Ethernet cabling

Various names, but often named wrongly

The proper name for the plastic 8-pin plug that we use for most ethernet connections is an 8P8C modular connector.  “8P8C” refers to a connector with 8 positions, and 8 connections (all eight pins are present and connected).  There are various different sizes of similar connectors, with different numbers of pins.  And some connectors might have places where pins could be fitted, but may only have metal pins fitted into some of them.

Ethernet cables typically use 8P8C connectors with all eight pins being used.  But some cheaper 10/100 megabit cables only bother to wire up four of the pins.

Phones might have a 6P6C plug (with all six pins fitted, using RJ12, RJ18, or RJ25 wiring schemes), a 6P4C (the same plastic plug, but with only the middle four pins fitted, with RJ13 or RJ14 wiring) or a 6P2C connector (the same plug with only the middle two pin positions fitted, using RJ11 wiring).  Phones can also use a 8P8C connector with RJ61, T568A, & T568B wiring schemes.  There's a variety of Registered Jacks, and they refer to how various different connectors are wired up.  And there are many other wiring schemes that use these plugs, that aren't official specifications.

It is not an RJ45 connector.  RJ45 is a wiring standard (the method for how one these connectors might be wired up—originally for telephone use) specifying which pins are used for which purpose (and really refers to the wiring, not the plug), and originally used a slightly different 8-pin plug and socket (the plug had a plastic peg sticking out the side, and wouldn't fit into an ethernet socket).

It's a bit like calling a ¼″ jack a headphone plug.  It's not, it's a ¼″ jack.  It can be used for headphones, microphones, line signals, etc, and can be wired up in various different ways.  It's only a “headphone plug” when it's wired to a pair of headphones, you call it a “microphone plug” when its on a microphone, etc.  They're also, often called “phone jacks” because they're used in headphones and microphones, but the kind of plug used with old telephone exchanges and professional audio equipment patchleads is mechanically different.

And these same 8-pin connectors can, also, be wired in different manners for other purposes (keyboards, multi-channel analogue audio, serial ports, etc) that are incompatible with ethernet wiring.  So an 8 pin modular plug cable isn't necessarily always going to be an ethernet cable.  So pay attention to the fine details when buying connectors and cables.

T568A and T568B are wiring configurations, again with their origins in telephony.  They specify which coloured wires are used on which pins.  Although it's essential that the twisted-pairs are wired up correctly to the right pins in the connectors, electrically speaking it doesn't really matter which coloured wires are used where.  However, using a standard makes it easier to consistently wire everything the right way, the same way at each end, and for terminating cables when you can't see both ends at the same time (e.g. between different rooms and buildings).

So, essentially, it's an 8P8C modular connector, RJ45 is which pins the twisted-pairs of wires connect to, and T568A & T568B are which colour wires to use for each pair.  And the cables are generally known as an ethernet cable, but even that's over-generalising (there are other kinds of ethernet cable, such as coax and optical-fibre ethernet, and there other types of connectors).

Does it matter if you call it the wrong thing?  Nine times out of ten, it doesn't.  Most people make the same mistake (calling them RJ45 plugs), and are unaware of it.  All they deal with are ethernet connections, and that's the only stock they carry.  But if you were ordering parts from a distributor that carried many different types of modular plugs, you could end up buying proper RJ45 connectors, which won't be of any use for ethernet.  And on that note of ordering the right parts, some plugs are only suitable for solid core wires, others only stranded wires, and yet others say they're universal, so it's important to order the right kind for your wiring (the design of the blades that spike through the insulation are different for each), and there are different plugs designed to fit onto round or flat cabling.

Pinouts

Holding a cable (as per the below photo), with the cable coming out the left, the connector pins on the right, and the plastic latch underneath, the pins are numbered 1 to 8 from the top down to the bottom.

