A CAT5 question

[Dave Plowman]

The outlets I've bought mention T568B and T586A using what appears to be the same wiring colours but in a different sequence. What's the difference?

[Derek]

I came across this when I was wiring my house, before wireless LAN, now I'm wondering if it was worth it! They are two (very similar) standards. I think I read before that it's to do with backward compatibility with old standards. I don't think it should matter at all, once you are consistent with the sockets and the patch panel, and of course if there is no CAT5 wiring in the building already.

Derek.

[David W.E. Roberts]

difference?

Just looked in my Black Box catalogue, which has very useful wiring diagrams for CAT5 cable, and it mentions 3 variants:

568A 568B USOC Unfortunately it doesn't explain why there are 3 different standards.

568A and 568B are very similar; 3 pairs stay together and the 4th pair splits and divides them (outside;pair 1;single;pair 2; single; pair 3; outside). Only difference seems to be that for A the orange is split and for B the green is split.

So (as mentioned elsewhere) it doesn't really matter as long as both ends of an individual cable (plug to plug, plug to box, box to box) are wired to the same standard.

AFAICS you could use a mixture of standards on patch cables and they would all interwork fine.

USOC is different in that it splits 3 pairs and only keeps the blues together. See

no reason why an individual cable or cable run can't use this standard.

Google just turned up

which looks like a good read :-)

Cheers Dave R

[stefek.zaba]

Google's your friend on this one. What's vital to correct functioning is that the following four pairs of pins are each wired with a twisted-pair of their own: 1-2, 3-6, 4-5, 7-8. Note this is indeed a bizarre pairing: for the four pins in the middle of the connector it goes 'innermost to outermost', with 4-5 innermost and 3-6 surrounding it; but for the 4 pins at the edges it goes 'adjacent'. (There's an utterly anoraky-historic reason reaching into the historical evolution of the US phone system standards). And within each pair, pin positions should not get swapped - i.e. 1 goes to 1 not to 2, 3 goes to 3 not to 6, and so on.

The electrons themselves don't give a monkeys about the colour of the jacket insulation, and for any one length of wiring between two connectors performance will be unaffected provided you get the pairing-up right. So you could wire a patch-cord as follows: 1-2 in white-green/green;

3-6 in brown/white-brown; 4-5 in blue/white-blue; and 7-8 in white-orange/orange; while wiring a socket-to-patch panel as 1-2 in white-brown/brown, 3-6 in white-blue/blue, 4-5 in orange/white-orange, 7-8 in green/white-green. Any of the (24 pairings * 16 solid-or-tracer-first) = 384 combinations will work just as well as any other, electrically.

Maintenance-wise, though, it'd be complete idiocy to vary the colour coding like that. So, we have a Standard. But to confuse the world, we have *two* almost-but-not-quite identical standards, roughly being "the old Bell Telephone way adhered to in many but not all US installations and increasingly widely used the whole world over" and "the international-standards-org-blessed just-a-little-bit-different way", with the latter having one variant identical to the Bell Telephone way (so as not to have the US delegation voting against the standard) and the other being "The One True Way" which swaps the colour of the pairs used for pins 1-2 and 3-6 - which are the only two pairs used for most 10BaseT and 100BaseT interfaces, leaving 3-6 and 7-8 unchanged. Thus: pin T586A T586B 1 wh/green wh/orange 2 green orange 3 wh/orange wh/green 4 blue blue 5 wh/blue wh/blue 6 orange green 7 wh/brown wh/brown 8 brown brown

(You have to accept an absence of symmetry somewhere given the mixed 'inner-outer' vs 'adjacent' pairing. The Committee(s) seem to have preferred the 'striped-plain striped-plain striped-plain striped-plain' consistency over the 'always put striped into the lower-numbered pin' consistency.)

So out of the 384 electrically feasible combinations, standards bless just two. One would've been a much nicer number than two; but two is what we've got. To maximise confusion, the international-standard flavours whose names Dave mentions - 568A and 568B - have, as even the casual reader will notice, a single alphabetic letter suffix to distinguish them. The AT&T standard which is identical to one of them also has a single-letter suffix after three digits. It's called 258A. Guess which of 568A and 568B it's identical to? Why, 568*B* of course. Just remember: A is B, and B is A. It's quite simple. It's like in the popular panel game 'Just A Minute', where Humph asks the teams to sing 'One Song To The Tune Of Another'. It's really very simple (etc. etc. etc. ;-)

Many punchdown sockets you buy will be marked with both colour conventions, e.g. the 568A colours up one side and the 568B (that's the 258A, remember ;-) up the other. If it's got only one marking, chances are high that it's the increasingly dominant 568B marking. Since most sockets come with their own little internal printed-circuit board or other way of carring the signal that last inch or so from punchdown point to springy-fingers-where-the-plug-goes, you can't rely on the physical layout of the 8 punchdown points to be all that simply related to the pin numbering. Indeed, in an effort to make it easy for you to do the VERY IMPORTANT thing of keeping those pairs TWISTED at their 'natural' twist pitch up to AT MOST 10mm from the punchdown point, decent sockets will nearly always put the punchdown points for the two wires in each colour/tracer pair right next to each other, rather than spacing out the 3-6 punchdowns in their 'obvious' way.

