26
1
I have an Ethernet cable running from a Huawei ADSL router to a computer> The total length is 61 meters (200 ft).
When connected to the computer, a cycle of identifying, disconnected, identifying, disconnected keeps looping with a second interval in between. A shorter, 2 meter cable works just fine. I checked the wiring (T568B) and it's OK. The cable looks physically OK too.
Is the length of that cable causing this?
42
It's not the length (100 meters -approximately 330 feet- is the max for ethernet UTP), but most likely the quality of the cable is bad. Maybe the cable is damaged internally or at some point was pinched severely. This is not always visible from the outside. Could also be that one or both ot the RJ45 connectors is crimped on badly or is incorrectly wired. (The cable should be 8 wires 1 on 1 and the ordering of the colored wires IS important.)
EDIT: Some clarification regarding cable categories: For normal ethernet over UTP cable:
Tonny
Posted 2014-06-30T09:50:04.047
Reputation: 19 919
could you please cite a source for that 100m max. length? thanks – avalancha – 2014-06-30T12:50:04.980
1
@avalancha this might help http://en.wikipedia.org/wiki/Fast_Ethernet 100BASE- TX "Each network segment can have a maximum cabling distance of 100 metres (328 ft)." http://www.cisco.com/c/en/us/support/docs/routers/10000-series-routers/46792-ethbase.html "for 100-Mbps Transmission...Maximum segment length..100M" but those links aren't to the standard itself. and also there are different types of ethernet. like 100 Base T, or gigabit ethernet. so I don't know exactly but I see 100M mentioned there.
– barlop – 2014-06-30T12:59:18.8432@avalancha 100 meters is in the official specs for ethernet UTP cables. Look up CAT 5, 5e, 6, 7 on Wikipedia for further information. (Links to official specs are at the bottom of the wiki pages.) – Tonny – 2014-06-30T13:37:18.803
1@ArtB Thanks for the edit. Typing on my phone makes it hard to write something decent.. – Tonny – 2014-06-30T13:49:37.453
@Tonny NP, glad to help. Btw, how many metres would you count a single male/male connector to be? (ie I've heard that for wireless range a single dry-walled wall counts as 10 feet, but a cinderblock one is 30 etc). – Sled – 2014-06-30T14:09:08.287
1@ArtB A single connector maybe 0.10 to 0.20 meters. Doesn't make that much difference for RJ45's. For wireless any estimates are ball-park figures. A lot depends on local conditions and the exact material. (Humidity and absorption by material are MAJOR factors and vary with the weather. I have done warehouse deployments with lots of cardboard boxes. Needed a lot of extra AP's.) Wire mesh (as in fences) can also be a BIG issue for wireless. It can reflect the signal and even create a dead-zone several meters away from the mesh where the direct and reflected signal cancel each other out. – Tonny – 2014-06-30T15:40:32.407
1"up to" is two words. – David Conrad – 2014-06-30T21:02:50.120
Cat5e does gigabit for 100 meters. Not sure why you are spreading misinfomation about that. – Ecnerwal – 2014-07-01T13:58:53.557
@Ecnerwal Technically Cat-5e supports Gigabit (I'll update the answer, you are right). But: In practice: If you are extremely lucky you might be able to get 100 meter on Cat5e with Gigabit. In most cases the speed is reduced to effectively 400 Mb/s or even 200 Mb/s. Latency is often affected by the retries. 30 meters is the rule-of-thumb safe distance of Gigabit on Cat-5e. – Tonny – 2014-07-01T20:32:48.383
5
I have had a cable of 30M be faulty, and a cable tester (albeit a cheap one) showed it as OK / didn't detect a fault.
It may be that longer cables can be trodden on and that makes them less reliable(by damaging them).
The kind of ethernet cables that I find to be reliable, have been one with a rugged jacket, they are expensive, and maybe a better quality build too.
