[Back to Previous Page]
[ Forward to Next Page]
Electrical Power Transmission
The electrical power being delivered to a device or part of a circuit can be calculated by multiplying the voltage measured across it by the current flowing.
Some of the power delivered to the circuit is lost as heat in the cable. This means that:
The total power that must be drawn from the mains supply is more than that required by the powered device. [this can as much as double the power requirement of the circuit.]
There is a voltage drop across the cable. In other words, the voltage across the powered device is lower than the voltage delivered by the power sdource.
The power loss and voltage drop are caused by the cable's resistance. Resistance is proportional to length and reduces with wire thickness. The longer a cable is, or the lighter its gauge, the greater the power loss and voltage drop across it.
The power loss in a cable is proportional to the square of the current flowing through the conductor. It is more efficient, therefore, to transfer electrical power at a higher voltage, as the current necessary to deliver the same power will be lower, so less power is lost in the cable. This is the reason that power grid transmission lines operate at such high voltages.
Cat 5 Cable and Power Transmission
The principles of carrying electrical power over Cat 5 are no different to those of other power distribution systems, such as the AC mains wiring, but as the power is being transferred over light-duty cable for long distances, the effects of the power loss and voltage drop become significant.
The arrangement and connection to the cabling used for PoE also differ slightly from conventional power wiring, in order to work around the existing standard for Ethernet data. Cat 5 network cables contain a bundle of eight wires, arranged as four twisted pairs3.
In the most common type of Ethernet, 100BASE-T or Fast Ethernet4, only two of the four pairs are used to carry data; each pair carrying a signal in one direction. These are known as the data pairs, and the remaining two are unused and are referred to as the spare pairs.
PoE power transmission using the data pairs of Cat5 cable
Although each data signal can be carried within a single pair, PoE treats each pair of wires as a single conductor (a reason for this is that using both wires halves the overall resistance). As electrical current must flow in a loop, two pairs are required to allow power to be carried by the cable, and either the data or spare pairs can be used for this. The PD must be able to accept power from whichever pairs the PSE delivers it to.
Normally, the data pairs are used to carry the PoE power, and the spare pairs remain unused. The example shown in the figure above illustrates how electrical power can be connected to the data pairs in the cable, and the path it follows. As the PoE current, carried in common by the wires, is DC5, while the data signal carried within the pair is very high frequency, both can exist on the same cable without degrading performance, and electrical transformers can be used to separate them at either end. Bridge circuits in the PD correct the electrical polarity, and allow power to be received from either the data or spare pairs, while preventing current from being able to flow back down the unused wires.
Upgrading a Cat5 connection to PoE, using a midspan to inject power on the spare pairs
The current flow when using a midspan PSE to deliver PoE over the spare pairs of the cat 5 cable is shown in the figure above. No transformers are required to inject the DC current as the pairs are not shared with data, and the original Ethernet signal on the data pairs can pass straight through from one cable to the other (typically the mid-span will be patched in straight after a non-PoE network switch).
Although this method of power injection is defined in the standard, many midspans connect their power to the data pairs using transformers6 – the method employed will usually vary from brand to brand. Furthermore, it is possible to support gigabit Ethernet (which typically uses all four pairs for data), by always injecting power via a transformer or equivalent arrangement.
Notes
3 Twisting together the two wires used for a signal greatly reduces common mode interference
4 Specifically 100BASE-TX, in full-duplex mode
5 Direct current, which typically does not change over time, as opposed to constantly-changing alternating current (AC)
6 This is because IEEE 802.3af, like so many standards, rather conveniently incorporates matter that has already been patented by contributing manufacturers