What matters in any home wiring you do is *internal* consistency, so that all the fixed wiring you put in at various times is done to the same colour code. And I'd suggest you stick to the 568B=258A variant if you have no compelling reason to do otherwise.

HTH, Stefek

[romic]

I had a similar problem when I bought some outlets from TLC - they had the two choices of colours, whereas the Maplin ones I'd used had only one. On investigation, it turned out that I needed to use the T568B colour scheme to match the maplin ones. The wiring goes to a switched router, hub and cable modem and all are quite happy using that configuration.

Info below copied and pasted from somewhere that I'd saved it from in the past.

Pin # T-568-A T-568-B

1 White/Green White/Orange 2 Green/White Orange/White 3 White/Orange White/Green 4 Blue/White Blue/White 5 White/Blue White/Blue 6 Orange/White Green/White 7 White/Brown White/Brown 8 Brown/White Brown/White

Roger

[Chris O]

difference?

Lots of threads on this in comp.dcom.cabling

C.

[Dave Liquorice]

The colour of two of the pairs is swapped around.

As you say the wiring is the same. I don't know if there is any affect from the lay of the pairs within the cable or not, in one you be using opposite pairs the other side by side. This is assuming that all CAT5 cable pairs are laid up the same in relation to the pair colour...

[Stephen Gower]

A "crossover cable" wired with variety A at one end and B at the other is used to directly connect two computers.

[Dave Plowman]

Thanks for the explanation Stefek and everybody - at at least for once I seem to have picked the right one out of two.

I must get into the habit of Googling for things now I'm not paying per minute. ;-)

[bweebar]

The T568B configuration is what you end up with on the second RJ45 if you wire the first in the T568A configuration in a crossover cable.

[stefek.zaba]

Yup - at least for the "important" two pairs (1-2 and 3-6). The Web has contradictory claims on how the usually-unused pairs (4-5 and 7-8) should be wired for a general-porpoise crossover cable - most diagrams suggest wiring these two straight-thru, but a vocal minority suggests the other. Makes b-all difference in normal use, since (a) only a small fraction of deployed structured wiring makes use of these other pairs, and (b) of those, the great majority will use a "real" device to which all nodes link (hub/bridge) - which increasingly do autodetect of which pins are presenting R+/R- to entirely eliminate the need for crossover cables at all...

Stefek

[Lurch]

On Tue, 11 May 2004 09:03:31 +0100, in uk.d-i-y Dave Plowman strung together this:

ISTR it being something to do with interconnection with voice systems and not using the same pairs.

[bweebar]

Aren't pairs 4-5 and 7-8 being used for power over ethernet more often now and wouldn't crossing them cause problems? I'm sure POE will become more popular especially with applications like IP cameras.

[stefek.zaba]

No - the differences between 568A and 568B are only in the colour of wires used, but both specify "straight-through" (when used at both ends of any single run, natch) and keep the pairs just right for Ethernet.

There is a third wiring scheme which *is* incompatible with Ethernet over distance, called USOC. That runs inner-to-outer across all 8 pins, so pairs are used for 1-8, 2-7, 3-6, and 4-5. Since "normal" Ethernet uses 1-2 and

3-6, and for noise immunity over non-trivial cable lengths it really rather does matter that these use a single twisted pair each, using USOC (which was a voice "standard" which has almost-completely faded into history, and is just about never seen on any new installs) knackers the reliability of Ethernet over twisted pair... the wannabe balanced/out-of-phase Tx+/Tx- signals which so very much wanted to be running over a single twisted pair to-and-from pins 1-2 are now (sob, sob, feel the heartache) split up (oh horror, horror, horror, tongue nor heart cannot conceive nor name thee; a pair of star-crossed lovers; etc. etc. etc chiz chiz) across (gasp) two different singles in two *different* pairs (never the twain shall meet; die, die, die). Worst of all is that simple electrical continuity tests as done by multimeter or "el cheapie" network testers don't reveal split-pairs (it's the thousand-quid-and-upwards PentaScanner and cousins which have the requisite smarts), and networking might still "work", but be slow and unreliable because of all the retransmissions (as higher-level network layers notice that there's Something Wrong and cause resends). The original cable installer or "oh yes, this place is all wired up" handwaver disappears, leaving a headscratching puzzle for whoever ends up with network management responsibility....

Stefek