Ethernet is a family of specifications that governs a few different things: It covers all the different wiring specifications (10BASE-T, 100BASE-TX, 1000BASE-T, etc…). It describes how to send bits (1s and 0s) across each wire. It also determines how to interpret those bits into meaningful frames.
Initially, this article was meant to just cover the basic differences and use-cases for Crossover cables and Straight-through cables. But in light of our mission statement, we thought the topic of Ethernet Wiring deserved a bit more depth.
We’ll start off with a disambiguation of all the terminology that gets thrown around when discussing physical cabling, then answer a couple basic questions: Why do we need crossover cables vs straight-through? What exactly is Twisted Pair? How is a single bit transmitted across the wire? Finally, we’ll wrap things up with a look at the standard for Gigabit Ethernet.
If you’ve been around the networking world for even a short duration, you’ve heard lots of terms that are thrown around referring to cabling. Terms like Ethernet, Twisted Pair, RJ45, Shielded, and Unshielded.
But what do each of these terms mean? How are they different from one another? Are any of these terms being misused? To put it bluntly, yes – these terms are often misused. Let’s take a look.
This is the specification that governs the physical connector on either end of an Ethernet wire. This is what regulates that there are 8 Positions and 8 Contacts. It also defines the design and dimensions of the clear plastic plug that terminates the cable.
Registered Jack standard number 45 specifies the amount of wires in the cable, the order in which they appear, and the usage of the 8P8C physical connector.
Specifically, RJ45 defines two wiring standards: T568a and T568b:
Notice the only real difference between the two standards are the colors of wire pair 2 and pair 3.
Twisted Pair wiring is a type of cable which uses eight individual wires in a bundle. The eight individual wires are paired in sets of two, and each pair is twisted around each other. This creates four pairs of wire, each of which serve as a channel through which data can be transmitted.
The pairing of the wires is very important, and we will look at why later on in this article, but the short version is it helps negate and minimize the effects of Crosstalk and Electromagnetic Interference (EMI).
There are two prominent types of Twisted Pair wiring, a Shielded variant and an Unshielded variant:
Notice in both cases, the pairs of wire create four distinct channels or lanes through which data will be sent.
Unshielded Twisted Pair (UTP)
This is the more commonly deployed variation. There is no additional shielding against electromagnetic noise, but none the less, UTP can carry a signal reliably due to innate features of twisted pair wiring. We will explore these in more depth later on in this article.
UTP is less expensive, more (physically) resilient, and more flexible. These attributes typically make UTP the preferred choice.
Shielded Twisted Pair (STP)
STP has additional shielding around each pair of wires and then one more shield around all four pairs. This helps contain and isolate the electromagnetic noise that occurs when signals travel through a wire.
That said, if any part of the shielding is damaged, or if the wires aren’t perfectly grounded on either side of the connection, the shielding can act as an antenna and introduce additional electromagnetic noise from stray radio waves and Wi-Fi signals in the air.
Moreover, the STP wire must also be coupled with shielded 8P8C connectors to ensure the additional shielding is present throughout the full end-to-end spectrum of the wire.
As you can imagine, STP is the more expensive variant. STP is also more fragile than its UTP counterpart – the shield is prone to tear if the wire is bent excessively. As a result, it hasn’t seen as much widespread use as UTP.
STP is typically reserved for use in areas with extreme levels of electromagnetic interference. For instance, in wiring that has to pass over or near any sort of power generator or heavy machinery.
As was said before, Ethernet is a family of specifications that governs a few different things. One of those things are all the different wiring specifications: 10BASE-T, 100BASE-TX, 1000BASE-T, and so on.
Ethernet also describes how to send bits (1s and 0s) across each wire, as well as how to interpret those bits into meaningful frames. For example, Ethernet states that the first 56 bits of every frame must be alternating 1’s and 0’s (known as the “Preamble”). The next 8 bits must be “10101011” (known as the Start of Frame delimiter). The next 48 bits are the Destination MAC address. The next 48 bits are the Source MAC address; and so on, until the entire frame has been transferred.
