As well as enabling the basic function of delivering power over network cables, the PoE specification incorporates some more advanced features in order to enhance efficiency, reliability and safety. Classification is one such feature: as part of the start-up process when a PoE connection is made, the PD can advertise its power class, which is an indication of how much electrical power it requires to operate.
There are five power classes in total, although only three different bands of power level that devices can fall into. These are listed below.
Min PSE Power
Max PSE Power
VoIP phoneExample PD
3, 4 or 0
Wireless access point
As can be seen, each class specifies two different values for power: one for the PSE, and one for the PD; and this is to account for the power that is lost in the cable. Hence it is important to remember that only devices of up to 12.95 watts can be powered using PoE, even though the PSE specifies a 15.4 watt output – and this is complicated further as these numbers may frequently be rounded to 13 watts and 15 watts respectively.
Note also that these figures are upper limits to power. Power cannot be “forced” down the cable – a surprisingly common misconception. The PD simply presents a load to the cable and draws as much current as it needs. Most PoE-powered devices will draw a fixed level of power, so for example a 5 watt IP camera will advertise itself as Class 2, as it will never need more than 6.49 watts, and then draw enough current at its local voltage to provide the 5 watts it requires to operate.
Classification is an optional stage of the power-up procedure: to save cost, some PDs will leave out the classification circuitry, and will be detected as unclassified, which is Class 0. Such devices may draw any power level up to the limit, in contrast to Class 3 devices which are known to require more than 6.49 watts. In either case it is perfectly possible that the device may normally draw a level of power far lower than indicated – for example an IP dome camera may only require 4 watts when stationary, but take 10
watts when moving. Class 4 is a reserved class and will be used for future standards; see the later section on PoE Plus.
The figures in the table are also based on the worst-case operating conditions for PoE. PSE voltage is nominally 48 volts, but its allowable range is between 44 and 57 volts: because efficiency increases with
voltage, the losses are based on a PSE supplying the 44 volt minimum. A worst-case value for cable resistance is also assumed, and this is based on a maximum length7 of light-gauge cable, with an allowance for connectors, patch panels, or bad wiring.
Classification is just one feature of PoE which enables enhanced power management of network devices. For example, most PoE network switches and mid-spans have a power budget that is not sufficient to deliver full PoE power to all of their ports. A small network switch with four PoE ports will often have a power budget of 30.8 watts; a limit set by the capabilities or restrictions of its own power supply. It can allocate its power to any combination of devices whose requirements, indicated by their power classes, do not exceed this limit, so for example it could enable power to four Class 2 devices with
no risk of overload, as the total power needed by them will never exceed 28 watts. However only two Class 3 devices could be powered safely, so any further PoE devices connected would be denied power. Such operation is essential for correctly-designed PSEs to be able to guarantee continuous power
delivery to devices whose requirements may change over time.
At the other end of the scale, remote power management of complex installations is possible, using SNMP or other technologies to control and monitor switch and midspan ports, in a similar manner to and often in parallel with the control of network data traffic in managed installations. This allows
devices' power draw and budget allocation to be monitored, and devices can be switched on or off, reset and prioritised as desired. The PSEs can also be connected to an uninterruptible power supply (or UPS), to provide a power backup for all or selected network devices.
This intelligence and control means that PoE-enabled network installations can be designed to be far more power-efficient than the mains-powered alternative. However it is important to remember that as all the power is routed through the PSE equipment, the requirements in the wiring closet can be much higher, and AC power installations should be designed for the worst-case load.
With the greater power requirement in the wiring closet comes an increased amount of heat, along with associated concerns over cooling and ventilation, although it must be noted that most of the electrical power drawn from a PSE will be directed to remote devices rather than converted to heat in the PSE
itself. For example, a midspan that can deliver 24 channels of maximum-class PoE has a budget of 370 watts (15.4 watts × 24 ports). If its power input is rated 4 amps at 120 volts AC, then it requires a maximum of 480 watts (4 amps × 120 volts8). This means that, at maximum load, 480 watts is required
from the AC supply, of which 370 watts is routed to external equipment, and the remaining 110 watts goes towards heating up the mid-span.
The PoE standard incorporates a number of measures to ensure its safe operation, both to protect legacy equipment from damage and to protect users from the hazards of electrical power. The main safety features of PoE are as follows:
Signature detection, as mentioned previously, prevents damage to legacy devices by ensuring that PoE voltages can only be applied to the cable once a compatible PD has been detected. All 802.3af PDs must display a signature when unpowered, whose electrical characteristics can be detected by the PSE using a safe, low-voltage technique.
Current limiting is employed to protect the PSE from overload and to quickly disable malfunctioning PDs. The current drawn on each enabled port is continuously monitored, and power is disconnected if it increases beyond the allowed limit. This limit can be set to correspond to the connected device's power class for added protection.
Polarity protection provided by the bridges in the PDs power input circuitry means it can safely receive power regardless of the configuration and polarity of the voltage on the cable. This makes it immune to reverse connections on crossover cables or different types of equipment. Badly-wired or incompatible equipment will either work safely or will not connect at all.
Automatic disconnection of devices is performed when the monitored current falls below a minimum level, returning the cable to its original unpowered state, ready for the safe connection of another PoE or non-PoE device.
Low voltages are used to protect installers from electric shock hazards. The voltage used by PoE (up to 57 volts) is high enough to be efficient but low enough to be safe. This has the further advantage that cabling does not have to be installed by a qualified electrician.
Isolation of the cable from other devices protects further against electrical hazards. Both PD and PSE should provide a 1500 volt isolation barrier between the cable and ground or any accessible parts (although some PSE manufacturers flout this rule).
In addition to these designed-in features, the fact that PoE is standardised mitigates the risk of using different and incompatible local power supplies, with exposed power connectors that cannot be securely fitted.
7 100 metres, which is the maximum distance for a twisted-pair Ethernet connection
8 AC (RMS) and DC voltage levels can be treated the same way in power calculations (although peak AC voltage is higher)