T568A or T568B? Check out what Fluke has to say

Thanks for the link. I would have been a bit more convinced if the customer repeated that termination -A & -B to convince me it wasn't a soft/skewed punchdown. And tried a different run of cable that hadn't been kinked :)

I notice they didn't report Cat3, Cat5, Cat5e or length. I'm surprised that the mfr would admit which are the worst pairs or would even know without a lot number. If compromises have to be made on a cable, I'm very surprised they'd color the worst pair orange. Blue or Brown should get the boot.

I'm not entirely surprised that 3,6 are worst on the jack (USOC straddle might be good for the wires, but the PCB traces will be longer).

In short, I'm not counting on it.

-- Robert

Well, that and published amount of tests that were run and the ambient temperature and ... OK, let us get real here: glorious days of big bucks put into cabling R&D are over, at least for a while. We gotta learn how to survive on that limited information that is possible to scavenge from free sources, and basically use what's available. So, this will have to do :-(

That was CAT6 link as the test tag states. Also, judging by the attenuation I'd say it was exactly 100 meters of a decent (but not CAT6+) CAT6 cable. No way to guess on origins of the jacks though. If I were to guess, I'd say they used jack-to-jack setup without a patch panel simply because it's easier to put together.

These days they make pins twist so much before they even hit the PCB that I would not be so sure about which pair is better or worse.

Here I go with the link again: I don't think it came out right last time.

I'm curious, was this a one-time test? The article doesn't really give an indication. That, along with so many other factors that can affect a termination like the pressure applied to the wire, the angle of the punchdown tool and so many other things can affect a termination. One test result, no matter how dramatic, is statistically insignificant. Rodgers Platt

All true, no questions asked. But then, again, has anyone else conducted/published a statistically valid one? Gotta work with what's available...

On a side note: I know it was not a R&D project on a government's grant ;-), but I do not expect Fluke guys to be absolutely silly and not exercise at least some degree of decency conducting the test. I would assume things like randomly picking the parts, carefully doing the terminations and maybe even multiple (but not statistically significant, I agree) tests have been done.

Nope. This wasn't their test, but a one-off report from a customer of theirs.

-- Robert

Does anyone ever test cables from both ends?

I don't think I am convinced, anyway. There are a number of possible differences between cables, the exact mechanics of crimping the connector on, the positioning of the pairs inside the cable, that one should not be surprised about small differences between cables.

For any cable less than about 80m there should be plenty of margin to go around. For those making 100m cables, keep trying until it is perfect.

-- glen

Rats! You're right, it's right there: "collected by a customer". I guess, you cannot expect any decency from a customer, conducting a test ... Just kidding ;-)

It does not sound right that Fluke cannot perform a simplest test and has to rely on a customer. Tough times...

Does anyone ever test cables from both ends?

Actually, the testers do it now automatically. The "smart remotes" in use since '95 or so have all tested from both directions. If you had one of the original Pentascanners or one of the other early Cat 5 Testers you had to test each cable twice. Once from the patch panel, the other from the workstation in order to be compliant with TSB-40A.

Given that you're using the exact same pairs, I'd be suspicious of any test that says one or the other method is bad. That sort of test should be repeated with several different cable samples, to reduce chance of error.

I believe there's an ancient saying: "The plural of anecdote is not data".

If someone wants to make a such a claim, they should run multiple tests, with many samples of wire, from different spools and manufacturers.

Thanks for the infor, certainly it was a good reading

Zeb

From the author of the article in question:

If you doubt the conclusion, please try it yourself and let me have the data. I decided only to publish one set of data in order not to overwhelm the reader - a common complaint I receive. This conclusion is also known by cable manufacturers as well - they may just decide not to tell you.

There are indeed variables in this experiment. However, when these are taken into account, the conclusion is still the same. If the cable has significantly less performance on the orange pair for RL and you connect it to an outlet that has its worst pair for RL on 3,6; wiring it to T568-A is not the first choice.

The shorter the length of the link, the less impact this will have of course, since the contribution from the cable is reduced and the contribution of the connectors become an issue.

In recent months I have started seeing connectors with equally good performance on all pairs. I believe this is where we want to be. And no, I am not telling you which ones - but they are not cheap.

Kind regards

Adrian Young Fluke Networks

##-----------------------------------------------##

Article posted with Cabling-Design.com Newsgroup Archive

comp.dcom.cabling - 1599 messages and counting!

##-----------------------------------------------##