Below, we’ll describe some of the wiring standards specified by the Ethernet standard.
### BASE T* Terminology
This set of terms all refer to how the wires are used inside the cable. For instance, which ones are transmitting, which ones are receiving, how they transmit signals, and at what voltages?
There are three parts to this term, so let’s discuss them each individually first before we look at any specific standard:
The number at the beginning simply refers to the speed of the wire in Millions of bits per second, or more often referred to as Megabits per second (Mbps). A wire rated at 100 Mbps can theoretically transmit 100,000,000 bits per second, which equates to roughly 12.5 MegaBytes per second (MBps). Notice the capital B vs the lower case b to refer to Bytes vs bits.
An Ethernet cable rated at this speed is sometimes also referred to as Fast Ethernet. This is in contrast to a regular Ethernet cable which is rated at 10 Mbps, or Gigabit Ethernet which is rated at 1000 Mbps.
The term base is short for baseband signaling. Its counterpart is broadband signaling. When these terms originated, the difference between them was baseband signaling sends digital signals across the medium, whereas broadband sends analog signals across the medium.
The difference between a digital signal and an analog signal is the number of possible interpretations of each signal.
An analog signal can represent a theoretical infinite amount of values. For instance, a certain voltage on a wire might represent a green pixel, and another voltage might represent a red pixel, and so on and so forth until every pixel in an image is transmitted across the wire.
A digital signal can represent a finite amount of values – typically just two: 1 or 0. If the same image from above were being sent across a digital wire, a stream of 1’s and 0’s would be transmitted. The receiving end would be able to interpret the binary values as a series of numbers, perhaps based upon the RGB color codes, to represent each colored pixel.
The main difference being, at any given time on an analog wire, a plethora of signals (and therefore values) can be read. Whereas on a digital wire, at any given time the signal can either only represent a value of 1 or a 0, and nothing else.
This allowed digital transmission to be more error resistant as the wire’s entire voltage range at any given time is only divided into two possible values (1 or 0). Whereas an analog signal is more prone to transmission errors because any slight distortion will change what the other end interprets entirely.
For example, consider a case where some sort of interference or degradation of service caused the voltage received on a wire to be slightly different than what was initially sent.
In an analog world, since each voltage value could represent any one of millions of colors for an individual pixel (for example), a modification of the received voltage would create a distorted image.
Where as with Digital transmissions, the entire possible voltage range on the wire is divided to transmit only two values: 1 or 0 — which means it would take significantly more interference to modify the voltage transmitted enough to turn a 1 into a 0, or a 0 into a 1. The slight difference in received voltage would likely still be in the range of voltage values which communicate what was initially sent.
This image illustrates the effect very plainly. Notice as the signal quality degrades, the digital transmission can still interpret a 1 or a 0, and therefore still display the image without any visible distortion. Whereas with analog, as the signal degrades, a slight degradation in the signal causes the receiver to interpret the wrong colors for given pixels, causing the image distortion (the image is from a blog post by Antenna Direct in Australia.
The “–T” stands for Twisted Pair. This is in contrast to other wiring standards like -2 and -5 which indicate Coaxial wiring with maximum ranges of 200~ and 500 meters, or -SR and -LR which are Short Range and Long Range Fiber Optic wiring standards.
With each individual part defined, we can now look at the two prominent specifications for Fast Ethernet (we will look at two specifications for Gigabit Ethernet later on in this article):
100BASE-T4 uses all four pairs in the bundle (all eight wires). One pair is used solely for Transmitting signals (TX). One pair is used solely for Receiving signals (RX). The remaining two pairs can be used for either RX or TX, and it’s up to both sides of the wire to negotiate which of the remaining pairs are used for what.
T4 is one of the earlier specifications for Twisted Pair, and doesn’t see much modern use due to unnecessary complexity in the design for very little gain over the 100BASE-TX iteration described next.
