1

EtherCAT Tutorial, an introduction for real-time

hardware communication on Windows
Kevin Langloisa , Tom van der Hoevena , David Rodriguez Ciancaa , Tom Verstratena , Tomislav Baceka , Bryan
Convensa , Carlos Rodriguez−Guerreroa , Victor Grosua , Dirk Lefebera and Bram Vanderborghta
a
Robotics and Multibody Mechanics Research Group (R&MM), Vrije Universiteit Brussel,
Pleinlaan 2, 1050 Elsene.
The R&MM Research Group is a partner of Flanders Make.

Abstract—Setting up a real-time hardware communication ideal case, the reader should be able to have a simple test-setup
for applications such as precise motion control can be time- running in a matter of hours and with limited hardware costs.
consuming and confusing. Therefore, a tutorial on the deployment The fully detailed step-by-step tutorial can be obtained from
of an EtherCAT protocol is introduced. In this article, the authors
situate EtherCAT, briefly discuss the origins and working prin- the online appendix of this paper. Additionally, the tutorial
ciples, and mention advantages over other widely used protocols. contains successful implementations on several systems, such
Additionally, the main objectives of the tutorial and the required as the Cyberlegs Beta-prosthesis (figure 1).
software to complete it are presented. Online supplements are
included with this article, explaining all steps to run a Simulink
model in real-time on a Windows machine within a few hours. II. E THER CAT IN LAYMAN ’ S TERMS
Index Terms—EtherCAT, TwinCAT, Robotics, Realtime, Hard-
In this section the functionality of the EtherCAT protocol is
ware communication. presented. To begin with, the general concepts and origins of
industrial communication are briefly recalled, after which, the
general working principles that apply specifically to EtherCAT
I. I NTRODUCTION will be mentioned. All of this will be discussed with a certain
Particular application domains in the field of robotics and level of abstraction, as the goal of this tutorial is only to
industrial automation, require precise motion control capabil- give the reader a basic understanding of the system in order
ities. To achieve this, real-time control systems are key to to successfully deploy the protocol. Specific resources are
develop highly dynamic and intelligent robotic systems such conveyed to the reader interested to delve deeper into the field.
as prosthetic devices [1], [2], exoskeletons [3], legged robots
[4], humanoids [5], etc. A. EtherCAT origins
In these high-end motion control applications, there is a need
for very fast low-level control loops with data rates in the Technologies such as programmable microcontrollers and
region of 250 microseconds to 1 millisecond [6]. digital signal processors allowed for the replacement of purely
Furthermore, these systems require accurate time synchroniza- analog control loops with digital controllers such as PLCs.
tion and high data throughput to which Real-Time Ethernet This led to the core of industrial networking, termed fieldbus
(RTE) protocols have recently emerged as the leading solution protocols.
in industry. Many fieldbusses were developed in parallel, and essentially,
EtherCAT (Ethernet for Control Automation Technology) is a every company designed their own protocols, which naturally
widely used RTE protocol, which has shown excellent perfor- led to confusion for consumers and developers [11].
mances at a relatively low cost [7], [8], [9]. To conclude, the As a reaction to this, the Fieldbus War; a web of company
simplicity with which it can be deployed and run a Simulink politics and marketing interests which in 2000, led to the
model in real-time, makes EtherCAT a practical solution for establishment of an international fieldbus standardization, the
robotic prototyping. IEC 61158 [12], [13], [14], [15].
This article serves as an introduction to a detailed tutorial on In the mean time, Ethernet protocols were already well es-
how to deploy an EtherCAT master running a Simulink model tablished in the office world. The increased data rates of
in real-time on a Windows machine. In the following section, newer Ethernet standards (for example 802.3u Fast Ethernet)
the general working principles and the origins of EtherCAT made it easier to create real-time Ethernet protocols, as the
are summarized. This is in order to situate EtherCAT in the transmission and retransmission times are significantly shorter.
vast family of RTE protocols. Following, the advantages of the While primarly Fieldbus systems such as Profibus, SERCOS,
protocol relative to other RTE systems and CAN are presented. CAN and many others have allowed for distributed industrial
And to conclude, a summary of the objectives covered in the automation systems, performances of these protocols were
tutorial are explained, and the required software is listed. In the considered as too limited when compared to the available
performances (mainly in terms of data throughput) of non real-
Kevin Langlois is corresponding author. E-mail: kevin.langlois@vub.be time networks such as Ethernet [16].
 2

The variance of the response time of a signal is often referred

to as jitter. And low jitter is required due to the fact that
variance in time has a negative effect on control loops (the
derivative and integral portions of a control loop are affected
by time variation).
Secondly, the precise clock synchronization of the network.
Real clocks drift and will differ in time over long periods. As
pointed out in [20], unsynchronized networks usually suffer
from non-negligible jitters.
In conclusion, a family of RTE protocols has in recent years
evolved into a large number of solutions and standards. Two of
the most prominent representatives of this group are EtherCAT
and Profinet where the former is believed to offer the best
performance in terms of communication efficiency and short
cycle times [8].

B. Working principle
Today, there are over 25 different RTE solutions on the mar-
ket. They are offered by diverse manufacturers and academic
community [19], [9], [16].
These solutions integrate different working principles, such as
the method for encapsulation of process data into Ethernet
frames, the limitations on network topology, synchronization
of the network, implementation costs, etc.
First of all, EtherCAT is Industrial Ethernet and utilizes
standard frames and the physical layer as defined in the
Ethernet standard IEEE 802.3 [21]. Of the fast Ethernet
standards, 100BASE-TX is the most common and the one
used in EtherCAT networks. By utilizing the full duplex
(data transmission in two directions) features of fast Ethernet
allows effective data rates of 100 Mbit/s [17]. Furthermore,
the EtherCAT protocol employs the master/slave principle to
Figure 1. The CYBERLEGs (CYBERnetic LowEr-Limb CoGnitive Ortho- control access to the medium. The master node (typically the
prosthesiS) Beta-prosthesis, which actuates two degrees of freedom (knee and control system) sends the Ethernet frames to the slave nodes
ankle) is controlled via an EtherCAT protocol. The prosthesis integrates series (such as sensors and actuators), which can extract data from
elastic actuators, controlled by custom made electronics boards. [10]
and insert data into these frames.
These process data (of all network devices) are carried to-
gether in one or more Ethernet frames. This is the so-called
In addition, Ethernet bandwidth enables bus cycle times in the summation frame principle as opposed to the individual frame
microseconds range instead of the ms range which, together approach in which every frame carries process data for only
with the superior performance of modern PC-based control one device [8].
systems, allows one to close the control loops over the fieldbus With EtherCAT the standard Ethernet packet is no longer
that previously had to be closed locally in the peripheral received, interpreted and copied at every node, instead, slave
systems [17]. And finally, the large amount of research that devices process frames on the fly, reading and inserting data
has gone into developing Ethernet as a standard, as well as while the frame is passing through the device. This is handled
the cheap and readily available hardware, and the Internet of by hardware-integrated EtherCAT Slave Controllers (ESC) in
Things (IoT) [18] with which it becomes possible to connect the slaves.
almost anything with everything, anywhere, illustrates the wish The process data in industrial networks are relatively small
to develop RTE communication protocols. in quantity (only a few bytes), so that the summation frame
Nonetheless, many challenges presented themselves as the method, combined with the “processing-on-the-fly” feature of
requirements between those two domains - office world and the EtherCAT slaves, offers good system performance [8],
industry - are very different, as discussed in [11], [19]. The two [22]. Moreover, network topology plays an important role
main differences in requirements are related to communication when performance of the system is evaluated [22]. Important
time determinism and precise clock synchronization. aspects are not only cycle time or efficiency, but also cabling
Firstly, time determinism, is the requirement for the transmis- effort, diagnostic features, redundancy options and plug and
sion times of data packets to be known. This means that the play features. EtherCAT networks have no practical limitations
latency of a signal must be bounded and have a low variance. regarding the topology, line, star, tree, ring and all those
 3

