Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This question is basically a continuation of this previous question. Never mind, here's the deal:

I've made a LAN cable that goes through a wall, but it doesn't work. The cable is roughly 10m/30ft long. I crimped both ends myself according to this detailed explanation, and the job looks to be well done; all the wires are all the way in the plug.
I bought a cable tester after not being able to fix this. I thought perhaps the cable this is what I bought has a bad kink somewhere. But the cable tester says all wires are okay! The cable tester flashes its lights nicely in the correct sequence. According to the half-chinese instruction page, this indicates that everything is wired correctly. Even if I might have the wrong color sequence, the copper itself seems to be correct.

But when I unplug the LAN cable sneaking through the hallways, and connect this new LAN cable instead, Windows reports that there's no network.

What can be wrong? How can I find out?

Updated: Italic text above, in response to the first few answers.

share|improve this question
Do you get a link light on the NICs? Are the individual conductors in the same order as the known-good cable, taking into account all four plugs? Do you have another piece of cable left over (even if it's relatively short) and a couple of extra plugs? Make up another cable, test it with your tester, test it between two devices. – Dennis Williamson Sep 20 '10 at 20:30
Just to be clear, you are not trying to connect PC to PC, correct? – Beaner Sep 20 '10 at 21:17
@Beaner, I am connecting pc to router, not pc to pc. – Torben Gundtofte-Bruun Sep 21 '10 at 6:02
up vote 10 down vote accepted

The twisted pairs matter.

The conductors connected to the following pin pairs must be twisted together:
1 & 2
3 & 6 <-- Note: NOT 3 & 4.
4 & 5 <-- Note: NOT 5 & 6.
7 & 8

A simple cable tester like that can't tell which pin pairs are on the same twisted pair within the cable. It can only tell if you've got the right pins connected to the right pins.

Each pair is a balanced transmission line that uses differential signaling to cancel out noise. If you get the pairs wrong, you lose a lot of the ability to reject noise, which can make the link unusable.

share|improve this answer
Comparing your explanation with the link in my question, it seems to me that I did it right. Are you saying that I have the sequence wrong? How? – Torben Gundtofte-Bruun Sep 21 '10 at 6:09
I finally got around to looking at this damn cable again. I cut off 1 meter of the cable and crimped a new plug on it. Before crimping, I noted the exact sequence used in the other end (factory plug) and used that in my new plug as well. Now it works, hooray! It would seem that the sequence in my previous plug was wrong, except that the cable tester correctly lit up in sequence. I'm suspecting that one of the leads must've had a damage in the part that I cut off. – Torben Gundtofte-Bruun Oct 17 '10 at 19:53
@torbengb Thanks for the follow-up. I don't know if other geeks are like this or not, but I always wonder if my answers or advice actually helped, so I'm always glad when someone follows up. – Spiff Oct 19 '10 at 1:33
I'm very curious about how questions I answer end, too, so you're not alone. – Torben Gundtofte-Bruun Oct 19 '10 at 18:13

I have the same tester!

I would recommend you look at both sides of the tester and make sure every light flashes in the correct order as it possible that there could be a wrong order at one end of the cable.

If however both all light up in order, I am out of ideas - sorry!

share|improve this answer
+1 - My guess would be all wires are good but not lined up correctly at one end. – JNK Sep 20 '10 at 19:38
that is fairly common, most novices don't expose enough of the wires initially and then when they put them in order and try to insert them into the plug the wires don't stay in place. – MaQleod Sep 20 '10 at 19:57
Both ends of the cable tester indicate that all wires are good, and that the order is correct. – Torben Gundtofte-Bruun Sep 20 '10 at 20:23

The instructions you listed that you followed give the wrong wire order for a patch (router-to-PC) cable. The order on that cable appears to be for a crossover (PC-to-PC) cable (but I have not made one of those in a while). The correct order is, left to right with the clip side down:

  • Orange-white stripe
  • Orange
  • Green-white stripe
  • Blue
  • Blue-white stripe
  • Green
  • Brown-white stripe
  • Brown

Try re-crimping both ends of the cable in that pattern.

