Leveraging the compatibility of Ethernet with additional Real-time protocol, EtherCAT has become an attractive and widely used replacement for proprietary field buses. With today’s multicore processors, it is also now possible to implement it almost completely in software.
BY DARON UNDERWOOD, INTERVALZERO INC.
For equipment and machine builders, Real-Time Ethernet standards deliver undeniable value because they enable performance far superior to previous field bus architectures and, better still, because they allow equipment manufacturers to replace proprietary components with commoditized parts, dramatically reducing costs of equipment, manufacturing and field support.
To date the Ethernet for Control Automation Technology (EtherCAT) standard has enjoyed the most rapid adoption, creating some separation from other real time Ethernet standards. According to EtherCAT Technology Group (ETG) benchmarks, EtherCAT offers the best performance of any available standard. And superior performance is only a part of the reason for EtherCAT’s success. EtherCAT’s incredibly rapid rate of adoption and gains in market share are achieved because it is viewed as the most open standard.
Most Real-Time Ethernet standards are still largely controlled by a single industrial automation supplier that did the hard work of developing the standard in the first place. However, because of its standards independence from the industrial automation vendors, more motion drive vendors have embraced and support the EtherCAT standard than any other Real-Time Ethernet standard. Coupled with its superior performance, this support and commitment from the ecosystem make EtherCAT the leading standard.
EtherCAT’s ecosystem support is derived from its importance to machine builders. Indeed, EtherCAT tips the balance of power toward the machine builder and away from the industrial automation vendor by eliminating vendor lock-in and giving machine builders more freedom. It allows machine builders to optimize the cost and performance of their machines by mixing and matching motion drives and other components rather than being forced to buy a bundled package from a single vendor.
First, let’s understand the value of Real-Time Ethernet and then from that context we can better understand the differentiated value of EtherCAT and how to best implement it on a PC. The cost of a traditional proprietary field bus I/O card and field bus cable far exceeded the cost of a NIC card and Ethernet cables. Because Ethernet-based TCP/IP had become the de facto standard for corporate networks, the hardware components used in those networks – copper and fiber, CAT5 cables, RJ45 jacks, NIC cards, Ethernet switches – were seen as essential for reducing the costs of a machine’s bill of material if they could be used in a machine that demanded real-time control in a distributed setting. These components had become ubiquitous, high quality and most importantly because of the volume, extremely low cost.
Yet while Ethernet seemed like the obvious choice for cost and quality reasons, machine builders had a major concern. The Ethernet protocol (CSMA/CD) was not deterministic, raising issues of scalability, performance and safety. The collision detection model could not satisfy the real-time needs nor the tight time deadlines required by most industrial machines – particularly those relying on motion control. Without those capabilities the breakthrough architecture would remain just an idea.
In recent years, Industrial Ethernet solutions have begun to supplant conventional field buses in next-generation machine designs, as newly developed standards overcome concerns about efficacy, performance and safety. And Real-time Ethernet – and EtherCAT in particular – have enabled a more efficient architecture that can be implemented entirely in software and that can run on commodity Industrial PCs.
There is a chicken and egg problem with most standards and the Real-time Ethernet standards are no exception. For a standard to be effective, there must be a critical mass of vendors that adhere to the standard, but vendors do not want to invest in a standard until it has critical mass. Consequently, while many Real-time Ethernet standards are solid, they lack that critical mass of adoption by an ecosystem of vendors. And this is why EtherCAT is differentiated when it comes to servo controls, motion and I/Os. An examination of the EtherCAT architecture and it’s key value components explains why.
Like most field buses, EtherCAT uses the concept of master and slaves. The EtherCAT master does not require special hardware and can be designed solely in software on a COTS system with one or more standard Ethernet ports. Slave devices use an EtherCAT Slave Controller (ESC), which processes the frames in hardware, ensuring predictable performance.
The EtherCAT master sends a telegram to the first slave device, which reads from/writes to the appropriate area and then passes the telegram to the next slave (Figure 1). This process continues until no more device nodes are detected and the telegram is then sent back to the master. Some of the main features that distinguish EtherCAT from standard Ethernet include:
In EtherCAT, slave devices read the data addressed to them while the frame passes through the node. Similarly, input data is inserted while the telegram passes through. The frames are only delayed by a few nanoseconds.
Fewer Network Constraints: In many Industrial Ethernet solutions, the use and topology of switches and hubs limits the network. With EtherCAT, the application defines the topology, removing the need for switches and hubs and thereby eliminating network constraints.
Real-Time Performance: Importantly, EtherCAT has exceptional real-time performance. It uses a single telegram which all nodes can read/write to as it passes by. Node latencies are constant and very small – less than 500ns regardless of the size of the telegram.
Built-in Safety: Functional Safety over EtherCAT (FSoE) is supported as part of the network architecture. The EtherCAT safety technology conforms to IEC61508. Safety applications can support up to a SIL 3 level of integrity.
Node Synchronization: EtherCAT uses a mechanism based on distributed clocks (DC). The clocks are calibrated in hardware based on the first slave device. This mechanism achieves system jitter that is less than us.