combined with up to 65535 nodes per segment. - Data throughput (currently 100 times higher)
Then, for synchronization, EtherCAT relies on a clock syn- - Unlimited network length
chronization mechanism which is known as distributed clock - Increased system performances
(DC). - Use of established hardware components
Basically, clocks of all the DC-enabled slaves in the network
are synchronized with a common timing reference under direct IV. E THER CAT DEPLOYMENT
control of the master. [7]. Despite being rather simple and An EtherCAT master runs the EtherCAT network and
straightforward, the DC mechanism enables accurate synchro- communicates with all slaves. This master needs to be
nization (in small-to-medium systems, clock deviations are implemented on a real time operating system. Several
well below 1 microsecond). solutions have been developed, the one demonstrated in this
tutorial is TwinCAT (The Windows Control and Automation
III. E THER CAT: A DVANTAGES AND D ISADVANTAGES Technology).
In this section the advantages of setting up an EtherCAT Beckhoff provides the TwinCAT program, whose essential
network over other popular industrial communication proto- functionality is to reserve a number of physical cores on your
cols are discussed. personal Windows computer, and run the EtherCAT network
from these cores. Windows OS does not run on these cores
anymore and only operates on the cores specified by the user.
A. Ethernet based solutions
As an example, in this tutorial a quad-core laptop was used
There are more than 25 Ethernet based industrial protocols where two cores are reserved for the EtherCAT protocol and
on the market, but the list of protocols that have a considerable two cores are used for running Windows.
impact on industry is much shorter [23]. They are: The driver running on the EtherCAT reserved cores is a
- EtherCAT compiled version of a program that can either be written in
- Powerlink PLC language or C/C++ code. Beckhoff provides functionality
- Ethernet/IP to run compiled Simulink models (in C++) as drivers in the
- Modbus/TCP kernel space.
- ProfiNet
A multitude of performance evaluations of the remaining Below, a list is presented with the required software to
systems are reported in literature. The conclusions of these successfully turn a windows computer into a real-time target
studies are that EtherCAT is an overall highly performing real- running an EtherCAT network.
time protocol, when compared to the aforementioned protocols - Microsoft Visual Studio 2010 or higher. Required if
[24] [25] [8]. ProfiNet has advantages over EtherCAT in programming is directly done in C++, otherwise the shell
specific conditions, such as efficiency in asynchronous provided with TwinCAT can be used.
communication [26]. Ethernet/IP neither Modbus/TCP are - Matlab 2011 (or newer) including Matlab Coder.
deterministic and by consequence, are not suited for hard- - TwinCAT 3.1 (Free for non-commercial use).
real-time control. Major advantages of EtherCAT over the - Beckhoff TE1400 module (Free for simulink models with
ProfiNet and Powerlink are the costs of implementation and 5 inputs, 5 outputs and 100 blocks. Larger models require
the commercial diffusion of the technology [27]. Studies a license).
also predict the future pervasiveness of EtherCAT in the - Microsoft Windows Driver Kit 7 (Free) or higher.
industrial automation and robotics fields [23]. This suggests The tutorial guides the reader through the implementation of
EtherCAT to be the leading communication protocol for these a real-time control loop developed in Simulink and compatible
applications. with widely used hardware components (Maxon drivers and
motors). The reader is directed through the installation and
A major other communication protocol which is not part of configuration of all the necessary software on a Windows
Ethernet based systems are Serial protocols, of which CAN is machine. Debugging tools for TwinCAT and Simulink, as
widely used in the robotics field. well as instructions on the activation of Beckhoff licenses
are included. Additionally, experimental implementations are
B. Controller Area Network (CAN) shown, such as the active prosthesis depicted in figure 1, a
Controller Area Network (CAN) still is an adequate frameless brushless motor, depicted in figure 2, and more.
choice for low-cost industrial embedded networking. How-
ever, Ethernet-based protocols are now able to overcome the V. C ONCLUSION
shortcomings of CAN, such as limited baud rate and limited In this article the main outline of a broader hands-on tutorial
network length (1Mb/s at 120 feet). Although the very low- on the EtherCAT communication protocol for real-time hard-
cost implementation as well as the low resource requirements ware communication in robotics and automation applications
of CAN protocols still make it an adequate choice in certain is presented. This includes an introduction to fieldbus systems,
applications (such as automotive industry, small embedded and a general description of the EtherCAT features. Literature
solutions, aerospace), the overall advantages of EtherCAT over concerning the assessment of the protocol’s performances were
CAN are: conveyed to the reader and the advantages of the protocol were
 4