Since the first two pins have the orange pair on it, you can immediately see that it's wired in the T568B specification.  If it were wired to T568A, they'd have the green pair on them, instead.  That is, assuming it's wired correctly.  Some people wire cables up with any coloured wiring on any pair.  It'll still work fine, but is confusing if you ever have to re-terminate a cable.  See the table, below, for the full wiring list for the specifications for normal ethernet wiring.

There are two standard wiring schemes (T568A and T568A) for ethernet using these connectors, both are completely interchangeable.  A T568A-wired cable has the pins wired to each other at each end of the cable in exactly the same way that a T568B cable has (pin 1 connects to pin 1, pin 2 connects to pin 2, etc), the only difference is which coloured wires are used on four of the pins.  You can patch a T568A-wired cable into T568B-wired wall socket, and vice versa.  You can wire up an entire building as T568A or T568B without it making any difference.  You could have mixed wiring standards within a building, and so long as both ends of each cable are wired the same as each other it'll be fine, but the potential for getting that wrong is why you shouldn't use two conflicting standards at the same location

I started out using the scheme that matched existing cabling in my location.  But as time went on, and more pre-terminated cables were added, it became a mixture of both schemes, as there was no consistency to the pre-made cables, and I didn't want to be dismembering perfectly fine cables to rewire them all to the same colour scheme.  There doesn't need to be, T568A or T568A make identical connections as far as the electronics in the networking hardware is concerned.  Each pair of wires is just as capable as all of the other pairs at carrying data, or power.

It's when you're fitting connectors onto a cable that you have to ensure that both ends of the same cable are wired to its connectors using the same scheme (unless you want a cross-over cable).  That applies to patch cables, and to socket wiring.  A normal cable should either have both ends of it wired to T568A, or both ends wired to T568B.  To avoid wiring mistakes, it's best to adopt one scheme in an installation, and never use the other.  In some countries it has become conventional to use one of those schemes in preference to the other, but considering that ready-made cables are often sourced from various different countries, you can't really expect everything to be the same.

The wires are used in pairs.  e.g. Pin 1 & 2 form a pair, and they're both used together to carry one signal.  The connector can have up to 4 pairs fitted to it.  The pair numbering scheme, and the odd way that pair two straddles pair one instead of sitting beside it, harks back to the days when they were used with telephones (the third and fourth pair don't follow the same pattern of straddling the middle pair, further and further apart, because it was realised they'd suffer too much signal interference if they did that).  The first phone line would use the first pair, if there was a second line it'd be on the second pair, etc.  But since most phones were only a single line, some of the other pins were often used for non-standard purposes (such as supplying power to a handset).  For reasons like that (very different signal types, and potentially destructive voltages and currents) it's essential to not mix up phone and ethernet connections (using 8P8C connections for ethernet and 6P6C connectors for phone helps in that regard—you can see if you're plugging the wrong plug into the wrong socket, although that doesn't stop people forcing the wrong ones together).

On that note of pair two straddling pair one, it is a shame that they didn't just have each pair wired directly beside each other (a pair on pins 1 & 2, another on 3 & 4, another on 5 & 6, and the last on 7 & 8), it would have made fitting the plugs on the cables much easier to do, and noise rejection better (the pair straddling the centre two pins has worse performance).

Each pair of wires are twisted together (it's why they're called a “twisted-pair”), with one wire coloured and the other wire white (or, if you're lucky, the white wire has a coloured stripe on it so you don't lose track of which wires are paired together when you untwist them to wire them up).  On that note, leave pairs twisted together as close to the point where you have no choice but to separate them to fit into the connector.  The twist is an important part of their electrical characteristics (mainly to do with noise rejection, minimising incoming and outgoing interference).  Each separate pair just sit alongside each other within the cable sheath, the pairs are not wound around each other.

10 & 100 megabit per second ethernet only used two of the pairs (pairs 2 and 3, the green and orange ones), gigabit ethernet uses all four pairs.  I can only guess that the early 10/100 mb/s ethernet skipped pair 1 in an attempt to avoid accidents with telephone wiring (which normally only used pair 1, for a single phone line).  That was the thought behind digital telephone systems using the same 8-pin jacks as analogue telephones, with analogue wired up to use the middle two pins, and digital not using them.

