Gabriele Ribichini

Monday, February 13, 2023 · 0 min read

What Is EtherCAT Protocol and How Does It Work?

In this article, we will learn about EtherCAT protocol - what it is and what it does, with enough detail that you will:

  • See how EtherCAT applies to real-time control and DAQ systems

  • Learn about the key EtherCAT features and capabilities

  • Understand how EtherCAT differs from ethernet and why

Are you ready to get started? Let’s go!

EtherCAT introduction

EtherCAT stands for “Ethernet for Control Automation Technology.” It is a protocol that brings the power and flexibility of ethernet to the world of:

  • industrial automation,

  • motion control,

  • real-time control systems, and

  • data acquisition systems.

The EtherCAT protocol is maintained by the EtherCAT Technology Group and is standardized under IEC 61158

EtherCAT technology group logo

EtherCAT and EtherCAT logos are trademarks or registered trademarks, licensed by Beckhoff Automation GmbH, Germany.

A brief history of Ethernet

Developed in the 1970s at Xerox’s Palo Alto Research Center (PARC), Ethernet was designed as a low-cost and fault-tolerant network interface for both local and wide-area networks. At the time of its invention, there were other networks, such as TokenBus, TokenRing, ARCNET, CDDI, and a variety of lesser-known or proprietary network interfaces. 

The old PARC (Palo Alto Research Center) facility sign

PARC scientist Robert Metcalf was assigned to figure out how to interconnect the company’s hundreds of computers so that they could share the world’s first laser printer, which had recently been invented by Xerox. 

This might seem like a trivial problem today, but in the early 1970s, few companies had more than two or three computers. There were no personal computers, notebooks, cell phones, tablets, etc. Existing networks were not scalable or fast enough to connect so many machines, so Metcalf and his colleagues needed a new approach to solving this problem. 

They combined some internet technology with their own ideas and gave birth to the network that is today everywhere in the world, connecting millions of devices to each other and to the internet itself.

Under IEEE-802.3 issued formally in 1985, ethernet became the de facto standard interface for networks large and small, and even for individual instruments. It’s a combination of hardware and software developed to be fault-tolerant and fast. 

Information is broken into “packets” or “frames” known as datagrams. Each datagram contains not only the data itself, but with identifying header and address information so that it can be reconstructed at the receiving end, and a 32-bit CRC (cyclic redundancy check) at the end to prevent errors.

Devices on the network have an ethernet interface, each of which has a unique address. This is important because, with so many devices potentially transmitting and receiving across the same network, each device needs to know which data is intended for it. 

How data is sent through the Ethernet protocol

Imagine hundreds of letters and boxes from the post office flowing down your street -- most of them are for your neighbours, and a few of them are for you. But which ones? Well, the address printed on the face of each letter ensures that those with your address are automatically put into your mailbox.

Business reply mail

Simple, right? It has worked like this for hundreds of years - long before computers.

But now imagine that each letter coming down the street has actually been cut up into thousands or even millions of tiny pieces, and each piece is just one word from inside that letter. 

Furthermore, these words are not necessarily in order. In fact, they are mixed together with billions of words from your neighbour’s letters, too. Suddenly it’s a lot more complicated. 

But with ethernet, each “word” (datagram) contains the info that your mailbox needs in order to grab the words intended for it, and then reassemble them perfectly into each unique piece of mail that was sent to you. So when you open your mailbox, the letters are perfectly reassembled, and they look just the way they did when they were mailed.

We don’t need to get deeper into how ethernet exactly works, but it is important to understand the nature of this interface, and why it has become so ubiquitous today.

A brief history of EtherCAT

EtherCAT was originally developed by Beckhoff Automation, a major manufacturer of PLCs (Programmable Logic Controllers) used in industrial automation and real-time control systems. 

They had developed their own version of Fieldbus called “LightBus” in the late 1980s, to address the bandwidth problem of other interfaces. Additional work on this protocol eventually resulted in the invention of EtherCAT.

Beckhoff introduced EtherCAT to the world in 2003. And then in 2004, they donated the rights to the ETG (EtherCAT Technology Group), who is responsible for the promotion of the standard. ETG has a very active developer and user group. EtherCAT is standardized under IEC 61158.

Why EtherCAT?

Why can’t we simply use ethernet to interconnect data acquisition and control systems? Ethernet is fast. It’s inexpensive. And it’s easy to implement in today’s computer-based instruments. So what is missing? 

The answer lies mostly in determinism or time accuracy. The imaginary mailbox in our example above has all the time it needs to collect all of the datagrams from your letters and reassemble them into your mail. 

