one to the other. THAT's your goal, and it's important because the receiver in your Ethernet card [or switch, or..] must hear the far weaker signal from the far end over its own transmit signal.

But "split" those pairs, and you have in fact coupled most of the transmit signal into the receiver... and that's bad news [tm]...

Does that help?

[1] If you are up to the integral calculus, you can prove this. I emphasize the "you" in this part...cuz I have long since forgotten..

Weird huh? When I was first introduced to twisted pair data cabling I couldn't understand why it mattered and you couldn't just use straight through termination. Keeps some of us employed!

--Dan

That's a *BIG* topic. However, the idea is that a single wire can act as an antenna, radiating to and picking up signals from other wires. If two wires are twisted together, and used in a balanced circuit, the current in the two wires cancel out the antenna effect, so that they're less likely to suseptible to interference and also less likely to create interference.

A balanced circuit, is one where neither conductor is grounded. An unbalanced circuit, is one where one side of the circuit is tied to ground and often uses coaxial cables, such as those used for cable TV.

You might want to read up on transmission lines. A good reference for beginners, is "The Handbook for Radio Amateurs", published by the American Radio Relay League. These books are in many public libraries.

That only applies, if the transformers are connected to balanced circuits. There are many situations, where they are used with unbalanced circuits. Twisted pair (10baseT etc.) ethernet is a balanced circuit. Coaxial cable ethernet (10base2 & 10base5) are unbalanced.

What do you think is the cause of crosstalk, between the pairs? A twisted pair, even when used in an unbalanced circuit, provides a significant amount of shielding, though not as much as an overall contiuous cover shield, such as you'd find in coaxial cable. When a pair is twisted, there is that shielding behaviour at work, along with the rejection of common mode signals, in balanced circuits.

Transmission lines are a very complex topic, which most people have not studied. I have.

For the reasons, read up on transmission line theory. Some amateur radio books are a good resource for beginners.

unshielded.

intolerable.

There _is_no_shielding_. You don't seem to understand the concept of a balanced pair.

There is NO shielding behavior. The (common mode) interfering signal can be very high level on both conductors (because there is no shielding!), but since the interfering signal is _equal_ on both conductors, there is *no* difference of potential _between_ the conductors, so there is _no_ transfer of interference to the receiver.

You need to go back and study to learn the basic concept of a balanced line.

There is _no_ shielding in a UTP cable.

connector, but

Oh, gee, you make this mistake so often that you have a usual number for it? ;-)

A 50-meter cable WILL fail for that reason. The pin layout is important, and not just because cabling standards guys said so. The NIC cards and its hub/switch counterpart port are designed to expect a balanced signal to come out of the transmitter and come into the receiver. Even the RJ45 jack inside the NIC card is not just a jack these days but a pretty complex part with impedance matching transformers inside, and the end of the coils are connected to the particular pins of the jack, and it is expected that both conductors of a twisted pair will connect to a particular coil. So, if you mess around with such basic principals of the signal transmission, you run a risk of degrading the performance so much that it would only work within a room or two, and then fail inevitably.

Here is the correct pin layout for your cable repairing pleasure ;-)

BTW, a not-so-sophisticated cable tester will not pick up the pin layout problem if all the conductors are wired straight through: it only measures connectivity between the pins, and not the proper impedance of the circuit between the pins.