[3] S. Wang, L. Wang, C. Meijneke, E. Van Asseldonk, T. Hoellinger,

G. Cheron, Y. Ivanenko, V. La Scaleia, F. Sylos-Labini, M. Molinari,
et al., “Design and control of the mindwalker exoskeleton,” IEEE
transactions on neural systems and rehabilitation engineering, vol. 23,
no. 2, pp. 277–286, 2015.
[4] C. Semini, V. Barasuol, J. Goldsmith, M. Frigerio, M. Focchi, Y. Gao,
and D. G. Caldwell, “Design of the Hydraulically Actuated, Torque-
Controlled Quadruped Robot HyQ2max,” IEEE/ASME Transactions on
Mechatronics, vol. 22, pp. 635–646, Apr. 2017.
[5] F. Negrello, A. Settimi, D. Caporale, G. Lentini, M.Poggiani,
D. Kanoulas, L. Muratore, E. Luberto, G. Santaera, L. Ciarleglio,
L. Pallottino, D. Caldwell, N. Tsagarakis, A. Bicchi, M. Garabini, and
M. Catalano, “Walk-Man Humanoid Robot: Field experiments in a post-
earthquake scenario,” Robotics and Automation Magazine, 2017.
[6] P. Neumann, “Communication in industrial automation—What is going
on?,” Control Engineering Practice, vol. 15, pp. 1332–1347, Nov. 2007.
[7] G. Cena, I. C. Bertolotti, S. Scanzio, A. Valenzano, and C. Zunino,
“Evaluation of EtherCAT Distributed Clock Performance,” IEEE Trans-
actions on Industrial Informatics, vol. 8, pp. 20–29, Feb. 2012.
[8] G. Prytz, “A performance analysis of EtherCAT and PROFINET IRT,”
in Emerging Technologies and Factory Automation, 2008. ETFA 2008.
IEEE International Conference on, pp. 408–415, IEEE, 2008.
[9] J. Jasperneite, M. Schumacher, and K. Weber, “Limits of increasing the
performance of industrial ethernet protocols,” in Emerging Technologies
and Factory Automation, 2007. ETFA. IEEE Conference on, pp. 17–24,
IEEE, 2007.
[10] V. Grosu, C. Rodriguez Guerrero, B. Brackx, S. Grosu, B. Vanderborght,
and D. Lefeber, “Instrumenting complex exoskeletons for improved
human-robot interaction,” Instrumentation & Measurement Magazine,
IEEE, vol. 18, no. 5, pp. 5–10, 2015.
[11] B. Galloway, G. P. Hancke, and others, “Introduction to industrial control
networks.,” IEEE Communications Surveys and Tutorials, vol. 15, no. 2,
pp. 860–880, 2013.
[12] International Electrotechnical Commission, International Electrotechni-
cal Commission, Technical Committee 65, International Electrotechnical
Commission, and Subcommittee 65C, Industrial communication net-
works: fieldbus specifications. Part 1, Part 1,. 2014. OCLC: 957247601.
[13] M. Felser and T. Sauter, “The fieldbus war: history or short break
Figure 2. A Kollmorgen RBE type 12 pole frameless motor as used in the between battles?,” in Factory Communication Systems, 2002. 4th IEEE
test setup. The sensors in the motor include hall sensors for the commutation International Workshop on, pp. 73–80, IEEE, 2002.
of the motor and an optical incremental encoder for the position of the motor. [14] J.-P. Thomesse, “Fieldbus Technology in Industrial Automation,” Pro-
A small ring is clamped to the main axle to be able to turn the motor axle ceedings of the IEEE, vol. 93, pp. 1073–1101, June 2005.
manually. [15] J. R. Moyne and D. M. Tilbury, “The Emergence of Industrial Control
Networks for Manufacturing Control, Diagnostics, and Safety Data,”
Proceedings of the IEEE, vol. 95, pp. 29–47, Jan. 2007.
[16] J.-D. Decotignie, “Ethernet-Based Real-Time and Industrial Communi-
discussed. Conclusively, the technical features of EtherCAT cations,” Proceedings of the IEEE, vol. 93, pp. 1102–1117, June 2005.
make it an outstanding communication protocol for appli- [17] M. Rostan, J. E. Stubbs, and D. Dzilno, “EtherCAT enabled advanced
control architecture,” in Advanced Semiconductor Manufacturing Con-
cations requiring precise motion control, while keeping the ference (ASMC), 2010 IEEE/SEMI, pp. 39–44, IEEE, 2010.
implementation cost to a minimum. Additionally, it provides [18] F. Xia, L. T. Yang, L. Wang, and A. Vinel, “Internet of Things,”
compatibility with common hardware and software. With the International Journal of Communication Systems, vol. 25, pp. 1101–
1102, Sept. 2012.
appended tutorial the reader should be able to deploy a reliable [19] M. Felser, “Real-Time Ethernet - Industry Prospective,” Proceedings of
real-time communication system and run a Simulink control the IEEE, vol. 93, pp. 1118–1129, June 2005.
loop in a matter of hours. [20] P. Ferrari, A. Flammini, D. Marioli, A. Taroni, and F. Venturini,
“Experimental analysis to estimate jitter in PROFINET IO Class 1
networks,” in Emerging Technologies and Factory Automation, 2006.
ETFA’06. IEEE Conference on, pp. 429–432, IEEE, 2006.
ACKNOWLEDGMENT [21] “IEEE Standard for Ethernet,” IEEE Std 802.3-2015 (Revision of IEEE
Std 802.3-2012), pp. 1–4017, Mar. 2016.
This work was supported in part by the ERC-grant [22] M. Knezic, B. Dokic, and Z. Ivanovic, “Topology aspects in Ether-
”SPEAR” (no.337596) and the Strategic Basic Research CAT networks,” in Power Electronics and Motion Control Conference
(EPE/PEMC), 2010 14th International, pp. T1–1, IEEE, 2010.
project ”Yves” of Flanders Make. [23] W. Voss, The Future of CAN/CAN open and the Industrial Ethernet
Challenge. President esd electronics, Inc USA, 2011.
[24] X. Wu, L. Xie, and F. Lim, “Network delay analysis of EtherCAT and
R EFERENCES PROFINET IRT protocols,” in Industrial Electronics Society, IECON
2014-40th Annual Conference of the IEEE, pp. 2597–2603, IEEE, 2014.
[1] L. Flynn, J. Geeroms, R. Jimenez-Fabian, B. Vanderborght, N. Vitiello, [25] J. Robert, J.-P. Georges, E. Rondeau, and T. Divoux, “Minimum cycle
and D. Lefeber, “Ankle–knee prosthesis with active ankle and energy time analysis of Ethernet-based real-time protocols,” International Jour-
transfer: Development of the cyberlegs alpha-prosthesis,” Robotics and nal of Computers, Communications and Control, vol. 7, no. 4, pp. 743–
Autonomous Systems, vol. 73, pp. 4–15, 2015. 757, 2012.
[2] P. Cherelle, V. Grosu, A. Matthys, B. Vanderborght, and D. Lefeber, [26] R. Schlesinger, A. Springer, and T. Sauter, “New approach for improve-
“Design and Validation of the Ankle Mimicking Prosthetic (AMP-) ments and comparison of high performance real-time ethernet networks,”
Foot 2.0,” IEEE Transactions on Neural Systems and Rehabilitation in Emerging Technology and Factory Automation (ETFA), 2014 IEEE,
Engineering, vol. 22, pp. 138–148, Jan. 2014. pp. 1–4, IEEE, 2014.
 5

[27] D. Orfanus, R. Indergaard, G. Prytz, and T. Wien, “EtherCAT-based

platform for distributed control in high-performance industrial applica-
tions,” in Emerging Technologies & Factory Automation (ETFA), 2013
IEEE 18th Conference on, pp. 1–8, IEEE, 2013.
 EtherCAT Tutorial, an introduction for real-time
hardware communication on Windows
Kevin Langlois* a , Tom van der Hoevena , David Rodriguez Ciancaa , Tom
Verstratena , Tomislav Baceka , Bryan Convensa , Carlos
Rodriguez−Guerreroa , Victor Grosua , Dirk Lefebera and Bram
Vanderborghta
a
Robotics and Multibody Mechanics Research Group (R&MM),
Vrije Universiteit Brussel,
Pleinlaan 2, 1050 Elsene.
The R&MM Research Group is a partner of Flanders Make.

*
Kevin Langlois is corresponding author. E-mail: kevin.langlois@vub.be

1
1 Introduction
Robots are a growing group of devices that can be used for an ever increasing
variety of tasks. Typically robotics is a profession where many different areas
of expertise are to be combined into a device that for example interacts with
a human or works in a factory. Unfortunately, for the majority of robotics re-
searchers excellence in all areas simultaneously is not easy. Often, excellence in
mechanical engineering appears to go hand in hand with the slight neglect of
practical programming of control architectures.
It is typically in a prototype lab, as the example above illustrated, that many
mechanically oriented engineers need ’that one control guy in the lab’ to get their
mechanical structures to move, let alone be controlled properly. Figuring out
the best hardware control strategy, learning to code in a new language and many
other practical problems often discourage the mechanically oriented engineer to
properly control their own test-setups. Naturally you will have to annoy your
fellow researchers with questions on ‘how to move your setup or robot’.
If we break the problem of ’moving your setup or robot’ down a little there are
three different building blocks that need to be considered.
Firstly, on the most mechanical level there are motors that actuate your setup
and sensors that measure the state of the motor. These motors naturally need
to be regulated with a voltage, and this task is performed by a low-level mo-
tor controller. The low-level motor controller generally reads the sensors on the
motor axis to determine the current motor state and accepts a position/veloci-
ty/torque setpoint from a high level control program to set a desired motor state.
The low-level motor controller then moves from the current state to the desired
state by applying the required voltage to the motor.
Secondly, on the most control oriented level there is the high-level control pro-
gram. This control strategy could for example be that your motor has to track
a certain reference trajectory with the use of a PID controller.
Thirdly, there is the communication between the high level control program and
the motor controller. This ’hardware communication layer’ can be done in many
ways. A widely used high performance solution is to use an EtherCAT network.
Such a network can easily run at 1kHz on your own computer to ensure real-time
communication between your measurement/actuation setup and your personal
computer. Therefore this tutorial will focus on exactly what EtherCAT is, what
make up the key components of an EtherCAT network and most importantly:
how you can setup your very own network for your measurement setup or your
robot.
A tutorial is presented on how to get a Simulink measurement model running on
a Windows machine with a real-time EtherCAT network. This tutorial includes
real-time tracking of a reference trajectory with a motor and reading values real-
time from sensors.
Throughout the tutorial additional information is presented on the components
that make up the EtherCAT network. As you will read in this instructive docu-
ment, an EtherCAT network consists of a master and several slaves. The tutorial
part focuses on the master side of the EtherCAT network, or in other words the

2
software that runs and maintains the communication in the network.
The remainder of this document is structured as follows. Section 2 contains
an instructive and comprehensive overview of what an EtherCAT network is
and what the key components are in such a network. In section 3 the tutorial
for a prototyping test setup is presented. All required (mostly free) software
is clearly listed and the main installation sequences are illustrated with the es-
sential ’print-screens’. The basic control systems developed in this tutorial are
based on Maxon driver boards and controllers, however any EtherCAT slave
module or device can also be used to achieve a running setup. Debugging tools
in Simulink and TwinCAT are introduced in section 4. Section 5 is dedicated
to the licensing of TwinCAT products, allowing the user to integrate more com-
plex models. Additionally, we added some experimental implementations of the
EtherCAT protocol in section 6, illustrating possible application domains. Lastly
in Appendix A an informative section is presented on the different EtherCAT
slaves. Throughout the tutorial, references are made to any available additional
online material and examples.

