This FAQ will review the cost, efficiency, and other performance tradeoffs for PoE and USB PD implemented with flyback, forward, and active-clamp forward topologies, emphasizing key power transformer characteristics and specifications.

In a PoE application, the typical input voltage for the power converter will be a nominal 48 Vdc from the Ethernet cable, and the converter will be a dc/dc topology. On the other hand, USB PD is typically an ac/dc power supply with an input range of 85 to 265 Vac. Flyback and forward topologies are versatile and can implement either a dc/dc converter or an ac/dc power supply.

At first glance, a flyback converter and a forward converter look similar. Both are typically single-switch topologies, but they operate in fundamentally different ways. It’s the magnetics that makes a big difference. The transformer in a flyback converter can be modeled as two inductors with windings of opposite polarities sharing a single core. When the switch is “ON” energy is stored in the magnetic field in the air gap. When the switch turns “OFF,” the stored magnetic field collapses, and the energy is transferred to the output. The transformer in a forward converter is modeled as a true transformer with same-polarity windings; there is no air gap to store energy, and energy is transferred directly to the output when the switch is “ON.”

Flyback topology
Flyback converters can provide a low-cost solution for low-current and low-voltage output power supplies (Figure 1). They provide good input-to-output isolation and do not need an output inductor. However, they have a higher output ripple current and need a larger and more expensive output capacitor. They have higher peak and RMS currents than a forward converter, which reduces efficiency.

When the main switch is ON, the transformer’s primary is directly connected to the input voltage source. The transformer’s primary current and magnetic flux increase, storing energy in the transformer. The induced voltage in the secondary winding is negative, so the diode is reverse-biased blocking current flow, and the output capacitor supplies energy to the output load. When the main switch is OFF, the primary current stops flowing, and the magnetic flux drops. The transformer secondary voltage is positive, forward biasing the diode, and current flows from the transformer, recharging the capacitor and powering the load.

Standard flyback transformers are available for PoE and USB PD applications. For example, flyback PoE transformers can support 1500 Vac isolation from the input to output. Standard power levels include 4 W, 7 W, and 13 W, and the transformers accept input voltage ranging from 29.5 to 60 Vdc (Figure 2).

When considering a flyback transformer for either PoE or USB PD, designers should select a transformer with a peak primary current or a primary saturation current that is well above the expected primary current peak for the application. The peak current requirement occurs at the worst-case condition of maximum load and minimum input voltage. If the peak current flow exceeds the transformer rating, core saturation will occur, and the primary inductance will drop, resulting in a loss of regulation.

In addition, voltage is reflected to the primary side from the secondary side when the switch is turned OFF, and the stored energy in the flyback transformer is transferring to the load. The increased voltage stress experienced by the switch can result in damage and lower reliability. A passive resistor-capacitor-diode (RCD) clamp on the primary side can help protect the switch from excessive voltages. The drain-source voltage (V_DS) is clamped at a relatively high level with a passive clamp, and the switch turns ON when V_DS is high. A high V_DS increases turn ON losses (reducing efficiency) since switching loss is proportional to (V_DS)².

In an RCD clamp, the leakage-inductance energy absorbed by the capacitor is dissipated in the clamp resistor. This loss of energy limits the switching frequency, which results in the need for a larger transformer. A passive clamp provides low-cost protection for the main switch but increases losses and reduces switching frequency. Those limitations can be addressed with a higher-cost active clamp solution.

Active clamp flybacks
An active clamp flyback recirculates the energy instead of dissipating it, improving efficiency (Figure 3). When the main switch is OFF, the clamp switch sends the clamp capacitor energy to the secondary just before the main switch turns ON. And the clamp ensures that the main switch turns on when V_DS is near zero, reducing switching losses and reducing EMI generation. The use of an active clamp can also enable a lower voltage rating synchronous FET on the output, which can reduce costs and/or lower ON resistance, further increasing efficiency.

Forward topology
Forward converters are essentially buck converters that use a forward-mode transformer. In PoE and USB PD applications, the transformer steps down the voltage and provides a safe dielectric isolation between the input and output. The main limitation of forward converters is that the maximum duty cycle is about 50 percent. The transformer in a forward converter includes a ‘reset winding’ (Figure 4). The core is driven in a unidirectional manner, and the core can saturate after a few cycles unless it is reset or ’emptied’ of excess magnetization energy. When the power switch is OFF, and the rectifiers are not conducting, the transformer is reset by drawing current from the reset winding. In most designs, the reset winding is a separate winding, but the primary winding is sometimes used to reset the transformer core. Current from the reset winding is returned to the input capacitor and reused during the next cycle of operation.

