6

I've installed a bunch of cat-5 and when I check it for continuity and polarity it seems fine, but when I hook up a pair of laptops with a x-over cord and try to ping between them, they can detect there is something on the other end but I get 100% ping failure (mixed between "hardware failures" and "timeouts"). I punched a few jacks on the end of a length of cable and that works just fine, so I know I can make good connections.

Any idea what might be causing it?

My current best guess is that it has something to do with my the wires being bundled together. How tight can you bundle cat-5 without causing problems?

What it turned out to be:

In short: bad test equipment.

The long version: I borrowed some a better tester and it said that all of the cable was good but I still couldn't get data through it. So I yanked the shortest segment out of the wall (man I love nice spacious attics), re-terminated it and brought it to a guy with a better cable tester and even it said I was doing things right (or at least I was enough of the time that I shouldn't be seeing the failure rate I was). Then I tested the patch cords I was using to do the test... Bingo! They were bad. After getting a known good replacement (and disposing of the old one with extreme prejudice) everything works.

I still don't know why the patch cords consistently worked on a few ports and none of the others but I now have tested everything in the wall and it all works!

BCS
  • 1,065
  • 2
  • 15
  • 24
  • 1
    It's not totally clear what you've done, it sounds like you've installed some cable runs and terminated them and they aren't working. Putting jacks on some of the cable runs doesn't prove anything, you might have done that right but not punched them down properly. If you haven't pulled them out of the jacks and punched them down again, I'd do it over. Personally, I prefer to test with a switch on one end and a laptop on the other. – Ward - Trying Codidact Nov 23 '09 at 23:31
  • I've re-punched some of then to no effect. (The short length with the jacks I mentioned is about 3ft long and just used as a test.) – BCS Nov 23 '09 at 23:51
  • 3
    And this is why Cable Testers were invented – Mark Henderson Nov 24 '09 at 00:29
  • @Farseeker: I'm looking for someone I can borrow one from. OTOH that will still not tell me how to fix the problem (short of tare it all out and replace it) – BCS Nov 24 '09 at 01:59
  • 2
    A good cable tester will tell you everything you need to know. Whether the pairs are in the correct order, how far down the cable a break is, quality of signal, etc. – Dennis Williamson Nov 24 '09 at 02:34
  • The only fundamental thing you haven't confirmed, then, is that it's really Cat5 (or better) cable. You've got 4 pairs, properly twisted? – Ward - Trying Codidact Nov 24 '09 at 04:48
  • Speaking of that, what speed are you testing at? You said Cat5 not Cat6, but lots of people use "Cat5" as a synonym for any UTP cabling... If you're trying for Gigabit, you're unlikely to get it from your own terminations. – Ward - Trying Codidact Nov 24 '09 at 04:50
  • half the stuff says "Cat-5e" of the sheathing, the rest says "Cat-6" but I'm only running 100Mbit (I ended up with the 6 because it was all the vendor could get in a third and forth color). – BCS Nov 24 '09 at 22:03

2 Answers2

7

You don't say how you've bundled them together. Did you use zip ties and pull them tight? Did you loop the cable and cinch the loop (like a figure eight)? Are there any other places where there are sharp bends? How small is the radius of the loops or bends?

You don't say whether they're pulled through walls or drop ceilings. You don't say what the conditions are that you're working in. Stray nail tips and sheet metal edges, etc. Are you running parallel to any fluorescent lights?

Besides squirrels and rats (which can be very real problems), most of the time it's rough handling or improper termination that cause problems.

You say "I punched a few jacks on the end of a length of cable". I presume that's just an imprecise way of saying "I punched one jack at each end of a length of cable" because you can't punch a few jacks on a cable for networking and get away with it like you might for multiple phone jacks on one cable.

Also, continuity and "polarity" mean essentially nothing. Your test with two computers or a computer and a switch or best-bet rent a proper Cat 5 or greater tester will be the only ways to approach testing. Lots of things conduct current that won't work for networks.

Also, don't trust patch cables. Try another one.

Dennis Williamson
  • 62,149
  • 16
  • 116
  • 151
  • The tight spots are wire ties (tightened until the cable won't slide through then) and a pipe coming down thought the ceiling (that's got 24 wires with enough room for maybe one or two more). -- Yes, 2 jacks, one on each end of a cable, just like all the runs. -- Continuity and polarity are good indicators that I at least hooked the correct wires to the correct spot; it's necessary, but not sufficient. – BCS Nov 24 '09 at 01:55
4

In addition to having correct cabling (which you can verify by checking that the link light is showing active on both laptops), you must configure the TCP/IP stack to communicate on the same subnet with the same subnet mask and Unique IP's for both computers. Since there is no switch present, you must configure TCP/IP manually with STATIC IP addresses. Can you confirm that you have completed this step? Note use 192.168.1.0/24 (255.255.255.0 mask) if you want to conform to private network standards.

Update: If you've already confirmed the X-over cable alone allows pings, then confirm adding a non- x-over (regular) patch cable and simple cat5 cable connector (back to back rj45 plugs) into the line also passes the ping test. If not, get one of each that you can confirm work.

If those work, and you now try patching over the cable run. If it fails at this point you have positively confirmed you have a bad cabling install. This can be the result of either bad punchdowns, or a break inside the wire run itself. You can use a continuity tester to check each pair of wires by doing the follwoing:

Bare the ends of all the wires on one end, then twisting each colored pair together making sure to keep them isolated from the other pairs. Go to the remote end and apply the two sides of your tester to each of the wires in the pair. if the light lights the wire run is intact (not necessarily good, but intact). Test each pair this way.

If all your pairs test good, punch down your ends again. Now purchase a cheap $10 wiring tester. Hook the two ends up and watch the tester confirm your wiring is correct. If you get green lights on all 4 lights, your punch downs are good.

Now hook your 2 comps up again using the known good cables. If you ping test still doesn't pass, I would guess that you have some major interference in your cable run. At that point I would consider calling up a professional cable installation company for advice, as they will need to inspect your cable run.

Zak
  • 1,032
  • 2
  • 15
  • 25
  • (+1) I prefer 10.* but I've got that setup and tested by using just the x-over cable (it works until I add in the installed cable). – BCS Nov 24 '09 at 01:57
  • You've got speeds set properly at both ends? You're using two computers to test, so turn off auto-negotiate and set speed and duplex manually and identically. – Ward - Trying Codidact Nov 24 '09 at 04:49
  • The exact same setup of computers work with a cross over cable, so I know it's not a settings issue. – BCS Nov 24 '09 at 22:04