In a world with increasing reliance on wireless communications, the wired world seems rather passé. However, in the industrial IoT (IIoT), wires are still the norm. There are several reasons for maintaining wired connectivity in the industrial setting including RF interference, crowded radio bands, licensing requirements, and simple responsiveness.
Traditionally, industrial applications have been the target of older fieldbus technologies such as Profibus, Modbus, CAN, and others. These technologies are typically based on twisted pair wiring and are generally in the 1 Mbps or less performance level. In the industrial world, cost is everything. So, wiring to sensors and actuators needs to be inexpensive and long-lived. Hence, single twisted pair wiring is the norm.
Nonetheless, we are seeing the expansion of the fourth industrial revolution otherwise known as Industry 4.0 (I4). I4 is characterized by the use of large-scale machine-to-machine (M2M) communications and the deployment of the Internet of Things (IoT) for increased automation. The goal of this trend is the deployment of smart devices that can operate, analyze data, and take action without human intervention. The key to the success of I4 is based on connectivity and performance with a minimum of intermediate translations of the signals that could introduce latency. In addition, the IoT element often centers on being able to seamlessly communicate data from the sensor all the way to the cloud. In our current world, this implies the use of the IP protocol stack.
Unfortunately, none of the traditional fieldbus technologies typically pass IP frames. This means that there is a requirement for translator boxes that translate from Profibus or CAN frames to IP and vice versa. This introduces latency and increases cost. And, since most of these fieldbus technologies are relatively slow in comparison to modern communications such as Wi-Fi or 5G Cellular, a new approach is needed to support the growing communications requirements of I4.
Leveraging Automotive Advancements
In the automotive world, we are seeing an explosion in the amount of data needed for sensors found in Advanced Driver Assistance Systems (ADAS). Like traditional industrial applications, automotive applications have relied on fieldbus technologies such as CAN for communications between the Electronic Control Unit (ECU) and sensors/actuators such as anti-lock brakes, emission controls, and others. However, the 1-Mbps rate limits for CAN or even the enhanced data rates for CAN-FD (up to 5 Mbps) are not sufficient for the multiple camera feeds, radars, and lidar of the modern ADAS-equipped vehicle. What automotive applications need is a reliable, high-speed networking capability that shortens time to market for new capabilities while reducing weight. In addition, the ability to support IP within the vehicle would ease software development and minimize the need for protocol converters.
If we take a page from the information technology world, we would see that Ethernet is very much a dominant technology even in light of wireless protocols. Ethernet is the backbone that delivers guaranteed performance at data rates up to 400 Gbps within the cloud infrastructure. However, traditional Ethernet typically uses two-or four-pair wiring or fiber optic. When compared to the single-pair CAN bus, traditional Ethernet cabling is more expensive, heavier (in the case of two-or four-pair wiring), or less robust (in the case of fiber optic). So, a single-pair Ethernet that could handle data rates in the 1-Gbps range would be ideal. Enter xBASE-T1 Single Pair Ethernet (SPE).
The IEEE 802.3 working group is responsible for the standards associated with Ethernet. Originally standardized in 1983, Ethernet in its many forms has eclipsed competing technologies including ARCNET, FDDI, and Token Ring. Originally based on coaxial cable, Ethernet evolved to the use of both shielded and unshielded twisted pair cabling and by the early 1990s the ubiquitous RJ45 (8P8C) connector became a common fixture on many computing devices in IT.
The original 10BASE-T implementation relied on two wire-pairs with one differential pair being used for transmit and one differential pair being used for receive. Limited to 10 Mbits/sec, this standard was much faster than the original coax-based approach, but used a star rather than the bus topology of the coax solution. This star-wired approach mandated the use of a centralized hub known as an Ethernet switch that could handle data movement between devices connected to the switch’s ports. This same two-pair cable solution was continued with the introduction of Fast Ethernet, otherwise known as 100BASET, which supports data rates up to 100 Mbits/sec. With the introduction of Gigabit Ethernet (1000BASE-T), the number of wire-pairs doubled to four and data rates jumped up by an order of magnitude.