On a day when only 100 letters came down your street, your mail would appear quickly in your mailbox. On heavy mail days, it would take longer, but the difference is irrelevant, right? 

So, ethernet is absolutely fine for sending documents around the office or from one device to another. Your “mail” is delivered quickly, and it really doesn’t matter if a letter was received at 10:00:02.123211 or at 10:01 - again, the exact instant does not matter in that application.

EtherCAT article email letters

But Control systems are all about timing. It really DOES matter what time something happened and with as much time axis resolution and accuracy as possible.

Factory automation control systems are by definition real-time systems. Turning machines on and off requires very low latency. You don’t want your emergency shut-down message mixed in with a gigabyte data backup stream, for example - the real-time messages should always get priority. 

But in a conventional ethernet system, there is no protocol for this - all data is essentially “equal.” This works fine with your office computers sharing network bandwidth to access servers and printers, but not so well with real-time applications.

Typical Industrial control system
Dewesoft logo

Learn more about EtherCAT

In this video, Martin Rostan, Executive Director of the EtherCAT Technology Group, explains in 20 minutes how EtherCAT generates competitive advantages for users.

EtherCAT physical layers

EtherCAT uses the same physical and data link layers of ethernet, shown below:

By skipping OSI layers 3-6, EtherCAT achieves cycle times better than 100 µs and communication jitter better than 1 µs

The physical layer is the hardware that physically conveys the data across the network. This is the core electrical, i.e. “mechanical” level of the network.

The data link layer is where the data is encoded into packets. The ethernet implementation here is fine, and EtherCAT uses it. But then the other layers well-known by ethernet users like the Network (IP) layer and Transport layer (TCP and UDP) are skipped completely by EtherCAT in the interest of cycle time. 

This and other aspects of the protocol are how EtherCAT can reduce the 10 ms cycle times of ethernet by several orders of magnitude. This yields an effective data rate of 100 Mbps.

Ethernet vs. EtherCAT differences

EthernetEtherCAT
Common Physical and Data Link LayersYesYes
International StandardIEEE-802.3IEC 61158
Deterministic TimingNoYes
Master/Slave OperationNoYes
Ring-based TopologyNot requiredYes
Optimized for real-time controlNoYes
Optimized to avoid data collisionsNoYes

The layers higher than the physical and data link layers are different between Ethernet and EtherCAT. Let’s take a look at the differences, and then the resulting advantages of EtherCAT for real-time and DAQ system applications:

Standard Office Ethernet network

In a typical ethernet network at your office or home, multiple devices are connected essentially at the same level. Any device can send data across the network, and any device can receive data. The network probably has a switch that connects it to an internet device that provides access to the outside world.

This is very flexible but is prone to data overload when multiple devices send or request a large amount of data at the same time. Time-critical messages may be slowed down or even blocked in extreme circumstances.

EtherCAT master and EtherCAT slave devices

EtherCAT, on the other hand, operates in a very different way:

EtherCAT network with ring topology

The EtherCAT MASTER device is the only one allowed to transmit data across the network! The master sends a string of data across the bus, eliminating the data collisions of an ethernet system and optimizing speed as a result.

EtherCAT frames are embedded within a standard Ethernet frame and are identified in the EtherType field by the value 0x88A4. The master is the only device in an EtherCAT segment that is allowed to send messages - the slaves can add data and send the frame along, but they cannot create new messages on their own.

These frames are received by the EtherCAT slave devices (nodes) that it is addressed to. The slave devices process data and add back whatever was requested by the master and send the frame along to the next node in the ring. 

The next node does exactly the same thing, taking in the data intended for it, putting the required data back into the EtherCAT frame, and sending it along to the next node.

Speed is increased from conventional ethernet not only because there is just one device sending data, but also because of a technique called “processing on the fly.” In conventional ethernet, each device has to read the header of each message to determine if the data is intended for it, then ingest the data and process it in some way. But with processing on the fly, the node reads the header and sends the data along simultaneously, saving time and improving efficiency.

Finally, unlike conventional ethernet, EtherCAT allows incoming and outgoing data from more than one device on the network to be combined into single frames. This again optimizes speed. 

Interestingly, if a particular node does not have the processing power to handle the data the bus speed can be adjusted by the master, ensuring that no data is lost by any device on the network.

Learn more about EtherCAT Topology

EtherCAT time-stamped data

One of the most important aspects of EtherCAT is the distributed clock. Each node timestamps the data when it is received, and then stamps it again when it sends it on to the next node. So when the master receives back the data from the nodes, it can easily determine the latency of each node. Every data transmission from the master gets an I/O time-stamp from every node, making EtherCAT far more deterministic and accurate on the T axis than ethernet can be.