100BASE-TX uses only two pairs, one dedicated to TX, and the other dedicated to RX. The other two pairs on the wire are unused. You could very well construct a 100BASE-TX wire which only had 4 of the 8 wires in the correct pin-positions (1,2,3,6), but often the other four wires are still included mostly as place holders for the remaining pin-positions, as well as for future compatibility.
100BASE-TX (with all eight wires) is the commonly used Fast Ethernet specification today. However, it is often (lazily) referred to as simply T. Again, T is meant to refer to the category of Twisted Pair options, and TX is the specific standard that calls for using the pairs at pin-positions 1&2 and 3&6.
The point of defining each term above, independent from the others, is to give each reader a practical and technical understanding of what each term means. In practice, despite knowing the true meaning of the terms, it is often far easier to simply use the common term, even if it might be slightly incorrect — a little inaccuracy can sometimes save a lengthy explanation.
There are many guides on the internet that describe when you need to use a Crossover wire verses a Straight-through wire. But very few sources really explain why it matters, or exactly how it works. In this section, we will explore these concepts with more depth.
The 100BASE-TX and 10BASE-T specifications both call for 8 wires in a twisted pair cable to be grouped into four pairs.
Of the four pairs, only two will actually be used: pair 2 and pair 3. Each individual wire in the pair is a simplex medium, which means the signal can only ever cross any one wire in one direction.
In order to attain full-duplex communication, some wires are permanently set aside for communication in one direction, and the other wires are permanently set aside for communication in the opposite direction.
The configuration of the Network Interface Card (NIC) will determine which pair is used to transmit and which pair is used to receive.
A NIC that transmits (TX) signals over pair 2 (pin 1&2) and receives (RX) signals over pair 3 (pin 3&6) is called a Media Dependent Interface (MDI) NIC. While a NIC that does the opposite (TX on pair 3, and RX on pair 2) is called a Media Dependent Interface Crossover (MDI-X).
PC to PC
A PC uses an MDI NIC, which means PCs always transmit on pair 2, and receive on pair 3. But if two PCs connected directly to each other are both trying to transmit over pair 2, it would lead to a collision of their signals. And worse, neither PC would receive anything on pair 3.
As a result, the pin-pairs need to be crossed on the wire, so that what is sent from one PC on pair 2, arrives on the other PC on pair 3, and vice versa.
Here is a simplified illustration (the colors below are irrelevant, they simply indicate two different paths, for two different directions of the communication):
Notice both PCs can transmit signals through a dedicate channel, and due to the cross of the pairs in the wire (represented by the giant X), both PCs can receive what the other transmitted from a dedicated channel.
Hence, a connection from a PC directly to another PC requires a crossover cable.
PC to Switch to PC
A switch is a device that is meant to facilitate communication between two PCs on the same network. To that end, a switch NIC uses the MDI-X specification, which means a switch always transmits on pair 3, and receives on pair 2 (the exact inverse of an MDI NIC on a PC).
This causes the switch to have a built-in crossover function. The wire doesn’t need to cross the pairs, because the switch will take care of it:
As you can see, a PC connected to a switch can simply use a straight-through cable, and let the Switch deal with crossing the pairs. The end to end path remains consistent: every device is transmitting on its TX ports, and receiving on its RX ports.
PC to Switch to Switch to PC
We discussed earlier that two PCs connected directly to each other require a cross in the wire since they both use the same wire pairs for TX and RX. Similarly, two Switches connected to each other also use identical wire pairs for RX and TX.
As a result, we have to account for this by introducing yet another crossover between the switches:
From the diagram above, we see that a switch connected to another switch requires a crossover cable.
In this way, the end to end path remains consistent. The PCs are both transmitting and receiving on the expected wire pairs. And each direction and step along the path always goes from a TX pair to an RX pair.
Routers and Hubs
But what of routers and hubs? What type of NIC do they use?