In addition to changes in the data rates, a technique known as Power over Ethernet (PoE - IEEE 802.at-2009) introduced several alternative means to deliver power on the same Ethernet cable as the data. Supporting the delivery of up to 25.5 W at 48 VDC, PoE allowed for powering remote devices such as surveillance cameras and wireless access points. A later version is known as Power over Data Lines (PoDL – IEEE 802.3bu-2016), which allows for delivery of up to 50 W at 12, 24, or 48 VDC. PoDL was developed specifically for the xBASE-T1 SPE market and allows for both data and power to be delivered via a single wire pair.
The xBASE-T1 standards can be further broken down into 10BASE-T1L (IEEE 802.3cg), 100BASE-T1 (IEEE 802.3bw), and 1000BASE-T1 (IEEE 802.3bp). The following table summarizes the primary features of these variants:
The “L” in the 10BASE-T1L stands for “long reach” owing to its 1 Km length (many implementations can actually exceed 1 Km depending on cable quality and connector types). One of the many advantages to the SPE specification is that it can use the existing single twisted pair fieldbus cable runs. This is a huge savings for industrial applications. And, with the addition of PoDL, the remote device can support communications and be powered over the same cable segment. As an added benefit, the 10 Mbit/sec data rate is significantly faster than the fieldbus implementations it is meant to replace.
Whereas the T1L variant is targeted at point-to-point applications, there is also a “short reach” variant (10BASE-T1S) that is wired as a multi-drop implementation to replace common fieldbus versions such as 20 mA current loop and CAN. The reach of the T1S flavors is significantly shortened to 25 m. But, the use of multi-drop allows for a single cable run with a single port interface for the media access PHY.
In order to support the multi-drop access and avoid potential contention on the cable, there are two approaches. The first is using Carrier Sense Multiple Access/Collision Detection (CSMA/ CD) that harkens back to the original Ethernet implementations from the 1980s. In this approach, each station first listens for traffic on the bus before transmitting. In the event that multiple stations attempt to transmit at the same time, the collision is detected, each station stops transmitting, waits a short random amount of time, and then the process repeats with listening for an idle bus and trying the transmit again. This is a simple approach, but it adds a stochastic element to the communications, which will introduce latency.
If the application is particularly latency sensitive, an alternative Physical Layer Collision Avoidance (PLCA) mechanism can be added to the CSMA/CD, where one station is designated the master, which sends out a beacon that allows only the station designated in the beacon to transmit (somewhat like a token ring implementation). This facilitates determinacy by providing a designated slot for each station and avoiding the potential for collisions as the number of stations increases to the 31-station maximum. It should be noted that PoDL is not yet defined for use in multi-drop applications.
Ethernet is Ethernet
One of the significant advantages to SPE is that in the final analysis, it is Ethernet. So, as far as protocol stacks are concerned, SPE is just like every other Ethernet segment. This means that you can easily use IPv4/IPv6 protocols on top of the SPE implementation. This is a huge software development time-saver, as the software teams can use standard IP-based APIs for communications. There is no requirement for protocol conversion from one of the fieldbus variants to IP and back, thereby reducing latency and eliminating the cost of the protocol converter device.
Auto speed detection in the SPE switch is also a possibility just like the typical four-pair Ethernet switch. Therefore, one switch could handle 10BASE-T1, 100BASE-T1, and 1000 BASE-T1 segments as well as supporting more traditional two- or four-pair Ethernet interfaces for debugging or interfacing to traditional IT hardware such as panel PCs. The switch could also support PoDL to power remote devices if needed.
In order to avoid inadvertently mixing up SPE and traditional Ethernet segments, SPE uses the IEC 63171-6 connector. This connector is an open standard and is available in both IP20 and IP65/67 versions. Manufacturers are providing the connector in standard insertion, push/pull, and screw-type mating faces. In addition, there are options for both M8 and M12 connectors found in existing fieldbus implementations.
SPE is poised to play a significant role in Industry 4.0. The SPE Industrial Partner Network already consists of over 30 manufacturers that are providing cables, assemblies, PHY silicon, Ethernet switches, and evaluation devices. With the ability to provide greater performance, utilize existing cable plants, provide power, and be software compatible with the traditional IT-type Ethernet implementations, SPE provides a relatively low-cost upgrade path for replacing wired sensors and actuators as the existing systems age out. Is SPE the future of industrial automation and automotive? It certainly has the potential.
This article was written by Mike Anderson, Sr. Project Leader — Embedded Systems Architect, The Aerospace Corporation (El Segundo, CA). For more information, contact Mr. Anderson at michael.e.anderson@ aero.org or visit here .