Normal equipment connections (e.g. computers to routers, printers to routers, etc) have the inputs and outputs on different pins in the equipment, and straight-through cables connected them together correctly:  Outputs go into the inputs (or to use networking signal parlance, transmitters connect to receivers).  Unusual connections (e.g. computer to computer, directly) used to need a cross-over cable, so that one computer's output pins connected to the other's input pins, and vice versa.  But that requirement was overcome with some later equipment designs.

Normal cables are straight-through connections, pin 1 at one end connects to pin 1 at the other end, likewise for the rest of the pins (pin 2 to pin 2, pin 3 to pin 3, etc).

But with cross-over cables, pairs 2 and 3 are swapped over at one end, but pairs 1 and 4 are not modified.  In essence the cable is wired T568A at one end, and T568B at the other.  It doesn't matter which end of the cable is used at which end of the equipment.  Most modern equipment doesn't require cross-over cables, they'll detect which pins have signals on them, and internally switch the input and output functions around to suit.

Cross-over cables were used with 10 & 100 megabit ethernet (where needed).  Later on, 100 megabit equipment could support automatically dealing with the signal direction by switching inputs and outputs internally (making cross-over cables redundant).  Gigabit ethernet equipment was an even later invention, and all gigabit ethernet equipment is supposed to use automatic handling of data direction, and not supposed to need or use cross-over cables.

NB:  Although the specifications say a feature is supposed to be supported that doesn't guarantee that a manufacturer actually will do so.  You can mostly expect gigabit ethernet to support it, you may still have to use cross-over cables with 100 megabit ethernet.

Cross-over cables should be labelled so they don't accidentally get re-used with something else that needs a normal cable.  And they're rarely needed any more, anyway.

Since the move away from dial-up networking, almost all networking is done via ethernet or wireless, and that often entails more than two devices being networked together.  A single PC-to-PC network is rarely useful, any more, and just about all networks have a switch or router that acts as a hub between everything.  The cabling between all of them is made with straight-through cables.  About the only thing that may still need a cross-over cable is when linking switches or routers together that don't handle automatic direction sensing, and that's a rarity, too (most do handle automatic sensing).

Be aware that some manufacturers supply cheap low-quality pre-made leads, and you may be better off discarding them and buying decent quality leads.  Some are so cheap that they only bother to wire up pairs 2 and 3, since that's the only pins that their equipment uses.  Those two-pair leads will only be suitable for 10 & 00 megabit connections.  And I've come across leads that didn't even twist the pairs, they're really only suitable for analogue telephone signal patching.

Power over Ethernet (PoE)

PoE allows a single ordinary ethernet cable to supply power and data to devices.  It's used for things like ceiling mounted access points, security cameras, digital telephones, remote sensors, etc.  It allows the simplicity of only having to run one cable, not having to install a power source for remote devices, and avoids potential problems caused by connecting signal lines between things running from different mains power sources at each end.

Some PoE (power-over-ethernet) equipment uses pairs 2 & 3 (the orange and green pairs) to carry power, and since someone might use a cross-over cable they should design the equipment to handle reversed power polarity.  However, since manufacturers often don't follow best practices and standards, make sure that you don't use a cross-over cable unless it's specified that you must by the actual equipment in use.  Cross-over cables are for special purpose usage, they should be clearly labelled so they're not accidentally connected to the wrong equipment.  Normally, all ethernet cables are straight-throughs.

Normal straight-through cables have each pin wired like-for-like (pin 1 goes to pin 1, pin 2 goes to pin 2, etc), and it shouldn't matter whether the cable is wired using T568A or T568B, as the same pins are connected together, in the same manner, no matter what colour wires are used.  Each pair of wires is just as capable as the any of the other pairs for carrying data, or power.  You should be able to use any four-pair straight-through cabling completely interchangeably.  i.e. Any normal ethernet cable.  (I consider the cheapskate 2-pair/4-wire cables to be abnormal.)