Even before the EtherCAT goes to work, however, the master sends out a broadcast to all of the slave nodes on the network, who latch it when they receive it and when they send it back. The master will automatically do this as many times as necessary to reduce jitter and keep the slave nodes synchronized with each other. 

This timing accuracy is extremely important in real-time control and factory automation applications. It further allows DAQ systems such as the ones available from Dewesoft to be integrated so easily into control systems.

EtherCAT’s built-in distributed clock provides excellent “jitter” performance far smaller than one microsecond (1 µs), which is equivalent to IEEE 1588 PTP (Precision Time Protocol), without the need for any additional hardware.

EtherCAT’s fault tolerance

If the last node’s output is not connected to the master, the data are automatically returned in the other direction via the EtherCAT protocol. Timestamping is maintained.

Highly fault-tolerant EtherCAT network

This fault tolerance means that EtherCAT networks do not have to be arranged in a ring like the diagrams shown above, but can be configured in a variety of ways, including tree topology, ring topology, line topology, star topology, and even combinations.

Of course, there has to be a path between the slaves and the master. If you literally unplug them they cannot work, but the point is that the topology of the network is highly flexible and tolerates faults to an exceptional degree.

Switches like you find in ethernet systems are not required by EtherCAT systems. Cable lengths up to 100 meters (328 ft)  between nodes are possible. The LVDS (low-voltage differential signalling) on the twisted pair copper lines runs at high speeds and with very low power consumption. It is also possible to use fibre optic cables to increase speed and add galvanic isolation between devices.

Types of EtherCAT devices

With few exceptions, systems using EtherCAT are divided into two groups:

  • Real-time control devices

  • Measurement devices

Control devices such as PLCs are EtherCAT masters on the EtherCAT network, whereas measurement devices have historically been EtherCAT slaves.

However, the third type of device has been invented by Dewesoft - a data acquisition device that combines high-speed data acquisition and real-time control capabilities in parallel (real-time data feed to a PLC or master controller software/hardware).

Until recently, when engineers wanted real-time data from a DAQ system, they would take several analog outputs (separate outputs for each of the analog input channels) from the DAQ system and bring them into the PLC controller. This required multiple analog inputs as well as redundant conversion from analog to digital data.

The old analog method of sending DAQ data to a PLC

By installing an EtherCAT slave port into their DAQ systems, however, Dewesoft eliminates the redundant analog inputs on the PLC completely, as shown in the graphic below:

Dewesoft DAQ system sending real-time data via a single EtherCAT line

But this is actually just the beginning because DAQ systems like IOLITE Rack and IOLITE can even eliminate the PLC hardware completely in many applications. It is possible to connect IOLITE to a computer host running real-time PLC software. These include systems like:

  • CLEMESSY Syclone®,

  • Beckhoff Twincat®,

  • MTS FlexTest®,

  • Acontis EC_Master®,

  • National Instruments LabVIEW®,

  • and others.

Combining IOLITE with one of these real-time PLC software systems, connected via EtherCAT, and IOLITE can even drive the actuators, providing the real-time hardware control needed for the system.

IOLITE can serve as the hardware backbone of a Control/DAQ system via EtherCAT

Dewesoft EtherCAT data acquisition and control hardware

Dewesoft manufactures data acquisition and control systems with EtherCAT compatibility. Let’s look at each one and look at how they apply EtherCAT technology.

IOLITE data acquisition and control systems

IOLITE is a DAQ system and real-time control system for industrial applications. It uniquely provides both real-time control and feedback monitoring. Each IOLITE data acquisition system is equipped with two fully independent EtherCAT buses that run in parallel.

IOLITE R8 and R12 data acquisition system

The primary data bus provides perfectly synchronized data acquisition via DewesoftX DAQ software. IOLITE can stream any number of input and output channels at high speeds to the computer's hard drive. IOLITE racks are available in two models:

  • IOLITE R12: 19-inch rack-mounting mainframe chassis with 12 slots for IOLITE input/output modules.

  • IOLITE R8: benchtop mainframe chassis with 8 slots for IOLITEinput/output modules.

  • IOLITE R8R: extremely rugged benchtop rugged mainframe chassis with 8 slots for IOLITEinput/output modules.

IOLITE's secondary EtherCAT bus can be used in two ways: 

  • A low-latency front-end interface for any third-party EtherCAT-compatible real-time controller.

  • A redundant data acquisition system bus for critical data acquisition applications.

Channel expansion via EtherCAT master port

The dual redundant EtherCAT ports on IOLITE can also be used to receive synchronized data from other EtherCAT-equipped DAQ systems, like the SIRIUS and KRYPTON models.

