How to Extend the Industrial Edge with SPE and PoDL

Designers have growing opportunities, and perhaps increasing pressure, to add more sensors and intelligence at the industrial edge. Industrial systems are moving toward continuous monitoring, predictive maintenance, and distributed decision-making. While instrumenting at scale can be difficult with traditional architectures, a convergence of technologies is making it more practical to deploy intelligent edge nodes with far simpler physical and system-level infrastructure.

The ability to embed intelligence directly at the point of measurement is becoming just as important as centralized control, enabling industrial systems to respond in near-real time rather than through delayed, aggregated data. Problems such as equipment degradation, energy waste, and process drift typically emerge at the machine level, where local sensing and processing can identify anomalies before they propagate through the broader system.

Traditional industrial Ethernet architectures are typically centralized, gateway-driven, and heavily reliant on cables. Architecturally, sensors and actuators connect through separate power wiring, local I/O modules, or fieldbus networks that aggregate data before it reaches the broader network.

That traditional model has worked well for factory automation and control, but it's not as effective for embedding intelligence at the edge. With each new sensor, designers must consider additional wiring, power infrastructure, gateways, and integration complexity. Too often, valuable data stays trapped at the periphery of systems that could benefit from real-time visibility and local intelligence.

Conventional Ethernet physical layer (PHY) transceivers are often unsuitable for resource-constrained edge sensors due to their high pin count and power consumption. These requirements exceed the tight thermal envelopes of compact, sealed IP67 sensor enclosures. Additionally, many low-power microcontrollers do not include Ethernet media access control (MAC) functionality, making conventional PHY implementations impractical for compact edge nodes.

Extended data and power runs

Designers can advance industrial edge applications with a combination of Single-Pair Ethernet (SPE) and Power over Data Lines (PoDL) for Ethernet communications and power delivery on a single lightweight pair. This simplifies installations and reduces the costs of enabling Ethernet connectivity. By comparison, conventional Power over Ethernet (PoE) deployments are generally optimized for shorter Ethernet links and higher-power endpoint devices such as wireless access points, IP cameras, and switches.

The SPE/PoDL approach makes it practical to network smaller edge devices directly rather than routing everything through layers of intermediate infrastructure. Instead of concentrating intelligence and connectivity at centralized control points, designers can distribute intelligent sensing nodes much closer to machines, processes, and physical assets.

Utilizing that lighter-weight infrastructure with MEMS sensors and devices that integrate Ethernet PHY and MAC functions into a single component, designers can connect compact, lower-power industrial edge devices and intelligent sensors directly to Ethernet networks, enabling simpler, more cost-effective distributed sensing systems.

Analog Devices, Inc. (ADI) offers such a device, the ADIN1110 transceiver (Figure 1). Designed to enable systems that provide operational visibility at the edge, the low-power, single-port MAC-PHY component implements the 10BASE-T1L long-reach SPE standard for data and power runs extending up to 1,700 meters at 10 Mbps. Standard Ethernet is typically limited to 100-meter cable runs.

Figure 1: ADI's ADIN1110 transceiver incorporates MAC and PHY functions to enable data and power runs up to 1,700 meters. (Image source: Analog Devices, Inc.)

The 10BASE-T1L Ethernet standard facilitates seamless connection to converged IT/OT networks, providing maintenance teams with access to real-time asset health information from anywhere, at any time. That's essential for the practical implementation of applications such as predictive maintenance, energy monitoring, environmental sensing, robotics, and modular automation.

ADIN1110 nodes are designed for distributed intelligence and condition-based monitoring in industrial environments. Each node operates from a single power supply rail of either 1.8 V or 3.3 V.

This architecture eliminates the need for the host processor to have an integrated MAC, allowing system designers to select simpler, lower-power microcontrollers, like the MAX78000 or MAX32670, to achieve the lowest possible overall system-level power consumption. Consequently, designers can streamline Ethernet integration while reducing overall system complexity and power consumption.

The device incorporates four distinct power supply domains: AVDD_H, AVDD_L, DVDD_1P1, and VDDIO. Although it can operate on a single supply (e.g., 3.3 V), segregating these domains enables developers to optimize power dissipation by using lower-voltage external rails.