The wiring-colour standards T568A and T568B (what colour wires are used where, as per the table above), and PoE power modes A and B (which pins are used for carrying power, as per the table below), are two completely independent things.

In general, you don't get to choose which power mode the equipment uses.  Which modes the switch or router can supply, and what mode the devices use or need, is determined by the way that they're built, though it's possible that some managed routers or switches might allow you to selectively enable certain modes.  But, otherwise, you need to purchase your equipment with compatibility in mind.

The Mode A wiring schemes (see the fourth and fifth columns in the table above) uses the two pairs that will be swapped over in a cross-over cable for the power wiring, likewise with PoE++ or 4PoE (see the last column).  If a cross-over cable is used, the PoE voltage polarity will be reversed.  PoE equipment is supposed to be able to handle this (work either way, and not get damaged), but manufacturers are notorious for not properly supporting standards.

The 10/100 Mode A wiring scheme (in the fourth column) is the only one that can work with those ethernet cables which only wired-up pairs 2 & 3.  All the other PoE wiring schemes require all four pairs to be wired up.  The practice of only wiring up two of the pairs, and leaving the others disconnected is extremely outdated and shouldn't be done any more.  There's a strong chance that, later on, you may try to use the lead with something else that needs all the pairs wired up.

PoE+ (a 2018 standard) violates older standards of NOT having power on all four pairs.  This could be a destructive problem if connecting older PoE equipment to a new PoE-supplying router/switcher.  Though the devices should have been built to handle there being power on the wrong pins, even if they don't use it.

Some manufacturers (e.g. Swann) who supply cheaply made ethernet cables that only wire-up two of the pairs will often use very thin wiring that's really not up to the task of supplying PoE, nor tough enough for being dragged through the infrastructure of a building.  I had continual equipment disconnects with their supplied cables that went away when I used decent cables.  It's best to discard cheap pre-supplied cables and use better quality cabling, and one which has all four pairs wired up.  It future-proofs an installation, you won't need to rewire anything if you upgrade equipment.

Wiring

Ethernet cables are made of twisted pairs.  That's a pair of wires that are twisted around each other, bundled together with other twisted pairs of wires.  And each has a different number of twists per foot.  This is done to minimise signal interference between each other.

The pairs should not be untwisted.  When wiring up connectors, keep the wires twisted together all the way up to the wiring terminals, only straightening enough of them to be able to fit the wires into the terminals.

Untwisted wires (straight parallel wires) are telephone cable not ethernet, they're not suitable for ethernet use.  And not particularly good for telephones, either.

The same connectors can be used for phone services as well as internet.  Quite apart from it being essential to not mix their wiring up between the two of them, some of the wall-panel connectors meant for phone usage are not suitable for ethernet usage (primarily because they split the pairs of wires apart far too much, may use screw terminals, and some don't mate well with the plugs).

If using ethernet cabling while wiring up network and phone services, it's a very good idea to use a different jack for the phone (e.g. 6P6C, 6P4C or 6P2C).  A phone plugged into an ethernet network will do nothing but, perhaps, mangle the pins in the socket.  A computer plugged into a phone service is likely to get its network circuitry damaged by the voltages from the phone service.

Wall sockets should be mounted so the connector doesn't fill up with muck, contaminate the pins, and make the connector unreliable.  It shouldn't be mounted so the opening is facing upwards.  And sideways facing openings should have the metal pins at the top, so dust can't easily land on them (this leaves the plastic latch at the bottom, when a plug is fitted).

A proper crimping tool should be used to fit plugs onto cables, it's unlikely you'll be able to make a reliable connection without one, and it's highly likely that you'll damage the metal pins with the wrong tool.  You should also keep your fingers off the metal pins, contaminating them can make a connector unreliable (a wipe with cleaning alcohol should suffice if you have got them mucky).  But once the plug is crimped on, the pins are recessed below the plastic of the plug making them okay to handle, as long as you don't dump the ends into something dirty.

