Power-over-Ethernet (PoE)

Electricity from the Ethernet Cable – Multiple Advantages

IEEE 802.3 af/at In this article, find out why Power-over-Ethernet (PoE) offers a simple solution to imminent changes in the law that will affect every user. In fact, there are economical reasons indicating it will also be a solution of the future. ROI times of less than 12 months, a contribution to environmental protection resulting from more efficient use of limited energy resources, and not the least of all cost savings, all speak in favor of PoE. For any little technical problem that may arise, our specialists are here to support you with ideas and measures.

Power-over-Ethernet (PoE) is not really anything new. While turning to Wi-Fi in an attempt to limit the amount of wiring necessary, for example, in historic buildings, other technologies were developed at the same time to use structured IT wiring for various purposes. Whether telephony or Internet, only four of the eight leads are used for communication. The possibility of a duplicate assignment from two end devices per cable remained an exception, and didn’t conform to applicable standards. The use of a data cable for the simultaneous supply of power to a device, however, was not out of the question.

The first products were soon developed in which the distributor side supplied power using an injector for coupling. At the end device a splitter was used to decouple. The term ‘active Internet’ became common in practice, but there was no standard for it at the time. Because manufacturers used different voltages for their devices (mostly 5, 12, 24, or 48 volts), users were generally forced to use proprietary solutions from the same manufacturers. Mistakenly using the wrong voltage often resulted in fried input circuits.

The breakthrough came in June 2003 with the development of the finalized standard 802.3af, which regulated voltage and maximum current loads, among other things. Because it’s not obvious to users whether a PoE is available on the supply side and because, for example, an ISDN device can be damaged by PoE power, a process known as resistive power discovery was implemented as a protective mechanism. As soon as a PoE injector recognizes that an end device has been connected, it repeatedly provides a minimal feed of up to 10 volts to the leads, which does not threaten to damage any appliance. In accordance with standards, PoE end devices must be equipped with a load resistance of 15 to 33 kOhm, which is detected by the supply device. An identification process then occurs at 14.5–20.5 volts, which determines the required performance class of the end device. The PoE injector then provides the necessary electricity for this performance class (36–57 volts, normally 48 volts). Only then can the appliance begin to operate.

The permitted value for a PoE device is up to 350 mA in continuous operation. At startup, a maximum of 400 mA are permitted. At 48 volts, the maximum output is 15.4 watts, which when taking into account line loss results in a value of at least 12.95 watts to supply the device. Within this range, wireless access points, Bluetooth devices, IP cameras, IP telephony, as well as hubs and small servers can all be supplied with power.

So where are the advantages?

1) No doubled wiring

Because we assume we are dealing here with IP devices, they will require an Ethernet cable depending on the system. Because of the PoE, there is no need for a power cord. Despite the rising costs of raw materials, a cost savings based on materials alone is only a decisive factor for 230-volt cable at very long distances. However, an electrician is needed for work on electrical circuits, whereby an IT technician can handle Ethernet cable connections. No technical knowledge is required because the power supply at the PoE distributor is achieved with a standard network cable. The system is almost completely plug-and-play. If the PoE power supply is retrofitted, each port requires an additional patch cable and a PoE distributor. No modifications to the wiring are necessary. If PoE is planned for a new installation, the PoE distributor replaces the otherwise standard TP patch field. Otherwise a space of 1 to 2 rack units is required for 24 ports. The Ethernet norm also limits lengths, so there is nothing special to consider for PoE.

2) Central UPS emergency power supply

Another advantage can be found in the option to supply emergency power to end devices. Anyone who has developed and implemented a proper emergency response concept understands the significant additional costs resulting from wiring an additional power circuit and the installation of a second row of supply outlets for a central or peripheral UPS. In practice, however, outages still occur when the end device to be protected is mistakenly connected to the wrong outlet. It also happens that devices are connected to the UPS circuit that were not planned for. The increase in power use leads to an interruption of the planned power failure bridging time, and the entire concept fails.

The injectors are connected to the distributors together with the UPS system for emergency power supply using PoE. There is no need for additional wiring. Maintenance of the USP system is carried out centrally and therefore economically. If individual ports or individual end devices need to be specially handled, manageable PoE devices are used. Intelligent software management allows devices to be recognized using an IP or MAC address and then turned on or off.

3) Energy savings from management

Energy efficiency and the accompanying decrease in CO2 emissions are no longer only a question of being ‘green.’ With the ratification of new EU guidelines, all users including those in B2B are legally required to achieve specific efficiency coefficients for power supply units. If you leave the discussion of energy politics to others and trust that the saving of resources by reducing power usage is all in the hands of the manufacturers, you don’t realize that you are the one paying the bills, particularly for electricity and operating costs.

At the same time, saving energy is not so difficult and for two reasons a responsibility of management: first, operations management, which must make a decision based on ROI calculations, and second, IT management, which has to carry out the work. Manageable PoE devices are used so that SNMP can be applied in order to activate or deactivate targeted devices. Why should a device be supplied with power long after a worker has left the workspace? On the other hand, special functions, emergency response installations and devices, and monitoring installations all need to continue to operate. In intelligent systems, functions are available such as remote on/off, time-controlled power switching, and web monitoring of the uninterrupted power supply.