As a final note, it should be stated that there is no part in this work which
can be considered really ‘novel’. The majority of the work can be constructed
by combining various documents and films on the web and the EtherCAT tech-
nology exists already for quite some years. The tutorial is merely intended as an
introduction to a high performance hardware communication protocol for hard
real-time applications and which is compatible with Simulink. In this way we
hope to give the reader a shortcut to a functional system with low hardware
costs.

3
2 EtherCAT, TwinCAT, master and slaves
This section contains a start to the background information on EtherCAT net-
works. It is not the intention of this tutorial to explain the detailed working
principles of the protocol, but merely give the reader a general understanding
of the technology in order to complete this tutorial. For a more detailed expla-
nation, the reader is referred to the Beckhoff Information System website [1].
First, a small section is devoted to what EtherCAT is and what makes up an
EtherCAT network. Secondly, a section is devoted to TwinCAT as a master
environment to an EtherCAT network.

2.1 What is EtherCAT?

EtherCAT is a way to communicate between a computer and motor drives and
all sorts of analog and digital inputs and outputs (IO). The main advantage
compared to any other ways to do the same communication like USB, RS232,
CAN, etc. is that this type of communication is Industrial Ethernet (and utilizes
standard frames and the physical layer as defined in the Ethernet Standard IEEE
802.3 [2]) and can achieve real-time communication. Realtime communication
generally means that information (like setpoints for the motor or readings of the
sensors) can be sent over the network at a high frequency (1kHz or higher).
Today, there are over 25 different Real Time Ethernet (RTE) solutions on the
market offered by diverse manufacturers and the academic community [3], [4], [5].
These solutions use different approaches regarding the method for encapsulation
of process data into Ethernet frames, which offer different advantages.
EtherCAT protocol is based on so-called summation frame principle in which
the process data of all network devices are carried together in one or more Eth-
ernet frames, as opposed to the individual frame approach in which every frame
carries process data for only one device [6]. Also, with EtherCAT the standard
Ethernet packet (containing data) is no longer received, interpreted and copied
at every slave, instead, slave devices process frames on the fly, reading and in-
serting data while the frame is passing through the device. This is handled by
hardware-integrated EtherCAT Slave Controllers (ESC) in the slaves.
Process data in industrial networks are relatively small in quantity (only a few
bytes), so that the summation frame method, combined with the “processing-
on-the-fly” feature of the EtherCAT slaves, allows to achieve good system per-
formance when compared to other protocols [6], [7].

2.2 Master/Slave Architecture

EtherCAT employs the master/slave principle to control access to the medium,
see figure 1. The EtherCAT master (typically the control system) uses a standard
Ethernet port and network configuration information stored in the EtherCAT
Network Information file (ENI). The ENI is created based on EtherCAT Slave
Information files (ESI) which are provided by the vendors for every device. The
master node sends the EtherCAT frames to the slave nodes, which can extract

4
data from and insert data into these frames, schematically represented in figure
2. Slave nodes are devices such as EPOS3 motor drives which have the Ethernet
ports on it to communicate with an EtherCAT master. An EtherCAT master is
a computer device which maintains the data communication between the master
and the different slaves.

Figure 1: The EtherCAT network architecture, consisting of a master/slave pro-

tocol.

Figure 2: The EtherCAT frame simply replaces the IP frame of a standard Ether-
net message. Thus, the Ethernet frame does not need modification, contributing
to flexibility for EtherCAT.

5
 The EtherCAT frame starts with a standard header which tells the nodes
how long the EtherCAT portion of the frame will be. After the header, the
frame contains Process Data Objects (PDOs).
The PDOs correspond to the number of nodes and contain data for a node. The
PDOs are individually addressed, telling the nodes which PDOs to take. This
is an important term as we will later see, when our control model’s inputs are
linked to the hardware outputs. The final portion of the frame is the working
counter which ensures the entirety of the frame was received by the node.
Network topology plays an important role when performance of the system is
evaluated [7]. Important aspects are not only cycle time or efficiency, but also
cabling effort, diagnostic features, redundancy options and plug and play fea-
tures. EtherCAT networks have no practical limitations regarding the topology,
line, star, tree, ring and all those combined with up to 65535 nodes per segment.
For synchronization, EtherCAT relies on a clock synchronization mechanism
which is known as distributed clock (DC). Basically, clocks of all the DC-enabled
slaves in the network are synchronized with a common timing reference under
direct control of the master. [8]. Despite being rather simple and straightfor-
ward, the DC mechanism enables accurate synchronization (in small-to-medium
systems, clock deviations are well below 1 microseconds).

2.3 TwinCAT
On the software side, different master environments exist, such as Etherlab,
Labview-realtime and SOEM/ebox, but in this tutorial we will use TwinCAT
(The Windows Control and Automation Technology). The reason being that it
does not require a Linux computer and is generally easy and user-friendly to op-
erate. Basically, TwinCAT allows to to reserve some cores in a personal windows
computer, make a link to the kernel space of the computer and turn it into a
realtime target that can run and maintain an EtherCAT network. TwinCAT 3
eXtended Automation Engineering Environment (XAE) allows the integration of
additional programming languages or tools, such as MATLAB and Simulink, see
figure 3. Users can design their control algorithms in MATLAB and Simulink,
and use Simulink Coder to generate TwinCAT modules. The integration of
MATLAB and Simulink into TwinCAT 3 facilitates a connection to the scien-
tific field by integrating the TwinCAT real-time execution of tasks and modules
into Microsoft Visual Studio. In conclusion, we know that TwinCAT is a pro-
gram to implement a realtime target that runs an EtherCAT network between
a master, such as a Windows machine, and the slaves, such as motor drives and
IO. In the next section, we present the implementation of a Simulink model on
this realtime target created in TwinCAT.

6
Figure 3: TwinCAT 3 eXtended Automation Engineering Environment (XAE)
allows the integration of MATLAB and Simulink .

7
3 Tutorial
In this section the implementation of a Simulink model on the realtime target
created with TwinCAT is detailed.
To start, in section 3.1, the required software to complete this tutorial is listed.
This is assuming a computer with only blank Windows 7 or 10 running on it.
Secondly, in section 3.2, the software installation sequence is explained.
Thirdly, section 3.3 introduces a simple Simulink model along with the instruc-
tions on how to compile the model into a module that can be run on the Twin-
CAT target.
Fourthly, in section 3.4, the realtime TwinCAT target is set up, the com-
piled Simulink model is imported and the inputs and outputs of the (compiled)
Simulink model are connected to the variables of the EtherCAT network slaves.
Lastly, in section 3.5, an instruction is presented on how to run the compiled
Simulink model on the TwinCAT target.

3.1 List of required and provided software

This is a list of the required software to complete the tutorial. The following pro-
grams must already be preloaded on the computer or installed from the various
software companies that provide them:

- Microsoft Visual Studio professional 2010 or newer.

