The IEEE recently published a specification which defines how 48V of power is to be delivered over standard Ethernet cables. CE-E asked George Thomas, president of Contemporary Controls, to have a look at it.

Commonly referred to as Power over Ethernet (PoE), the 121 page IEEE 802.3af amendment was approved in June and will be included into the voluminous 1538 page 802.3 standard. The intention of the standard is to define a unified method of powering devices such as IP telephones, web cameras and wireless access points from the same four-pair cable that handles Ethernet data.

The controls industry is hoping that this standard can be used to power Ethernet control devices similar to AS-i, which specifies a single two-wire cable for both power and data which is capable of carrying 24 VDC (8A).

With PoE, the Ethernet star topology is retained. Only two of the cable pairs carry power and it is possible to have power and data on the same pairs. There were concerns such as guarding against destruction of legacy devices, life safety and compatibility with existing equipment. Therefore, the RJ-45 connector was retained as well as category 3 and 5 cable. Not all 802.3 twisted-pair standards were supported, only the most significant. Specifically, 10BASE-T, 100BASE-TX and 1000BASE-T can be powered through their media dependent interface (MDI) but with limitations. The amendment introduces us to more acronyms common in IEEE standards.

The device to be powered is called the Powered Device (PD) while the source of power comes from Power Sourcing Equipment (PSE). The Link Section separates the two. The PSE can be an endpoint device, such as a switching hub, or a midspan device allowing the use of conventional switches. The figure shows two methods of powering a device.

Alternative A uses the two data pairs for carrying power. Each conductor within a pair would be at the same potential. For example, the transmit pair would act as the positive power leg while the receive pair the negative. However, the polarities may be reversed so the PD must be able to handle either polarity. Alternative B simply uses the unused data lines instead.

Attaching a PD to a PSE is much more complicated since all devices must follow the approved protocol. First it must be determined that the Ethernet device can be powered which is accomplished by the PSE, in power-down state, sensing a signature (a 25 Kohm resistor) within the PD. If the signature is correct, the PSE powers the PD. Once powered, the PD must continue to represent itself as a valid device otherwise the PSE will remove power. A PD can also advertise its power consumption requirements during the initial power-up sequence. There are four power classes, from 4 watts to 15.4 watts. The voltage provided the PD is a nominal 48 VDC consistent with telephone equipment.

Can this standard be used for device-level Ethernet? It is possible. DC to DC converters could provide the 24V. The 15.4 watt power capability is quite sufficient for most field devices, but the power requirement for a multi-port PSE is very significant.

Integrated circuits are now available to support the 802.3af protocol in anticipation of the huge demand for IP telephones. However, do not assume that existing Ethernet equipment can be easily adapted to PoE. The pulse transformers used in the Ethernet MDI must be re-evaluated for PoE applications since the presence of a DC current in its centre-tap. Suppliers of Endpoint and Midspan PSEs must also be qualified.

Finally, there is the issue of the RJ-45 connector with its limited acceptance on the plant floor. The standard addresses no other connector so it is up to our industry to qualify alternative connectors.

Power over Ethernet involves more issues than simply defining the cable connections and the 802.3af effort provides valuable guidance on how it should be accomplished. Perhaps this is the last barrier to device-level Ethernet.