Transcript
Guard PoE systems from lightning surges, electrical hazards Read about evidence-based design methods that protect PoE equipment from various electrical hazards. By Phillip Havens Technical Marketing and Application Manager Littelfuse Inc. and Chad Marak Product Marketing Manager Littelfuse Inc. Power over Ethernet (PoE) is becoming popular at a rapid pace. It eliminates the need for separate power supplies for connected equipment, the need to locate equipment near an AC outlet, and the need for separate power cables. Recent increases, in the amount of power that can be delivered by the PoE power supply equipment (PSE), from 15.4W (PoE) to 30W (PoE+), have only increased the number of potential applications. Ethernet now delivers sufficient power for VoIP telephony, to power extended-range wireless access points, and to operate surveillance cameras, where it is used to power “pan and tilt” functions. Increased use of PoE broadens the range of locations in which Ethernet may be used. It ranges from strictly indoors to campus layouts (building-to-building) or even first-mile/ last mile telephony applications. These outside environments greatly increase exposure to lightning-induced surges and electrostatic discharge (ESD), not to mention accidental power faults due to inadvertent shorts to the AC power line. This article provides guidance on evidence-based design methods that protect PoE equipment from these electrical hazards.
Table 1: PoE PD power classifications.
Table 1a: PD power classifications and their signatures for PoE+ (IEEE 802.3at) * The PoE+ Type 2 returns a Class 4 classification signature.
EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 1 of 6
PoE basics Power is fed into the cable by power sourcing equipment (PSE), either through a switch (also called an endspan) or a midspan (if located anywhere other than at the end of the cable). Devices on the cable that consume power are powered devices, or PDs. These may also have auxiliary ports for backup in case PoE power fails. PoE power levels. The IEEE 802.af PoE standard limits PD power consumption to 12.95W (360mA), which corresponds to a PSE output limit of 15.4W per port (400mA) after accounting for cable losses. This standard takes into account line losses for maximum loop lengths of up to 100m, thereby allowing up to 57VDC from the PSE. The nominal level is 48VDC. The PoE+ standard, IEEE 802.at, allows the PSE to deliver up to 30W and the PD to accept up to 25.5W for Type 2 equipment (PoE+ Type 1 is equivalent to PoE); with the PSE supplying up to a maximum of 600mA. PoE+ also requires the use of low impedance wiring (12.5 ohms per loop pair compared to 20 ohms per loop pair for PoE), such as CAT5e or CAT6. Companies are working to increase this even more; there are PSEs available that promise 60W per port, and one vendor sells midspan PSEs claimed to provide 95W per port using a proprietary discovery process. This may be approaching the physical limits on CAT5 cable, however, which means that for higher power (there has been talk of 200W PoE), some way around this will have to be found. One simple approach is to increase separation between bundled cables, to allow for improved heat dissipation. Beyond that, cable with heavier gauge conductors would be required. These higher voltages are neither IEEE 802.3af nor IEEE 802.3at compliant. Determining power requirements. It is important for the PSE to deliver the amount of power required for the PD (or PDs) it feeds without causing damage. To determine the power level required, the PSE and PD engage in a back-andforth signaling handshake routine at turn-on that involves voltage pulses from the PSE that determine the impedance signature of the connected PDs; this discovery process sets the system to one of five classes (table 1). Table 2 below shows the PoE PD classifications for PoE+. PoE modes (phantom power technique) Power can be provided in one of two ways over the Ethernet cable. In Mode B power is applied over the “spare” data pair (pair 4-5 & 7-8) found in 10BaseT or 100BaseTX systems since only two pair are used for data delivery (RJ-45 pins: 1-2 and 3-6 are used for data). This leaves RJ-45 pins, 4-5 and 7-8, available for power delivery (figure 1). Notice that PoE uses the phantom powering technique so that a single pair carries a zero DC volt potential difference between its leads. The power supply voltage is derived as the difference between the two center tap connections of the different wire pairs.
