I have a 25 meter Cat5e ethernet cable connected to a FritzBox and a computer. If I'm using a short cable I will get an 100 and 1000 Mbps connection respectively. But with the 25 meter cable I will not get any IP of the router. Otherwise if I set ethtool -s p4p1 speed 10 duplex full autoneg off it works. Obsolete: Is there any reason for that? I pretty sure cable length of ethernet is not limited to 25 meters but 100 meters.

I could also live with 10 Mbps. But I don't want to fix that so if there are other communication partners and shorter wires respectively I don't want to limit that speed too. So the easiest way if the configuration is correct, to have a "fallback" routine not only to stay connectionless but also use 10 Mbps as a fallback. I'm not familiar with the negotiation process.

I wonder about setting 10 Mbps manually will work but the devices don't get it themselves. Maybe the negotation just avoids to attend an unstable connection?


The ethernet cable is in a UV protected tube within the power supply. However it doesn't work with unplugged power supply too. And it cannot be easily replaced due to time and cost reasons.

I noticed that the cable is split up and soldered again so that it can be plugged. I think thats the weakest link. So testing the cable would be the best way to go. Btw. what output do I expect of an cable tester? The frequency to which it will work correctly?

Finally How can I get 100 Mbps and 1 Gbps connections and if that is not possible force 10BaseT? Right now the negotiation process just hangs. Or how can I see clearly why it hangs?

  • 2
    It seems highly likely that your 25 meter cable is badly or incorrectly made. If you have a cable tester, then test it. If you don't have one, then simply buy a new cable.
    – Zoredache
    Oct 20 '14 at 23:36
  • @Zoredache I already test it with an multimeter to exclude electrical contact problems. The problem is that the cable is packed in a UV protected hose with power cable etc. It can not easily be replaced. So I wanted to know if there is a way to get closer to the error and eliminate configuration errors. The easiest way if the configuration is right, to have a "fallback" routine not only to stay connectionless but also use 10Mbps as a fallback. Automatically it doesn't seem to work.
    – wenzul
    Oct 20 '14 at 23:59
  • You tested that all 4 pairs were properly terminated? How exactly did you test? A multimeter really isn't very useful for checking network cables.
    – Zoredache
    Oct 21 '14 at 0:31
  • @Zoredache Yes I know, therefore I wrote I had tested with a multimeter to get a beep/not beep conclusion. I don't have equipment to measure impedance termination. Maybe there is a way from the software side to see whats going on in more detail? Because I wonder about setting 10 Mbps manually will work but the devices don't get it themselves.
    – wenzul
    Oct 21 '14 at 0:42
  • The quick way to see if this cable is the problem is to substitute it with a known good cable. 25m is not that long. This is a more sensible step than purchasing a cable tester.
    – sawdust
    Oct 21 '14 at 1:12

As Zoredache has said, this is probably a wiring problem. A cable-tester should confirm you've got one or more wires in your cable that aren't connected (or connected incorrectly).

I recommend you get a proper RJ45 ethernet tester from ebay. You should be able to get one for around $10. You may as well order some replacement RJ-45 connectors and a crimping tool so you can replace your connectors yourself.

Here, I'll give you a recipe....

My recommendations, in order

  1. Order an RJ-45 cable tester, replacement RJ-45 connectors and a RJ-45 crimping tool from ebay (etc).

  2. If the tester reports the cable is fine, then your problem isn't likely to be the cable. If the tester says the cable's fine, then stop following these instructions and look at your other equipment.

  3. I'm betting the tester will indicate there's a problem with one or more wires. Pick one end and replace the connector. You can find endless guides on how to do this with a quick search.

  4. Re-test the cable. If there's still a problem, then replace the other RJ-45 connector.

  5. Re-test the cable. If there's still a problem AND you're certain you didn't screw up either end (just keep replacing until you're certain you got it, maybe practice on (s)crap cables before working on your important one), then I wager one of the wires in the cable is physically snapped somewhere and you'll probably have to just replace the whole cable.

  • Thank you for your step-by-step recipe. :) That would be the hardware approach. I also think this could be a cable issue. But before I buy a cable tester I want to know if there is a deeper look in the logs than syslog "interface not ready" possible. In the first step I don't wanna replace the cable. So I need to set 10Mbps for this cable but don't wanna restrict it for all connections. But in the second step if there is more time and budget I have to investigate the cable.
    – wenzul
    Oct 21 '14 at 1:03
  • Investigate the mains power cable first. AC current induces voltage in any nearby conductive material eg wire, which is why you should never run mains & data [or audio] in parallel & only ever cross at 90 degrees.
    – Tetsujin
    Oct 21 '14 at 6:56
  • @Tetsujin I plugged power off and just used the network. Also works only in 10Mbps mode. Or does the power cable by itself has an effect?
    – wenzul
    Oct 21 '14 at 11:12
  • mains cable without power should have no effect; so long as you are certain there was no power on it at all. Back to it being a cable issue, I think
    – Tetsujin
    Oct 21 '14 at 11:14
  • @Tetsujin Yep thanks, there was no power. And I think it's the cable. But I'm still looking for an answer to my second question.
    – wenzul
    Oct 21 '14 at 14:11

Posting this late response for the benefit of future readers:

Double-check the wiring order of the twisted pairs in the cable. If they have been terminated so that the pairs run in the order 1/2, 3/4, 5/6, 7/8, you will get substantial cross talk between the wires giving an unreliable link at speeds like 100BASE-T or 1000BASE-T.

10BASE-T is far more tolerant of marginal cables, and this might explain why your connection behaves as you described even though all the individual wires buzz out correctly with a multimeter.

The correct wiring order for each twisted pair is:

  • pair 1 - orange - pins 1/2
  • pair 2 - green -  pins 3/6
  • pair 3 - blue -   pins 4/5
  • pair 4 - brown -  pins 7/8

The odd numbered pins are always white wires and the even numbered pins are coloured wires. In the US, you may find orange and green pairs swapped in position, but as long as both ends are done following the same pattern, it will work fine for normal networking at all supported speeds.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.