Channel expansion via additional EtherCAT equipped DAQ systems

SIRIUS RT data acquisition and control systems series

Like the IOLITE model, the SIRIUS RT series models feature dual-mode functionality that combines high-speed streaming to a host via USB AND low-latency data streaming over EtherCAT to any third-party host. They combine flagship SIRIUS DAQ signal conditioning and ADC technology with all of the advantages of deterministic data over EtherCAT.

SIRIUS R4RT data acquisition system

Available with 2, 4, or 8 slots for SIRIUS modules, they can be equipped with from 8 to 128 channels, as shown in the table below:

SIRIUS RT models showing max channel count
R2rt R4rtR8rt
SIRIUS DAQ module slots248
Max total channels using only 8-ch modules163264
Max total channels using only 16-ch modules3264128

Road load application case study using Dewesoft’s R8rt

One interesting example of how engineers are putting this capability to good use is outlined in our case study.

Road Load Data or Durability testing is well-known within the automotive industry. Cars and trucks are mounted onto a multi-axis road simulator, which then simulates the loads and stresses of actual and extreme driving conditions. In this way, engineers can subject their vehicles to accelerated life cycle analysis in a way that is more efficient than using human drivers (no one can drive for 100 hours straight, for example).

Typical multi-axis road simulator

The data that the multi-axis road simulator uses comes from real road tests. Dewesoft DAQ systems have long been used for these tests, riding along inside the vehicle and collecting dozens of channels of dynamic data from a variety of sensors, especially accelerometers.

Previously, however, the DAQ system used on the test track and the controller of the road simulator had no connection at all. Data collected by the DAQ on the test track had to be converted offline for use within the road simulator.

But with the dual data buses of the SIRIUS R8rt, it is now possible to completely integrate the DAQ system with the controller system, achieving much greater efficiency and cost savings. Kilometres of cables are eliminated by not having to re-digitize data and by sending it via EtherCAT, and the same DAQ system used on the test track can also be employed on the test bench for the load simulations.

In the case study, you will learn how the SIRIUS R8rt system was integrated with the MTS road load testbench rig. By combining two data buses (USB and EtherCAT), and with MTS also introducing EtherCAT communication on their FlexTest® series controllers, the instruments acted essentially as one. 

Using the same R8rt measurement system for test track and on the road simulator testbench.

The same SIRIUS R8rt DAQ instrument is used both in the vehicle on the test track to gather the real load data, and also on the testbed for real-time streaming of this data to the controller. 

Using its unique dual-mode capability, data samples from the SIRIUS DAQ modules are transmitted in real-time over EtherCAT to the MTS controller at 10 kS/s per channel. At the same time, the test results from the vehicle on the multi-axis road simulator are recorded by the R8rt internal computer and SSD hard drive at high speeds up to 200 kS/s per channel.

Complete case study:

KRYPTON rugged data acquisition systems for field testing

Most DAQ hardware on the market is designed for office or light industrial use. But there are applications where the DAQ equipment must be located in harsh environments. For example:

  • High temperatures up to 85°C (185° F)

  • Cryogenic temperatures down to -40° C/F

  • High shock and vibration up to 100 g

  • Water spray or immersion

  • High dust concentration

KRYPTON rugged data acquisition (DAQ) systems

Once again, most DAQ instruments are not designed to withstand even one or two of those extreme environmental conditions. Based on these requirements, Dewesoft developed the KRYPTON series of DAQ instruments.

Distribution via single EtherCAT cables

KRYPTON modules are interconnected via a single rugged EtherCAT cable that carries data, power, and synchronization. The cables can be up to 100 m (328 feet) long so that engineers can distribute the modules where the signals are. This might be across a bridge or a large factory floor.

Distributed KRYPTON modules interconnect via EtherCAT

The fast EtherCAT interface runs at 100Mbps Full-Duplex bus speed, providing for 6MB/s to 10MB/s data throughput per chain. Modules intended for dynamic measurements can sample up to 40 kS/s. All modules on the chain, which can be spread out as far as 100m (328 ft) between nodes, are precisely synchronized. 

Many modules can be connected this way. KRYPTON is available in multi-channel modules as well as single-channel KRYPTON ONE modules for ultimate flexibility.

Single-channel KRYPTON ONE modules are available for a wide variety of signals and sensors

Waterproof and dustproof

KRYPTON systems meet IP67 standards for water and dust, meaning that they can withstand not only water spray but total immersion in 1 meter (39 inches) of water for 30 minutes. They are completely sealed against dust and other small particulates.

Extreme temperature tolerant

Being filled with a thermally insulating rubber, KRYPTON modules can withstand a wide range of temperatures from -40° to 85° C (-40° to 185° F). KRYPTON modules are found in hot and cold weather automotive tests, on rocket engine test stands, and in factories where ice cream and other cold (or hot) foods are made. 

