3
\$\begingroup\$

Is telephone and 10/100 MBps LAN wiring electrically symmetrical? I mean is it allowed to swap wires in the pairs, or is one of them like GND so they cannot be swapped? It's important to know when I plan longer wires if I can use just two wires of the same color.

I am unable to find some information on how telephone and 10/100Mbps LAN communication works on this physical layer because I don't know the correct technical terms I should search for.

(I am not talking about symmetrical speed of digital lines.)

\$\endgroup\$
8
\$\begingroup\$

You cannot swap wires in the pairs for either, but for different reasons.

In telephony, the AC voice signal is overlayed on a DC supply, which is used to power telephones. Research starting point would be hybrid. If you filtered out the DC part, and only transmit the AC part, you could actually swap the wires.

In 10/100 Mbit/s LAN you cannot swap the wires, because the signal on the wires is differential, the signal of the second wire gets subtracted from the signal on the first wire. If you switch the wire, you reverse the polarity of the signal, effectively munging the bits. Start out here: physical layer

\$\endgroup\$
  • 4
    \$\begingroup\$ Postel's Law of Robustness: "Be conservative in what you do, be liberal in what you accept from others." Although new PHYs may be able to figure out and cope with wrong wiring arrangements (they're liberal), always stick with convention and standards (be conservative). \$\endgroup\$ – shimofuri Jan 11 '13 at 9:59
  • 1
    \$\begingroup\$ Most phones will work either way round but it's better if you just get the wiring right, it may affect line-tests as the automated test equipment can/does identify A/B reversal. \$\endgroup\$ – John U Jan 11 '13 at 10:02
  • \$\begingroup\$ Interestingly, I just found out that one of my home phone cables is 6P2C with wires switched. And it works. So the posipiet's answer now doesn't seem to me very good. \$\endgroup\$ – Al Kepp Jan 13 '13 at 10:00
  • \$\begingroup\$ @Socrates: Please write your comment as an answer to let me accept it. \$\endgroup\$ – Al Kepp Nov 7 '14 at 0:34
  • \$\begingroup\$ @AlKepp done as requested. \$\endgroup\$ – Socrates Dec 12 '14 at 21:41
1
\$\begingroup\$

I am a telco repairman. ALways go with the convention. Standard is Blw = first line Orw = second line Grw = third line Brw = fourth line Ethernet = Blw and Grw = standard for data and Orw and Brw are spares, unless using Cat 6 or something exotic. You can lay wire on the ground and carry voice a LONG way when necessary (miles). The pairs contain different values of twist. VDSL is preferrred to be wired from the NID (protector) to the modem location on the GrW pair. The twist has been determined to be the best for VDSL applications. From experience whatever you need to repair a circuit is what you do for service. Changing pairs is not an issue. But keeping continuity isolated and straight ahead is the key. You cannot combine circuits. That is a crossed line. Buzz and humm will be your problems as well as shorts. As you should know ethernet has a 350 foot limit. However it can be longer, it just isn't shown in the specs and you could have a problem tommorrow after it works today.

\$\endgroup\$
  • 1
    \$\begingroup\$ This doesn't answer my question at all. \$\endgroup\$ – Al Kepp Jan 24 '13 at 19:12
1
\$\begingroup\$

The wiring is not symmetrical.

However most phone systems are indifferent to the polarity of the line. Some older PABX and enhanced PSTN systems did require the correct polarity before enhanced functions would operate. The bulk of modern electronic phone line termination circuits incorporate a bridge rectifier before the DC clamping/hook/flash circuits so do not know which polarity is connected.

10 and 100 Mbps UTP LAN cabling makes use of a RJ45 8pin connection. Only 4 of the circuits are used even if all 8 are connected. The pinout requires the transmit pins of one device to connect to the receive pins of the other device. This requires proper pin selection. The transmission is a differential drive so the transmitter has a positive and a negative pin and these must match the appropriate positive and negative on the receive side. The LAN interfaces were made with the transmit either on pins 1 and 2 and receive on pins 3 and 6 but on devices like network switches and hubs the transmit was on 3 and 6 and receive on 1 and 2. This allowed straight through cabling in this standard configuration. The cable pairs had to be on 1 and 2 and the other on 3 and 6.

Now when there were more unusual connections cross over cables were required. All this was a bit confusing so automatic detection of which pairs were connected to which pins was implemented. Modern devices appear to also tolerate crossing of positive and negative as well as swapping of pairs.

Modern Gigabit LAN uses 4 pairs and does not need cross over cables as it establishes the wiring in the hardware and communicates appropriately. I am not certain if the standard requires them to ignore pair crossing but this may be possible.

So the answer to the LAN question is modern systems also seem to be increasingly indifferent.

\$\endgroup\$
  • \$\begingroup\$ The vast majority of flat phone cables of the RJ12 type terminated with a flat cable were of indeterminate crossing. It was rare to find cables that were marked and even more rare to find cables with the connectors in any predictable orientation even if the cables were marked (with the ridge that is). \$\endgroup\$ – KalleMP Aug 25 '17 at 21:37

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.