It's also difficult, though not impossible, to wire a socket onto a cable without using a punch-down tool, but it is better to use a tool designed for the job.  Basic ones only cost a couple of dollars, and are often included with some wallplate mounting hardware.

Cable testers are a valuable tool, not to be dismissed as unnecessary.  The basic ones simply test that each pin is connected to the right pin at other end of the cable, fancier ones test signal integrity down the cable, some can even tell you how far down a cable a fault is.  You can have a fault that messes up data transmission, yet all the pins pass continuity tests okay, caused by savage kinks in cables.  You may be able to unbend a kink in a cable and fix thing up, or at least salvage a long faulty lead into two shorter reliable ones.  Having a tool that tells you where to look for a fault saves a lot of hunting around when a cable is hidden in the infrastructure of a building.

Cat 3/5/5e/6 etc

Ethernet cabling comes in different categories.  Cat 3 to 6 are the common ones used for ethernet as we generally know it.  There are even higher categories for specialist purposes (e.g. within data centres).

Cat 3 being the first twisted-pair scheme for ethernet, designed in the 10 megabit/second era, and not really suitable for anything else (other than voice telephony).

Cat 5 being the next cab off the rank, the standard of cable used for 100 megabit/second.  And, later on, for 1000 megabit/second (aka 1 gigabit/second), even though it was originally thought that it wouldn't be good enough for it.

Cat 5e is an improvement on Cat 5 (better crosstalk rejection).  Often cables sold as Cat 5 are actually compliant with the Cat 5e specification.

Cat 6 is next, it can be used 10 gigabit/second ethernet, albeit with a reduced maximum cable length (55 metre or 180 feet).

Cat 6A is an improvement to Cat 6, allowing a longer cable length the same as other ethernet cables (100 metre or 330 feet).

Cat 6 and 6A have very stringent requirements on properly fitting the connectors and not being kinked or bent tightly.  It's used to carry very high frequency signals, which are easily disrupted by such things.

Best practice is to use the best quality (highest category) cable available for installations, for future proofing (if the equipment is replaced with faster networking hardware, you shouldn't need to replace the cabling).  For domestic wiring, Cat 5 cable is quite adequate (in 2023).  The fastest network devices in home equipment is still 1 gigabit/second on computer hardware, more than enough for most things, and other domestic equipment on ethernet is often only 100 megabit/second (e.g. Bluray players), as that's all they need.  Cat 6/6A may be the best practice for business installations, who may be prepared for the extra expense of higher speed networking (now or in the future).  Here, in Australia, most domestically sold cable is still Cat 5 or Cat 5e, you'll need to go to a proper supplier for better cable.

Most ethernet cabling is designed for permanent fixed cabling.  Both the rolls of cable sold for installing into the walls, and the patch cables between wall sockets and equipment.  But for cables that will be trailed across floors and moved about (such as demonstration or audio/visual equipment on trolleys), you need a more flexible cable, one that will survive being moved around, dragged about, and will lay flat on the floor instead of being a trip hazard.  But most cable will be damaged by being walked on, especially if the stepped on part was not flat on the floor.

Patch panels

Patch panels allow the wiring of an installation to be re-arranged from the front of the equipment, instead of having to delve into the back.  While this provides a lot of convenience, it does increase the expense of an installation, and puts more interconnections in the middle of a signal path.  Potentially this adds more possible points of failure, and losses of going through each connection could affect data or PoE.

For security equipment there's advantages in not going through any patch panels and having a single cable go directly between equipment (minimising any signal and power losses, reducing the chance that someone might unplug something, and making it harder to connect unauthorised equipment).  If you do use a patch panel, it really should be inside a locked enclosure.

Earth loops / ground loops

An earthloop is a wiring problem in some situations, where interconnected equipment develops a hum (1), buzz (1), or erratic behaviour (2), because a circuit has formed between its signal connections and its main power connections.  This can also happen with equipment that have multiple signal interconnections, and therefore multiple signal ground connections, without the mains earth being a part of the problem.