Shock and vibration resistant

KRYPTON modules are designed for and rated to handle 100g shock and high-vibration environments. This can be a serious requirement in many applications where large vibrating machinery is found, or testing rockets and other large force-producing machines.

SIRIUS modular data acquisition system

In specific applications, very high-end DAQ is required to supplement the relatively low-speed data that a PLC is capable of recording. For these applications, the SIRIUS and SIRIUS Waterproof from Dewesoft is a perfect solution.

SIRIUS has both high-speed USB and EtherCAT data buses, so it can provide the data to the EtherCAT master at a rate that it can handle and in parallel, record much higher-speed data to a separate computer running DewesoftX software.

SIRIUS DAQ system running fault-tolerant dual-mode data to an EtherCAT master and to a computer host

SIRIUS can also be connected to an IOLITE system, for channel expansion. The dual-mode operation of sending data to the EtherCAT master and to a separate computer running Dewesoft X software in parallel is also fault-tolerant

Even if the Windows computer running Dewesoft X software were to completely fail, SIRIUS would continue sending data to the EtherCAT master.

Please read the testing of Ariane V rocket solid-propellant boosters case study and learn about how 800 channels of high-speed, isolated Dewesoft DAQ and control systems are used and fully integrated into the SYCLONE control system using EtherCAT. 

DEWESOFT and CLEMESSY combine the best of both worlds: high-end data acquisition capacity and control front-ends in a single device with an efficient and full-performance control-command software solution. The high quality required by the National French Space Agency and more generally the European Space Agency has increased the Dewesoft solution's maturity and robustness for large test benches with hundreds and thousands of channels.

EtherCAT cables for harsh environments

Well, if the DAQ hardware is waterproof, dustproof, and can withstand -40° temperatures, the cables must also be able to withstand these environmental extremes, too. Otherwise, the system will fail simply because of the cables.

Accordingly, Dewesoft has developed EtherCAT cables that are used to interconnect its KRYPTON and SIRIUS Waterproof series of harsh environment DAQ systems. In the video below we show the cables being frozen to -40°C and how they maintain their flexibility.

Dewesoft compatibility with third-party EtherCAT masters

Dewesoft EtherCAT-based DAQ systems are compatible with a wide range of third-party EtherCAT hardware and software masters on the market today, including:

  • Clemmesy Syclone®

  • Beckhoff Twincat®

  • MTS Flextest®

  • Acontis EC_Master®

  • National Instruments LabVIEW® with PLC

What is the difference between EtherCAT and CANopen? 

CANopen is a high-level protocol based on the CAN (controller area network) hardware bus. It also includes a specification for any devices on the bus. While EtherCAT uses the lowest two layers (data and physical) of the ethernet protocol (see this section for a reminder), CANopen uses the lowest two layers of the CAN OSI. 

The 7-layer OSI model as implemented by CANopen

The leveraging of readily available and proven CAN bus hardware made by a wide assortment of manufacturers is a major advantage of CANopen. CAN may have started out as a way to reduce electrical wiring in automobiles, but it has evolved over the decades to be used across virtually every industry and thousands of applications, from industry to aerospace, energy, and more. Putting the high-level CANopen protocol on top of proven and reliable CAN hardware makes system development and deployment far easier than ever before. 

CANopen provides a high-level and thus simplified way for engineers to integrate devices using CAN as the hardware communication layer. The protocol handles many hardware-specific low-level tasks, simplifying and speeding up development. CAN-specific hardware issues like acceptance filtering and bit timing are handled by the CANopen protocol. CANopen provides communication objects (COB) for time-sensitive processes and other hardware management tasks. 

EtherCAT is a single master / many slave architecture. The master assigns addresses to each slave, controls the transmission rate of the network, and performs an initial time-synchronization of all devices, which can be repeated as needed. Furthermore, the master is the only device that is allowed to transmit messages. Slaves are responsible to react to messages, inserting their time-stamped responses, and then returning them to the master.

CANopen networks can have more than one master. However, the integrator must ensure that each device has a unique address and that all devices are set to the same bit rate. The highest recommended bit rate of a CANopen system is 1000 kbps.

A CANopen system can have up to 127 devices on it, one of which must be the master. An EtherCAT segment can have up to 65,535 devices on it. The maximum distance between devices is 100 m (328 feet).

EtherCAT is highly deterministic, achieving better than 1µs jitter at 100 Mbps. The CANopen SYNC telegram is limited to one frame length and may jitter with 130μs at 1 Mbps.