I used to get cables from a company that specialised in cables, but I just found them bad quality! But when I got a few rugged cables (and it would have been a specialist company making them), I found those were very reliable. So, maybe as a rule of thumb to get a quality cable, you could look for rugged jacket ones from a company that makes the rugged jacket ones, if you're willing to pay the price for such cables.
barlop
Posted 2014-06-30T09:50:04.047
Reputation: 18 677
3The only way to truly test a cable is push data through it. Just because it works at a lower frequency doesn't mean it works at frequency. – Loren Pechtel – 2014-07-01T04:45:36.767
@LorenPechtel pushing data through is how I found my cable was faulty. Are there any cable testers that do that? And at a good price? That said.. There isn't much point in a cable tester 'cos they require you to get both ends of the cable anyway(and it may be a long cable). So changing cable for a working one, large or short, is the best test. – barlop – 2014-07-01T06:36:52.177
1
Cheap cable testers generally are only testing continuity (i.e. is it a closed circuit so electrons can flow). Continuity is no where near enough for most data communications. Better testers come in a wide variety of ranges and capabilities (and corresponding prices). The lowest one I would recommend is the Fluke LinkRunner which will cost about $600 new (be careful of used as they may need maintenance).
– YLearn – 2014-07-01T22:01:05.7604
It could be that you are using an incorrect category cable. You didn't mention what speed you are using. Here is a link to Cisco's explanation of cable types. At a minimum you should be using CAT 5 cable, this gives minimum error rate for 100BT and lower. With Gig-E widely available, CAT 6 cable is needed for the length run you state.
DougB
Posted 2014-06-30T09:50:04.047
Reputation: 41
3
Well, since no one has mentioned it...
There is no reason to believe that a particular NIC's line drivers will actually push electrons all the way out to the Standards's maximum lengths. I have had exactly the problem you describe fixed by replacing one instance of a NIC with another instance of the same NIC -- same part number, same lot. I've also found that some NIC models just cannot push signal the full length of a perfectly good maximal cable. Discovering this can be done with an oscilloscope and some messing about.
Remember: the standards only promise maximum lengths. Cheap card manufacturers are not obligated to reach those maxima.
Eric Towers
Posted 2014-06-30T09:50:04.047
Reputation: 151
They are obligated to (since they advertise their products as IEEE 802.3 Ethernet)... but in the real world it's not rare that products developed with a race-to-the-bottom-cost mindset may fail to meet their obligations. – Ben Voigt – 2014-07-01T17:12:59.353
@BenVoigt: They could well be meeting their obligations from the spec. Just because they produce the required voltage with adequate (instantaneous) current doesn't mean they can actually drive the electrons all the way down the wire. But I know what you mean and I totally agree. (Especially since most actual wire runs are much shorter than the maximum. Except between wiring closets, where more expensive equipment is usually involved.) – Eric Towers – 2014-07-01T22:47:00.337
0
I had the same thing once when I used a dual IDSN cable instead of a normal LAN cable. The length of an ISDN cable is shorter than a cat5e UTP Cable, but they look exactly alike. I believe this dual IDSN cable was cat3, 20m. It only worked if we had a hub or pc at either ends, and the cable could not be made longer by just extending it with a plug and cable.
On top of that, the connection itself matters as well. If your connection is 100BaseT it is shorter than 100BaseTX
LPChip
Posted 2014-06-30T09:50:04.047
Reputation: 42 190
0
There are a number of issues that could cause problems with an ethernet cable. Induced noise might be the most common. When you have a long cable (61 M is long) there is a good chance it will run into problems. Pulling that cable to put it in place can stretch the copper inside (not very likely, but possible). But more likely is that the cable is running parallel to other EM conductors (Power Lines). If it remains close to a power line for a long enough distance, the Power line will induce capacitance and voltage onto the ethernet cable. Cat 5, 5e and 6 cables are twisted pair, 8 strands of copper twisted in pairs with the four pairs twisted slightly around each other. This is highly noise resistant, but if exposed to a strong signal for long enough, the signal will intrude. Florescent lights were notorious for this, but they have gotten better. Anyway, check the cable path and make sure the cable is not running parallel to power cables or next to anything like a florescent light fixture. And check the ends of the cable too. If they were not made properly, there could be cross talk (normally be discovered by a tester), or maybe they are not seating properly in the jack.
Rob
Posted 2014-06-30T09:50:04.047
Reputation: 1
0
1) use patch cord to ensure the devices are ok 2) look through the cable and check if it's tensed, compressed, constrained, twisted or turned with small radius. straighten it if required 3) remember the time and reliability is money and just replace it
Anonymous
Posted 2014-06-30T09:50:04.047
Reputation: 1
2What category? What's the quality of the cable? Solid copper or copper-clad aluminum? – Mike Guthrie – 2014-06-30T13:37:54.737
1Did you check if the cable is a patch cable or a cross-over cable? – Sled – 2014-06-30T13:49:36.933