The IEEE 802.3af standard for Power over Ethernet (PoE) specifies the transmission of DC power along with 10/100/1000 Mbps data, bringing a new look to Ethernet. However, PoE introduces a series of unique challenges, and many engineers experienced in designing Ethernet devices are unfamiliar with the new ways of thinking required to address these issues. PoE is now commonly used in VoIP phones, wireless access points, and security cameras. As PoE evolves, it's essential to enhance understanding of the standard to enable new applications. A review of the standard reveals that a PoE link allows a powered device (PD) to draw up to 12.95W of power from a power supply equipment (PSE). The PoE link or port is controlled by the PSE, which identifies the PD and monitors the port (ICUT, ILIM, and disconnect) through pre-power-on detection and classification. The PSE bears a significant portion of the PoE burden; it must detect PoE and seamlessly disconnect power to avoid damaging existing equipment. If the PSE fails to adequately perform its classification, power supply, and monitoring functions, intermittent failures and power instability may occur. The PSE (Power Sink) cannot control everything; when it supplies power, it assumes the PD (Power Distribution Unit) is compliant and switches on in a non-oscillating manner, thus avoiding drawing more power than required. Because both types of devices must collaborate, PD and PSE design engineers must consider design issues from the perspective of both devices. Emerging applications require greater power . 13W is sufficient for basic IP phones, but it is severely insufficient for applications such as motorized cameras, multi-point wireless access, and large-screen color displays. IEEE is currently developing a higher-power standard called PoE+ (officially IEEE 802.3at) that will coexist with currently available 802.3af devices. The final power level defined by the new standard has not yet been determined, but so far, we will likely see 30W two-pair power supply systems and 60W four-pair power supply systems. The IEEE 802.3at committee has been given the daunting task of defining a secure, higher-power, backward-compatible, and interoperable global standard for currently deployed 802.3af devices. Due to the high complexity of writing such a specification, we expect it is unlikely that we will see the final specification within one to two years from now. While a typical CAT 5 cable has four twisted pairs, the 802.3af standard only allows two of these pairs to carry current for a given time. One option is to allow the third and fourth pairs to carry additional current, thus doubling the available power. A second option is to increase the current limit, allowing the same pairs to carry more power. These techniques have already appeared in dedicated PoE systems. However, each approach has drawbacks, making the choice between them more complex. Implementing a near-standard high-power PSE during the transition period is necessary for applications requiring high power, and waiting for the new standard is clearly impractical. Several solutions exist for this purpose. The following are circuits built upon the basic 802.3af standard: Figure 1a shows a PSE circuit using the LTC4258, and Figure 1b shows a PD circuit using the LTC4257. If the application requires AC disconnection, the LTC4259 can be used instead in the PSE circuit; if the application requires integrating a switching regulator, the LTC4267 can be used instead in the PD circuit. Figure 1a: PSE circuit compliant with basic 802.3af standard implemented using LTC4258 Figure 1b: PD circuit compliant with basic 802.3af standard implemented using LTC4257 High-power operation The following circuit examples demonstrate several ways to achieve high-power operation. Note: In some of the PSE circuits below, channel 4 is used to describe circuit variations; however, any other channel can be used. Two pairs of high-current schemes The power level of the PSE can be increased by simply changing the values ​​of the sensor resistors (RS1 to RS4 in Figure 1a). RSn is set to 0.5Ω, conforming to the 802.3af standard (375mA ICUT, 425mA ILIM). For example, reducing RSn to 0.25Ω doubles the current limit (750mA ICUT, 850mA ILIM). This doubles the PD power when using a short cable; if the cable is longer, its losses increase, thus limiting the power delivered to the PD to less than twice the original. Figure 2a: Dual-current high-power, 802.3af compliant PSE. Note that the LTC4258 also uses a sensor resistor to detect DC open circuits. Reducing this resistor to 0.25Ω doubles the DC open circuit threshold, technically disqualifying it from compliance with the standard. Other 802.3af parameters remain unaffected: detection and classification still meet the standard requirements; while the AC open circuit threshold (only for the LTC4259) is unaffected by the sensor resistor change. Because the increased DC threshold carries the risk of disconnecting very low-power 802.3af PDs, although this risk is small, AC open circuits are recommended for 802.3af PD interoperability. The other two components in each channel need to be changed to handle the additional current. Typically, MOSFET Q4 needs to be replaced with a larger device to handle greater power during current-limited periods. In this application, a D2PAK packaged IRF530-class device is sufficient. Additionally, the PoE data magnetic module should be specified to handle higher currents. Several magnetic component suppliers have recently introduced devices with sufficient current capability. By adding two new