The active power consumed by the transceiver varies based on the transmit amplitude and the power supply configuration selected:

• With a 1.0 V p-p dual supply, the power consumption is 42 mW.
• Using a 1.0 V p-p single supply, the power consumption is 50 mW.
• With a 2.4 V p-p triple supply, the power consumption is 78 mW.
• Using a 2.4 V p-p dual supply, the power consumption is 87 mW.
• Using a 2.4 V p-p single supply, the power consumption is 119 mW.

ADI’s transceiver incorporates on-chip cable diagnostics, including a time-domain reflectometry (TDR) engine. The engine allows for the detection and location of cable faults, thereby improving signal integrity over extended distances.

These capabilities, combined with the advantages of low-power microcontrollers and MEMS sensors, enable the development of more cost-effective Ethernet-connected sensor nodes with real-time monitoring capability.

In cases where deployments involve dozens or even hundreds of interconnected devices, designers need to implement intelligent traffic forwarding between network segments. The ADIN2111 (Figure 2) addresses that need by combining dual 10BASE-T1L interfaces with integrated switching capabilities, enabling daisy-chain and other multi-node network topologies.

Figure 2: An ADIN2111 Ethernet switch with two 10BASE-T1L PHYs can be used to extend network connectivity with intelligent traffic forwarding between network segments. (Image source: Analog Devices, Inc.)

Simplifying industrial sensor deployment

Beyond reducing the number of components and power consumption, SPE significantly simplifies the deployment of large-scale sensors in industrial environments.

Traditional condition-monitoring installations often necessitate separate Ethernet, fieldbus, and power cabling routed through conduit, junction boxes, and local control cabinets. Consequently, installation labor and cable management frequently become a more significant cost factor than the sensors themselves.

SPE’s ability to combine Ethernet communications and power delivery onto a single twisted pair enables designers to minimize cable bulk and simplify routing through limited industrial spaces. Additionally, the long 10BASE-T1L link distances allow sensors to be deployed across expansive facilities without the need for additional switches, media converters, or remote I/O infrastructure.

These characteristics are particularly advantageous in retrofit applications where adding new monitoring points to existing equipment may otherwise necessitate substantial rewiring or control-system modifications. Instead of redesigning the broader automation architecture, engineers can often add distributed sensing nodes directly at the machine level while maintaining connectivity to standard Ethernet-based networks.

CbM application design

A typical SPE vibration sensor for condition-based monitoring (CbM) combines four key design elements: MEMS sensing, low-noise PoDL power extraction, mechanically robust packaging, and lightweight edge networking software.

Traditional industrial vibration monitoring systems often relied on bulky single-axis piezoelectric sensors. In contrast, modern SPE edge nodes increasingly use integrated triaxial MEMS accelerometers such as ADI’s ADXL357 series. A triaxial MEMS device can simultaneously monitor vibration across multiple axes, allowing the system to detect conditions such as bearing wear, shaft imbalance, or rotor eccentricity without requiring precise sensor orientation.

To minimize installation complexity, many SPE sensors derive both data and power from the same twisted-pair cable using Power over Data Lines (PoDL).

In a typical implementation, coupled inductors and AC-coupling capacitors separate the DC supply from the Ethernet data path. The extracted 24 V rail is then converted to the lower voltages required by the sensing electronics. To reduce noise, designers commonly pair a high-efficiency buck converter such as the LT8618 with an ultralow-noise LDO such as the LT3042. This combination maintains efficient power conversion while delivering a clean supply rail to the MEMS sensor and analog circuitry.

The sensor enclosure can significantly affect measurement accuracy. If the housing resonates within the operating bandwidth of the accelerometer, it can introduce distortion or false vibration signatures into the captured data. Rigid aluminum or steel enclosures with secure mounting points help ensure that machine vibration is accurately transferred to the sensing element without adding structural artifacts.

The ADIN1110 system architecture accommodates multiple types of sensors, including temperature, pressure, sound, and position. Analog or digital output can be implemented with minimal changes to the microcontroller firmware.

Conclusion

By integrating the 10BASE-T1L capabilities of the ADIN1110 MAC-PHY with SPE networking, PoDL power delivery, and compact MEMS sensing, designers can deploy Ethernet-connected monitoring nodes much closer to industrial assets without the infrastructure overhead associated with traditional architectures. The result is more scalable condition-based monitoring and greater operational visibility across distributed industrial systems.