Subject: stds-1394: 1394b UTP pinouts
Date: Wednesday, February 5, 2003 7:05 AM
From: Les Baxter <retxabal@hotmail.com>
Reply-To: les@baxter-enterprises.com
To: <stds-1394@ieee.org>

At the UTP working group meeting last Monday (1/27/2003), we discussed the 
possibility of moving the 1394b UTP signals to different pins on the R-45 
jack.  Specifically, the proposal was to change from 1,2 and 7,8 to 1,2 and 
3,6 -- in other words, to move to the same pins as Ethernet. One of the 
concerns was that the crosstalk performance of the inner pairs is not as 
good as the outer pairs which are currently specified. Here is some 
information that may be useful. This note is not advocating any particular 
position on the proposed pair switch, it is simply addressing the crosstalk 
issue.

CATEGORY 5 CROSSTALK
Although Cat 5 is now obsolete, the formula for pair-to-pair NEXT (Near-End 
Crosstalk) in a Cat 5 channel is:

NEXT >= 27.1 - 17log(f/100) dB
where f is in MHz.

For low frequencies, this formula saturates at 57 dB. At 62.5 MHz, the 
maximum NEXT is 30.6 dB, and at 100 MHz, it is (obviously) 27.1 dB. This 
formula applies to the worst pair-to-pair reading out of the 6 possible 
combinations. Generally, this will be found on the two center pairs, but the 
spec is the same for all the pairs. In practice, some pairs just exceed the 
spec more than others. But the amount of this excess margin will vary 
unpredictable from one manufacturer to another and among different product 
lines from the same manufacturer. For a robust design, you should use the 
formula above for the worst case NEXT regardless of which pairs you are 
transmitting on. I hope this is what the 1394b committee did when developing 
the S100 UTP spec. If so, then there is no NEXT issue when changing from one 
pair to another.

For reference, I have attached a plot of crosstalk measured on a sample 
patchcord (courtesy of Masood Shariff). It shows a large variation among the 
different pair combinations, but all meet the requirements (in this case, 
the Cat 6 requirements.)

PINS VS. PAIRS
Note ö if you are not interested in the details of the cabling system, stop 
reading here.

As in the sample crosstalk plot, cabling guys often talk in terms of pairs 
rather than pins. The pairs are also frequently referred to by color -- 
1=blue, 2=orange, 3=green, 4=brown. (This, of course, represents the color 
of the wire pair connected to those particular pins.) The only fly in the 
ointment is that there are two sets of pair definitions. The T568A pinout is 
as follows:

Pair 1 = pins 4,5
2 = 3,6
3 = 1,2
4 = 7,8

The T568B pinout is similar, except that pairs 2 and 3 are swapped:
Pair 1 = pins 4,5
2 = 1,2
3 = 3,6
4 = 7,8

Or, in other words, the orange and green pairs are swapped. In the sample 
crosstalk plot referred to above, I think it was using the T568B pinout, but 
for this example it doesn't really matter ö the purpose is just to show that 
the crosstalk varies quite a bit among the various pair combinations.

The entirely reasonable question of why there are two standard sets of pair 
assignments has a historical basis and requires quite a few expletives to 
properly explain. If you are interested, ask me at the next meeting and I 
will give you my (admittedly biased) explanation of how and why it happened. 
Or maybe I've already told you more than your really wanted to know ...

One final point regarding the T568A vs T568B pinouts -- from the user's 
perspective they are virtually indistinguishable, assuming that:
1)all 4 pairs are wired up
2)the building is uniformly wired with either A or B jacks.
The only difference is that your signal will be carried by an orange wire in 
one case and a green wire in the other. The place where it makes a 
difference is if you are the guy wiring up the back end of the jack -- then 
you have to punch the wires down in the correct sequence and do it 
consistently throughout the building. Most manufacturers make a single jack 
for both T568A and T568B applications and just stick a different decal on it 
showing where the pairs should go.

In the unlikely event that a channel has an A jack on one end and a B jack 
on the other end, then the orange and green pairs will appear to be switched 
and the 1394 auto-crossover function should straighten them out.

The result of all this is that equipment specs usually list pin numbers on 
the jack (like 1394b does) and do not get sucked into the pair-naming 
debate.  I think this is the proper approach.  Just specify what pins you 
are transmitting and receiving on and let the various cabling providers call 
the pairs whatever they want.

Les Baxter
les@baxter-enterprises.com
732-212-1400




_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online  
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963