Power Electronics News explores Power Over Ethernet (POE) and provides an overview at Power Source Equipment (PSE) and a Power Device (PD).
Power Over Ethernet (POE) is a growing field that allows for the transmission of DC power over existing Ethernet cables. Ethernet data transmission can range from the older and slower 10base-T and 100Base-T to the newer Gigabit speeds like 1000Base-T, 2.5GBase-T, and 5GBase-T. They typically use Cat5e cables that comprise of 8 wires arranged as 4 pairs, with data transmitted and received in a differential mode. The gigabit speeds require all 4 pairs to be used for the data, while the slower formats use 2 pairs with the others as spare. 100m is the maximum device-to-device cable length.
POE at the basic level consists of a Power Source Equipment (PSE), which supplies the power to the cable, and a Power Device (PD) at the receiving end that receives this power and drives the load or application. Figure 1(a) shows a Mode A POE where the data and power are sent on the same wire pairs, while Figure 1(b) shows a Mode B POE where the data and power are sent on separate pairs.
The power is sent as a common mode voltage between two of the wire pairs. The easiest method is to have an always-on power delivery on the spare wire pairs available on the 10Base-T and 100Base-T, with no communication/negotiation between the PSE and the PD. This is called passive POE and can be used, for example, for a 24V voltage supply for a radio antenna. It’s also the least flexible and potentially unsafe if the load is changed or has faults. Several standards have been developed for an active POE method of delivery starting in 2003, called IEEE 802.af, 802.at, and the most recent, 802.bt. These standards specify the entire environment for the POE, including power levels, negotiation and handshaking protocols between the PSE and PD, etc.
Table 1 gives a basic guideline on some of the characteristics on the various POE modes as classified in these IEEE 802 standards. As can be seen the power capability on the recent IEEE 802.bt standard has been significantly increased. Earlier standards were primarily used for basic IP phones, static security cameras and Wifi access points. The need for expansion of its usage prompted the higher power categories. These can now serve a much wider application usage such as full-video security cameras with tilt/zoom, Point of Sale (POS) terminals, network printers, lighting and even computers. The 48V DC transmission over standard Cat5e Ethernet lines with the enhanced POE standards could facilitate the realization of smart buildings with LVDC built into the structure thereby allowing for a more efficient, greener power delivery that is based on real time data feedback from the room where the power needs to be delivered .
Along with the enhanced power capability, the 802.bt also makes the power delivery smarter and more flexible. For example, some new features now include:
- Auto classification: The PSE can dynamically adjust the delivered power level based on actual power usage from the PD
- Two interoperable PD classifications are called “single signature” and “dual signature.” As shown in Figures 2(a) and 2(b), these figures also show the basic block diagram of the use of transformers with center taps to generate the common mode voltage transmitted over the twisted pairs. For example, in the dual signature mode, two different load classes can be fed over the Ethernet connection, for example, a camera and a network printer.
3. The Maintain Power Signature (MPS) allows a pulse used by the PD at a certain duration to signal to the PSE that it is maintaining power. This MPS pulse duration has been lowered to allow lower standby power
4. Another improvement is the extended power capability whereby the PD measures the real time power loss through the transmission and hence the minimum power needed, thus saving power over a hard defined specification.
Let’s look at the key blocks that define the POE in some more detail:
In many ways, this can be considered the master device as it controls the power delivery detection and classification based on feedback it receives from the PD and also its own capability. It ensures that proper handshaking is done and that the PD is an actual POE device. In general, there are 2 categories of PSE:
a) Endpoint: where the PSE is an Ethernet switch that allows both Ethernet data and POE power to be transmitted from the source.