It turns out, a Router, like a PC, uses the MDI specification – TX on pair 2, and RX on pair 3. As such, you can replace any picture of a PC in any of the illustrations above with a Router, and can easily determine which connections would require a straight-through cable and which would require a crossover cable.
Furthermore, a Hub’s ports use the MDI-X specification – TX on pair 3, and RX on pair 2. You can replace any picture of a switch above with a Hub and can also easily determine what cables are required.
Ethernet Cable Wiring Diagram
Recall that there are two standards for the colors in the RJ45 specification: T568a and T568b. The standard being utilized on either side of a Twisted Pair wire is what determines whether the cable is straight-through or crossover.
To make a Straight-through cable, simply order the wires on both sides of the cable to one specification (either both T568a or both T568b):
To make a Crossover cable, simply use one standard on one side, and the other standard on the opposite side:
Note that wire pair 1 and pair 4 are not used (the blue and brown wires). You could, theoretically not include the wires in the cable at all, but this would make keeping the remaining wires in the proper order rather difficult.
Moreover, since they are not used, they do not need to be crossed in a crossover cable. However, the Gigabit specification does require using all 8 wires, and often all pairs are crossed for consistency. We will discuss Gigabit Ethernet later in this article.
And lastly, remember that the signal doesn’t really care what color the wire is. As long as the correct pins are connected to each other, communication will work. You could use all green wires, and as long as Pins 1&2 are connected to Pins 3&6 on the other side (and vice versa), you would have a fully functioning cross-over wire. But just because it works, doesn’t mean it is a good idea – such a cable would be a nightmare to maintain.
Easy Memorization Chart
We can aggregate everything we learned above regarding crossover wires and straight-through wires into a simple chart:
A benefit to how the graphic above is displayed is that it makes it very easy to sketch out. Simply draw L2/L1 on the left and right, and L3+ on top and bottom and connect everything to each other. The lines that cross each other require a crossover cable when connecting devices that operate at those layers of the OSI model. The lines that connect straight to each other require a straight-through cable.
An L1 or L2 device connected to another L1 or L2 device requires a crossover cable.
An L1 or L2 device connected to a L3+ device requires a straight-through cable.
An L3+ device connected to another L3+ device requires a crossover cable.
Or even simpler:
Like devices require a crossover cable.
Unlike devices require a straight-through cable.
Despite the simplicity of knowing when to use a straight-through cable verses a crossover cable (after it has been properly explained, of course), the fact that a choice exists at all has caused all sorts of downtime and headaches for network engineers across the industry.
As a result, a feature was created which allows the two devices to dynamically determine and switch their TX and RX wire pairs if necessary. This feature is known as Automatic MDI-X, or Auto MDI-X.
Auto MDI-X allows the use of a straight-through cable for every connection, and lets the two endpoints dynamically determine whether they need to inverse their TX and RX pairs.
Auto MDI-X is an optional feature for 100BASE-T implementation, and a required feature for all Gigabit Ethernet devices.
How does Auto MDI-X Work?
But how does Auto MDI-X work? How do the two parties determine which pairs of wires should be used for TX and which pairs should be used for RX? Which of the two parties should switch the TX and RX pairs if it is determined to be necessary? We will look at the inner workings of Auto MDI-X in this section.
Remember, the goal of the Crossover cable is to ensure one party’s TX pins are connected to the other party’s RX pins. For successful communication down a cable, a TX wire cannot be connected to another TX wire. Essentially, one NIC must use the MDI specification, and the opposing NIC must use the MDI-X specification. Here is how Auto MDI-X accomplishes this:
Both parties start by generating a random number in the range of 1-2047. If the random number is odd, that party configures their NIC to the MDI-X standard. If the random number is even, that party configures their NIC to the MDI standard. Both parties then start sending link pulses through their elected TX wire pairs.
If both parties are successfully receiving the other’s link pulses on their RX wires, then both sides do nothing further, as they are successfully transmitting on their TX wire pairs, and receiving on their RX wire pairs.
