It feels like just yesterday we were marveling at how our Ethernet cables could do more than just connect us to the internet – they could actually power our devices. Now, three years on from the official nod to IEEE's latest standard, known as PoE 2 or 802.3bt (and formerly PoE++), the momentum is undeniable. Even with the shift towards remote work, the deployment of Power over Ethernet ports continues to climb year after year.
Think about it: offices are being reimagined. Employers are looking at those empty desks and seeing opportunities to upgrade their IT infrastructure, building workplaces that are ready for whatever comes next. And to truly create a 'smart' office, you need a whole ecosystem of internet-connected IoT devices – think digital signage in meeting rooms, sophisticated conferencing equipment, and all sorts of sensors. The benefits are pretty compelling: energy savings, streamlined operations, and, perhaps more importantly these days, enhanced safety for everyone in the building. The pandemic really accelerated our need for controllable HVAC systems and touchless public amenities, pushing facilities and IT managers to collaborate on deploying PoE-enabled systems.
Market watchers are predicting a significant surge, with global shipments of switch/PoE ports expected to surpass 150 million by 2025. What makes PoE 2 so special? When it was approved back in 2018, it could deliver up to 71.3W to a powered device (PD), nearly tripling the power of previous standards (25.5W). This means the same gigabit Ethernet cable can now carry substantially more power, laying the groundwork for a host of applications, both current and future, that demand serious wattage alongside data. We're talking about things like remote temperature monitoring systems and infrared cameras at entry points, ready to screen people before they even step inside.
Historically, a single power channel was enough for each PoE port. But with 802.3bt, we're looking at two power channels per port for medium and high power levels, and the need to pack more power density into each channel. Across the global Ethernet market, the number of PoE-enabled ports is steadily increasing. All these factors are pushing IT departments to deploy systems that are not only high-power and high-port-count but also demand that critical 99.999% uptime and reliability. The demand for a truly scalable PoE subsystem that simplifies the deployment of high-port-count, PoE-enabled switches has been building for a while.
A Platform-Based Approach to PSE Design
Modern network switches are incredibly complex, often operating in challenging environments with surges and electrical discharges, all while needing to maintain exceptional reliability. The old way of designing Power Sourcing Equipment (PSE) subsystems involved looking at individual components and making incremental improvements. But by stepping back and viewing the PSE subsystem at a higher level, design teams can rethink the entire paradigm and offer system-level solutions. This is precisely the approach taken with the LTC9101/LTC9102/LTC9103 chipset and its future iterations. It brings together digital and analog components to tackle the challenges system integrators face.
This chipset is part of a self-isolated PSE controller family, built on a substrate specifically for PoE 2 systems. It simplifies the architecture by placing the LTC9101 on the non-isolated side, providing an isolated digital interface to the PSE host. Meanwhile, multiple LTC9102 and/or LTC9103 devices handle the high-voltage analog Ethernet interface on the isolated side. This clever arrangement replaces several expensive optocouplers and an isolated power supply with a more cost-effective and reliable 10/100 Ethernet transformer. The result? A more stable, reliable, and easier-to-manufacture PSE design.
This scalable solution offers flexibility, supporting large PSE systems with port counts ranging from 4 to 48, depending on the power requirements per port. Each design needs at least one LTC9101 digital controller and one or more LTC9102/LTC9103 analog controllers. The LTC9102 offers 12 power channels, capable of powering anywhere from 12 x 30W ports to 6 x 90W ports. Similarly, the LTC9103 provides 8 power channels, supporting 8 x 30W ports down to 4 x 90W ports. A single LTC9101 can manage up to four LTC9102 and/or LTC9103 devices, allowing for a mix-and-match approach. For instance, you could build a 24-port PSE with 4 x 90W ports and 20 x 30W ports using a combination of these chips.
IT and facilities managers will appreciate the LTC9101's advanced digital features, including onboard eFlash for firmware updates and custom configurations, backward compatibility with previous PSE drivers, and an I2C serial interface. The firmware is stored in a dedicated flash partition, with two complete copies maintained for maximum data protection. Once the chipset boots, users can configure and communicate with it via the I2C interface. Each port can be individually set to one of four PSE operating modes (auto, semi-auto, manual, or off), and system power can be managed using telemetry readings for port current, PoE voltage, and power.
The LTC9102/LTC9103 devices, acting as the 'branches' to the LTC9101's 'core,' ensure high efficiency and durability in the high-voltage power paths. Each power channel features dedicated detection and classification hardware, allowing all ports to detect, classify, and power up simultaneously, significantly reducing power-up latency. Unlike less advanced PSEs that can be susceptible to delays from the powered device, these chips use external MOSFETs for each power channel, giving users the flexibility to select low Rds(on) devices for reduced power loss and fault tolerance. Using 0.1Ω sense resistors further minimizes power consumption.
