Creating the Power Subsystem for Power Over Ethernet Designs
With Power over Ethernet (PoE) technology, we can send data and power over existing CAT5/5e/6 Ethernet cables. PoE systems have proven beneficial for applications like surveillance cameras, home/building controls, digital signs, VoIP phones within enterprises, and Wi-Fi access points. This application note provides guidance on PoE system design, with special focus on the powered device (PD) power subsystem. A similar version of this application note originally appeared in the June 2019 issue of Circuit Cellar; see pages 38-41.
PoE systems consist of some form of power sourcing equipment (PSE), such as a network switch or a midspan injector switch, along with a PD as shown in Figure 1.
Figure 1. PoE system with power sourcing equipment at the top and powered devices at the bottom.
PoE is beneficial for reducing the number of cables running to the end equipment. Using a single Ethernet cable simplifies installation and also provides for centralized power management – the PDs can be remotely powered off and on and can provide continuous operation in spite of an AC power outage if the PSE has uninterruptible power source (UPS) power.
Of course, there are limitations to the amount of power that can be efficiently provided to an end-point and the distance of that end-point from the network switch that provides the power. The maximum distance specified is 100m (333 ft); this is the distance from a PoE-capable switch to the PoE PD. A PoE Ethernet extender, however, can lengthen that span.
Power is delivered using the twisted data wires in an Ethernet cable. Each Ethernet cable has four pairs of data wires. In Gigabit Ethernet (by far the most common type), all four pairs carry data. Power can be delivered using either two pairs of the wires as shown in Figure 2. One is called Alternative A and the other is called Alternative B.
To be IEEE standard compliant, a PD must support both Alternative A and Alternative B, whereas a PSE can support either Alternative A or Alternative B, or both.
Figure 2. PoE ecosystem showing power connections for Alternative A and Alternative B.
The IEEE standard dictates the amount of power that can be delivered to the end-point:
- IEEE802.3af, the first standard ratified in 2003, was rated to put out 15.4W/port using two pairs of wires in the Ethernet cable. At a 100m distance, this means that a PD would get 12.95W of power.
- In 2009, the IEEE ratified the PoE+ standard – 802.3at. With this new standard, PDs can get 25.5W of power delivered at 100m. This standard is backwards compatible so that the older PDs would work with the new PSE.
- IEEE 802.3bt, approved in September 2018, delivers 71W to a PD over 100m. Now, the PSE can put out 100W over a single Ethernet cable, which helps expand the PoE market to LED lighting, large-screen displays, etc.
The voltage range sourced by the PSE and received by the PD is shown in Table 1.
Table 1: Voltage Range Sourced by PSE and Received by PD
|IEEE802.3af (V)||IEEE802.3at (V)||IEEE802.3bt (V)|
|Output by PSE||44–57||50–57||50–57|
|Received by PD||37–57||42.5–57||42.5–57|
PoE System Design: PD Power Subsystem
The power subsystem of a generic PD may look like the image shown in Figure 3.
Figure 3. PoE PD Power Subsystem. Top: Non-Isolated. Bottom: Isolated.
The power subsystem of a PD comprises of a PD interface controller that accepts power from the Ethernet cable and a DC-DC converter that then regulates the power down to the supply rails required for circuit functionality. Table 2 provides the maximum power amounts available for the PD to draw from the PoE connector. The PD must classify itself, via the PD interface controller's classification, to receive the right amount of power according to its class. More details on PD classification IEEE802.3bt standard.
Table 2: Power Draw from PoE Connector
|PD Class||Maximum Power PD Can Draw PClass_PD (W)|
Before choosing a suitable power solution, answering these questions can help you determine your PD requirement:
- How much power does your PD need?
Knowing the maximum power your PD requires, choose the appropriate class to match this need. It is good practice to not overclassify your PD power. One reason is that higher power adds cost to your power solution. Another reason is that it would reduce the remaining available power the PSE can allocate to other PDs connected to the same PoE network.
- Is isolation needed?
PDs and PSEs provide isolation between all accessible external conductors, including frame ground (if any) and all media-dependent interface (MDI) leads including those not used by the PD or PSE. There are two electrical power distribution environments to be considered that require different electrical isolation properties: Environment A: When a local area network (LAN) or LAN segment, with all its associated interconnected equipment, is entirely contained within a single low-voltage power distribution system and within a single building. A multiport network interface device (NID) complying with Environment A requirements does not require electrical power isolation between link segments. Environment B: When a LAN crosses the boundary between separate power distribution systems or the boundaries of a single building. In this environment, equipment with multiple instances of PSE, PD, or both should meet or exceed the isolation requirement of the medium attachment unit /physical layer (MUA/PHY) with which each is associated. To state it in a simple way, if your PD is a single device without any external connector and is fully enclosed in plastic casing (e.g. security camera, PoE-LED bulb, low-cost IP phone, etc.), then isolation is not required. In this case, choose a non-isolated power solution for simplicity and lower cost.
- Does your PD need to also be powered from a wall adaptor?
An IP phone, for example, is most likely to have an AC power adaptor input for usage in a building where PoE is not yet available. If your PD is to be used where PoE is not yet available, then choose a PD interface with a wall adapter interface feature.
- Does your PD need low-power standby mode to comply with some agency requirements?