Forward-mode transformers are fundamentally different from flyback transformers. A specific magnetizing inductance and a gapped-core construction are needed in a flyback, allowing the transformer to store high levels of energy without going into saturation. A forward-mode transformer has high primary magnetizing inductance to support efficient energy transfer between the windings. Lower secondary RMS current in a forward converter than a flyback results in lower losses, contributing to higher efficiencies in forward converters. In addition, the output capacitor in a flyback must handle the higher RMS currents, increasing the capacitor losses in a flyback compared with a forward topology.

Like flyback converters, forward converters can benefit from an active clamp. In a forward converter, the active clamp is used to implement transformer reset (Figure 5) efficiently. An active clamp provides a good balance between high performance and moderate cost. Zero-voltage switching (which reduces switch voltage stress), extended duty cycle range, reduced EMI, and significant improvement in efficiency result from the use of an active clamp transformer reset.

Summary
Flyback and forward converters are suited for PoE and USB PD isolated power solutions. They are both cost-effective single-transistor solutions but offer different mixes of features and benefits. The energy storage is in the (air-gapped) transformer in a flyback converter. A forward converter uses a transformer without an air gap, requiring an additional energy storage inductor. A basic forward converter is more complex than a flyback and more efficient. In addition to the basics, designers can choose to include an active clamp for energy recirculation, further enhancing the efficiency of the designs.

References

Magnetics for Power over Ethernet (PoE), Coilcraft
Power-over-Ethernet (PoE) application notes, Eaton
Power over Ethernet PD Power Converter Transformers, Coilcraft
The One−Transistor Forward Converter, ON Semiconductor
Understanding and Designing an Active Clamp Current Mode Controlled Converter, Texas Instruments=
What is Active Clamp Flyback?, Silanna Semiconductor

The post Flybacks, forwards, active clamps, PoE, and USB PD appeared first on Power Electronic Tips.

This FAQ begins with an overview of how The IEEE 802.3 working group defines PoE as part of the Ethernet standard. It then looks at the cabling needed to implement PoE and then discusses the variety of magnetic components required by PoE installations. Magnetics are required to provide the isolation, filtering, and power conversion functions needed to ensure that the primary mission of Ethernet, delivering high-speed and robust data connectivity, is not compromised by the added function of power delivery.

PoE is designed to operate with a nominal voltage of 48Vdc (36 to 57 Vdc), increasing to 73Vdc in some implementations. The standards define both the communication protocols for device identification and the physical requirements for PoE systems. PoE has evolved to address the desire for higher power levels to be delivered to PDs:

• IEEE 802.3af–2003 up to 12.95 W
• IEEE 802.3at–2009 up to 25.5 W (2–pairs, medium power)
• IEEE 802.3at–2009 up to 51 W (4–pairs, high power)
• IEEE 802.3bt–2018 up to 71 W (also called 4PPoE)

Each new revision is backward compatible with earlier standards and sets the maximum power delivered to a PD, taking into account the power loss over a 100-meter-long cable (Figure 1). Cabling requirements also evolved with the increases in power delivery. Cat 5 cabling is the minimum requirement for Type 3 (60W) and Type 4 (90W) PoE.

PoE equipment is categorized by its function in the system:

• Power Sourcing Equipment (PSE) provides power over the Ethernet cable to a powered device, including endspan equipment such as an Ethernet hub or switch
• Powered Devices (PDs) are the various devices being powered.
• Midspan Equipment are power sources between a non–PoE capable switch or hub and one or more PDs. Midspan equipment enables legacy Ethernet systems to be retrofitted to support PoE.

The power delivery architecture of PoE PSEs and midspan equipment is divided into two methods; Mode A and Mode B (Figure 2). PSEs most often use mode A, while Mode B is commonly used by midspan equipment, but there can be exceptions.

Mode A, sometimes referred to as Alternative A, uses two pairs of data pinouts 1-2 and 3-6 to transmit or receive power and data, and the other two pairs 4-5 and 7-8 pinouts are unused.

Mode B, also referred to as Alternative B, sends data using pinouts 1-2 and 3-6.  Power is sent using data pin pairs 4-5 and 7-8.

