9

I know that Ethernet over twisted pairs have two standards as the figure below: Ethernet standards over twisted pairs

The pairs are twisted to avoid cross-talk.

But why in both standards there is one pair that is not adjacent (pin 3 always pairs with pin 6)?

Once I have got a wrong long cable that was working intermittently. It was made with all rings adjacent to their corresponding tips. After correcting it worked fine.

But I am sure that I have already seen a short cable that worked fine too. Hence there must be an effect that this configuration avoids which I have googled and not found.

3

Data transmission use differential signalling, in pair two wires carry opposite polarity. It lower sensitivity to external interference and to cross interference (betwen different pairs in cable).

Pair 3,6 not adjacent only in jack. All other way it adjacent. In-adjacency in jack make low difference in interference, but make compatibility with phone, who use primary central pair (4,5)

As Tinti comment. If for transmission of differential signal used not twisted pair (it named split pair) cross talk raise dramaticaly and such link not work more that 10m. (UPD: not more that 10m in Full Duplex mode, that is auto detected in most cases)

  • 1
    Thus the problem might be that by not wiring in TIA/EIA 568 standards (A or B) one does not get the TX/RX on the same twisted pair. In other words you are not avoiding cross-talk. – Tinti Dec 10 '14 at 17:51
8

But why in both standards there is one pair that is not adjacent (pin 3 always pairs with pin 6)?

This goes back to compatibility in phone system wiring. It is based on the TIA/EIA 568 standard. The middle pair (4,5) was used for the first voice circuit. the next pair (going out from the middle - 3,6) was for the 2nd phone line. The reason it makes sense is that twisted-pair ethernet was first used on phone system (category 3) cabling.

There is a slight difference between the two TIA/EIA 568 standards (A or B), but I will leave that exercise up to you.

  • Nice! But is there any electrical effect that makes building the cable this way more noise tolerant or is it a pure historical reason? – Tinti Dec 6 '14 at 0:07
  • 2
    @Tinti - It's historical but practical. This allows an installed cable to be used either for phone or data. Phones use the innermost pair as the first line, and the next most-outer pair for the second line. An installed cable system is very expensive (especially if done correctly), and the flexibility avoids the need for two separate cable plants. – Ron Maupin Dec 6 '14 at 1:24
  • 2
    @Tinti The most important thing is that you have TX+/TX- and RX+/RX- actually on pairs rather than split across two pairs; then you will get crosstalk problems. It's like driving on the left or right; the reason why you do in your particular country is not nearly as important as that you drive on the correct side. – richardb Dec 6 '14 at 11:23
  • @richardb thanks! I think my problem was really a cross talk since I was in a long cable. – Tinti Dec 23 '14 at 16:54
2

But why in both standards there is one pair that is not adjacent (pin 3 always pairs with pin 6)?

Phone standards traditionally put the first pair on the middle, then the second pair straddling it, then the third pair straddling that and so-on. RJ series connectors were designed so that you could plug a smaller plug into a larger socket and it would end up in the center and this wiring arrangement meant that plugging a smaller plug into a larger socket would result in a sensible connection.

This worked fine at the low frequencies of voice and old low speed data systems but was poor for crosstalk. noise susceptibility, reflections etc at higher frequencies. The greater the number of pairs in this arrangement the worse the high frequency performance gets.

A wiring standard with each pair of wires on a pair of adjacent contacts would be best from a high frequency performance point of view but would be lack any backwards compatibility with the existing telco standards.

The standard used for ethernet is a compromise. The two central pairs are arranged in the same way they would be in a traditional phone wiring giving compatibility with phone applications that only use the first two pairs. The two outer pairs are arranged in a way that optimises their high speed performance.

Once I have got a wrong long cable that was working intermittently. It was made with all rings adjacent to their corresponding tips. After correcting it worked fine.

When doing differential signaling on twisted pairs it is important that a pair of signal connections are actually on a twisted pair of wires. If the two lines from a signal pair are split between two different twisted pairs then this is known as a "split pair" and will have far worse crosstalk and EMC performance than it should have.

Therefore your cable must match the pair arrangement your devices are expecting.

But I am sure that I have already seen a short cable that worked fine too. Hence there must be an effect that this configuration avoids which I have googled and not found.

Signal integrity problems become worse with increasing cable length.

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.