- MATLAB 2015b (including MATLAB Coder, which in some cases is part

of the academic license).

- The ’general target model’ as included with this tutorial. This can however
easily be replaced by any other model.

- Microsoft Windows Driver Kit 7 or higher (Installation “Microsoft Win-

dows Driver Kit (WDK)”).

Additionally the following software should be downloaded from the Beckhoff site
when specified in the tutorial, so not right now.

- TwinCAT 3.

- Beckhoff TE1400 module.

3.2 Installation and downloading of software

Please follow the instruction sequence for all programs presented in this section
carefully, it does matter in which order programs are installed.

8
1. Install Microsoft visual studio 2010 professional
- In order to build C++ programs with TwinCAT a version of Visual studio
professional 2010 or newer is needed. If the (future) intention is to code
directly in C/C++ in TwinCAT install the professional version of Visual
Studio.

- If you will only use PLC programs or compiled Simulink models, you may
use the Visual Studio shell that comes with TwinCAT.

2. Install TwinCAT 3 eXtended Automation Engineering

TwinCAT 3 eXtended Automation Engineering Environment (XAE) allows the
integration of additional programming languages or tools, such as MATLAB and
Simulink. Users can design their control algorithms in MATLAB and Simulink,
and use Simulink Coder to generate TwinCAT modules.

1. Navigate to the Beckhoff website to download the TwinCAT 3.1 - eXtended

Automation Engineering (XAE) program:
https://www.beckhoff.com/english.asp?download/tc3-downloads.htm

2. Select TE1xxxx Engineering.

3. Select TwinCAT 3.1 - eXtended Automation Engineering (XAE).

4. Start the download and follow the instructions. When asked whether you
want to integrate TwinCAT into Visual Studio 2010 professional, do it.
Don’t use the Visual Studio 2010 shell.

5. TwinCAT 3.1 should now run inside Visual Studio 2010 professional.

3. Install Microsoft Windows Driver Kit 7 (WDK)

Implementing TwinCAT3 C++ modules requires parts of the Microsoft “Win-
dows Driver Kit 7 (WDK)”. The steps to install the driver kit can be found on
the Beckhoff Information System website [1] and are represented here:

1. Download the “Windows Driver Kit 7.1” from the Microsoft Download
Center.
http://www.microsoft.com/downloads/en/details.aspx?displaylang=
en&FamilyID=36a2630f-5d56-43b5-b996-7633f2ec14ff

2. After download burn a CD from the downloaded ISO image or make use
of virtual CD software.

3. Start “KitSetup.exe” from the CD / downloaded ISO image (on Windows7

machines please start installation with “Run As Administrator...”).

4. Select option “Build Environment” - all other components are not required
by TwinCAT 3 - and “OK” to proceed. See figure 4.

9
Figure 4: Installation of the Windows Driver Kit 7.1. Select option “Build
Environment” - all other components are not required by TwinCAT 3

5. After agreeing to the Microsoft EULA license please select a target folder
for the installation.

6. Start installation with selecting “OK”, such as depicted on figure 5.

Figure 5: Installation of the Windows Driver Kit 7.1. Select option ”Build
Environment” - all other components are not required by TwinCAT 3

7. Navigate to “Start”-> “Control Panel” -> “System” and select “Advanced

system settings”

8. Select tab “Advanced” and click on “Environment Variables...”

10
 9. In the lower area for “System variables” select “New..” and add following
information:
Variable name WINDDK7
Variable value C:\WinDDK\7600.16385.1
The path may differ due to a different version of Windows Driver Kit or a
different installation path during installation. See figure 6.

Figure 6: Environment variable ‘WINDDK7’ added.

10. After installation re-login or reboot the machine for establishing the new
environment variable settings.

Test signing

- For the implementation of TwinCAT 3 C++ modules on x64 platforms

the driver must be signed with a certificate.
11. Enter the following in the Visual Studio prompt with administrator rights:
makecert -r -pe -ss PrivateCertStore -n CN=MyTestSigningCert MyTestSigningCert.cer

- This is followed by creation of a self-signed certificate, which is stored in the file

“MyTestSigningCert.cer” and in the Windows Certificate Store.

11
 - The result can be verified with mmc (Use File->Add/Remove Snap-in->Certificates).
See figure 7.

Figure 7: Verify the self-signed certificate was created.

- Configuring the certificate such that it is recognized by TwinCAT XAE on the

engineering system.

12. Set the environment variable “TWINCATTESTCERTIFICATE” to “MyTest-

SigningCert” in the engineering system or edit the post-build step of “Debug—TwinCAT
RT (x64)” and “Release—TwinCAT RT (x64)”.
The name of the variable is NOT the name of the certificate file, but the CN name
(in this case MyTestSigningCert). See figure 8.

Figure 8: Set the environment variable “TWINCATTESTCERTIFICATE” to

“MyTestSigningCert”.

13. Activate signing mode, so that Windows can accept the self-signed certificates.
Use a command prompt to execute the following:
bcdedit /set testsigning yes

14. Restart the target system.

12
 - If test signing mode is enabled, this is displayed at the bottom right of the desktop,
such as depicted on figure 9. The PC now accepts all signed drivers for execution.

Figure 9: If test signing mode is enabled, this is displayed at the bottom right of
the desktop.

15. TwinCAT is now configured to be able to run C++ modules.

4. Install MATLAB
- Install a version newer than MATLAB 2011a. The most important thing
is that the version of MATLAB includes the MATLAB/Simulink Coder
app.

5. Install Beckhoff TE1400 Module

The TwinCAT MATLAB/Simulink Target offers System Target Files for the
use of the MATLAB/Simulink coder. It enables the generation of TwinCAT 3
runtime modules, which can be instanced and parameterised in the TwinCAT 3
engineering environment.

1. Navigate to the Beckhoff website to install the TE1400 module:

https://www.beckhoff.com/english.asp?twincat/te1400.htm.

2. Press the download button and follow all instructions.

3. Start the ‘TE1400-TargetForMatlabSimulink’ Setup.

4. Start MATLAB and execute %TwinCAT3Dir%..\Functions\TE1400- Tar-

getForMatlabSimulink\SetupTwinCatTarget.p such as on figure 10.
Do not forget to run MATLAB with administrator privileges.

13
 Figure 10: Installation of the TE1400 module

All required software is installed and should be operational

3.3 Simulink model and compilation
In this section we analyse the setup of the Simulink model file that is provided
online with this tutorial. We will compile the Simulink model and import it in
TwinCAT which is running in Visual Studio.

Main model
In figure 11 the main Simulink model is depicted. Some text is provided to
function as a quick memory-help on how to use the model.
In the top section of the model the inputs that have to be read from hardware
can be distinguished. The ‘TC Module Input’ blocks can be found in the library
browser under ‘beckhoff’. The data type of the input blocks must be configured
to match the data type that the real hardware device emits over the EtherCAT
network. For example, a motor drive sends the current motor position as an
INT16, the input block in Simulink has to be configured to read an INT16. The
‘To file’ block has to be configured for arrays. Also make sure to change the file
location to .../general target/temp data. Further reading on this topic can be
found in the document http://download.beckhoff.com/download/document/
automation/twincat3/TwinCAT_3_Matlab_Simulink_EN.pdf on page 24.
In the middle section the motor trajectory is read from the work space and
written to a model output. This ‘TC Module Output’ block can also be found in
the model library browser under ’beckhoff’. The trajectory has to be generated
with the ’generate model trajectory’ m-file that is supplied with the Simulink
model. Do not forget to assign a valid data type for the trajectory output.
The bottom section handles the enabling of the motor drive and the starting and

14
stopping of the data saving.

Figure 11: The main Simulink model for this tutorial. The top section are the
three inputs read from the hardware (two encoders and one torque sensor), the
middle section is where the desired position is sent to the driver’s input, and the
lower section enables or disables the logging of the data and the motor driver.