More and more agencies are requiring green power features where equipment such as an IP phone during idle time (not used during the day) and sleep time (outside of work hours) consumes as little power as possible. Pick a PD interface controller whose features include maintain power signature (MPS) and low-power sleep mode to guarantee compliance and also help contribute to a greener world.
- Is high efficiency important?
High efficiency results in lower power loss, easing the heat dissipation requirement for your PD. Lower heat dissipation means lower operating temperature, which translates to higher reliability. In cases where your PD needs a lot of power to operate, 60W for example, an 80% efficiency PD will require an input power of 60W/80%=75W, which exceeds the maximum PoE power of 71W, rendering it non-PoE compatible. However, a 90% efficiency PD requires input power of 60W/90%=67W, which falls nicely under class 8 (71W). In this situation, high efficiency is a must. Also, it is always desirable to classify your PD at the lowest power class, so that the remaining PoE power in the system can be allocated to more PD devices. High efficiency could bring your borderline PD to the next lower power class.
Selecting the PD Interface Controller
When selecting a PD interface controller, consider the following important features:
- IEEE 802.3af/at/bt compliance
- Type 1~4 PSE classification indicator or an external wall adapter indicator output
- Simplified wall adapter interface
- Multi-event classification 0–8
- Intelligent MPS
- Sleep mode and ultra-low-power sleep mode
These are features that would satisfy most of the PD requirements mentioned before. The rest of the requirements will be addressed by the DC-DC controller, which will be discussed later. The selector guide presented in Table 3 shows some recommended PD interface controllers and their key features.
Table 3: Recommended PD Interface Controllers
|802.3af/at Compliant||CoC Compliant||70W|
Selecting the Non-Isolated DC-DC Converter
If your PD doesn't require isolation, then a high-voltage buck converter would be an appropriate choice for your DC-DC converter need. Efficiency, total solution size, and cost are important considerations. Devices with features such as synchronous rectification, a wide input voltage range, and a high level of integration support these considerations. Figure 4 shows an example of a non-isolated DC-DC converter solution that addresses efficiency and size requirements. This is a class 3 PD. The MAX5969B is the PD interface and the MAX17503 is the DC-DC buck converter (non-isolated). The output is 5V/2.5A with 92% peak efficiency.
Figure 4. Class 3 PD, non-isolated using the MAX5969B and the MAX17503.
Figure 5 provides an example of another non-isolated DC-DC converter, a class 1 solution using the MAXM15064 uSLIC power module at 5V/300mA. The MAXM15064 is in a tiny uSLIC 10-pin 2.6mm×3.0mm×1.5mm package.
Figure 5. Class 1 PD, non-isolated using the MAX5969B and a tiny uSLIC power module, the MAXM15064.
Selecting the Isolated DC-DC Controller
If your PD requires isolation, then a flyback converter would be appropriate, providing power of approximately 40W (class 5 and below). A device that minimizes the number of components required can save board space and cost. For example, an isolated flyback controller that doesn't require an optocoupler to provide feedback for output voltage regulation can save multiple external components and the associated board space and cost. Furthermore, an optocoupler degrades overtime, so not using one also results in higher reliability.
Figure 6 shows an example of an isolated DC-DC converter solution. It is a class 2 PD, 5V/1A output, and can operate with 12V to 57V wall adaptor input voltage. The solution uses the MAX5969B as the PD interface and the MAX17690 for the no-opto flyback DC-DC converter.
Figure 6. Class 2 PD, isolated using the MAX5969B and the MAX17690, featuring no optocoupler.
To further improve the efficiency of the flyback DC-DC converter, you can replace the output rectification diode with synchronous rectification. The schematic in Figure 7 shows an example of this.
Figure 7. Class 2 PD, isolated using the MAX5969B and the MAX17690, with the MAX17606 optional output synchronous FET driver for highest efficiency.
For output power beyond 40W, even though the flyback converter can still work fine, an active clamp-forward converter is recommended for higher efficiency. At higher output power, efficiency is very important to reduce the amount of heat dissipation in the PD. An active clamp-forward converter also has a lower electromagnetic interference (EMI) signature due to its soft switching edges. Figure 8 presents an example of an active clamp-forward DC-DC converter. The converter delivers an isolated output voltage of 57V at 700mA, totaling 40W at peak efficiency of 91.5%.
Figure 8. Class 5 PD, isolated high power using the MAX5969B and the MAX17599, active clamp-forward DC-DC for high efficiency and low EMI.
As more devices are networked, running them on just one Ethernet cable that provides both connectivity and power is a convenient option. Besides, having the power delivered via a centrally managed switch allows for other enhancements such as remote ON/OFF and uninterrupted operation even in the case of a local power outage.
Thanks to the newest IEEE 802.3bt standard, even more types of devices can now be powered via PoE. As the amount of power delivered increases, so does the need for higher efficiency and wider input voltage ranges to accommodate the complex power system designs. This requires careful consideration regarding the selection of both the PD controller as well as the DC-DC converter to generate the required regulated voltage to be used by the system.
- PoE Chipsets Market Size Worth $1.22 Billion by 2025 | CAGR: 12.6%: Grand View Research, Inc. 14 Aug. 2018, www.prnewswire.com/news-releases/poe-chipsets-market-size-worth-1-22-billion-by-2025-cagr-12-6-grand-view-research-inc--843035168.html.