Top-level comparison: EtherCAT and CANopen

EtherCATCANopen
Data and Physical Hardware LayersEthernetCAN bus
Bus Speed100 Mbps1 Mbps (max.)
Transfer ModeFull duplexHalf-duplex
Determinism (jitter between devices)As low as 1 nsTypically 100-200 ns
Max. Devices65,536127 (0 is reserved)
Max. Distance between Devices 1)100 m (328 ft.)Depends on the bus speed: 1.5 m at @ 1 Mbps2.5 m @ 800 kbps5.5 m @ 500 kbps11 m @ 250 kbps
Secondary Communication PortUSBRS232
Master/SlaveSingle master with one or many slavesSingle or multi-master, with one or many slaves
Automatic addressing of slaves by the MasterYesNo
Automatic time sync of devices by the MasterYesNo

1) For more details about CANopen please visit this page.

The CANopen standard is maintained by the CAN in Automation (CiA) International Users and Manufacturers Group. 

The EtherCAT standard is maintained by the EtherCAT Technology Group and is standardized under IEC 61158. 

Both EtherCAT and CANopen are used in a wide variety of industries and applications, including: 

  • Automotive

  • Transportation and Rail

  • Industrial

  • Healthcare

  • Industry

  • Factory Automation

  • Agriculture

  • Aerospace.  

Both protocols are modern, well-maintained, and useful. For low- and medium-speed distributed applications, CANopen is an excellent choice. For higher-speed applications, especially those that require high accuracy, deterministic time synchronization, and incorporate control (with or without DAQ), EtherCAT is a better choice.

How does CANopen over EtherCAT (CoE) work?

As we have seen throughout this article, EtherCAT is a robust system that takes advantage of ethernet hardware, which allows up to 100 meters between devices, accommodates flexible network topologies, and provides highly deterministic data flow, automatic slave time sync by the master, and more.

CANopen also has its advantages, of course. It’s been around longer than EtherCAT and has been adapted thousands of times because of its low hardware cost and easy implementation. Considering the strengths of both CANopen and EtherCAT, it’s almost obvious that there should be some way to combine them and take advantage of both systems. This protocol is called CANopen over Ethernet (CoE), and it allows engineers to use the entire CANopen feature set over fast and robust EtherCAT.

The CoE protocol includes process data objects (PDO) and service data objects (SDO). Nearly all existing CANopen stacks can be used without modification because the SDO protocol is implemented directly. 

PDO stacks are transferred by the faster, deterministic EtherCAT hardware, but are no longer subject to the 8-bit limitation of CANopen. The similarities between the EtherCAT and CANopen state machines are such that few changes need to be made when adapting a CANopen profile to run on EtherCAT. CoE supports all CANopen device profiles and includes the CAN state machine.

The higher bandwidth of EtherCAT allows the entire Object Dictionary to be uploaded across the network. Numerous device profiles can also be reused, reducing development time and cost.

Multi-protocol EtherCAT is a powerful, fast platform for running CANopen devices, and serves as the bridge that allows them to be easily migrated to industrial Ethernet.

What are the differences between EtherCAT and standard Ethernet?

First, EtherCAT is built on the first two layers of the Ethernet protocol, as we described in this section. So there are very strong similarities at the lowest levels - but not at the network, transport, or application levels, so there is no TCP/IP or UDP in EtherCAT.

EtherCAT has been optimized as a real-time, deterministic master/slave system. 

EtherCAT frames (messages) are built inside of standard Ethernet Frames. However, there are many big differences, too:

  • On an EtherCAT network, only the master can send messages, directing them to the appropriate slaves, and then getting the time-stamped data back from them. This is very different from an ethernet network where every device can send messages, and data is not time-stamped in a deterministic way.

  • EtherCAT is deterministic, meaning that extremely very low latency real-time data comes back to the master from the slaves.

  • In an EtherCAT network, the master is in charge of time-aligning all the slaves upon start-up and at intervals, in order to facilitate low-latency time-stamping. This is built into EtherCAT, but must be added to a standard Ethernet network.

  • EtherCAT was designed from the ground up to allow real-time control over devices and systems with extremely low latency. Ethernet was designed primarily for office applications and interconnecting computers, printers, and other network peripherals.

  • Unlike a standard Ethernet network, in an EtherCAT network data collisions are not possible because of the restrictions mentioned above.

  • Data transmission on an EtherCAT network is faster than a standard Ethernet network: 100 Mb/s in effect with very low jitter.

  • Like Ethernet, EtherCAT networks can be arranged in a wide variety of topologies: line, ring, star, etc.

How can I deploy EtherCAT using Ethernet?