Depending on the application, this can lead to an energy savings of up to 30%. Here is one small example, based on a medium-sized company with 50 end devices, including 40 voice devices and 10 Wi-Fi access points. At an assumed rate of 15 watts usage per device and an availability of ten hours (8 hr + 2 hr flextime), there is a savings of 14 hours. 0.015 kW x 50 devices x 14 hr = 10.5 kWh per day or >3,800 kWh per year. If there is no work on the weekends and holidays, that adds 60 x 10 hr = 450 kWh. The total adds up to 4,300 kWh, which at 20 cents including fees equals €850 annually. At this rate, the PoE investment amortizes completely within the first year. PoE injectors are also available in one-port versions.

New developments: 802.3at

The market shows a desire for PoE devices with even higher performance, which is not surprising given what has already been discussed. In terms of transmission speeds, Gigabit Ethernet devices should also be able to be used. Even though GE necessitates using all eight lines, the PoE standard makes this a possibility. To do so, Ethernet data is transmitted together with electricity. There are no limits to application here either.

In order to be able to use devices with increased performance requirements, the 802.3at standard, which has not yet been fully completed, but which for some time has been successfully implemented in various devices, was developed. The maximum values have been determined at an output of 32 watts, which is more than double the 802.3af norm. Typical end devices that require this output are access points based on 802.11n, GE devices, IP cameras with expanded functions like zoom, motor-driven functions, and color, WiMax transmitting stations, thin clients, video telephones, and others.

Because the price of these products is less in relationship to their performance, even shorter amortization times can be achieved.

PoE in Ethernet switches

Depending on the design, one either speaks of PoE devices as end-span, meaning Ethernet switches with PoE ports, or mid-span, meaning devices that are placed between the switch and end device.

Which of these solutions is preferable? Should we believe the manufacturers who tell us theirs is the one? It’d be better if we weighed out the advantages and disadvantages.

For end-span, the advantages are:

  • No need for an additional patch cable, although the costs for a 0.5-m patch cable are negligible.
  • No need for additional installation space. Only in rare cases are comparable types of switches with and without PoE of different heights.
  • No need for an additional device, eliminating an additional potential point of failure.

For mid-span, the advantages are:

  • For most environments, not all Ethernet ports require PoE. In this case, only ports that need PoE are injected.
  • The price difference between switches with and without PoE shows that mid-span solutions for the same number of ports cost only half of this price difference.
  • Devices are available for the various standards. In this way, products can be procured that are truly needed.
  • For example, a one-port device is easily integrated for the retroactive expansion of an end device requiring 802.3at. For an end-span solution, the entire injector side needs to be replaced.
  • The energy saving management discussed is available here. For end-span, this function is totally unavailable at this time.
  • The output of the device is designed for the maximum number of ports. With end-span, there is only the option of partial output (see following paragraph) almost without exception.

In the end, the user must decide which solution he will turn to. In any case, the price and feature advantages speak for the mid-span solutions.


Typical PoE Mid-span Installations
Typical PoE Mid-span Installations: Power-over-Ethernet integrates data and operating power (15 watts, 48 watts) into a single Ethernet cable. The technology enables IP telephones, Wi-Fi access points, network cameras, and other company-supplied devices to enjoy a secure supply of electricity using cables produced in accordance with categories 5 and 5E without having to make any changes to the existing infrastructure. (image from Microsemi)

Power ratings for PoE switches

In practice, we repeatedly determine that users experience problems with power supply through PoE switches. In many cases, this is due to a misunderstanding of manufacturer data and is discovered most often during the process of expansion. Users are confused that they are buying a PoE switch, compounded by statements like ‘PoE functionality on all ports,’ and falsely assume that a 24-port PoE switch can inject 24 PoE devices.

What is meant is that PoE end devices can be connected to any port, but not necessarily to all at the same time. Very often a power supply is designed only for the operation of PoE devices on, for example, half the ports, while remaining Ethernet devices require their own power supply. One absolutely needs to be aware of the maximum available output.
The indication that a 24-port PoE switch is designed with a 200-watt power supply also confuses more than it informs. The 802.3af standard provides for different performance classes of end devices: 0 = 0.44–12.96 W; 1 = 0.44–3.84 W; 2 = 3.84–6.49 W; and 3 = 6.49–12.95 W. The line loss also needs to be calculated. Now if we calculate the maximum number of Class 3 devices for the 200 W power supply, the result is 200 W/12.94 W, which allows for only 15 devices and not 24. The 12.94 W receptor corresponds to a maximum of 15.4 W at the switch port, depending on the length of the lines. It is also important to take into account that the switch itself also requires power from the supply in order to operate.
In such cases, one should rely only on details that have been confirmed in writing. When in doubt, make sure to ask questions, because phrases like ‘802.3af compatible’ probably indicate more of a limitation than a guarantee of performance. Otherwise, you may be disappointed when a PoE device suddenly stops working in the middle of some operation because the output is compromised by too many devices or by the aging of components and the switch’s supply turns off – or in an extreme case is destroyed.


In terms of imminent legislation as well as simple economic considerations, Power-over-is a solution of the future that is available today and amortizes within a short time.

Our specialists will be happy to guide you through planning, selection, ROI, and calculating the amount of power you require.