Compiling the model to C++ code

In order to run the Simulink model on the real-time target created in TwinCAT,
the Simulink model must be build into C++ code and compiled onto the target.
In this section the main screens are presented for setting up MATLAB Coder to
do the job.

In the Simulink model press ’Ctrl+H’ to open the model explorer. The screen in
figure 12 will appear. Click on ’Configuration (active)’ in the menu on the left.
There are three important items in the list in the middle. These three items are
’Solver’, ’Data Import/export’ and ’Code generation’.

15
 The solver has to be set to a fixed step solver because the model runs at a
fixed frequency. The stop time does not really matter, the model will continue
running after the specified ten seconds. Under ´additional options’ the fixed
step time can be set to 0.001 s, which equals a cycle frequency on the EtherCAT
network of 1kHz, such as in figure 13.
In figure 14, the ‘Data Import/Export’ item is configured (midpane). Saving
data to the MATLAB workspace is not possible via this route, but has to be
done in a different way that we will see later.

Figure 12: The main screen of the model explorer, with on the left the ‘Configu-
ration (Active)’ pane and in the middle, the ‘Solver’, ‘Data Import/Export’ and
‘Code Generation’ items.

Figure 13: The solver has to be set to a fixed step solver. The stop time does
not really matter, the model will continue running after the specified ten seconds.
Under ‘additional options’ the fixed step time can be set to 0.001 s.

16
Figure 14: In the ‘Data Import/Export’ block, switch off the ´save to workspace’
boxes.

In figure 15 the target is selected. Click ‘Browse...’ to select Twincat.tlc.

Move over to the tab ‘Tc build’, see figure 16. Here, we publish the Simulink
TcCOM module. This module is what we will import in TwinCAT to run on
the real-time target. In this case we publish both a Realtime(RT) 64 bit (x64)
and a Realtime(RT) 32 bit (x86) module such that it can also run on a 32 bit
machine if required. This is not mandatory, you can also build the module for
your system only.

17
Figure 15: Enlarged screen for code generation where the target can be selected.
Browse for the Twincat.tlc file.

18
Figure 16: The screen for publishing the Simulink TcCOM module. If you run
a 32 bit computer make sure to publish the generated C++ module for a 32 bit
target (x86).

The last two tabs that need to be considered are the ‘Tc External Mode’ and
‘Tc Advanced’ tabs. In the ‘Tc External’ tab, figure 17, we can select whether
we want to control the Simulink model running on the real-time target with
the Simulink model running in MATLAB. Technically this means that an ADS
(Automation Device Specification) protocol is set up between MATLAB and
TwinCAT. It is a method for accessing the Bus Coupler information directly
from the PC. For the rest of the tutorial we will use this protocol, and thus,
check all boxes under the tab ‘Tc External Mode’.

19
Figure 17: The ADS (Automation Device Specification) protocol is set up between
MATLAB and TwinCAT. Check all boxes to allow the Simulink model running on
the real-time target to be controlled by the Simulink model running on MATLAB.

Lastly, move to the ‘Tc Advanced’ tab, depicted in figure 18. In this tab, we
can decide how the generated module should be called. Make sure the task
assignment is set to ‘ManualConfig’, the callby is set to ‘CyclicTask’ and step
size is set to ‘RequireMatchingTaskCycleTime’. This is because we require the
model to be called by a ‘task’ that has the same cycle time as we compiled for
the model (1ms).

Figure 18: This is the ‘Tc Advanced’ tab. Here, we can define how to call the
real-time generated module. In this case, we require the model to be called by a
’task’ that has the same Cycle time as we compiled for the model (1ms).

The remaining tabs are not considered important at this point. Having com-
pleted these steps you can return to the ‘General’ tab and press ‘Generate Code’.
MATLAB will now start compiling your model and publish it to a directory
where TwinCAT can read it.

20
3.4 Setting up the TwinCAT real-time target and importing and
connecting the compiled Simulink model
In order to set up the TwinCAT realtime target, we follow 4 steps:
1. Any EtherCAT slaves, like in our case an EPOS3 motor drive and a Beck-
hoff module, are connected to the EtherCAT network master.
2. Importing the compiled Simulink model into the TwinCAT environment
and setup to run on the real-time target.
3. Connection of the inputs and outputs of the Simulink model to the Ether-
CAT slaves in the network.
4. The ’ExtendedFileWriter’ module for writing data from the real-time tar-
get to the hard drive of the computer is added and connected in TwinCAT.
All these steps are explained in this section.

TwinCAT real-time target

Start up Microsoft Visual Studio. Create a new project (File -> New -> Project).
We make a TwinCAT XAE project, and name it ‘simulink realtime target’. We
now see the main TwinCAT screen as depicted in figure 19. On the left the
project solution tree can be found. Under ’System’ the parameters for the real-
time part of the system and the task execution speeds are set. Also, the Simulink
model will show up here. Under PLC, we can write programs to be executed
by our hardware in PLC language. Under C++ we can write programs to be
executed by our hardware in C++ code and under I/O all hardware in our
system is listed.

Figure 19: The main screen of a TwinCAT project with on the left, the project
solution tree.

21
In the tree, by expanding the ‘SYSTEM’ tab we can create a new task (right
click ’tasks’ -> add new item). Name it ‘SimulinkTarget’ and reduce the ’Cycle
Ticks’ to 1. This means the task cycle time is set to 1ms.

Simulink model import

Just like any program written in PLC language or C++ language, the Simulink
model will function as the ’high level control program’ that works with the
hardware listed under IO. To import the Simulink model, add a new TcCOM
object (right click TcCOM Objects -> add new item). Expand TE1400 Module
Vendor, like in figure 20, and select ‘general target model’.

Figure 20: To import the real-time generated Simulink model into TwinCAT, a
TcCOM object was added. If needed, multiple models can be imported and run
simultaneously.

By expanding ‘TcCOM Objects’ and double clicking ‘Object1’, the Simulink

block diagram can be found, see figure 21.

22
Figure 21: By expanding ‘TcCOM Objects’ and double clicking ‘Object1’ in the
left tree, we can find the Simulink block diagram under the ’Block Diagram’ tab.

Now select the ’context’ tab (double clicking ‘Object1’ in the left tree). In
the ’task’ dropdown menu select the ‘SimulinkTarget’ task, see figure 22. This
procedure links the model to the task that we created before. In this way, the
task is updating every 1ms which means the Simulink model will run at 1kHz.

23
Figure 22: This screen shows how to link the Simulink model to the task that will
run one loop of the model, every 1ms.

EtherCAT slaves
In the top menu bar, click TwinCAT and select from the drop down menu ’Show
Realtime Ethernet Compatible Devices...’ like in figure 23. Under ‘Compatible
Devices’ select your Local Area Connection and install, however in your univer-
sity/lab it may already be done. In this case you can skip this step. TwinCAT
only supports intel LAN cards. After installation, your Local Area Connection
should show up under ‘Installed and ready to use devices’, like in figure 24.

24
Figure 23: This screen shows the access to the configuration of the Local Area
Network card.

Figure 24: Here, the LAN card can be enabled or disabled. If TwinCAT does not
recognise your IO (later in this tutorial) or does not go into ’run mode’, reset
your LAN card.

Now we can add some hardware to the project. In the tree on the left expand the
I/O tab. Right click ‘devices’ and click scan like in figure 25. When all connected

25
hardware is powered and running, this wizard will find it automatically in the
network.

Figure 25: This wizard will automatically look for hardware connected in the
network.

If devices are recognised we get notified such as in figure 26. If a ‘hardware pro-
tocol’ was found, something is wrong. Check first if all the devices are powered
on. Another issue can be outdated firmware on the device.

Figure 26: When devices are recognised, you will get this screen.

Proceed to add the hardware to the project and scan the network for boxes.