In the event of an overcurrent fault or port short circuit, the LTC9102/LTC9103 can quickly disconnect power in about 1µs, protecting the PSE, MOSFETs, and downstream circuitry. Crucially, all port-facing pins can withstand voltage transients of up to +80V or down to –20V without damage. The chipset is designed to operate under surges exceeding ±6.5kV, meeting IEC 61000-4-5 surge immunity specifications with minimal external components. After a fault, the devices can safely re-enable the MOSFETs in a current-limited fashion, minimizing interruptions to the PD and maximizing network uptime.
PoE 2 Topology, Detection, and Power Classification
PoE 2 introduces two distinct PD characteristic configurations: single-signature and dual-signature PDs. A single-signature PD shares the same detection and classification characteristics across its pairsets, while a dual-signature PD has independent characteristics on each pairset, allowing for completely separate classification and power allocation. Dual-signature solutions are more complex and costly. It's important to note that despite sharing a similar architecture, 802.3bt dual-signature PDs are not the same as previous UPoE devices.
The LTC9101/LTC9102/LTC9103 chipset supports a robust PoE 2 PD detection process, including a new connection check sub-process to determine which PD signature configuration the PSE is connected to. Beyond just checking the connection, the devices verify that the connected PD is IEEE-compliant. While IEEE standards allow for either a 2-point voltage or 2-point current detection scheme to identify valid PD signatures (25kΩ), these devices employ both simultaneously for a more stable and reliable approach. This multi-point detection mechanism helps eliminate false positives and prevents damage to network equipment not designed for PoE DC voltage.
PoE 2 powers two pairs of conductors (4 wires) for up to 25.5W and four pairs (8 wires) for up to 71.3W. This not only enables higher power levels but also improves efficiency for older, lower-power systems. By using more conductors, power loss in the cable is halved. For example, to deliver 25.5W to a PoE 1 PD, a PoE+ PSE typically needs to supply 30W due to a 4.5W loss over a 100m CAT5e cable. With PoE 2 powering the same PD using four pairs, losses can be reduced to below 2.25W, boosting total power transmission efficiency from 85% to 92.5%. Considering the vast number of PoE PDs globally, this translates to significant power savings and a potential reduction in carbon emissions of 7.5% in many use cases.
PoE 2 introduces four new high-power PD classifications, bringing the total number of single-signature classifications to nine. Classifications 5 through 8 are new with PoE 2, corresponding to PD power levels from 40W to 71.3W. PSEs can still choose to classify via the physical layer (e.g., 5-event classification for 71.3W) or the data link layer (Link Layer Discovery Protocol, LLDP). PDs must support both classification schemes to be compliant. Remember, with dual-signature PDs, each pairset can have a different classification, allowing for combinations like Class 1 on one pairset and Class 2 on another, resulting in a dual-signature Class 1 and Class 2 (10.3W) PD.
An optional extension to physical layer classification in PoE 2 PDs, called Autoclass, allows the PoE 2 PSE chipset to measure the actual maximum power drawn by the connected PD. This power management feature enables the remaining power to be allocated to other devices if a particular PD is consuming less than its classified power due to lower brightness settings or shorter cable lengths.
Of course, PoE 2 is backward compatible with older 25.5W and 13W PoE 1 standards. Lower-power PoE 1 PDs can connect to higher-power PoE 2 PSEs without issue. When the situation is reversed – a higher-power PoE 2 PD connecting to a lower-power PoE 1 PSE – the PD can operate at a negotiated lower power state, known as downgrading. If a PD ignores downgrading and operates at its maximum power, it can cause the PSE to repeatedly cycle on and off, hitting its current limit and shutting down, leading to low-frequency parasitic oscillations. Both PoE 1 and PoE 2 PDs require downgrading, but unfortunately, it's often overlooked in many implementations.
Highly Efficient PDs
Analog Devices, along with Maxim Integrated (now part of ADI), offers a range of specialized ICs to maximize PoE 2 PD performance. The LT4321, an active diode bridge controller, replaces the traditional diode bridge rectifier. It uses low-loss N-channel MOSFETs, which not only increases the available power to the PD but also reduces heat dissipation. Since PoE 2 requires PDs to accept DC power of any polarity at their Ethernet input, the LT4321 smoothly rectifies and combines power from both sets of data line pairs, achieving end-to-end efficiency exceeding 94% and operating across a wide temperature range.
