0
\$\begingroup\$

I have an ethernet cable that is hooked up to a 20 foot long 24 AWG twisted pair cable. The twisted pairs are not twisted to the same wires like in the ethernet cable. Would this be a problem? I imagine that it could be, considering that the signal between the wires is differential. Edit: Both the number of twists, and the wires aren't twisted with the same wire on each cable. I will take both as an answer. Additionally, there are 6 different breaks in the cable connected by terminal blocks. I am unsure if this will cause significant reflections or differences in resistance?

\$\endgroup\$
2
  • \$\begingroup\$ Some answers assume you mean that the amount of twist for each pairs isnt the same as the ethernet cable, while others assume you mean that one wire from one ethernet pair is twisted with another wire from another ethernet pair - which is it? \$\endgroup\$
    – BeB00
    Aug 7, 2023 at 22:35
  • \$\begingroup\$ It may be a problem or not. It fully depends on for what you use it for. It definitely is not a good idea though. What do you use the cable for? \$\endgroup\$
    – Justme
    Aug 8, 2023 at 5:41

5 Answers 5

2
\$\begingroup\$

Three things to check:

  • (1) Are the pairs done up as... pairs?
  • (2) Are the characteristics of your 24AWG twisted-pair consistent with Cat5e (same differential impedance, reasonable loss / crosstalk, etc.)
  • (3) Do the pairs in your 24AWG cable have differing twist rates to reduce inter-pair crosstalk?

For (1), pair-wise connection is essential. Fix this first: make the wiring pair-to-pair. It will not work otherwise. If your cable uses punch-down connection this should be easy to fix.

For (2), this will depend on your required cable run length. The longer the run, the more important this will be owing to losses in the cable. While 20' is pretty long, 10/100 is pretty forgiving. 24 AWG twisted-pair is typically in the 100 ohm range anyway, even if it's not specified as impedance controlled. Toss-up.

For (3), this is again length-dependent. And again, 20' is kind of on the long side. Also a toss-up.

This all said, is it a big problem to secure an actual by-dog Cat5e cable for your run, lest there be any question of your link signal integrity? Cat5e wire is cheap; downtime not so much.

If you're facing pushback on getting the right Cat5e cable for the 20' run, get access to a network cable tester to evaluate the quality of the overall cable plant. This rules out any guesswork on your part.

\$\endgroup\$
3
  • \$\begingroup\$ For (1), it most likely will work otherwise -- but it won't enumerate at 1Gbps, and probably won't even do 100. 10 is robust enough to run on wet string (literally), I would be shocked if it doesn't work at all. It's tricky, because just getting a light on the port, and a connection on the network, can make it seem successful, but the proof comes when testing the actual link quality (bandwidth). This also affects immunity to ambient noise, and testing in a quiet location (the usual case) can be similarly misleading. \$\endgroup\$ Aug 8, 2023 at 6:33
  • \$\begingroup\$ Almost no one uses 10M anymore. OP didn't provide details, but I'm guessing they're trying to repurpose some as-is existing plant wiring (maybe old telco wire or suchlike.) Getting the pairs right is an easy problem to solve. That said, the mere threat of getting a tester would be enough motivation to get the electrical contractor to do their job and provision a correct Ethernet wire pull. \$\endgroup\$ Aug 8, 2023 at 17:29
  • \$\begingroup\$ Unless 10M has been widely deprecated and I'm not aware (which is possible--I don't work in networking), it's possible they get a connection, and don't realize possible issues until they're later uncovered (at which point this question will be long forgotten). That's all I was thinking. Definitely, it needs to be done right. \$\endgroup\$ Aug 8, 2023 at 18:43
1
\$\begingroup\$

Yes, that is a big problem - being untwisted would be very bad, but being twisted with the wrong pairs is extremely bad. The signals are differential to each other - there are 4 pairs of signals, and each pair has a plus and a minus. When the plus switches, the minus switches in the opposite direction, to cancel out emi and reduce interference. If you twist the plus from one pair to a signal from another pair, you remove all the benefits of the twisting, and add extra problems, like the wires interfering with each other. I wouldn't be surprised if this is a big problem for even the slowest common type of Ethernet.

\$\endgroup\$
1
\$\begingroup\$

AFAIK, the amount of twist isn't specified anywhere, other than all four pairs should have different rates of twist. The big problem would be the impedance mismatch, assuming your 24 AWG cable is not 100-ohm or not specified; and the step change in impedance at the junction between the two. 10/100 can be surprisingly forgiving though so you might be able to make it work but I wouldn't trust it to anything critical.

Edit: and yes, as @BeB00 says, if the wires aren't twisted to the same pairs (e.g. Rx+ twisted to Tx+) then I wouldn't even bother trying.

\$\endgroup\$
0
\$\begingroup\$

The twisted pairs are not twisted to the same wires like in the ethernet cable. Would this be a problem?

The bigger question is whether the cable characteristic impedances are the same.

If they are the same then having a different twist length isn't going to be a significant problem. If the cable characteristic impedances are different then you might hit significant problems.

\$\endgroup\$
0
\$\begingroup\$

Depends on how long the cable is untwisted for. I had a prototype that I bought a female ethernet extension cable to use for the already made female termination on one end and it had straight wire in it! I used the straight wire for 2 feet. It worked fine but I would not use this in a normal product.

That being said, CAT 5 runs at 100MHz, which is a wavelength of ~120in or 3m. So a small fraction of this is probably not going to create an issue because if you create a stub that is a small fraction (like 1/5th) it's probably not going to create reflections. You'll also get some attenuation and the wire will be more succeptable to noise because you are making the loop area between wires larger.

\$\endgroup\$
2
  • \$\begingroup\$ the question says 20 feet, but I think that it also means that the wrong pairs are twisted - waiting on clarification from OP though \$\endgroup\$
    – BeB00
    Aug 7, 2023 at 22:55
  • \$\begingroup\$ It says 2o feet so yeah, waiting on clarification \$\endgroup\$
    – Voltage Spike
    Aug 8, 2023 at 1:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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