26
Append the axis to NC axis. Activate free run. In the model tree on the
left you will now see the EtherCAT hardware in your network. As depicted in
figure 27 in our case, this is an EPOS3 drive, a MIR device to read encoders
and a Beckhoff EK1100 module. TwinCAT can only recognise devices if it has
their .ESI file. This file contains information about the device functionality
and settings. In this case, you need to download the ESI (EtherCAT Slave
Information) file for the EPOS3 drive from the Maxon website and copy it to
the folder Twincat\3.1\Config\Io\EtherCAT. After doing this, TwinCAT should
be able to see your device. Note it is possible that an etherCAT enabled device
requires more than a .xml (ESI) file for proper operation. Always ask the device
manufacturer if you have all the right files.

Figure 27: Overview screen with on the left all hardware listed in the model
solution tree.

Double click the EPOS3. Select the ’slots’ tab like in figure 28. In here, the
EPOS3 can be configured in different modes. Each mode has a specific PDO
configuration so if you change mode it is likely that links you make to these
PDO entries (as explained later in the tutorial) will be lost. A solution can be
CST/CSV/CSP mode where you can toggle between any mode you want and
the PDO’s contain all entries for current position/velocity/torque and curren-
t/velocity/position setpoints. Further reading on the EPOS3 can be found in
the application notes of the Maxon website.

27
Figure 28: This is the ‘slots’ tab where the EPOS3 can be configured in different
modes.

Put the drive in Continuous Cyclic Position (CSP) mode by selecting whatever
is in the left screen (in this case PP/PV mode), pressing the ’x’ button in the
middle, selecting the CSP mode in the right screen and pressing the ’<’ button
in the middle. This means a new position setpoint can be sent every 1ms to the
controller and the axis will move there. In figure29 you can also see that if you
expand the MIR device you can see the content of one of the TxPDO’s where
the encoder values are clearly indicated.

Figure 29: The content of one of the TxPDO’s where the encoder values are
clearly indicated.

28
PLC project
The next thing to do is to make a PLC program to enable and disable the motor
drive. This could also be done in the Simulink model but for illustrative purposes
it will be achieved with a small PLC project.
Right click PLC in the tree menu on the left and click ’Add new item’. Name
the new standard PLC project ’EnableMotor’ and click ok. You have a situation
like in figure 30 now if you expand the menus in the PLC tree.

Figure 30: Right click PLC in the tree menu on the left and click ’Add new item’.
Name the new standard PLC project ‘EnableMotor’.

Right click POUs and select ’add’ and then ’POU’ like in figure 31. Name
the POU ´EnableMotor’ and select ´program’ and ´Structured text’ for the
implementation language. Click Open. Double click the new ´EnableMotor
(PRG)’ program. Copy the content from figure 32 to your own program. Go to
the MAIN (PRG, in the left tree) and copy the content of figure 33 to your main
program. Save all and press ´F7’ to build your solution.

Inputs and outputs connection

We can connect the variables in the hardware PDO’s with the in- and outputs of
the Simulink model. As depicted in figure 34 right click for example RelPosition
and click ’change link’.

29
Figure 31: In this screen we add a Process Organization Unit (POU). As can be
seen from the screen, a lot more, such as VISU’s (visualizations/frontscreens)
can be added if desired.

Figure 34: Screen for establishing the connection between a hardware entry (in
this case the RelPosition of an encoder and an entry on the control program.

Like in figure 35, select ‘position end encoder’ and click ok. The link is now
established between the position of the end encoder in the hardware and the
input of the Simulink block that receives the position of the end encoder.

30
Figure 32: Here, the commands to enable the motor can be written. The neces-
sary and specific commands for a Maxon EPOS3 controller can be found in the
application notes of the controller.

Figure 33: This is the program ‘EnableMotor’ from the main program (MAIN),
which is executed by the task ’PlcTask’.

31
Do the same for the other entries as well.

Figure 35: Screen for establishing the connection between a hardware entry (in
this case the ‘RelPosition’ of an encoder and and entry on the control program
(in this case ’Position end encoder’).

Specifically, change the links under ‘EnableMotor Instance’ to their appropriate

connections. In figure 36 ’Enable Motor’ is linked.

Figure 36: This screen shows the link of the ’enable motor’ PLC variable.

In figure 37 ’mode of operation’ is linked.

32
Figure 37: This screen shows the link of the ’mode of operation’ PLC variable.

In figure 38 the ’controlword’ is linked.

Figure 38: This screen shows the link of the ’controlword’ PLC variable.

ExtendedFileWriter module
A last thing that needs to be done is to make sure that the ‘to file’ block in
the Simulink model can write data from the real time part of the computer
to MATLAB which is not running realtime. This can be achieved with the
TcExtendedFileWriter object. Right click ‘TcCOM objects’ and select ’add new

33
item’. Referring to figure 39, select the ‘ExtendedFileWriter’ module and click
ok.

Figure 39: The TcExtendedFileWriter object makes sure that the ‘to file’ block
in the Simulink model can write data from the real time part of the computer to
MATLAB which is not running realtime.

Similarly as for the TcCOM Objects, go to the context tab and link the task to
the SimulinkTarget task. Now go to ‘Object1 (general target model)’ and select
the tab ’parameter (init)’. Like in figure 40 connect the ExtendedFileAccessOID
to the ‘Object2 (TcExtendedFilewriter)’ object. Under ‘Object2 (TcExtended-
Filewriter)’, change the link of the ‘pause’ entry to ‘stop logging data’ like in
figure 41. That’s it for setting up the hard- and software.

34
Figure 40: Screen for connecting the ExtendedFileWriter module to the Simulink
module.

Figure 41: This screen shows the connection between the ’pause’ entry of the
ExtendedFileWriter module and the provided ’Stop logging data’ entry of the
Simulink model. This was done to make sure that the .mat file that is created by
the filewriter module only contains the data from the experiment that we want.

35
3.5 Running the real-time target, the Simulink model and the
m-files
This section handles the final and most practical part of the tutorial: doing a
measurement with the real-time Simulink model. If all previous sections were
completed with success, this section should not give any problems and presents
merely the execution sequence of running a realtime Simulink model on your
computer.

Figure 42: This figure shows the main buttons needed for running the TwinCAT
target. When updating parameters in TwinCAT, do that in ‘Configuration Mode’
(the blue button). Use ’Reload Devices’ to make sure the connection between the
master and the slaves is running, when TwinCAT asks ‘activate free run?’ click
yes and the lights on the ethernet connectors should start blinking fast. Use
’Activate Configuration’ to restart TwinCAT in runmode, meaning that your
real-time part is running.

Make sure all hardware is connected and supplied with the appropriate power
sources. Open you general target Simulink model. In TwinCAT, referring to
figure 42, click the ‘activate configuration’ icon. TwinCAT will ask whether it
should overwrite old configurations and after that, if it should restart in Run-
mode. Click ok in both cases. If TwinCAT asks you to generate a 7 day trial
license, do it. You can regenerate this license for free an unlimited number of
times. When the red and blue icon in the right bottom corner has turned green,
TwinCAT is in runmode. Press ‘login’ and press ‘play’. In Simulink, set your
model in external mode as shown in figure 43.

36
Figure 43: This figure shows you to put your Simulink model in ’External’ mode
in order to be able to make a connection with the real-time Simulink module
running on the TwinCAT target. The button highlighted in red is the ‘connect to
target’ button, and establishes a connection between the two. You can start the
simulation by pressing the ‘play’ button (next to ‘connect to target’.

In MATLAB, run ‘generate motor trajectory.m’ and open the general target
model in MATLAB. Press the ‘connect to target’ button to connect to the tar-
get1 . When you press play, the drive will enable and the motor will start to move
the trajectory that you set for it in the Simulink model. When the motor has
stopped after 15.3 seconds, the simulation stops but you still need to press the
stop button. Click the ‘restart TwinCAT in configuration mode’ button. Run
‘save and plot data.m’. You will see the data that you just recorded.

