Baffling structured cabling issue

humaxf1

Hi all,

A strange one here for you...

I have been asked to (in an office) patch a patch panel to a network switch. The data points at the desks are in sunken floor boxes. Structured cabling runs under a false floor. Straightforward stuff...yeah right!!!

I ordered short patch cables (1.5m) for the cabinet and longer ones (5m) to connect the PCs to the floor. Here's where it gets strange...

** 5m cable passes a cable test on a basic tester and on a Fluke analyzer

** The structured cabling passes the test too

** A 5m cable between a router and laptop, no problem getting an IP

** A 5m between the same laptop and a floor point, it struggles to get an IP, connects/disconnects, connects/disconnects constantly

** If I use a totally independent cable (20m), no problem getting an IP address (I don't think the max CAT5 length is an issue in this case)

I even used 2x 1.5m patch cables at random, 1 at the patch panel and 1 at a floor point with my laptop. This combo successfully got an IP address. At this stage I'm stumped, although it looks like the issue is the the 5m cables.

Looks like I'll have to buy a new batch of 5m cables at this stage as I can't put the blame on the structured cabling.

Any ideas would be gratefully appreciated. Thank you

Static M.e.

Very strange.. It sounds like the 5m cables all right, perhaps the wires within the cable are in the wrong order which the cable tester isn't picking up on it.. (No idea - guess). Could they be 5m crossover cables?

if you had a crimping tool you could always chop off both ends of a 5m and test one

humaxf1

The thought had crossed my mind (pardon the pun) but the test says they are straight through. Looking carefully through the RJ-45 and at the colours, PIN 1 is to PIN 1, 2 to 2, 3 to 3 and so on.

The company who did the install were baffled also. Ah well...that's IT for ya!!!

I could try putting new RJ-45's on the 5m for the craic and see what happens.

johnmacward

Static M.e. wrote: »

Very strange.. It sounds like the 5m cables all right, perhaps the wires within the cable are in the wrong order which the cable tester isn't picking up on it.. (No idea - guess). Could they be 5m crossover cables?

if you had a crimping tool you could always chop off both ends of a 5m and test one

I work with him and had the exact same conversation. Crossover or length could have been the only issues, as the switch at the end was only a Linksys router, that means no MDI-X for the crossover issue and probably slightly underpowered LAN ports for long distance. The dudes who ran in the cables are pro's, no doubt about it and there Fluke analyser (all seven grand's worth) showed up no crossover issues and all cabling was certified to TIA standards according to the Fluke report.

axer

So has the link been tested using the fluke tester from the 5m cable all the way to the end of the 1.5m cable as opposed to each segment separately?

humaxf1

@ axer;

Your suggestion was not carried out, but I would suspect it would fail.

There is another part of the building to be structured cabled, so I may be able to get the test carried out just out of curiosity.

BaconZombie

Did you check the wiring on both cable and the patch panel ?

Are you mixing 568A & 568B {or even a cross-over cable } ?

BigEejit

cat5e is good for 100meters isnt it?

If everything is working ok using just the 1.5m cables connected to either end of the structured cabling and is not working when you subsitute in a 5m cable then it pretty obvious that the 5m cables are your problem ....

Is the cable properly terminated? i.e. where the individual wires go into the rj45 connected there should be an absolute minimum of straight wire, the pairs should be twisted until just when they are pushed down on the posts in the connector.

William Powell

I think there is one variable here that everyone has missed.

The one thing that is different when the 5M cables are tested and when they are used and thats the CAT5 sockets they are plugged into.

You can somtimes get weird (cheap?) CAT5 plugs that have the pins pressed very deeply into them or the plastic inbetween the connectors is wider than normal (even damaged and pushed over to one side by the crimping tool).

So just maybe the sockets on the fluke tester made good contact with the cheap plugs but an average floor socket didn't make good contact?

Take a look with a magnifying glass and compare the pins and the plastic on the plugs on a good lead with the ones on a bad one?

humaxf1

An update...

I got a new bunch of cables to replace the 5m & from a different supplier.

Absolutely zero problems now. I went to each data point and my laptop got an IP address as you would normally expect.

johnmacward

No it's not a crossover issue, the fluke report shows everything straight through.

Some of the other's mentioned highly depressed pins, cheap sockets and so on but unfortunately when the fluke is in place everything works fine - with the problematic cables. When there's a DHCP router/switch on the end, nothing works. It's like it's a case of mis-frequency or lack of power, something along those lines.

As far as I've heard we we're back out again with different brand cables and everything is perfect. It's horribly strange, there's obviously an explanation, but still - horribly strange.

humaxf1

BigEejit wrote: »

cat5e is good for 100meters isnt it?

If everything is working ok using just the 1.5m cables connected to either end of the structured cabling and is not working when you subsitute in a 5m cable then it pretty obvious that the 5m cables are your problem ....

Is the cable properly terminated? i.e. where the individual wires go into the rj45 connected there should be an absolute minimum of straight wire, the pairs should be twisted until just when they are pushed down on the posts in the connector.

Yes, 100 metres. I have seen RJ-45 terminations that are DISGRACEFUL, hanging on with a prayer and they still work. A 5 year old could do a far superior job after instruction on how to terminate! I've also seen CAT5 that has been trodden on, office chairs wheeled over it constantly and it still works!

humaxf1

William Powell wrote: »

I think there is one variable here that everyone has missed.

The one thing that is different when the 5M cables are tested and when they are used and thats the CAT5 sockets they are plugged into.

You can somtimes get weird (cheap?) CAT5 plugs that have the pins pressed very deeply into them or the plastic inbetween the connectors is wider than normal (even damaged and pushed over to one side by the crimping tool).

So just maybe the sockets on the fluke tester made good contact with the cheap plugs but an average floor socket didn't make good contact?

Take a look with a magnifying glass and compare the pins and the plastic on the plugs on a good lead with the ones on a bad one?

I tested at random, 5-6 of the 5m cables on our structured cabling in the workshop and they worked fine. The jacks used in the office are "excel" who I would regard as high quality but are allergic to the 5m cables LOL

I'll compare the old and the new plugs just out of curiosity. Main thing is, the wired LAN is now working as it should.

thegen

humaxf1 wrote: »

I tested at random, 5-6 of the 5m cables on our structured cabling in the workshop and they worked fine. The jacks used in the office are "excel" who I would regard as high quality but are allergic to the 5m cables LOL

I'll compare the old and the new plugs just out of curiosity. Main thing is, the wired LAN is now working as it should.

Would not consider Excel as high quality. They are in the same league as Austin Taylor and Connectix. A level up would be AMP/Tyco, Hubbell, next level would have Brand Rex, Leviton, Otronics and premium brands would be Systimax and Krone.

Excel will do a job. Your problem was faulty 5m leads.