For most ethernet connected equipment you shouldn't encounter this kind of problem because it's conventional to use coupling transformers inside the equipment between the ethernet sockets and the rest of the device's circuitry.  Those transformers make an indirect connection (they're magnetically coupled between output and input).  Another technique is optical coupling (internal to the device).  Either method is a form of “galvanic isolation” (where there is no direct electrical connection between things).  Though they may only have enough isolation to reduce nuisance interference, and not enough to withstand electrical faults or nearby lightning strikes.

However, if you use shielded twisted-pair (STP) ethernet cabling, the shielding will directly connect the equipment together.  And some ethernet devices don't use coupling transformers, nor optical isolators, or they may implement problematic grounding techniques internally, so they may be prone to this problem, even if you don't use STP cabling.

In general home computing, you're not likely to see this issue that much.  But people who connect their HiFi systems to computers and network playback devices may encounter it (buzzes and noises in the background).  There's a couple of typical remedies for them; the traditional audio coupling-transformer between their audio gear and the digital equipment, or using WiFi instead of cabled ethernet.  So-called special audiophile switches and routers are generally a complete con job.

In the professional world (such as broadcasting), another option to reduce ground loop noise was to connect a heavy-gauge ground wire directly between all the equipment.  The idea being that it would shunt any problem currents through this wire rather than the signal cabling.  This can even benefit some rack-mounted equipment, as painted metalwork acts as an insulating barrier, and the equipment casing may not provide a good ground through the rack.  While balanced audio wiring does reduce interference problems, equipment performance can still be upset by current on the audio wiring's ground, and modern equipment rarely uses audio transformers any more.  The typical transformer-less input (or output) stage is ground-referenced, somewhere in the circuit, even if it does have a balanced input (or output).  And lifting the audio ground wire on a directly-coupled balanced input can create noise problems in itself.

Premises wiring

A much more serious problem is connecting equipment together via their ethernet wiring throughout a building, or between buildings, where there is a potential difference (PD) between the mains earth wiring, or neutral wiring.  There nearly always is a small one, as no wiring is perfect.  But at a very low PD it probably won't be noticed, a bigger PD can damage ethernet ports and equipment.  A very large PD due to mains wiring faults can cause fires and electrocutions.

Good ethernet device design in ethernet NICs, switches, routers, etc, uses some kind of isolation (a barrier), between the ethernet wires and the internal circuity (transformers, optical) as mentioned above.  But that's mainly about signal integrity, it's often not going to be robust enough to protect against extreme conditions, such as electrical wiring problems, or electromagnetic radiation from nearby lightning strikes.  Even strong winds blowing over suspended cables can create damaging static electricity charges, though being strapped to a grounded guy wire ought to mitigate that.  For safe inter-building connections the best option is usually optical fibre, since it doesn't conduct electricity.  Even though it's more expensive, it may cost less than replacing equipment ruined by electrical problems, especially if you have to replace things several times.  It's also worth noting that fibre supports even higher speeds than copper cabling does.  And although WiFi obviously gives you isolation, it's prone to hacking, interference, dropouts, and speed limitations.

Externally run cable should either be in conduit, or be cable that's meant for outdoor use.  General purpose cable doesn't last well when exposed to sunlight and weather.  Outside cable that's under shelter (such as under the eaves of a house), will fare better, but will still deteriorate over time.  It's still subjected to wide temperature ranges, moisture, and reflected sunlight.

Ethernet cable should not run with, or cross over, mains power wiring.  Quite apart from electrical noise issues, most ethernet cabling insulation is not rated to protect it from mains voltages.  There should be physical separation, either by running it through conduit, or some other physical barrier should be between them.

Cable management

Label things.  Anything more than about 5 cables becomes a right pain to sort out which cable goes where by random experimentation.  Either label things with their purpose, or simply number them.