An EtherCAT network uses standard ethernet cables to interconnect devices. So simply in terms of the cabling, both EtherCAT and Ethernet networks use the same CAT5 cables.

However, if the question refers to something more, such as connecting other ethernet devices to an EtherCAT network, then we need to look at the protocol called Ethernet over EtherCAT, or EoE.

Ethernet data is tunneled into the EtherCAT system via a Switch Port

EoE is a protocol that allows a Windows client application to communicate with devices on an EtherCAT network. Ethernet packets are sent from the client into the EtherCAT network via a device called a Switch Port. A Switch Port tunnels the ethernet data into the EtherCAT protocol by inserting TCP/IP messages into the existing EtherCAT system messages in a way that does not interfere with the network. 

A Switch Port can be implemented as a separate device, such as the Beckhoff 6601, which supports all Ethernet (IEEE 802.3)-based protocols, and is electrically isolated to 500 V. It can be installed anywhere within the EtherCAT segment and does not require any configuration.

But it can also be implemented as a function within one of the slave devices on the EtherCAT network, or even as a software function on the EtherCAT master.

What is an EtherCAT master?

Every EtherCAT segment or network needs one EtherCAT master. This master is in charge of the network and is the only device that is allowed to send messages across it. It is also responsible for time-synchronizing all of the slaves on the network, and assigning addresses for each slave. It is responsible for requesting data from the slaves and receiving back the modified messages from them containing the requested data.

An EtherCAT master sends data by means of the MAC (Media Access Controller) at layer 2 (the data layer) in the standardized ethernet OSI model. No additional communication processors are needed, which means that EtherCAT master functionality can be implemented on any device that has an Ethernet port. 

As a result, EtherCAT masters are available both as a dedicated piece of hardware and in software form running on a computer, regardless of the operating system it runs. EtherCAT masters have been developed for Microsoft Windows, Linux, QNX, RTX, VxWorks, and more.

When using a computer as an EtherCAT master, the only requirement is the ethernet port. This can be a built-in port or an added NIC (Network Interface Card). Most NICs have direct DMA access, which means that the CPU is not involved with data access, making for a very high-performance system.

What is the maximum distance between EtherCAT devices? 

This is a question without a single answer because there are so many variables. First, the topology of an EtherCAT segment can be a line, ring, tree, star, or even a combination of these configurations. 

Second, EtherCAT supports up to 65,536 unique devices on a segment, which certainly implies that a very large number of devices are supported and can be connected simultaneously. Of course, the practicality of such a system is a question.

Using standard ethernet cables (100BASE-TX), the maximum distance between any two devices on the segment is 100 m (328 feet).

However, utilizing fiber-optic cables (100BASE-FX) this 100 m can be extended up to 2 kilometres (1.25 miles).

Practically speaking, nearly all EtherCAT systems are deployed across a typical factory building or a specific area within a factory.

Which is the best Ethernet-like protocol for real-time applications:  EtherCAT, Profinet, or something else?

In fact, EtherCAT and Profinet are not the only two Industrial Ethernet (IE) protocols in use today. There are several, including EtherCAT, Profinet, EtherNet/IPPowerlinkSERCOS IIIModbus TCP, and CC-Link IE.

The purpose of each of these protocols is to leverage low-cost and fast Ethernet hardware when creating monitoring and control systems in industrial process control environments … while dramatically improving time synchronization, determinism, and environmental ruggedness.

Introduced in 2003, Profinet is a technical standard for communication over industrial Ethernet. It should not be confused with Profibus, which is a process Fieldbus introduced in 1989, and which is managed under IEC 61158. Profibus is based on RS485 serial communication, whereas Profinet is based on Ethernet. A Profinet system can incorporate existing parts of a Profibus system without change, making it attractive as an upgrade path for those older systems.

All of these Industrial Ethernet systems can be characterized by how much of the Ethernet OSI they utilize. As we saw earlier in this document, EtherCAT uses only the first 2 layers of the Ethernet OSI, skipping the transport layer (TCP, UDP) and so on, in order to achieve the fastest possible cycle times and thus the highest determinism for real-time control applications.

