Driving High Power Ethernet Power Technology

Power over Ethernet (PoE) technology can provide power through standard Cat-5 cables, without the need for wall adapters or other external power sources to drive network devices. Such systems require two main components: power sourcing equipment (PSE) that provides power and a power device (PD) that receives and uses the power, as shown in Figure 1.

Figure 1 Schematic diagram of a typical PoE network with power supply equipment and power supply devices

There are many useful PoE applications that have been developed, however, new applications require more power than the power limit defined by the current 802.3af standard, which is about 13 W for the powered device. A new PoE+ standard is needed to meet the increased power requirements, but this new standard has not yet been defined. Until the new standard is available, it will be necessary to be able to power PD loads that require more power than what is currently available.

System Requirements for High-Power PoE
Any high-power solution using Cat-5 cable must meet the following basic specifications:
• Must not cause overheating of the Ethernet cable, power supply equipment (PSE), or PD
• Must not cause current imbalance to avoid the risk of transformer saturation
• Must maintain operating voltage within the 802.3af standard voltage range
Any solution should provide the following functions whenever possible:
• Compatible with the 802.3af standard: search, classification, current consumption, undervoltage lockout, in-rush current, current limiting, etc.
• Have power management functions
• Damaged wire detection during cable installation

The 802.3af standard specifies a PSE output voltage between 44 and 57 V, with a minimum PSE ICUT (indicating the degree of overload) of 350 mA and a minimum ILIM (indicating the maximum possible current consumption) of 400 mA. Using
a 100-meter cable with a worst-case feed resistance of 20 Ω, the PD input is limited to 12.95 W at the lowest allowed voltage.

To provide more power, simply operate at a higher bus voltage closer to the 57 V limit. The benefits of this solution are improved system performance and reduced losses in the cable and PSE and PD input stages.

Existing Cat-5 cabling installations can continue to be used as there is no increase in cable loss, and the risk of wire imbalance and magnetic saturation is no greater with this solution than with standard PoE installations, as long as design rules are followed to ensure no adverse conditions are introduced. In addition, this solution is compatible with the 802.3af standard and uses the same magnetics as 802.3af devices.

For this solution to work, the switching power supply providing the input voltage must be regulated with tighter tolerances. For example, if the deviation is between 53 and 57 V, a ±3.5% voltage tolerance is required in addition to the overvoltage protection of the power supply.

Today, Cat-5 copper communications wiring is the minimum requirement for broadband services, the standard used is EIA/TIA-568, and Cat-5 cable is designed with 24 AWG conductors, resulting in a worst-case feed resistance of 12.5.

Powering over four pairs
To achieve higher power requirements at higher voltages, using all four pairs with two PSE operating ports is a viable solution (see Figure 2). With this solution, all of the benefits listed above are still achieved, and the designer on the PD side can choose to connect the DC/DC power supplies in series or parallel.

Figure 2: Managing high-power devices using existing solutions

For the solution to work, the designer must ensure that each feed shares an acceptable current share (current sharing should not be confused with balance within each pair). Since each feed has its own diode and diode bridge (negative temperature coefficient), fuse, and resistor components, and each PD has a switching power supply with negative output impedance, care must be taken in implementation. Alternatively, two different voltage levels can be achieved at the PSE end by using two different voltage supplies to drive each pair. If the sharing is uneven, the current limit achieved will be much lower than the expected total current.

To increase power using two pairs
, a possible solution is to increase the voltage to increase the current per pair. In such applications, the current limits set on the PD and PSE are higher than the limits defined by the current 802.3af standard.

PD and PSE power components must have higher current capabilities, which is especially important for data transformer components. We can expect the DC current imbalance between the two wires in the same pair to be higher than the 802.3af standard specification, however, signal integrity must be maintained regardless of the DC current condition.

Existing Cat-5 cable installations can continue to be used, but it is important to ensure that the rated upper limit of the cable temperature is not exceeded, as the ambient temperature and cable temperature will rise as the heat dissipated by the wires.

It is strongly recommended that only a few wires in the cable carry higher currents to avoid high temperature rise in the internal wires of the cable. In addition, the connector must also be selected based on the mating force/disengagement force (capacitive and inductive loads) and current conduction capability.

For PD loads with power requirements higher than the 802.3af standard, two-pair or four-pair solutions can be used. However, these two solutions cannot allow the PSE to estimate how much power the PD still needs (beyond the 802.3af standard limit), nor can they estimate whether one port or two ports should be used. In addition, the PD cannot know how much more power the PSE can provide under standard conditions.