A rat's nest of wiring all over the place is hard to maintain and fault-find.  Neatness is useful, it doesn't just look better.

When trimming cables to the right length before terminating them, leave enough length that cables can still be handled, and equipment can be moved around.

Mangling cables can damage them, or cause signal problems.  Cables should have a straight a run as possible.  Gently curve them, don't right-angle bend them.  Cables that get kinked while being hauled around may be permanently damaged.

When tying cables in place do not pinch them tightly together.  The outer sheath should not be crushed.

Looming cables together can cause problems.  It'll increase any unwanted interaction between them, and bundled cables used with PoE can potentially get too warm.

Bogus arguments

Is T568A or T568B better?

No, as far as the equipment is concerned they're electrically interchangeable.  Simply pick one and stick to it, and follow whatever wiring scheme has already been used when working on existing installations, or the specifications given to you.  Label your racks with the standard that you want to use to encourage technicians to use the same scheme for any work done in the future.

It's often said that Australia and America usually use T568A, and Britain usually uses T568B.  That reason isn't technical, it's simply more memorable to use the one with the same letter as the name of your country.  Countries with names starting with different letters will have to invent their own reminders about any particular preference.

Other advice says use T568A as the colours also correlate to a telephony wiring standard.  But both T568A & T568B actually come from telephone company standards (ethernet jumped on the bandwagon afterwards), so that's a spurious argument.  But seeing as it's an unrelated purpose (telephony), that argument is pretty poor, you shouldn't be using one cable simultaneously for two different purposes (it can be done, but mistakes can be too easily made).  Yet another point of view is that T568B is newer, and we should be using the newest standard, with no other justification than that.  Some people claim that T568B is better, but say nothing to substantiate that claim.  And I've seen another mantra spread about of using T568A for analogue phone lines and T568B for binary data lines, so you can tell them apart by looking at them.  Though I seriously doubt anyone will be looking into the back of a patch panel as they hook up data and phone lines throughout an office's multi-purpose CAT5 cable installation.  And mixed standards in an installation is just asking for trouble.

Both standards come from telephone systems, predating ethernet, soo there's very little point in arguing superiority of one over the other for the highly different circumstances of ethernet.

To be blunt, ethernet is typically 4 pairs in one cable jacket, and one cable per connection, so internal wiring colouring only serves to patch that device correctly (each end the same as each other).  All you need to be able to do is plug the right cable into the right socket.  Telephony is typically one pair per phone line, with multiple phone lines carried in a trunk cable, and split out into separate lines at each end; cable colouring is how you identify each individual phone line.  But inside an office, you're unlikely to use a single CAT5 cable to carry four phone lines across the room to four wall sockets, one for each phone.  These days you're much more likely to use one CAT5 cable per wall socket even if you're only using one of the pairs inside the cable, because it becomes a universal tie-line (today you use it for a phone line, the next year it gets repurposed as an ethernet line, and no rewiring is necessary).  Though you may find that a small office with only four phone lines could have a CAT5 cable coming in from the street, to a patch panel that breaks it out into four separate lines.  Likewise, with single-line home installations; a CAT5 cable coming from the outdoor termination panel to the indoor phone wiring, even though it's only using one of the pairs, because CAT5 cabling has become so universal (it also means there's three spare pairs, if a cable fault occurs).

You'll find people arguing that one scheme is better than the other, or insisting that PoE only works on one of the standards, but there's no factual basis to support their argument, it's a modern-day superstition.  Electricity doesn't care what colour the plastic insulation coating is.  Other factors will be the actual difference (completely different cables or connectors with different quality, reterminating badly made connections, fixing up mis-wired cables, replacing poor or damaged cables, etc).

Out of the two pairs (2 & 3) that are swapped between the two wiring standards, one is a transmit pair and other is a receive pair, both are as important as each other, and everything suffers if either of them was sub-standard.