What magnetic components are needed for PoE?

Magnetic components are used for various isolation, filtering, and power conversion functions in PoE equipment. For example, a high level of electrical isolation is required between devices attached to an Ethernet cable and any circuitry used to transmit or receive data over that cable. The basic Ethernet specification requires that equipment meet one of two electrical isolation tests:

• 1500Vrms at 50 Hz to 60Hz for 60 seconds, applied as specified by IEC 60950.
• 2250Vdc for 60 seconds, also applied as specified by IEC 60950.

While these two tests may seem at odds, they turn out to be compatible. The 1500Vms requirement corresponds to an equivalent of 2121 Vdc. Common practice is to design 2250Vdc isolation into PoE equipment, meeting both isolation tests.

In the signal path, center-tapped 1:1 isolation transformers are used for injecting current, and common mode chokes are used to filter out EMI (Figure 3). Signal path isolation transformers for both the transmit and receive sections are usually incorporated into a single package. Common mode chokes for both transmit and receive sections are also incorporated into a single package. In some cases, the isolation transformers and common mode chokes are all co-packaged.

The dc/dc converter can be a source of common-mode interference in a PoE design. That usually requires using a common-mode choke designed to suppress as many broadband signals as possible from the switching frequency of the dc/dc converter without attenuating the PoE signal.

PoE applications place unique requirements on LAN transformers. Larger wire cross-sections are needed to handle the power, and the magnetic cores must be less susceptible to saturation in the presence of high PoE currents. In addition, PoE common-mode chokes are usually fabricated with trifillar or quadfillar windings, also to void saturation. Pairs of these magnetic components can support higher (double) power levels in higher Class PoE implementations. In some cases, PoE magnetics integrate signal path transformers, and common mode chokes into a single package for the most compact design (Figure 4).

Ethernet jacks with integrated magnetics that support PoE can significantly simplify PCB design efforts and simplify peripheral powering in the total system while still meeting Ethernet data transmission requirements. RJ45 jacks (and assemblies of multiple RJ45s) with integrated transformers and common mode chokes are available in various configurations (Figure 5). Some are compatible with industrial Ethernet systems, such as EtherCAT or Profinet.

PoE power transformers are available that support flyback, forward and active clamp dc/dc converter topologies. Flybacks are better suited for higher voltage/lower current output designs and are more cost-effective, while more efficiency can be achieved with forward and active clamp topologies (Figure 6). Some PoE power transformers include support for synchronous rectification.

Typically, a flyback converter provides power to relatively simple PDs, such as a basic IP camera. At the same time, a forward topology may be needed to support more functionality, such as a tilt/pan/zoom IP camera. Both topologies can support the isolation requirements for PoE. Flyback converters provide the lowest-cost solution for PDs that consume under about 6A. With the proper magnetics, they can produce multiple output voltages. An advantage of a flyback dc/dc converter is that no output inductor is needed. Still, the higher output ripple current results in higher output capacitor cost and only moderate efficiency due to higher peak and RMS currents.

In contrast, the forward topology is the lowest-cost solution for PDs that consume over 6A of current. It is more costly than a flyback since it requires additional components, including an output energy storage inductor and an additional rectifying device (diode or synchronous MOSFET). In return for the greater cost and complexity, a forward converter has a low output ripple current. It uses a smaller capacitor, and this topology delivers higher efficiency because lower peak and RMS currents can utilize synchronous rectification.

The PoE transformer inside the dc/dc converter is isolated like the LAN transformer, typically 2,250Vdc primary to secondary side isolation; separating the SELF and TNV-1 circuits, as required by IEEE 802.3. PoE dc/dc converters tend to be compact devices and have switching frequencies of 200 to 300kHz, occasionally even higher. Some PoE transformers feature split primary and secondary windings to minimize leakage inductance.

Summary

PoE eliminates multiple ac/dc power supplies, reduces installation costs and improves installation flexibility, enables monitoring and controlling of PDs over the same network infrastructure used for data transfer. With PoE, PDs can be turned on or off or reset remotely. A variety of magnetic components are needed to provide electrical isolation, filtering, and power conversion functions in PoE equipment as specified in IEEE 802.3 to ensure that the primary mission of Ethernet to deliver high-speed and robust data connectivity is not compromised by the added function of power delivery.