components, we can switch between 802.3af compliant operation and high-power conditions. In this case, RS4 should be set to its original 0.5Ω value, and RS4B should be selected so that RS4 II RS4B provides the desired higher current. Setting RS4B to 0.5Ω (the same value as RS4) allows the high-power mode to be set to twice the power level of 802.3af. When Q4B is off, the port operates in 802.3af compliant mode. Turning on the Q4B switch, the port operates in high-current mode. This switching can be done at any time: before detection/classification; after detection/classification, but before the port is powered on; or after power-on. Note that Q4B can be a low-voltage MOSFET, as only the drain of Q4 has a high port voltage. Q4B should be a MOSFET with very low on-resistance to prevent insufficient accuracy in higher current limits. For example, the IRLML2502 is a suitable device in an SOT-23 package. The changes to the PD are slightly more complex (Figure 2b) because the internal MOSFETs are pre-configured to operate at a 375mA current limit. However, adding external passive components controlled by the PWRGD pin allows operation in high-current mode; simultaneously, it maintains full 802.3af detection and classification characteristics while limiting instantaneous peak current. Figure 2b: Two pairs of high-power PDs and four pairs of low-current schemes To increase the power delivered to the PD, another technique is to power all four pairs of wires in a CAT-5 cable. Figure 4a shows a four-pair PSE circuit, where each pair has a standard 802.3af power supply. No changes are needed to the sensor resistor values. Figure 3a: Four pairs of 802.3af power supplies. The four-pair PD circuit is the biggest change (Figure 4b). Now, two LTC4257 devices are required, and the power supply circuit must be intelligent enough to limit the current drawn from each channel to within the limits allowed by the 802.3af specification. To achieve this, the current drawn from each pair of wires, or the power drawn from each pair, must be balanced until it approaches (but does not exceed) the ICUT limit before drawing current from other pairs. This circuitry can be quite complex, and designs vary considerably. Figure 3b: Four-pair low-current PD. The advantage of the four-pair technique is that it utilizes all the conductors in the cable, minimizing total cable impedance and power loss due to long cables. Any high-power technique utilizing standard current is also very close to 802.3af compliance, as it can be achieved simply by using the signal pair or spare pair. The main disadvantages are high complexity and high cost. Each port of a PSE requires a two-channel controller chip, halving the port density; while a PD requires two channels and additional current balancing circuitry to ensure that the current drawn from each pair does not exceed the maximum value. Furthermore, the four-pair technique is ineffective if only the signal pair is continuous, as seen in some CAT-3 building installations. Because of the high cost and complexity of the four-pair scheme, two-pair high-current techniques are preferred at medium power levels. The four-pair system is most suitable when the PD power exceeds 35W. Four-Pair High-Current Solution Combining high-current circuitry with four-pair connections allows for the transfer of more power along the cable than any other technology. Four-pair high-current allows for the transfer of 50W of power to the PD along a 100-meter CAT-5 cable, with even greater power transfer if the cable length is shortened. While this solution suffers from all the drawbacks of previous solutions, it delivers the highest power output. For power exceeding 50W, long cables quickly encounter impedance matching problems, where the cable consumes more power than is transferred to the PD. Shortening the cable length allows for further increases in current, ultimately limited by the bias current of the RJ45 connector, magnetic components, and the temperature rise in the CAT-5 cable. Extremely high-power (>50W) circuitry should only be used in systems where the entire solution is specified by the same supplier. Category: When to Apply High Power In special cases where the above circuitry is not used, one approach is to determine when to apply high power to the line. Under normal circumstances, all technologies will successfully power PDs compliant with the 802.3af standard. Dual-threshold circuits need information from the PD to know when to switch thresholds; while four-pair schemes need to know when it's appropriate to switch to the second set of wires. The IEEE 802.3at committee is working to resolve these issues, but a final solution has not yet been determined. During the transition period, special schemes are needed to identify high-power PDs. 802.3af defines an unused class (Class 4) that appears to be specifically designed for high-power PDs; both the LTC4258/59 PSE chip and the LTC4257/67 PD chip support Class 4. Fortunately, a Class 4 PD, if inserted into a standard 802.3af PSE, can be powered using Class 3 current-limited supply; however, if it attempts to draw more power, it will repeatedly turn on and off. Class 4 can be used as an "alarm" signal for a connected high-power PD; however, it is recommended to send an additional handshake signal before transmitting more power. Ideally, a high-power PD should receive some signal from a high-power PSE confirming that "operating in high-power mode" is acceptable. If no handshake signal is received, the PD should send some signal to the user indicating that it has inserted an incorrect type of PSE.