And out of the same two pairs, one pair carries positive power for Mode-A PoE, the other pair carries negative power, and either of them being sub-standard will cause problems.  For Mode-B PoE, neither of them are used for power.  And whether the equipment uses mode A or B depends on how that equipment was designed.  Mode A or B PoE is completely unrelated to T568A or T568B wiring standards.

Some will argue about the different twist rates on each pair (they're all twisted at different numbers of turns per metre).  The equipment doesn't care which twist rate is used on which pair, they're twisted differently to minimise interractions between each of the different pairs.  It's the twisted cables, themselves, that provide that function.  Inside the equipment each pair is connected to identical circuitry (same termination, same drivers, same amplifiers).

Other arguments about the twist rate claims that some of the pairs are thinner than other, to fit them in the same jacket.  I've seen no cables with different gauge wiring on some of the pairs (solid-core nor stranded wiring).  And every ethernet cable I've seen has the wiring slack inside the outer jacket, they don't need to make one pair thinner to fit inside it.  Also, different gauge wires would affect the ability to properly crimp the 8P8C connector pins onto the wiring, they'd need to have different construction on two of the pins.

Another twist-rate argument says the tighter twisted pair will have an overall longer length than other other slower twist rates.  But any tiny voltage drop by slightly different lengths are going to be insignificant, and PoE uses two (or four) of the pairs (at least one pair for each pole of the DC voltage).  It won't make any difference if you lose 0.001 volts on the positive or negative supply, it amounts to the same thing:  A tiny voltage drop of the total voltage supply, that's taken care of by the voltage regulators inside the equipment (no device runs directly off the power on the twisted-pairs, they run on a lesser voltage, internally regulated to be consistent).

About the only time T568A or T568B has any difference is with telephone wiring:  The pair that straddles the two centre pins will have slightly poorer noise rejection (because the twist has been undone further).  Whatever phone is on that line will have a tiny bit worse noise rejection (barely noticeable, and insignificant compared to any noise that may get in through the entire length of the phone line and equipment, as opposed to the tiny space inside the plug).  For a two line system, T568B doesn't have the second line straddling the centre pair, but T568A does (because the standard is to wire phone lines in first using pair 1, then using pair 2, then using pair 3, then using pair 4, when you have more than one line).  If you have a three-line phone system, one phone line unavoidably straddles the centre pair.  This is presuming that you're carrying several phone lines inside one 8P8C connector.

For ethernet, that concern is a nonsense.  There'll always be using a pair that straddles the centre pin.  It's unavoidable that one pair will will have barely measurable performance difference, and won't help you if that's the transmitting or receiving pair, ethernet is a continuous two-way conversation.

The other telephone aspect to T568A or T568B wiring is frame wiring within a building.  If one end is raw-wiring of phone lines into a frame without plugs and sockets, there are standard colours for which pair is line 1, 2, 3, or 4, and only the colours are the identifying factor.  But again, that doesn't really apply to ethernet, it just about always is wiring up plugs and sockets, and it just matters that pin to pin wiring is correct.

You can even completely ignore T568A and T568B, wire signals up with any coloured wiring pair, and it'll work.  Each pair of wires is just as capable as any of the other pairs of wires at carrying signals or power.  But it will make future wiring management confusing, that's all.

Various terms used on this page

CAT5 - CAT5 is a category of cables, referring to a specification it must adhere to.  Any manufacturer's CAT5 cable should be equivelent to anothers.  Or, at least, meet the minimum standard (some exceed it).  It's the lowest grade of UTP ethernet cabling that will work up to 1 Gb/sec, though CAT6 is recommended for that, at least.

UTP - Unshielded Twisted Pair, the usual kind of cabling used with internet or telephone wiring.

STP - Shielded Twisted Pair. the same as UTP but with a metal shield around the outside of all the pairs, that's connected to ground.  There's another less common STP cable, where each individual pair is shielded.  STP is not automatically a better cable choice.  For one thing, you need to be using equipment with plugs and sockets that actually make a connection to the shield.  Ordinary plastic plugs and sockets do not.