PoE is a natural offshoot of the need for networked peripherals to be powered independently of a dedicated source of power. Given that a network cable is already running to the peripheral, why not use that same cable to supply power to the device?
Ethernet frequently runs in Cat 5 UTP (Category 5 Unshielded Twisted Pair) cabling, which has four pairs of wires, each pair twisted within itself. The twists minimize the effects of outside electrical interference; in addition, each pair is twisted with a differing number of twists over a given length of cable to minimize crosstalk between pairs. But, as the table below shows, with only four wires needed for 10BASE-T and 100BASE-TX Ethernet signalling, four other wires are unused.
It didn’t take long for people to start using those unused wires to carry power, but there was no standardized way of doing so until 2003, when the IEEE 802.3af-2003 Power over Ethernet standard was ratified. This standard can provide up to 15.4 W of DC power per device (with a minimum of 44 VDC at 350 mA). Depending on the type and length of cabling, some power losses can occur, down to an allowed 12.95 W.
In 2009 the IEEE 802.3at-2009 standard came along, upping the power available to provide up to 25.5 W. Some implementations conforming to this standard and using all four pairs of the Cat 5 cable can supply up to 51 W of power.
Power over Ethernet requires a source of Ethernet power. Sources of PoE are usually either a PoE router or PoE switch (such devices are called “endspans”), or a power-injecting device (called a “midspan”) between a standard router or switch and the power-receiving device at the endpoint.
Devices receiving power over Ethernet can include:
• IP telephones
• IP cameras (where PoE can power pan, tilt, and zoon operation)
• wireless network access points
• network routers or switches
Advantages of PoE include:
• supplying more power than USB
• higher data rates than AC powerline networking
• inexpensive cabling
• standardized connectors (RJ-45)
• independence from local power standards, assuming the endspan or midspan will operate in that power context
• centralized protection from surges and spikes when the endspan or midspan is protected
The PoE standard can operate in a couple of modes, identified in the table above. Earlier I made it sound like unused wires were needed for PoE; that’s not in fact the case. In Mode A, the power is carried on the same wires as are carrying 10BASE-T and 100BASE-TX data: this mode is sometimes called “phantom power.” This mode is necessary for four-wire cabling and PoE on 1000-Base-T (gigabit Ethernet), which uses all eight wires in Cat 5 cabling for data.
In Mode B, the power is carried on the four unused (in 10BASE-T and 100BASE-TX) wires in a Cat 5 or similar cable. Powered devices must be able to use either mode. Power-supplying sources (endspans or midspans) must only supply power when they detect a powered device is on the line.
Finally, the IEEE 802.3at-2009 standard, apart from offering more power, redefined the powered device’s operation to permit two powered devices to co-locate on one powered port, with one device using Mode A power and one using Mode B power.
Owing to its simplicity, versatility, and use of existing cabling, PoE has the capability to play a large role in power distribution in the world of networked devices.