2
\$\begingroup\$

How many segments are permitted in an F/UTP CAT 6A cable run? Segment endpoints require peeling back the foil and untwisting the pairs, which I fear will affect data speeds. I heard that a cable run with a mixture of different cable CATegories will reduce to the speed of the slowest one. Will having so many segments, even if all are CAT 6A, keep compound-reducing the speed? Is the following run allowed and will it work at 10 Gbps, assuming cable, plugs, keystones, and couplers are all CAT 6A and shielded? (In each segment, besides the first, last, and the one with the coupler, I give the names of the endpoints.)

  1. router
  2. modular plug to modular plug (6 ft.)
  3. female-female coupler
  4. plug to plug (6 ft.)
  5. keystone to keystone (12 ft., wall, plenum jacket)
  6. Plug to plug (50 ft., attic, plenum jacket)
  7. keystone to keystone (12 ft., wall, plenum jacket)
  8. plug to plug (10 ft.)
  9. computer

If this is unreliable, should segments 5 through 7 be consolidated into one keystone-to-keystone segment, even though that makes replacement much harder?

What is the maximum segment and length limit? Does length limit depend on number of segments and vice versa?

Will this work with PoE or PoE+ at those speeds or any speed (replace computer with PoE/PoE+ device and replace router with PoE/PoE+ injecting router)?

\$\endgroup\$
  • \$\begingroup\$ This question is probably better served on SU.EE \$\endgroup\$ – laptop2d May 6 '17 at 3:48
  • \$\begingroup\$ @laptop2d , What's SU.EE? \$\endgroup\$ – CodeBricks May 6 '17 at 6:34
  • 2
    \$\begingroup\$ The F-F coupler will probably ruin a 10 Gbit connection by itself unless it's a very special one - I've had trouble making them work at 1 Gbps with a shorter, simpler run. \$\endgroup\$ – pericynthion May 9 '17 at 18:53
  • 1
    \$\begingroup\$ @pericynthion , +1. Did you use a shielded coupler? Was it wired like 2 back-to-back regular keystones? Do the wires have zero twists between the 2 jacks? \$\endgroup\$ – CodeBricks May 10 '17 at 4:59
  • 1
    \$\begingroup\$ @pericynthion +1 since this is important for the OP to know. That said, I believe there are CAT6 couplers on the market. \$\endgroup\$ – Dmitry Grigoryev May 10 '17 at 7:06
2
\$\begingroup\$

The standard tells you what characteristics your transmission medium has to provide in order to achieve a given error rate. For 10GB Ethernet which uses DSQ128 code, you will need an signal to noise ratio above 23 dB to get below \$10^{-7}\$ BER. Augmented CAT6 cables (used for 10 Gb) have specific requirements on various losses, reflections and noise levels at different frequencies to achieve this SNR level:

enter image description here

People doing network installations have network analyzers like this one to measure these characteristics. Whether your cable will work will depend on how your installation will be done: how much wire you will untwist, how well the wire will be crimped, at what angles you'll bend the cables etc.

What Evan seems to refer to is a typical installation described in the Telecommunications Cabling Standard TIA-EIA-568-B, which has these cable characteristics pre-calculated for typical commercial network topologies, like a single horizontal cable with patch cords on both ends:

enter image description here

For such topologies, maximum allowable lengths have been precalculated, which achieve the characteristics above with typical cables:

enter image description here

Now, having a "typical topology" doesn't mean that a different topology wouldn't work. It simply means that the above pre-calculated values don't apply to your case, and being under the maximum length doesn't guarantee that your installation will comply.

Practically, I suggest you cut, crimp and assemble your cable together before installing it, then test the network connection by actually connecting two pieces of equipment with that cable and checking if you get 10 Gbit/s. You're right that eliminating a few sections in the middle may help, but I see no way to know whether that will be necessary or sufficient without a measurement or a performance test.

\$\endgroup\$
  • \$\begingroup\$ Great answer! That analyzer is expensive. Is there any software suite to test link between computers to the same or better degree? \$\endgroup\$ – CodeBricks May 10 '17 at 10:24
  • \$\begingroup\$ @CodeBricks You can hire equipment like that. \$\endgroup\$ – Andrew Morton May 10 '17 at 10:42
  • 1
    \$\begingroup\$ @CodeBricks I know it's expensive, that's why I'm suggesting you simply try out your cable and see if you get 10 Gbit. If you do, great, if not, you'll need an analyzer if you want to know what is wrong exactly. netcat or a similar software would do, just make sure your computer can actually output data at that rate. \$\endgroup\$ – Dmitry Grigoryev May 10 '17 at 10:45
1
\$\begingroup\$

The various Ethernet standards assume that there are 3 cable sections per link: a patch cable going from the switch to a patch panel, a long horizontal run, and another patch cable going from a keystone jack or patch panel to the endpoint device or another switch. The cable and jack quality standards, as well as the echo cancellation requirements are based around achieving the needed SNR in this sort of configuration over the specified length in a typical environment (high EMI environments might require additional shielding for instance).

So by having so many junctions, you are operating outside what is expected and designed for. That doesn't mean it will or won't work, just that you are going beyond what is typical. Professional installers will generally do a link test, looking for crosstalk and SNR targets. If it passes test, it should work. That is really the only way to tell.

\$\endgroup\$
  • 1
    \$\begingroup\$ +1 Would you cite a source regarding the 3 sections per link? \$\endgroup\$ – CodeBricks May 10 '17 at 4:55

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.