share|improve this answer
while he wasn't following the standard (568B), as long as the wires are the same on both sides, the colors are meaningless, they're all just copper conductors. The only time it makes a difference is on long cat6 runs where the pair twists come into play. You can see some major resistsance issues if you don't follow standards. – MaQleod Sep 20 '10 at 19:46
Thank you for the proper color sequence. Given MaQleod's comment, and the apparently successful cable testing, would I need to re-crimp or can I leave the cable as it is? – Torben Gundtofte-Bruun Sep 20 '10 at 20:24
I see that your color sequence is different from mine in the way that orange and green are swapped, and orange-white and green-white are swapped. But the electrical result is the same, right? – Torben Gundtofte-Bruun Sep 21 '10 at 6:11
It seems like it would, but in the cable, the pairs are twisted to prevent crosstalk (signal bleeding over between wires). In order for this to work, however, a device has to send a signal over one wire and a matching inverted signal over the other wire in the pair. That way, interference from each wire in a pair is canceled out and doesn't affect the other wires. If the pairs don't match right, the computer will be sending different signals over two wires in a pair and actually increase interference. – TuxRug Sep 21 '10 at 20:41
The electrical result is not the same. The pairs matter. This is data we're sending, not DC. – David Schwartz Sep 6 '13 at 17:50

Obviously something is not ok. Did you nick any of the wires? are you sure you can see the copper touching the end of the plug? Are the lights on the tester flashing in the right order? Have you verified your tester is working correctly on a good and obviously not good cable? Try crimping each end multiple times, sometimes that helps if you bought a cheap crimper (it should be a full cycle ratchet for a decent connection).

Also of concern, if you are running it through a wall, why are you not punching it down into a jack and putting it into a faceplate? Its a much more secure and stable connection for that sort of thing.

share|improve this answer
Some replies: 1) I can see 8 copper ends on each plug when I look at them head-on. 2) All lights flash, in sequence. 3) My crimper is cheap (I will likely not crimp again the next 5 years) but if the cable tests okay it should be okay, right? 4) Jack, faceplate, sorry what? I've drilled a hole through the wall, from one room into the living room, with the router on one side and the PC on the other side. There's no special wiring or power lines or anything else going on. – Torben Gundtofte-Bruun Sep 20 '10 at 20:28
I've just checked the cable tester with a "known good" cable. It looks the same. I don't have a "known bad" cable. – Torben Gundtofte-Bruun Sep 20 '10 at 20:30
Jacks: Faceplates: – MaQleod Sep 20 '10 at 20:33

I have been crimping cables for well over 12 years, all kinds of cables (mostly cat5e) with standard connectors with the 568B configuration, pass trough and letting the switches take care of the cross over

i have been using a tester all this years, but also take the precaution of testing the cables with a laptop or pc as well when possible

Last night i brought home a cable i made which passed the tester ok, but for the life of me would not work on any of my home computers at all, light would not come on and pc's would not even recognize the cable being inserted

I ended up rewiring the ends 3 times each and it still didnt work, yet the tester said it was ok

I finally made it a cross over (swtiching green with oranges) and it worked. im still stumped since all my other currently working cables are 568B pass trough.

My only answer is that this cable is somehow damaged and crossing over strengthened the signal?

So far connection is good at either 100 and 1000 mbps.

Wanted to share this weird case.

share|improve this answer

Try reducing the network speed from 1000 Mb to 10 or 100 Mb and try it then. I think you can change this from network card properties in Control panel.

I had home made cable that could not handle 1 Gb signal and did not work because of that. Higher network speeds require high level cables and home made ones do not always work. If you wanna use 1 Gb network speed, you probably need to buy high quality cable (Cat-5).

share|improve this answer
gigabit network speeds require cat6 – MaQleod Sep 21 '10 at 6:19
@MaQleod the IEEE disagrees with you. 1000BASE-T (the standard for Gigabit Ethernet) only requires Cat 5. Perhaps you were thinking of 10-Gig Ethernet (10GBASE-T), which requires Cat 6a. – Spiff Sep 25 '10 at 3:30
In theory, but we don't live in a theoretical world. The standards define what should be plausible in under ideal conditions, not standard conditions of a data center or CO, and especially not some cable draped over a desk and zip tied in place. In any networking environment, cables are never perfect and I guarantee you will never see gigabit speeds on a cat 5, even if the link says 1000mb/s. Possibly a cat5e running a short distance will support it alright, but is guaranteed by a cat6. I've made, run, certified tens of thousands of cables, never rely on the minimums of standards. – MaQleod Sep 25 '10 at 4:12

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .