PoE Pinout and Wiring: Mode A, Mode B and 4-Pair Power

A question that comes up on every PoE commissioning job: which wires actually carry the power? The answer depends on the powering mode, and the fact that there are several valid answers is a deliberate part of the design. Standards-compliant equipment negotiates the mode automatically, but knowing the pinout matters the moment you pick up a multimeter or inherit a cable plant somebody else terminated.

Phantom Power: How Power and Data Share a Pair

The elegant trick at the heart of PoE is phantom powering. Ethernet signalling over twisted pair is differential - information is carried in the voltage difference between the two conductors of a pair, and each pair is transformer-coupled at both ends. DC applied equally to both conductors of a pair, injected at the centre tap of the transformer, produces no differential voltage at all. The data transceiver sees nothing; the DC passes straight through to the far-end centre tap and out to the load. Power and data occupy the same copper without interfering, and no extra conductors are needed.

Mode A: Power on the Data Pairs

Mode A, sometimes called endspan powering because switches usually use it, places DC on pairs 1/2 and 3/6 - the pairs that carry data in 10BASE-T and 100BASE-TX. Power rides along with the signal via the magnetics' centre taps. In T568B terms, that is the orange pair and the green pair.

Mode	Positive conductors	Negative conductors	Usually seen on
Mode A (alternative A)	Pins 1, 2	Pins 3, 6	PoE switches (endspan)
Mode B (alternative B)	Pins 4, 5	Pins 7, 8	Injectors (midspan)
4-pair (802.3bt)	Pins 1, 2, 4, 5	Pins 3, 6, 7, 8	Type 3 and Type 4 equipment

Mode B: Power on the Spare Pairs

Mode B uses pins 4/5 and 7/8 - the blue and brown pairs, which sit idle in 10 and 100 Mbps Ethernet. Midspan injectors historically favoured Mode B because it let a simple passive device add power without touching the data pairs at all. The word "spare" stopped being accurate with gigabit: 1000BASE-T uses all four pairs for data, so under Mode B at gigabit speed the power is once again sharing conductors with signal, phantom-fed exactly as in Mode A.

Why the PD Must Accept Either

The standard requires a compliant powered device to work with Mode A or Mode B, and to tolerate either polarity on each pairset. This is not a nicety - it is what makes the ecosystem work. A crossover cable swaps pairs and therefore swaps the apparent polarity, and installers cannot be expected to know which mode an arbitrary PSE uses. Compliant PDs put a diode bridge on each pairset, so whatever arrives is rectified to the correct polarity internally. A PSE, by contrast, is permitted to implement only one mode, which is why a given switch will consistently show voltage on one pair set and nothing on the other.

4-Pair Operation

802.3bt energises both pairsets simultaneously. Current divides across four conductors in each direction rather than two, which halves the effective loop resistance and therefore quarters the I²R loss for a given current - the reason Type 3 and Type 4 can deliver so much more usable power without raising the voltage. It also imposes a new requirement: the resistance of the pairs must be reasonably matched, because significantly unbalanced pairs will not share current evenly and one pair can be pushed past its thermal limit while another idles.

What This Means for Cable Plant

Three practical consequences. First, never use a cable with only two pairs terminated for anything except legacy 10/100 Mode A or Mode B links - a two-pair patch cord in a 4-pair PoE run is an immediate failure. Second, avoid "cable sharing" adapters that split one cable into two 10/100 links, since they leave no pairs for Mode B and no possibility of 4-pair power. Third, DC resistance unbalance becomes a real acceptance criterion for higher-power installations: within-pair unbalance is commonly limited to 3 %, and pair-to-pair matching matters for 4-pair operation. Field certifiers can measure both, and on a Type 3 or Type 4 job it is worth asking for those results rather than a wiremap alone.

Measuring in the Field

To check a live PSE port safely, measure DC voltage between pins 1/2 and 3/6 for Mode A and between 4/5 and 7/8 for Mode B, expecting roughly 44 V to 57 V - but only with a valid load attached, because a compliant PSE presents no operating voltage into an open port. Reading zero volts on an unloaded port is correct behaviour, not a fault, and misreading that is probably the single most common false diagnosis in PoE troubleshooting.