References

Electrical isolation requirements in Power-over-Ethernet (PoE) power sourcing equipment (PSE), Applied Power Electronics Conference
Magnetics for Power over Ethernet, Coilcraft
The LAN-PoE Connection, Würth Elektronik
Understanding Ethernet Magnetics Features and Design Considerations, Abracon
What Is Power over Ethernet (PoE)?, Cisco

The post Magnetics for Power over Ethernet appeared first on Power Electronic Tips.

and 1000BASE-T Ethernet networks. But the actual power levels that Cat 5e sees when used in these kinds of networking applications is relatively low, on the order of a few hundred milliwatts.

However, Cat 5e is now also used in Power-over-Ethernet applications where it must handle significantly more power than in ordinary networking. In contrast to ordinary Ethernet signals having power in the range of hundreds of milliwatts, the latest power-over-Ethernet standard, called Hi PoE (802.3bt Type 4), allows for delivering between 90 and 100 W through Cat 5e cable to a load. The operating voltage for this mode of operation is 50 Vdc. That implies that the Cat 5e cable is carrying a little less than two amps of current when it operates in a hi PoE environment.

Now you might wonder whether a cable originally designed to carry power levels measuring in the hundreds of milliwatts can safely work in an environment characterized by two amps of dc current and loads dissipating 90 W.

If you examine the cable design specifications, they look fine on paper. Cat 5e cables usually run between 24 and 26 AWG, while Cat6, and Cat6A usually run between 22 and 26 AWG. Smaller AWG numbers mean larger wire diameters and thus less wire resistance. The wire in Cat6 and Cat6a patch cables run slightly thicker than their Cat5E counterparts.

That means the AWG 26 wire in a Cat 5e cable should have a resistance of 0.041 Ω per foot. AWG 26 wire is rated to be able to carry 2.2 A maximum. Because Hi PoE divides its 100 W among the four twisted pairs in the network cable, that means any one AWG wire in a Cat 5e cable should be carrying a half-amp at most.

The margin is a bit smaller in one of the newest types of Ethernet cables called Slim Run Patch Cables. These generally contain 28 AWG wire but there are versions available on Amazon with 30 AWG. Use of smaller twisted pair wires lets these cords be at least 25% smaller in diameter than standard Cat5e, Cat6, and Cat6a cable, but the higher AWG limits the length of the cable. The current-carrying capacity of AWG 28 and 30 wire is 1.4 and 0.86 A respectively.

We should also point out that PoE equipment employs current limiting to protect PSE (power sourcing equipment) from overload and to quickly disable malfunctioning PDs (power devices). The current drawn on each enabled port is continuously monitored, and power is disconnected if it rises beyond an allowed setable limit. However, this safety feature may not prevent a network cable from heating up and perhaps melting until there is a short. It also doesn’t help much if a cable fails open.

To check things out for ourselves, we put ordinary Cat 5e cables that came with some consumer electronics into a simulated Hi PoE scenario. We assembled two RJ45 sockets and wired them to route power through the cable as defined by Hi PoE, with each of the four twisted pairs in the cable sharing the dc load. We then connected one end of the cable to a 50-V power supply and the other end to a load resistor that would draw close to the two-amp maximum involved with Hi PoE. Then we just waited and checked the temperature of the cable in various spots over the course of the test, paying particular attention to bends in the cable where there could conceivably be some compromise to the cable conductors.

Though this was an informal test, we figured any obvious cable problems were likely to show themselves after a few hours of handling a Hi PoE load close to the maximum spec. We’re happy to report that absolutely nothing happened. We checked all along the cables we used with a spot thermometer numerous times. We never saw even a one-degree temperature rise in our cables, even where they were bent, and even after a couple of hours of handling the max Hi PoE power level.

All in all, we’re pretty sure that problems due to poor cable construction would cause a temperature rise sooner rather than later, so we think it’s probably OK to put Hi PoE power levels through ordinary Cat 5e cable.

That said, we should point out this wasn’t an exhaustive test. We only ran our simulations for a few hours, not for days or weeks. And testing took place at room temperature whereas the maximum operating temperature for Cat 5e cable is 167°F. Further, our test setup divided the dc current through the cable evenly among the four twisted pair. It is possible to envision Hi PoE scenarios where current doesn’t divide evenly, potentially exceeding the 2.2-A limit of any one twisted pair. The situation would of course be worse for Slim Run Patch Cables.

The post How does Cat 5e cable handle Hi PoE? appeared first on Power Electronic Tips.