Protecting industrial single-pair Ethernet deployments
Since its development 40 years ago, Ethernet has quickly established itself as the defacto wired networking protocol. However, deploying Cat 6-style Ethernet cables and connectors in space-constrained applications proved challenging, leading to the development of single-pair Ethernet (SPE).
This article originally appeared in the Nov'23 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.
This article will briefly introduce SPE and investigate the challenges associated with deploying high-speed networking in electrically noisy industrial environments.
The Ethernet legacy emerges from the data centre
Initially developed in the 1980s, the Ethernet IEEE 802.3 networking protocol standard has become a trusted, highly reliable, resilient, and convenient method of connecting servers, computers, and data communication peripherals. Over the decades, Ethernet has gone from strength to strength and advanced in both bandwidth and use cases. Predictably, the networking architecture that connected an organisation's financial and operational systems would one day extend down to interconnect the manufacturing equipment. Industrial operational improvement initiatives such as Industry 4.0 demand robust connectivity, so Ethernet was ready when the technological concepts of the Internet of Things (IoT) became a reality.
Industry 4.0 and Industrial IoT (IIoT) are not the first technology deployments aimed at improving manufacturing efficiencies and effectiveness. Connecting manufacturing equipment and capital-intensive production assets has been in place for over four decades. Serial networking methods such as Modbus, RS232, RS422, and Profibus have existed for a long time. However, the IIoT has brought more immediacy to analysing data, requiring faster data transfer rates and a significantly larger utilisation of connected sensors, actuators, and control systems.
The development of single-pair Ethernet
Ethernet's capabilities made it a perfect candidate for various applications outside its legacy IT roots. Advancements such as industrial Ethernet delivered real-time, time-sensitive networking with deterministic behaviour and the ability to transport many of the legacy industrial control protocols highlighted above. Ethernet is also gaining adoption as the backbone for in-vehicle networking.
However, while Ethernet's flexibility, robustness, and simplicity are unquestioned, the physical cabling requirements for four twisted pairs and the widespread RJ45 connector are unsuitable for industrial use. In the industrial domain, with challenging environmental conditions and space-constrained control cabinets, the Cat6-style Ethernet cables are too thick and unable to accommodate the tight bending radius of metal conduits. In addition, the deployment of vast armies of IIoT Edge devices required a reliable source of power, which Power over Ethernet (PoE) can deliver but only using the eight conductors.
In 2019, the Ethernet Alliance announced the Ethernet solution that would satisfy all the cabling challenges and ensure Ethernet's future success in industrial and automotive applications.
Figure 1. The architecture of a SPE connection highlighting critical protection and EMI mitigation components. (Source: Bourns)
The gigabit SPE IEEE 802.bp 1000BASE-T1 standard provides gigabit connectivity using an unshielded cable consisting of a single pair of conductors. Compared to the traditional four-pair Cat 6 cable, an SPE cable is 60% lighter and significantly less in diameter. Most importantly, the SPE Power over Data Line (PoDL) standards 802.3bu and 802.3cg permit a maximum of 52 watts transferred over the single pair. In place of the RJ45 connector, SPE can utilise the popular industrial-grade IP65/67 rated M8 and M12 circular connectors.
Figure 2. A common mode noise signal on differential signal lines or via power rails. (Source: Mouser)
Networking communication challenges
Despite Ethernet's excellent link robustness characteristics, deploying any networking in an electrically noisy environment presents challenges. Lost packets increase the bit error rate, requiring resending, resulting in lower throughput and increased latency. Electromagnetic interference (EMI) and electrical transients are common sources of noise that can significantly impact high-speed data transmission across the network. Large motors, actuators, and variable frequency drives are likely sources of radiated (typically above 1MHz) or conducted EMI noise, reaching several volts in an industrial environment. Despite the twisted pair nature of the SPE cable, it is inevitable that with control systems close to electrically operated equipment, induced noise will result.
Figure 1 illustrates the architecture of a typical SPE path between two devices, for example, a switch or microcontroller host (left) and an edge-node sensor (right). The switch includes a power over data line injection point - power supply equipment (PSE). The sensor extracts the power from the twisted pair - power delivery (PD).