Figure 1: PoE Mode B power is applied over the “spare” data pairs found in 10BaseT or 100BaseTX systems or over pairs 4-5 and 7-8 of a 1000BaseT system. PoE uses the phantom powering technique so that a single pair carries a 0V potential difference between its leads; power supply voltage is derived as the difference between two different wire pairs. In a 1000baseT application, there is no “spare pair” and therefore power must be supplied over two of the active data pairs, either Mode A or Mode B. Mode A (figure 2) combines the DC voltage with the signal over the transmit (TX) and receive (RX) pair consisting of pins 1-2 and 3-6. An isolation transformer is connected across pair 1-2 with a center tap while a separate isolation transformer is connected across pair 3-6 with a center tap. These two center taps provide access to the DC power and the voltage across any single pair (i.e. 1-2 or 3-6) remains at zero volts DC. This phantom power technique for both Mode A and Mode B helps to prevent accidental shock hazards when single pairs are handled. EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 2 of 6
Note that both Mode A and Mode B connections can be used with any Ethernet application (10/100/1000BaseT) while Mode B over an unused pair is only applicable for 10 and 100BaseT, as they have unused wire pairs.
Figure 2: Mode A PoE uses the common signaling pair 1-2 and 3-6 combining the DC voltage with the signal over these data pairs. Finally, the PSE cannot provide power in both Mode A and Mode B simultaneously, but the PD must be compatible with both Mode A and Mode B powering techniques simultaneously since it cannot be predetermined what PSE mode will be connected to the PDs.
Figure 3: Lighting protection for a 10/100baseT application uses a combination of clamping devices. A lightning induced surge activates the TVS1 devices in the secondary position, providing a clamping function that routes the offending surge away from the sensitive Ethernet circuit. The tertiary device (TVS2) then provides another level of protection on the line driver side of the transformer. Power fault events, characterized as long-term 50/60Hz waveshapes, activate the fuses (F1F4). For a 1000BaseT system, this schematic above is duplicated for the other two data signaling pairs.
How to protect the system The details of the protective devices and how they are connected depends on the power mode and data rate of the Ethernet system. Both the PSE and PD must be able to continue operation after a lightning surge and also safely handle a power fault (connection or coupling to the AC line) for a specified time, per UL 60950-1 or EN 60950-1, EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 3 of 6
although operation after such a test is not required. To meet this requirement, a series current limiting device must be included. This can be a fuse, which will not open during lightning surge testing but will open appropriately for long-term AC power fault conditions, or a PTC device, as long as it is able to operationally survive the lightning surge testing. The advantage of the PTC, of course, is that it can self-recover after a power fault but PTCs are incompatible with higher data rate 100/1000 BaseT Ethernet systems due to their off-state resistance and hysteresis recovery characteristics (PTCs will not remain perfectly matched after multiple operations).
Figure 4: GR-1089 compliant solution for overvoltage and overcurrent events for 10/100/1000baseT applications in an outside environment subject to both lightning surge and power fault events (for 10/100BaseT, only the two data wire pairs would require this protection).
Where to apply protection Lightning Surge Protection. The choice of lightning protection depends on the expected exposure for the application. For an indoor, less severe application, figure 3 shows the use of TVS Diode Arrays between the RJ-45 and the Ethernet PHY in both the secondary position and the tertiary position. A lightning induced surge event activates the TVS diode array (TVS1) within nanoseconds, providing a clamping function that routes the offending surge away from the sensitive Ethernet line driver circuit. Any residual surge that gets coupled across the transformer is clamped by TVS2. For a more robust solution, a TVS diode array can be used for each wire pair; otherwise, a single TVS array can protect two wire pairs as shown by TVS2 in figure 3. Power fault events, characterized as long-term 50/60Hz waveshapes, activate the 1.25A fuses (F1-F4) after a TVS device provides a current path. Powered device protection. Since there is no way to know whether a particular PSE will be used in Mode A or Mode B on an Ethernet installation, 57V to 90V protection must be provided for all wire pairs at the PD end. Everything should be hardened to more than 100V. The surge protection device should have a trigger voltage greater than any steady-state voltage likely to appear on the cable; since PoE voltages can reach 57V, the device must not trigger at or below this voltage. This also prevents the surge protector from turning on during power classification testing or during the resistive power discovery test. And because some power systems supply +48V and others –48V, the protection device must not be polarity sensitive. Bi-directional thyristor surge protection devices are typically used. They are solid-state crowbar devices EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 4 of 6
that reset only when the available current falls below its holding current parameter. This is not a problem, as it will cause excess current to be drawn from the PSE, which will temporarily shut down during an overcurrent load condition and allow the thyristor surge protector to reset.