If both parties are not receiving the other’s link pulses, then they must have both picked an odd number or both picked an even number. Therefore, one of the parties must switch their TX and RX wire pairs to the other specification (MDI vs MDI-X).
But the parties can’t both switch to the opposite specification, because then their TX and RX wires would still not be offset. Instead, a system was devised that randomly switches the pairs at random intervals until they correctly match up.
That randomly generated number from earlier (1-2047) gets cycled forward in order for the parties to select a new specification (MDI vs MDI-x). But that number cannot simply be increased by one, because then both parties would go from odd to even, or from even to odd. In other words, if both parties had elected MDI originally, they would then both switch to MDI-X, which would still cause a TX wire pair to be connected to a TX wire pair.
Instead, that number is cycled forward through what is known as a Linear-Feedback Shift Register.
A Linear-Feedback Shift Register (LFSR) is an algorithm that cycles through every number combination in a certain range without ever repeating a number until every number has been reached. The numbers are cycled through in a predictable, but random order (aka not sequentially but in a consistent order).
For example, if the two parties picked a starting value of 1000 and 2000, whether their next number in the LFSR sequence would be odd or even would be completely random. However, if both parties randomly picked the same starting value, they would each have identical sequences through the LFSR.
This cycle happens every 62 milliseconds, with a random variance of +/- 2ms. If one of the parties switches their wire pair at 60ms, and the other party was planning to switch at 64ms, there would be 4ms where the TX and RX pairs are perfectly aligned, which stops further cycling and completes the AutoMDI-X process.
This process continues as many times as is necessary until the two peers have lined up their TX and RX wire pairs.
But this begs the question, what are the odds of both pairs picking the exact same number and the exact same intervals each time they cycle their number. We can determine this with a little math.
The odds of picking the same starting value are 1 in 2047. The odds of picking the same interval variance is 1 in 4. Which means the odds of both parties both switching their MDI/MDI-X specification at the exact same time twice in a row is 1 in 8188.
The cycle happens every ~62 ms, which means in a full second there are 16 possible intervals. The odds for the two parties to have the exact same cycle timing for the entire second are 1 in 4,294,967,296 (4.2 billion). The odds of that happening combined with both parties starting with the exact same random number are 1 in 8,791,798,054,912 (8.7 trillion). Pretty good odds, considering at worst this will only cost you an extra second of waiting for the link to come up.
Why Twisted Pair?
It is often simply accepted as fact that most networks use Twisted Pair wiring for their physical connections. But why? What about Twisted Pair has made it the predominant cabling method in computer networks?
There are two main reasons, and both have to do with Electromagnetic Interference (EMI): The first reason is that using a pair of wires greatly reduces the outbound EMI emission. The second reason is that twisting them around each other greatly reduces inbound, or induced, EMI.
Both of these are very desirable traits when the wire is often closely bundled with other wires over long distances (think data centers or wiring closets).
Reducing EMI Emission
It is a fact of life that any signal or electrical current running through a wire emits some degree of EMI that can affect neighboring wires – also known as Crosstalk. This EMI emission can be compensated for with additional shielding, but Alexander Graham Bell devised a clever method to negate the effects of Crosstalk.
His strategy was to use two separate wires — one of them sending the original signal, and the other one sending the exact inverse of the signal. This causes both wires to emit the exact inverse EMI from each other, thereby negating their effect.
To put it in simpler terms, if one wire transmits +10v of electrical voltage and leaks +0.01v of EMI, then the other wire will transmit -10v of electrical voltage and consequentially leak -0.01v of EMI. Their combined emissions cancel each other out.
This is referred to in the electrical engineering world as a Balanced Pair, and is represented in twisted-pair wiring with the TX+ and TX- wire.
This allows you to use wiring schemes that don’t require heavy investments in shielding, and is half the reason for the prolific use of Unshielded Twisted Pair (UTP) cabling in the networking world. However, so far we’ve only answered why we use a pair of wires, we will look into why they are twisted next.
