5

I'm doing course for CCNA and in one lecture teacher showed how to make a crossover cable for Gigabit Ethernet. But as I know Gigabit Ethernet uses all wires of twisted pair cable and each pair is used for both data transmission and reception, so devices that use this standard should be able to transmit and receive on every pin. Then why do we need a crossover cable in such case? Of course I know about Auto-MDI(X) and that it allows to use straight cable in all cases, but it doesn't cancel my question.

1
  • Did any answer help you? If so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you can post and accept your own answer.
    – Ron Maupin
    Jan 5, 2021 at 13:22

1 Answer 1

8

Short answer: you don't.

1000BASE-T requires Auto Negotiation to work - the fast link pulses are to be transmitted and received on specific pairs, compatible with 10BASE-T and 100BASE-TX. In theory, this requires the normal T586A/T568B crossover cable between two MDI ports or two MDI-X ports. Furthermore, the other two pairs would need to be crossed as well in order to match the expected transmitter-receiver pairs, requiring a special gigabit crossover cable as detailed in IEEE 802.3 Clause 40.8.2 Crossover function.

However, practically all 1000BASE-T interfaces support the Auto MDI-X option from Clause 40.4.4, making a crossover unnecessary. In addition, the 1000BASE-T PMA sublayer identifies each pair on each side (Clause 40.1.4), so 1000BASE-T simply works with or without single or dual crossovers.

3
  • Regarding "the 1000BASE-T PMA sublayer identifies each pair on each side", how exactly does the PMA do this? §40.1.4 mentions an "ability to automatically detect and correct for pair swapping", but I think that might just be referring to §40.4.4 (Automatic MDI/MDI-X Configuration), because I don't see anything in §40.4.2 (PMA functions) that suggests that the PMA identifies the pairs.
    – claymation
    Jan 8 at 23:25
  • @claymation It's a service primitive between PCS and PMA, with PMA using unique symbols per pair - see 40.2.5.
    – Zac67
    Jan 9 at 7:57
  • §40.2.5 defines PMA_UNITDATA.request, which is the PCS–to–PMA primitive for transmission. §40.2.6 defines PMA_UNITDATA.indication, which is the PMA–to–PCS primitive for reception. It says, "During reception the PMA_UNITDATA.indication simultaneously conveys to the PCS via the parameter rx_symb_vector the values of the symbols detected on each of the four receive pairs BI_DA, BI_DB, BI_DC, and BI_DD," but nowhere does it specify a mechanism for identifying or swapping pairs. It merely passes the symbol received on pair BI_DA to the PCS in rx_symbol_vector[BI_DA], etc.
    – claymation
    Jan 9 at 20:36

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.