b) Midspan: also called an injector, where the POE is injected into existing Ethernet data lines
Safety standards for the PSE have been defined in the IEEE 802.3xx as well as the IEC 609500-1 requiring 1500 Vrms isolation at 50-60 Hz for 60s as well as the ability to withstand 1500V, 10/700 us applied 10 times without any insulation breakdown. This is between the 48 V supply and the wire pairs that carry it and any chassis ground as well as logic circuits in the PSE. Several of the micro-controller (MCU) manufacturers have POE solutions that include PSE chipsets, where the digital portion connecting to the host is on an MCU chip, while the POE data out is controlled by a mixed signal chip that is electrically isolated from this MCU. Examples of this include Microchip Technology’s PD69210/PD69208 and Analog Devices’ LTC4291/4292. The isolation between the chipsets can be optical or with a 1:1 transformer. The PSE protects the PD circuitry during start-up by limiting the inrush current, depending on the class, as well as providing other safety features such as removing source power if it detects a fault on the line/PD or finds that the PD is not drawing the power. The Microchip Technology PD692xx PSE family does temperature monitoring of the die as well as the driver MOSFETs and turns the output port power off if limits are exceeded.
Since the POE power can be transmitted on either polarity (i.e. the green arrows shown in Figures 2(a) and 2(b) could be reversed), the PD provides polarity protection requiring a full diode bridge. The PD provides the signature to the PSE to allow for a valid detection. The detection is done through a series of low voltage test pulses from the PSE. A precision resistor at the PD (23.7 to 26.3 kOhms as well as an input capacitance of 50 to 120nF) ensures detection. The classification phase for the correct power level next entails further PSE to PD handshaking based on resistors used on the PD. The PD has to provide the same level of isolation as described above in the PSE section from ground to any user accessible parts. The PD would usually supply the 48V power to a DC/DC converter such as a buck converter, which should ideally be isolated from the PD. Referring to Figure 2(a) the diode bridge that used at the front end of the PD is preferably an Ideal Diode Bridge (IDB). The use of MOSFET’s instead of diodes improves the efficiency of the bridge significantly due to a much-reduced forward voltage drop. A typical NMOS based IDB is shown in Figure 3.
Eight MOSFETs would be used for the two bridges required for all four pairs. These would typically be rated for a minimum of 100V BVDSS. A controller chip would control the IDB gates. The HS NMOS FETs shown in Figure 3 would need to be driven with a charge pump to ensure proper Vgs for operation. An alternative, as used by Onsemi’s FDMQ8205A IDB, is the use of PMOS devices for the high side, which then would just need a diode and zener driven from the POE input to the bridge to maintain the correct Vgs and hence not need the charge pump. The IDB MOSFETs could also be integrated into the IDB controller, as is the case for the FDMQ8205A. The PD controller drives an output MOSFET, which can be internally integrated into the controller chip that supplies the DC/DC converter. Additionally, many PD controllers now integrate the DC/DC controller as well. An example of this is Texas Instrument’s TPS 2370 PD with an integrated DC/DC controller. This controller supports Class 1-6 (51W max) and has a PWM with flyback or active-clamp forward capability.
3) Surge and Inrush Protection: As the Ethernet cables are exposed to an outside environment, protection needs to provided by the POE setup to prevent damage to any user equipment as well as to the PSE/PD circuitry. Standards such as EN 61000 4-5, ITU-T K21 and GR-1089 have been defined for this. This includes both common mode protection (i.e. between the wire pairs and ground) as well as line-to-line differential mode. Protection devices would typically include 58V Transient Voltage Suppressor’s (TVS) as well as Metal Oxide Varistor’s (MOV) that can used to provide bi-directional transient surge protection such as those induced from lightning strikes. The PD output driver MOSFET for example needs to also handle inrush currents from the initial POE startup from capacitive charging (I = C dV/dt) and reliability tests need to encompass this aspect. References 2, 3 provide excellent details on the use of TVS and MOV devices these at various points on the POE circuitry to provide protection.
 Abdelwajid, Ouni, “Review of Low Voltage DC technology prospects in smart buildings,” 2022 5th International Conference on Advanced Systems and Emergent Technologies
 Application Note AN3410 from Microchip Technology “Design for PD System Surge Immunity PD701xx/PD702xx”
 Application Note SLUA736 from Texas Instruments “Lightning Surge Considerations for POE Powered Devices”