But let’s focus on EtherCAT and Profinet based on the question at hand, and look at how they use the Ethernet OSI:

EtherCATProfinet V1Profinet V2Profinet V3
Layer 7 ApplicationFieldbus Application Layer (FAL). Services and protocols
Layer 6 PresentationRSI (Remote service interface) or RPC (remote procedure calls) are used
Layer 5 SessionRSI (Remote service interface) or RPC (remote procedure calls) are used
Layer 4 TransportTCP/IP, UDP
Layer 3 NetworkIP can be inserted synchronously using Switch PortsIP servicesIP services are available but are async
Layer 2 Data LinkMAC ethernet
Layer 1 PhysicalFull-duplex 100 MBit/s copper (100BASE-TX) or fiber-optic (100BASE-FX) according to IEEE 802.3 can be used

an see from the table above that Profinet is available in three versions:

  • Version 1: Component-based (CBA): 100 ms cycle times

  • Version 2: Real-Time (software real-time): ~10 ms cycle times

  • Version 3: Real-Time (hardware real-time): < 1 ms cycle times

In Profinet version 2 and the transport (TCP/IP) layer is not used but is replaced with a dedicated process controller that is transported within the Ethernet frame. In version 2 this controller is implemented in software, whereas in version 3 it is implemented in hardware, achieving the best cycle times available within Profinet.

These changes do not prevent TCP and IP from being used by the protocol, but they avoid reliance on them.

Profinet Version 1, aka CBA (component-based automation), is not well supported anymore. It cannot handle a large number of variables and was seen as a bridge from the older Profibus to the Industrial Ethernet world. The focus today is on Profinet IO, formerly called SRT (software real-time) and IRT (asynchronous [hardware] real-time). 

The addition of special hardware in Profinet IO makes it fast and deterministic enough to be used for motion control applications. Since Profinet IO (IRT) is the closest to EtherCAT in performance, we will focus on it and not the other two versions in this top-level comparison:

Feature / ParameterEtherCATProfinet IO (IRT)
Max nodes per network65,53664 1)
Topologies SupportedLine, ring, tree, star, and combinations thereofLine, tree, star
Automatic network recoveryYesNo
Managed network switchesNot applicable to EtherCAT, which does not use switches or hubsRequired
Distributed clock technologyYesNo
Network Speed2 x 100 Mbps (full-duplex)2 x 100 Mbps (full-duplex)
Cycle Time100 µs< 1 ms
Synchronization precision~ 1 ns< 1 µs
Jitter< 1 µs< 1 µs
Susceptible to interruptions caused by traffic burstsNoYes. User must manage network loads 2)
Automatic slave device addressingYes, the master does this automatically, even when new devices are addedNo, the system integrator must ensure all devices have a unique address
Cost levelGenerally lower than most Fieldbus systemsHigher integration and software costs

1) More nodes are possible, but Siemens recommends 64 maximum. Siemens is the major manufacturer of Profinet devices and a supporter of the protocol.

2) Unlike EtherCAT, Profinet has no built-in way to restrict incoming traffic, so it is possible that network overloads can break the network. The user is responsible to manage network loads and ensure this does not happen.

By all major metrics, EtherCAT is a superior industrial ethernet solution. It is faster, easier to implement, more flexible, and low-cost in terms of topology, and is a completely open standard with no licensing costs.

Additionally, since we are focused on integrating DAQ systems with industrial process control systems, EtherCAT provides the required hooks, speed, low latency, and level of determinism that is required in order to make a successful DAQ integration. No other protocol that we have studied provides this.

EtherCAT summary

We hope that you have learned a lot about EtherCAT in this article and that you appreciate the benefits that it has in industrial control applications … especially those involving DAQ systems such as the IOLITE from Dewesoft, and expansion modules like KRYPTON and SIRIUS.

EtherCAT advantages

  • High performance - The fastest industrial ethernet topology at 200 Mb/s (full duplex = 100 Mbps x 2)

  • Deterministic - real-time operation with synchronization ~ 1 ns

  • Flexible topology - can be arranged in a ring, line, tree, a star without limit

  • No hubs or switches - virtually limitless network expansion

  • Easy operation - the master automatically assigns node addresses and sends clock sync messages to them

  • Simple configuration - there are no IP addresses or MAC addresses to maintain

  • Affordable - similar or lower price than a conventional Fieldbus network. Master controllers are relatively inexpensive.

EtherCAT applications

  • Factory automation

  • PLC networks

  • Servo drives control

  • Data Acquisition (DAQ) systems

  • Motion control and Machine control platforms

  • Robotics

  • Material and baggage handling systems

  • Weighing Systems

  • Printing presses

  • Semiconductor manufacturing

  • Metals and Pulp & paper manufacturing

  • Power plants

  • Test benches

  • Wind turbines

  • Farm machinery

  • Milling machines

  • Tunnel control systems

  • Safety systems

  • And hundreds more 

Advantages of DAQ and EtherCAT

  • High and low-speed data from advanced DAQ systems can now be integrated perfectly into a real-time control system

  • Eliminating redundant A/D processing means less complexity, better accuracy, and real cost savings

  • A separate very high-speed data file is available for advanced analysis if needed, and it is also synchronized with the PLC data