High Availability: EtherCAT supports cable redundancy by allowing a master to control two ports. The second port is connected to the end device of the line topology, thereby creating a ring. Multiple masters are also possible enabling automatic failover in the case of a master failure.
In summary, EtherCAT is superior to other field bus solutions currently available not only because of its greater real-time performance, but also because of its ability to scale across multiple use cases—from I/O-only systems to very high-precision motion control. With its unique architecture, EtherCAT can easily scale up to much larger numbers of devices while incurring relatively small additional latencies. This feature alone sets it apart from the other Ethernet-based field buses available today.
Importantly, the EtherCAT standard has managed to build the largest critical mass of vendors and that provides leverage. If a Servo drive manufacturer does not offer an EtherCAT- enabled solution, they are at a severe competitive disadvantage. Increasingly machine builders are demanding EtherCAT as part of the solution.
Via plugfests and working sessions for vendors, the EtherCAT standards body, the EtherCAT Technology Group (ETG), has done an excellent job of ensuring that the entire architecture simply works. The vendors that provide their qualified EtherCAT slaves – like Servo drives – can simply plug onto the network and work with any qualified EtherCAT Master. The goal is to make it as simple to plug a Servo drive onto the network as it is to plug a PC onto the corporate TCP/IP network,
Along with plugfests, which are informal, ETG also requires each device be tested by the manufacturer for conformance using the ETG supplied Conformance Test tool. Although it is not mandatory, device manufacturers are encouraged to submit their devices to one of the EtherCAT Test Centers (ETC) for formal third-party tests, which, if successful, result in a certificate of conformance. This is crucial to the customer base as it instills confidence in device interoperability and also signals the device manufacturer’s commitment to the technology.
It is important to note that EtherCAT is only part of the value and solution for machine builders. Today it is possible for machine and equipment builders to contain all of their control logic in software – ending the long dependence on proprietary chipsets, boards, and networks. A machine’s control logic and Human Machine Interface (HMI) can run on a single industrial PC and communicate using an open-standards communication and I/O interface, to any remote device that is part of that equipment.
And, as we’ve seen – with the arrival of real-time Ethernet – the solutions can also be deterministic. Tight synchronization of control messages between the PC and remote devices (e.g. motion control) can be done in real-time, with tightly bounded cycle times and event response. All real-time logic can run on commercial off-the-shelf PCs and network components. No proprietary communications boards; no proprietary control cards; no arrays of DSPs or FPGAs; no second PC with a stand- alone RTOS.
This architectural breakthrough allows machine builders to capture their intellectual property and value in software components, while also dramatically reducing costs by eliminating proprietary hardware. Also, because the solution on the PC is completely software based, patches or fixes can be downloaded remotely rather than performing a board replacement. This reduces support costs even further.
Two important technology advances – multi-core chipsets and Real-time Ethernet – have enabled this long-desired software-only architecture. There are four key elements in this deterministic architecture:
• Microsoft Windows OS
• A symmetric multiprocessing-enabled real-time extension to Windows, such as IntervalZero’s RTX64.
• Multi-core x86/x64 chipsets
• Real-time Ethernet capability.
An RTOS extension to Windows such as IntervalZero’s RTX64, which runs on multiple cores, is an essential component of this architecture. Critical to satisfying performance and safety demands, such an RTOS extension allows the control logic (e.g. motion control or digital music transformation) to run as a software component on a Commercial Off-the-Shelf (COTS) x86/x64 system across one or more cores.
RTX64, for example, has drivers that enable real-time communication to support any of the protocols listed above, eliminating the need for proprietary communications boards and field buses. And the Windows 64-bit OS is a critical component to protect the future investment for the equipment builder.
With Intel and Microsoft’s investments in the Internet of Things (IoT) and Industrial Ethernet standard for vision – known as GigEVision – the demands for memory will far exceed the capacity of the traditional 32-bit x86 systems. Machine equipment companies have a range of choices regarding how they take advantage of software-centric architectures’ capabilities. At one end of the spectrum, machine builders may want only Real-time Ethernet access, and at the other end of the spectrum builders may want a completely pre-integrated solution. Table 2 shows examples of actual deployment scenarios. Each has its strengths.
For example, the time to market and the flexibility to make field fixes dramatically improves as the machine builder moves closer to an all-software model. And while the upfront costs increase with that approach, it is important to recognize that the ongoing costs are dramatically lower. These tradeoffs need to be evaluated in context of each machine design and market.
Until recently, Real-time Ethernet has been the final, missing link to an architecture for machines and equipment with distributed deterministic requirements. Ethernet is familiar, universal and extremely cost-effective. Protocols like Profinet, SERCOSIII and EtherCAT have overcome the limitations that would prevent Ethernet from being deterministic.
By adding Real-Time Ethernet to an RTX64-enabled solution, customers can build breakthrough systems that outperform prior systems and do so at a fraction of the cost and complexity by using commercial off the shelf components like IPCs, NIC and CAT5 cables. The EtherCAT standard is a real win for the machine builder because they achieve breakthrough performance, eliminate vendor lock-in, establish a platform that supports emerging standards (like GigEVision), and dramatically lower all costs.
EtherCAT Technology Group
Port Orchard, WA