Hirose: Connecting the future
Industrial Ethernet Book Issue 89 / 20
Request Further Info   Print this Page   Send to a Friend  

Using Power over Ethernet procedures for effective designs

802.3at-2009 provides procedures for manufacturers that ensures compatibility, and serves as a guide for PoE applications. It also describes how to signal the power requirement from the powered device to power sourcing equipment for effective allocation of power.

INSTALLING A SEPARATE POWER LINE for devices such as IP phones, IP cameras, code readers or WLAN access points is not a requirement if the user supplies power and data via the same cable. Both can be provided by using an Ethernet switch and procedure called Power over Ethernet (PoE) which can help achieve the goal of distributing the required power to all consumers with fewer voltage supplies.

Power over Ethernet is used when no power outlet is available on-site such as with a camera mounted to a mast.

PoE standards and procedures

Because manufacturers do not always offer devices that supply and consume PoE at the same time, it is of utmost importance that all manufacturers adhere to one definition. The IEEE 802.3at-2009 standard for Power over Ethernet describes a procedure for all manufacturers that ensures compatibility with devices from another manufacturer.

A device supplied with power is a PD (Powered Device). A device supplying power is PSE (Power Sourcing Equipment). The practical challenge is to use the resources efficiently. It starts with dimensioning the voltage supply and ends with the required number of PoE-capable ports at the switch. In most cases, existing power is not used to its full potential.

There are two procedures in the standard describing how to signal the power requirement from the PD to the PSE to allocate this power. Electrical energy is not transmitted unless the PD has successfully registered as consumer.

Classification into power classes

The first procedure of classification takes place on the lowest layer, the physical layer. This procedure meets the older standard, IEEE803.2af, and is referred to as Type 1 in the current IEEE802.3at standard. Many devices on the market only support this type. Five power classes (0, 1, 2, 3 and 4) are defined for a PD. Because the value of class 0 corresponds to the value of class 3, we often only refer to four different power classes. Class 4 must be supported by a powered device according to IEEE802.3at Type 2.

The PD assigns itself to one of these power classes when connecting to the PSE. The PSE classifies the PD by measuring the classification current. A Classes 3 (6.49 W to 13 W) and 4 (13 W to 25.5 W) in particular cover relatively wide ranges, so that unused power must be reserved and is not available to any other PD.

Optimizing power demand signal

The second procedure is classification via the data link layer (layer 2 of the OSI layer model). The PD informs the PSE via LLDP (Link Layer Discovery Protocol) about its electrical requirement. This procedure has finer power resolution that goes beyond the defined power classes. The connected PDs usually reserve the maximum power they need at full load.

However, if this power is not used by the consumer device because devices seldom run continuously at peak performance, this power is not available to any other consumer according to the rigid scheme.

Because the current requirement values can be transmitted from the PD to the PSE even after initial registration, this procedure is superior to the one using power classes. Thanks to the fine granularity and the dynamic demand signal, more PoE consumers can be operated on one switch at the same available overall power than if each were to statically reserve the maximum power draw for its class. However, this procedure cannot be applied to devices of the "old" communications standard (IEEE802.3af).

Only a PD according to IEEE802.3at Type 2 must support classification via LLDP and via power classes. A PSE, on the other hand, must only support one of the two classification procedures. This ensures that each PSE can always be used with any PD.

Power management in extreme cases

But PoE power management involves even more. If all connected PDs require more power than is available at any given time, a prioritization has to take place. Three stages (critical, high and low) are available in this case. The standard stipulates that the PSE is not to distribute any electrical power to a PD if the demand exceeds the power that is available. As a result, the PD is switched off.

If a device with a higher priority is added to an existing system and if the total available power would be exceeded by this step, a PD with a lower priority is switched off so that the PD with the higher priority is supplied. A user should keep this in mind and come up with a useful prioritization that determines which device is most important and which device could be switched off first in case of need.

Because standards usually represent a compromise between the many parties involved, there is frequently a need for a solution that expands this standard for individual applications. Another option for power distribution is to specify a static maximum power limit for each port on the PSE.

The static PoE power distribution can also be used for PDs that only support the older standard, IEEE802.3af. The user sets the maximum power available per port at the PSE. Due to the finer granularity, this is better than power classification to the standard in which the range within a class can be rather wide. Static power distribution does have its limits, of course.

If a specific power is reserved for a port and a PD, it cannot be used by any other consumer even if the PD is not connected at all. Users need to be much more familiar with the requirement in this case, and may find information on power requirements in the technical specification of a PD or determine it by conducting their specific measurements.

Users do not have to install a separate power line for IP phones, IP cameras, code readers or WLAN access points if they supply power and data via the same cable.

Calculating the potential savings

Let us conclude with a practical example, which illustrates the savings potential of PoE power management. A major difference between the power that is drawn by the devices and the power that is reserved according to the power classification may add up to rather large amounts with multiple PDs.

The difference may be so large that additional PDs could be supplied with static power distribution compared to the standard power classification. If a WLAN Access Point consumes a maximum of 8W, for example, it must reserve 13W according to its classification (Class 3). The result is a difference of 5W that is reserved but never used. This means with a total available power of 30W with static power distribution a total of three instead of only two of these WLAN Access Points could be supplied. This also means that there is no need for an additional switch, which otherwise would possibly be required for the third access point. Scalance technology includes switches that are power sourcing equipment, as well as WLAN Access Points that are powered devices.

The standard is a good basis and enables easy compatibility across all manufacturers, but offers only limited options especially for devices that are not of IEEE802.3at Type 2. These devices are still widely used even though the standard was adopted in 2009. PoE Power Management enables optimal use of available power even when you use devices that do not support IEEE802.3at Type 2. The topic of efficient power distribution is becoming increasingly important due to growing power requirement caused by higher data rates.

Outgoing vs. incoming power

Let us now point out the difference between the power that is output by the PSE and the power that is received by the PD. We have to deduct the effective power loss caused by the transmission across the Ethernet cable with a maximum length of up to 100m. In addition, Ethernet cables have never been optimized for low power loss. While 15.4W have to be output for Class 3 at the power sourcing equipment, we are only expecting 13W to be received by the consumer. The loss is even greater for Class 4 where the output by the PSE is 30W while no more than 25.5W is received by the PD.

Anja Adling is Product Manager for Industrial Ethernet switches at Siemens.

Source: Industrial Ethernet Book Issue 89 / 20
Request Further Info    Print this Page    Send to a Friend  


Discover Cisco IoT
DINSpace fiber optic and Cat 6 patch panels
Siemens iWLAN
Japan IT Week Autumn

Get Social with us:

© 2010-2018 Published by IEB Media GbR · Last Update: 22.10.2018 · 25 User online · Privacy Policy · Contact Us