There are several categories of conducted and radiated electrical disturbances and noise artefacts that networking cables and interfaces may encounter. Figure 2 illustrates a conducted common mode noise signal superimposed on a differential signal. Common mode noise may also appear on the positive and negative power supply rails. Common mode signals flow in the same direction and typically have a return path via an earth (ground) connection and stray capacitances.
Common mode chokes, consisting of two windings around a ferrite core, are a convenient way of cancelling the common mode noise but allowing the required signals to pass unheeded.
Differential mode noise, also known as normal mode noise, signals flow in opposite directions (see Figure 3). Again, a suitable filter arrangement comprising an inductor and capacitor, or a differential mode choke will suppress differential mode noise. Electrostatic discharge (ESD) on networking cables and control equipment can damage semiconductors and other components. These transient voltage spikes (high dV/dt) can induce dangerously high voltages onto networking cables. A transient voltage suppression (TVS) component, such as a diode array, protects sensitive circuitry.
Implementing SPE protection
Figure 1 illustrates the recommended EMI Figure 2. A common mode noise signal on differential signal lines or via power rails. (Source: Mouser) COMMS: ETHERNET DESIGN 39 ELECTRONICSPECIFIER.COM filtering and ESD protection components for a SPE deployment.
A discrete chip LAN transformer (1), such as the Bourns SM4532xx series, is a small centre-tapped 1:1 transformer constructed on a compact drum core that can also be used as a common mode inductor. The SM4532 series is 802.3 compliant and offers a high degree of PCB mounting flexibility with its small dimensions of 4.7 × 3.3 × 2.9mm. It has a low insertion loss characteristic of typically -2dB up to 500MHz, features 1,500VAC (HiPot) isolation for 60 seconds and can be paired with a common mode chip inductor for EMI reduction. The SM4532 SPE series comprises six variants, accommodating 10Base-T1 to 1000Base-T1 PoE.
For PHY (chip) side common mode noise rejection (2), a common mode choke, for example, the Bourns SRF3216A common mode chip inductor, is ideal. Rated at 50VDC and capable of withstanding 125VDC, it is shielded and constructed with bifilar windings around a ferrite core. Typical DC resistance is 0.15Ω to 1.1Ω and a common mode impedance of between 90Ω to 2,200Ω (at 100MHz) and is part dependent. Differential mode impedance is typically less than 10Ω at 100MHz. Another PHY side common mode inductor example is the SRF2012AA series. The SRF2012AA has a common mode impedance range from 67Ω to 360Ω measured at 100MHz, a DC resistance of typically 0.35Ω, and is rated up to 400mA.
An example line side (3) common mode choke is the Bourns SRF6545A. This inductor features a common mode rejection of -43dB at 100MHz and a differential to common mode rejection of -50dB at 100MHz. Insertion loss at 100MHz is typically less than -3dB, and the choke is rated up to 350mA.
A dual differential mode choke (4) reduces differential mode noise in the power supply or power delivery circuits. An example is the Bourns SRF1260A series of dual-winding shielded power inductors. The SRF1260 series can be configured in a parallel or series configuration and is available in a range of inductances, from 0.47µH to 4,000µH.
Figure 3. Differential mode noise follows in opposite directions. (Source: Mouser)
dual-winding shielded power inductors. The SRF1260 series can be configured in a parallel or series configuration and is available in a range of inductances, from 0.47µH to 4,000µH.
For ESD protection, a diode array, such as the surface mount Bourns CDDFN6-3312P, provides up to 8kV, a minimum breakdown voltage of 4.5V, and a peak reverse voltage of 3.3V. The CDDFN6 is constructed in a DFN6 feed-through package measuring 1 × 1.2 × 0.45mm and has low capacitance specifications of 0.04pF (I/O to I/O) and 0.18pF (I/O to ground).
Building reliability and robustness into SPE deployments
SPE brings decades of reliability and respect for 802.3-based network connectivity gathered in enterprise IT and telecoms markets to emerging industrial and automotive applications. Maintaining link reliability and robustness without impacting bandwidth and latency depends on implementing trustworthy EMI filtering and ESD protection. In this short article, we've highlighted some of the disruptive sources of electromagnetic noise and voltage transients that impact link performance. The protection components highlighted in this article are all available from Mouser, an authorised Bourns distributor.