Negating Absorbed EMI
Despite strategies like the Balanced Pair described above, there is no getting away from all sources of Electromagnetic Interference (EMI). Stray radio frequencies, wireless internet, Bluetooth, spy satellites, and cell phones all contribute to stray EMI.
But Alexander Graham Bell came through for us again, and devised a brilliantly simple, yet effective, method to nullify ambient EMI.
The basic concept takes advantage of EMI being stronger the closer in proximity you are to the source. If the two wires take turns being closest to the EMI source, they will each absorb an equal amount of interference. Take a look at this simplified diagram:
The blue wire starts with +50v, and the green wire starts with the exact inverse, -50v. The source of the EMI is the red circle, and each wave that surrounds the EMI source impacts the wires progressively less and less. If you only add the EMI at each grey dot (the top and bottom of each twist), both wires end up receiving +22v of interference.
Even though the final voltage received on the right side of the wire is different, notice that the difference in voltage is consistent throughout the twisted pair of wires: it is always 100v apart. The EMI affected both wires identically. You could easily calculate the difference of the final values (100v), and display it on a number line to determine the starting voltages were +50v and -50v:
If you recall, data is sent across a cable in a digital signal, which is to say, as a stream of 1’s and 0’s. But how exactly is a Twisted Pair wire used to send actual data across the wire? We will use a bit of an over simplification to describe the basic premise.
Sending a signal down the wire is nothing more than applying voltage to the wire for a certain amount of time. The two parties will agree on a clock rate, also known as frequency, which determines how long each ‘instance’ of voltage must be applied. For the purpose of this simplified example, we will refer to this as the position. At any given time, each position can only mean either a 1 or a 0 being sent down the wire.
Different standards call for different voltage levels, and for the purpose of this simplistic description the true voltage doesn’t actually matter. But we will proceed to describe it using 100BASE-TX which prescribes a voltage range of +2.5v to -2.5v.
To send a 1 in a given position, the transmitter will send +2.5v down the TX+ wire. To send a 0, the transmitter will send -2.5v down the TX+ wire.
The TX- wire will always do the exact inverse: -2.5v to send a 1, and +2.5v to send a 0.
This is what it would look like to send a binary string of 110010101110:
Note that the graph above does not depict the physical layout of the wire (aka, this isn’t the twisting of the wire pairs). It just represents the alternating +2.5 and -2.5 volts being sent down the TX+ and the TX- wires. The twists in the twisted pair are (or should be) uniform across the length of the wire. As we pointed out before, you can see that the wires are always sending the exact inverse voltage of each other, and everything is neat and horizontally symmetrical.
Along the wire, noise is introduced from various EMI sources. We’ll apply a different amount of noise at each position of our bit stream and take a look at what is received on the other end:
Notice the graph is no longer as neat and symmetrical. The wires are still sending the inverse of each other, but offset by a constant value. Our nice and neat values of +2.5v and -2.5v are gone.
BUT, the receiver isn’t looking for exactly +2.5v or -2.5v. Instead, it is simply looking for which wire sent the higher voltage. If the TX+ wire sent the hire voltage, then the signal for that position must have been a 1, and if the TX- wire sent the higher voltage, then the signal for that position must have been a 0.
Or, to put it simply, on the graph above, if the blue line is on top, the transmitted bit at that position is a 1. And if orange line is on top, then the transmitted bit is a 0.
Notice also that even though the values were affected by EMI, they were both affected identically – they both went up or both went down by the same amount. At any time on the receiving graph, the value of the TX+ wire and the TX- wire are always 5v apart, just like they were in the sending graph. As we discussed earlier, this is due to the physical twisting of the TX+ and TX- wires.
In this way, the receiving end can piece together the signal, one bit at a time, despite whatever EMI might have affected what was originally sent. As you can see, UTP is not immune to noise, but it has functionality to negate the effect of noise.