Figure 5: Total PoE system protection example, showing both data pair protection and PD power connection protection. The TVS protection solution for the PD end power supply portion is compliant with both Mode A and Mode B PoE power. Typical TVS devices for this type of circuit are available in 400W, 1500W and 3000W ratings. The fuses (F1-F4) provide overcurrent protection compliant with GR-1089, Issue 6 and UL 60950-1. Application example Figure 4 shows a GR-1089 compliant solution for overvoltage and overcurrent events for 100/1000BaseT applications in an outside environment subject to both lightning surge and power fault events. The fuses in both data pair leads provide the necessary overcurrent protection that is insensitive to lightning induced overvoltage surges for First Level GR-1089 events. The bi-directional thyristor surge protection devices (U1-U4), also known as SIDACtor devices, provide an overvoltage crowbar protection solution compliant with both 1st Level and 2nd Level lightning surges of GR-1089, Issue 6. The two bias leads for U1-U4 are connected to any available voltage rails that are less than the turn-on threshold of the protective devices, which stabilizes their off-state capacitance and helps preserve signal integrity. The bidirectional thyristor surge protection device chosen for U1-U4 can be rated for a 58V minimum threshold for the case of a 48V PoE. For non-complying IEEE 802.3 systems that have a higher PoE voltage level than 57V, a higher minimum threshold protection device would be required. EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 5 of 6
The tertiary or chip-side solution is a TVS Diode Rail Clamp Array. This device (TVS3 in figure 4) provides additional protection after the coupling transformer. If Bob Smith terminations are used, they should be capacitively isolated so they do not load the PoE power supply. This combined metallic/differential and longitudinal/common mode protection solution requires a fuse on both leads of the TX and RX pair but a solution without the longitudinal mode may only require a single fuse per pair for lower data rate Ethernet such as 10 BaseT. For higher data rate 100/1000 BaseT systems, it is prudent to place the identical fuse element in both “legs” of a pair to maintain loop balance (figure 4). The single fuse solution is permissible if pins 3 and 6 of the protective thyristor device are not connected to ground, but instead remain open for a 10BaseT Ethernet low data-rate system. Since IEEE 802.3 does not strictly allow for a common-mode protection solution on the primary side of the coupling transformer for CDE protection reasons, the thyristor surge protection devices (U1-U4) are usually not connected to ground. Therefore, most Ethernet solutions will depend on the isolation rating of the coupling transformer for longitudinal/common mode protection on the line side, while the tertiary position (secondary side of the coupling transformer) may be connected to the Ethernet line driver ground reference (figure 4). A 2.5V TVS Diode Array (TVS3) is used as tertiary protection on the chip side of the coupling transformer. This solution will provide compliance with the surge and power fault requirements of GR 1089-CORE, Issue 6 intrabuilding and inter-building. A 0.3A PTC may be substituted for the fuses for compliance with ITU K.20/21 Enhanced and Basic, which contain coordination clauses for low rate 10BaseT Ethernet. However, for higher rate 100/1000BaseT, a pair of precision (1 percent) resistors appropriately sized would be used to force this coordination between the secondary and primary protector. Figure 5 shows a simple 400W TVS solution compliant with both Mode A and Mode B PoE power for the PD end of the system. More robust (1500W and 3000W) solutions are also available for harsher surge environments described in regulatory standards such as ITU K.20 (Enhanced) or GR-1089 (Port Type 5). Summary As the power delivered over PoE systems has increased, it has caused Ethernet equipment to be installed in areas that expose it to increased hazards from lightning-induced overvoltages and from 50/60Hz power line faults. Judicious use of bi-directional surge protection devices, transient voltage suppression diodes, fuses and PTCs can help ensure reliable operation despite these hazards.
About the authors Phillip Havens is the technical-marketing and application manager at Littelfuse and brings a wide range of electronics engineering expertise. He earned both bachelor’s and master’s degrees in electrical engineering from Louisiana Tech University (Ruston, LA), is a licensed professional engineer, and interfaces extensively with the technical teams of many leading electronics manufacturers to develop circuit protection for specialized applications. Havens represents Littelfuse at electronics-safety, circuit-protection, and telecom-related industry associations and helps define, direct, and support the company’s silicon-based-protection-product lines. Chad Marak is a product-marketing manager in the semiconductor-business unit of Littelfuse Inc. His responsibilities include providing strategic direction for the growth of the TVS-diode-array product line and managing the North America field-application-engineering team. Marak received a bachelor’s degree in electrical engineering from Texas A&M University (College Station, TX) and a master’s degree in electrical engineering from Santa Clara University (Santa Clara, CA). He has been in the semiconductor industry for 10 years and holds four US patents.
EE Times-Asia | eetasia.com
Copyright © 2012 eMedia Asia Ltd.
Page 6 of 6
