Help me diagnose a CAT5e UTP cabling gremlin
November 14, 2016 12:28 AM   Subscribe

I have two RJ45 outlets in one wall plate, labelled B-01 and B-02. An analog phone that worked for several years when connected via B-01 failed spontaneously; a quick test shows that it now works reliably via B-02 but cannot be persuaded to work via B-01. And yet, a PC with a gigabit Ethernet connection still works via either, and the TDR cable fault tester in the local network switch returns identical results for both. What am I missing?

Both outlets are wired back to a Krone Highway module in the comms cabinet via about 10m of CAT5e UTP in the walls. That cabinet also contains a D-Link gigabit Ethernet managed switch, and a phone pair from the PABX wired to pins 4 and 5 (the centre pair) of an RJ45 plug.

I plugged a 3m patch cable into a spare switch port and exercised the switch's inbuilt TDR cable tester against it. As expected, it showed an open on all four pairs at 2-3m distance.

Next, I shorted pins 4 and 5 (the centre two) on the end of that patch cable and tested again. This time it showed opens on pairs 1, 2 and 4 and a short on pair 3, again at 2-3m. As I was expecting the short to show on pair 1 and not pair 3, I tried again after shorting connectors 3 and 6, then again with 1 and 2, then 7 and 8. Each time, the test result showed a single pair shorted and three pairs open, all at 2-3m. Conclusion: the pair wiring is as expected but the switch software incorrectly swaps reports for pairs 1 and 3. OK, I can work with that.

Plugged that same patch cable into port 01 on the highway module and ran another test, with nothing plugged into outlet B-01. Four open pairs at 13m: so far, as expected.

Plugged the telephone cable (3 metres with an RJ11 plug on each end with only the two centre pins wired) into B-01 and tested again. Pairs 1, 2 and 4 open at 13m; pair 3 open at 16m.

Shorted the two centre pins on the free end of the phone cable and tested again: pairs 1, 2 and 4 open at 13m; pair 3 shorted at 16m.

Repeated all these tests on outlet B-02, with the same patch cable now connected to highway socket 02 in the comms cabinet. All results identical.

When I patch a switch port to highway socket 01 and plug a PC into outlet B-01, the Ethernet port reports a gigabit connection and works just fine. Same if I use highway socket 02 and then connect the PC to outlet B-02.

When I connect the telephone plug in the comms cabinet to highway socket 01 and plug the phone into B-01, I get no dial tone. With the comms cabinet phone plug in highway socket 02 and the phone plugged into B-02, it works.

I have taken the switch plate off the wall and looked at the back of the outlets. The UTP is correctly terminated on both (insulation displacement connectors with wires fully pushed home, pair colors arranged per T-568A). The cable connected to B-02 had some jacket damage - insect or mouse chew, by the look of it; there was a dead beetle behind the wall plate - but no visible pair damage.

So I am now completely baffled. What kind of connector or wire fault allows gigabit Ethernet to function properly but kills an analog phone?
posted by flabdablet to Computers & Internet (6 answers total) 4 users marked this as a favorite
 
Well, that was a waste of a question. Found it.

Turns out that the RJ45 in the comms cabinet that terminates the phone pair is exquisitely sensitive to the angle at which the wires come out the back of it. Plugging it into highway port 02 drags them a little further sideways than they go when plugged into highway port 01. I can make the phone work again via B-01 by wibbling and wobbling those wires.

Will come in tomorrow armed with a new connector and a crimp tool and fix it properly.
posted by flabdablet at 12:36 AM on November 14, 2016 [6 favorites]


This is top-level ask.metafilter teddy bear debugging in action.
posted by pharm at 1:39 AM on November 14, 2016 [9 favorites]


Wow.
posted by bz at 8:49 PM on November 14, 2016


Exquisite sensitivity to wire angles is generally a sign that the solid conductor has actually broken inside one of the wires, and is making intermittent contact.

Normally I see this as an issue when someone tries to crimp a stranded RJ45 connector onto solid cable. The two prongs on each pin are designed to pierce the conductor and push their way through the stranded wire inside. With solid cable, this typically crushes the conductor and may fracture the conductor, or at least make it sufficiently weak to become extremely susceptible to a future fracture when someone bends the wire a bit. Putting a crimp on solid cable requires a special type of crimp with three-prong pins, with the prongs arranged to surround the conductor, not to pierce it. These are still substantially more finicky than normal crimps. IMHO: don't use them unless no other option exists.

See pic: http://www.cableorganizer.com/images/articles/solid-stranded-rj45-connectors/identify-solid-stranded-rj45.png

To be thorough in this discussion, though, consider also the jack. With a punchdown style jack, there are two typical possibilities for failure, but they tend to be a little more unusual. One is that someone got a little too aggressive with the punchdown tool and placed undue pressure on the cable (tool moved back and forth) or damaged the jack IDC terminal itself (tool moved side to side) during termination. There are now some tools ("punchdown pucks") that can hold the jack while punching it down, making it somewhat easier. The other common failure is for someone to try to use stranded wire with a punchdown jack. This simply won't work right. The jack is designed for copper wire of a particular thickness, so that the blades on the terminal will pierce the insulation of each wire without damaging the solid conductor inside. Stranded wire will tend to flatten and will not make a good consistent contact.
posted by jgreco at 1:59 AM on November 15, 2016 [1 favorite]


the two prongs on each pin are designed to pierce the conductor and push their way through the stranded wire inside. With solid cable, this typically crushes the conductor and may fracture the conductor, or at least make it sufficiently weak to become extremely susceptible to a future fracture when someone bends the wire a bit.

Yes, this appears to have been the failure mode involved: on close examination, the two centre contacts in the old RJ45 plug (the only ones with wires behind them) sit higher than the others and have creased surfaces, so it seems that not only was a stranded-cable plug misapplied to solid wires but the crimp tool was wielded by a gorilla.

Reliability was also not helped by the fact that the pair in question has no jacket, just a blue wire and a white one that are not even twisted together much; these didn't engage with the RJ45's strain relief bar in any way whatsoever. The whole arrangement had Designed To Fail written all over it.

I've replaced it with an RJ12 that came from a pack of 10 sold as suitable for solid or stranded cable, and whose strain relief bar does clamp nicely on the wires. After crimping, all six contacts are at the same level even though only the centre two have wires behind them. I've also verified that the extra deflection the edges of this 6-way plug inflict on the outermost fingers in its 8-way socket does not permanently deform them. Krone makes good sockets.

The jack is designed for copper wire of a particular thickness, so that the blades on the terminal will pierce the insulation of each wire without damaging the solid conductor inside. Stranded wire will tend to flatten and will not make a good consistent contact.

All the actual CAT5e wiring has solid conductors and appears to have been competently terminated.

I've only ever seen two out of our hundreds of RJ45 wall outlets fail, and in neither case were the punchdown connections responsible. Both failures involved the whole connector body coming adrift internally due to extreme violence applied to the inserted RJ45 plugs by the hobnailed boots of small children; the installing electrician, who predates my arrival at this school, apparently believed that power and network ducting on walls under benches belongs at floor level.
posted by flabdablet at 7:49 AM on November 15, 2016


"coming adrift internally" ... you have a way with words. Thanks for the follow-up, it's always interesting to know how things like this turn out.
posted by jgreco at 9:14 AM on November 17, 2016


« Older Rebuild or replace my laptop   |   Help us move to the UK (London/Hatfield... Newer »
This thread is closed to new comments.