We’ve discussed Fast Ethernet (100 Mbps) in great detail. Now we move on to discussing Gigabit Ethernet (1000 Mbps, or 1 Gbps).
The first major difference is the gigabit standards require the use of all four pairs (all eight wires), unlike Fast Ethernet which only utilizes two pairs of wires. As a result, in Gigabit Ethernet, all four pairs must be crossed when building a Crossover cable.
If you recall, there are two wiring specifications proposed by the RJ45 standard: T-568a and T-568b. Below are images which depict what each of them look like when all four pairs are crossed:
That said, Gigabit Ethernet requires Auto MDI-X. As a result, you are safe to simply use straight-through cables everywhere and let the NICs determine whether they need to simulate a crossing of the wire pairs.
There are two wiring specifications within the Gigabit Ethernet standard:
This standard of Gigabit Ethernet uses all four pairs, but it dedicates two of the pairs for TX, and the other two pairs for RX.
Conceptually, this is a simpler process than how 1000BASE-T operates, but regrettably it requires upgrading all the twisted pair cables that have already been run from the common Category 5 or 5e to the more expensive Category 6. As a result, 1000BASE-TX has not seen much adoption in the industry.
This is the predominant Gigabit Ethernet standard. It uses all four pairs at the same time, in full duplex mode – each of the four pairs can be used for both RX and TX, at the same time. This is done with a process called Echo Cancellation, and we’ll explore that in more depth in the next section.
The primary benefit to this wire standard is you can achieve gigabit transfer on the much more prevalent Category 5e cables without being forced to upgrade all your twisted pair cables tot he more expensive Category 6.
Full Duplex on a Single Wire Pair
We learned in the last section that 1000BASE-T can send and receive signals on the same wire pair at the same time. We will discuss how this is possible in this section. First, we’ll start with an analogy to explain the premise.
Have you ever talked to someone on the phone and could tell that they put you on speakerphone because you could hear your own voice echoed back? This is an outcome of your voice being played on their speakerphone, bouncing around the room they are in, and being picked back up by their own phone’s microphone. This is known as an echo.
High end speakerphones can negate this effect by extracting the sound waves of what the speaker emitted from the sound waves of what the microphone is picking up — this process is known as Echo Cancellation.
Echo cancellation is also the basic concept which allows a Gigabit Ethernet wire to both send and receive data on the same wire pair at the same time. The basic premise is if you know what you sent, you can extract it from what you received.
Recall that sending a signal is nothing more than applying voltage to a wire. Conversely, receiving a signal is nothing more than reading the voltage observed on a wire.
If a sender applies voltage to a single wire in the following pattern:
+0.5v , +1v , -2v , -1v
And at the same time that same sender reads the voltage and observes the following pattern:
+1.5v , 0v , -2.5v , +1v
The sender can subtract the voltage it initially sent from the voltage it just received to determine what voltage the other end must have applied:
+1v , -1v , -0.5v , +2v
In this way, the same wire can be used to both send and receive signals (data) at the exact same time.
Again, these values are merely examples in order to explain the basic concept. In reality, the voltage levels are very different, and also account for induced EMI and electrical echoes along the copper wire itself. In addition, we are only showing Echo Cancellation from the perspective of a single wire in a twisted pair – the opposite wire would still be sending the exact inverse voltage, as discussed earlier.
Using this strategy, all four wire pairs can be used for both TX and RX at the same time. The wire pairs are still Twisted Pairs, and therefore still use the same strategies to negate the inbound and outbound EMI discussed earlier.
If you’ve made it this far, then you now know just how much there is to Ethernet and Twisted Pair wires. It was a little humbling to learn about it over the years and publish this article. So much technology goes into each wire, yet I have thrown away countless cables without a second thought.
Ethernet wiring is definitely full of technology that we easily take for granted. And to think, even this article left out significant detail in order to remain (relatively) simple.