3

How can 1000-BaseT (Gigabit ethernet twisted copper pairs) technology transmit over the Rx pins & receive over the Tx pins?

It's baseband technology (hence the name), right? Discrete digital pulses that occupy the entire bandwidth. Not broadband analog magic, where you can have multiple signals/channels propagate a single pair.

Why couldn't 10-BaseT (10Mb/s) or 100-BaseT (100Mb/s) do the same? What changed, specifically to allow this?

I know that 1000-BaseT uses all four pairs (8 conductors) of the RJ45 connector/Cat6 cable, and the earlier technology only used 2 pairs (4 conductors). But that alone couldn't facilitate this phenomenon. There must be more to it.

Improve this question
4
  • 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 provide and accept your own answer.
    – Ron Maupin ♦
    Commented Dec 25, 2018 at 9:36
  • Thanks @Ron, I know how it works. If and when I'm satisfied an answer is fully accurate, and I can understand it; I'll declare it. If you can improve or clarify an existing answer to remove all doubt, please do. I haven't abandoned or forgotten about it though. I think it's important we don't just pick something purely for the sake of closing it.
    – voices
    Commented Dec 25, 2018 at 16:45
  • OK. I was just doing year-end cleanup. Also, I would suggest that you really asked this question in the wrong place. You will probably get a better explanation on Electrical Engineering.
    – Ron Maupin ♦
    Commented Dec 25, 2018 at 19:14
  • @Ron. I think you're right. Can you transfer it to the other site?
    – voices
    Commented Dec 27, 2018 at 8:25

3 Answers 3

Reset to default
5

In 1000base-T there is circuitry called a hybrid which make the signal bidirectional. It's not really transmitting and receiving the wrong way round on the receive and transmit pairs, it's transmitting and receiving on all the pairs, which the other standards dedicate to transmit and receive.

IEEE 802.3 defines them as

"1.4.220 hybrid: A circuit (implementable with active or passive components) that enables full duplex trans- mission by allowing symbols to be transmitted and received on the same wire pair at the same time. It is often used together with an echo canceller to get adequate separation of transmit and receive signals."

Wikipedia entry describes it as

In a departure from both 10BASE-T and 100BASE-TX, 1000BASE-T uses four lanes over all four cable pairs for simultaneous transmission in both directions through the use of echo cancellation with adaptive equalization called hybrid circuits. -- https://en.wikipedia.org/wiki/Gigabit_Ethernet#1000BASE-T

In this case the bidirectionality is done with "voltage level analogue magic" (as it might be called) rather than "frequency-domain analogue magic".

It certainly could have been possible with other signalling, or have half-duplex turnaround (like two-wire RS-485), but the choice was made for pairs to go only in one direction. Like most standards decisions, it's a trade-off between cost, complexity, required-time-to-implement, and political manoeuvring from the interested parties.

Improve this answer
4
  • It's actually not a hybrid circuit that is used but a hybrid which is a special type of transformer.
    – Zac67 ♦
    Commented Oct 8, 2018 at 17:04
  • So, if you had some superfast multimeters, you would detect the same traffic on all pairs?
    – voices
    Commented Oct 8, 2018 at 17:13
  • Superfast multimeters, aka oscilloscopes, would show different bits on the different pairs, but yes, traffic going in both directions.
    – jonathanjo
    Commented Oct 8, 2018 at 17:17
  • 1
    @Zac thanks for picking up ambiguity, I reworded hope you agree it's clearer; also added IEEE definition
    – jonathanjo
    Commented Oct 8, 2018 at 17:21
3

Broadband (xDSL) uses frequency-division duplex (FDD) to separate receive from transmit direction - both use their own frequency sub-band.

Ethernet 1000BASE-T and faster use telecom hybrids and echo cancellation to cancel out their own signal in order to receive the other end's signal. Unlike 10BASE-T or 100BASE-TX, there are no dedicated receive and transmit pairs - the data stream is split into four lanes and transmitted over all four pairs simultaneously while they are also receiving.

Why couldn't 10-BaseT (10Mb/s) or 100-BaseT (100Mb/s) do the same? What changed, specifically to allow this?

When 10BASE-T was designed in the late 1980s, following StarLAN, running 1 Gbit/s over twisted pair cabling was pure science fiction. There weren't even computers capable of using that speed - it was generally limited by system throughput, not the cable.

Technology advanced and for 100BASE-TX, IEEE had to think up some way to reduce the frequencies on the cable (by copying FDDI's MLT-3 scheme). For 1000BASE-T later on they had to increase the voltage levels (PAM-5) and split the traffic to four lanes. 10GBASE-T requires a much better cable (Cat-6A) and uses very elaborate encoding to reduce frequencies even further.

Ethernet's popularity has always been significantly powered by its low cost. Oversizing the capabilities would have seriously hurt that.

Improve this answer
9
  • So, if you had some superfast multimeters, you would detect the same traffic on all pairs?
    – voices
    Commented Oct 8, 2018 at 17:14
  • No. Each pair carries 1/4 of total traffic. The stream is split in four lanes, these are transmitted separately and reassembled at the link end.
    – Zac67 ♦
    Commented Oct 28, 2018 at 11:55
  • is that what makes it faster ?
    – voices
    Commented Oct 28, 2018 at 16:29
  • Like a bucket with four holes drains quicker than one.
    – voices
    Commented Oct 28, 2018 at 16:32
  • Exactly. Instead of a single lane running 1000 Mbit/s (as with 1000BASE-X) you've got four lanes of only 250 Mbit/s each. (Very roughly) PAM-5 reduces that to 125 MBaud which Cat-5 can handle.
    – Zac67 ♦
    Commented Oct 28, 2018 at 16:37
2

Why couldn't 10-BaseT (10Mb/s) or 100-BaseT (100Mb/s) do the same? What changed, specifically to allow this?

The quality of the cabling improved drastically as well as the methods for transmitting data and the receivers used to “read” the signal from the cable. These days you could retroactively modoify the 10/100BASE-T standards to use the same technologies but why would you?

Improve this answer

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.