Congrats, you’re done! You can now use the knowledge to alter anything you
like in the Simulink model you would like and you can run it as often as you
want. Do however make sure that you complete the sequence for running an
experiment entirely each time you do it. Now start playing and enjoy!

1
Connecting to the target in Windows 8 or Windows 10 failed on certain computers in our
experiments. This was due to the system clock setup. To solve this, open command prompt
(Admin) and navigate to C:/TwinCAT/3.1/System, and execute win8settick.bat and restart
the PC.

37
4 TwinCAT debugging
This section gives a brief introduction to debugging tools available in Simulink
and TwinCAT environments. Different options are available in both of these
environments for detecting and analyzing errors within a TcCOM module cre-
ated in Simulink. Since Simulink may treat certain exceptions and/or errors
differently than TwinCAT, it is important to synchronize the two and the way
they handle relevant exceptions. For a complete overview of different TwinCAT
exceptions and debugging options, the reader is refered to the Beckhoff Infor-
mation System website [1]. In the sequel, a general debugging procedure will be
shown on the example of a floating point (FPU) exception. The whole procedure
is meant to enable debugging in the block diagram exported during generation of
the TcCOM module and displayed in the TwinCAT development environment.
(The most common way to debug C++ programs - setting the breakpoints and
stepping through the code - is also available in TwinCAT environment, but is not
subject of this tutorial.)

4.1 Setting up debugger in Simulink environment

The first step in synchronizing the two environments in error-handling is to
change the Simulink publish configuration to Debug, as shown in figure 44. When
the block model is in the testing phase, this option enables debugging of the
exported block diagram in TwinCAT environment, which is often needed. If
no debugging is required, e.g. in a release version, the option Release can be
selected here.

Figure 44: In the Simulink model, under Model Explorer (press Ctrl+H), go
to Configuration Parameters\Code Generation\Tc Build, and change Publish
Configuration to Debug.

Earlier in this document, in the section on setting up Simulink model param-

eters, it is already mentioned that options Export block diagram and Export block
diagram debug information in TcAdvanced tab of the configuration parameters

38
need to be enabled. Given the importance of these two options being enabled, it
is worth reminding here to make sure that they are enabled, as shown in figure
45.

Figure 45: Make sure that in the Simulink model, under Model Explorer (press
Ctrl+H), Configuration Parameters\Code Generation\Tc Advanced, the options
Export Block Diagram and Export Block Diagram Debug Information are enabled.

Above mentioned options are general ones, and need to be enabled regardless
of the error or exception one might experience. However, any one exception is
particular and might require different settings in order for diagnostics to work
properly. In the case of the floating point (FPU) error, used as an example in
this tutorial, it is necessary to change signal handling in Data Validity tab of the
block model Diagnostics tool as shown in 46. In the case of other exceptions,
signal attributes might need to be changed accordingly.

Figure 46: In the Simulink model, under Model Explorer (Ctrl+H), go to Config-
uration Parameters\Diagnostics\Data Validity, and change Division by singular
matrix, and Inf or NaN block output to error.

4.2 Setting up debugger in TwinCAT environment

TwinCAT C++ offers different mechanisms for debugging C++ modules run-
ning under real-time conditions, including the ones built within the Simulink
environment. To do so, the first step is to enable C++ debugger, as this op-
tion is switched off by default and must be activated before the configuration is
activated. Figure 47 shows how to activate C++ debugger in TwinCAT environ-

39
ment. As already described earlier, TwinCAT is integrated in Microsoft Visual
Studio 2010.

Figure 47: In Visual Studio, in the main solution tree of the project there is an
input called called ’C++’. Click on it and open the tab called ’C++ Debugger’.
There, enable the C++ Debugger.

If the above given steps are followed, and the C/C++ source code of the
generated TcCOM module is present on the engineering systems (so that Visual
Studio debugger is able to find it), debugging within the block diagram can be
executed. The debugging is carried out by assigning the possible breakpoints to
the blocks in the block diagram, which are then represented as points. The color
of the point provides information about the current state of the breakpoint. To
get to this point, one must first activate configuration (see figure 48) and then
attach the process to XAE debugger code (see figure 49). If exceptions occur
during processing of a TcCom module, the point at which the exception occurred
is shown in the block diagram - the block that caused the exception is highlighted
in the block diagram, provided the line of code responsible for the exception can
be assigned to a block. The name of the block is shown in red, and the block
itself is marked in bold.
Attaching the process to XAE debugger will automatically start the Twin-
CAT Live Watch, a tool that can be used to monitor process variables without
setting breakpoints. This tool can also be started independently, by going to
the main menu of Visual Studio and navigating the following: Debug - Win-
dows - TwinCAT Live Watch. If TwinCAT runs on a real machine with axis
movements, the user will probably not wish to set any breakpoints just for mon-
itoring variables. On reaching a breakpoint the execution of a task would be
stopped and, depending on the configuration, the axis would immediately come
to a halt or would continue to move in an uncontrolled fashion, which might
have hazardous consequences.

40
Figure 48: In the TwinCAT project in Visual Studio, select tab ’TwinCAT’ and
then ’Activate Configuration’. The status icon should turn green, meaning that
the system is in active mode.

Figure 49: In the main menu of Visual Studio, select ’Debug - Attach to pro-
cess..’. When ’Attach to Process’ window opens, select ’TwinCAT XAE’ as a
transport method, ’Localhost’ as a qualifier and confirm with ’Attach’ button.
The process in active configuration is now connected to the debugger.

41
5 Licensing TwinCAT products
For certain applications it might be necessary to purchase a commercial license
for the TE1400 module.More specifically, a license is required for models with
more than 100 Simulink blocks, or when the number of inputs of outputs is
higher than five. Of course, a license is also required in the case of commercial
use. This chapter contains detailed instructions on how to activate a TwinCAT
license.

Situation
Beckhoff TE1400 Module allows the use of MATLAB/Simulink to establish con-
trol systems that can be later exported into a C or C++ code that can be run
within the TwinCAT3 environment.This way, complex control systems can be
created in a fast and intuitive way without prior programming knowledge in C
or C++ language. This makes the TE1400 module a useful tool for rapid con-
trol prototyping, real-time simulation and model-based monitoring applications.
For testing purposes, the module generator of the TE1400 can be used in demo
mode without a license. However, this limits its application to models with a
maximum of 100 Simulink blocks, 5 inputs and 5 output signals, and only for
non-commercial purposes. Generally, robotic applications require larger models
with greater number of input and output blocks corresponding to sensors and ac-
tuators. In these cases, purchasing a commercial license for the TE1400 module
becomes a requisite.
Beckhoff offers three kind of licenses:

ˆ Trial license: some TwinCAT 3 Runtime products can be activated free

of charge for testing purposes for 7 days, as many times as required. The
license TC1320 (or TC1220 with PLC license) is required to start a Twin-
CAT configuration with a module generated from Simulink. Therefore, it
is necessary to get this license if the TE1400 module is used. Without
activated license, the module and consequently the TwinCAT system can-
not be started. In this case you get error messages relating to the license
violation. You can generate a 7-day trial license for the TC1320 within the
TwinCAT environment as many times as required without any limitation.
Unfortunately, this option is not available for the TE1400, and purchasing
a commercial license is mandatory.

ˆ Standard license: it consist of a commercial license that is tied to a unique

system ID. It is installed in a single IPC and cannot be shared with others.

ˆ Volume license: they offer simple handling of TwinCAT 3 licenses without

requiring activation and with a straightforward exchange of IPCs. The
TwinCAT3 license dongle with memory function consists of a new storing
device for license files and offers an attractive alternative to a volume
license. TwinCAT3 license dongles combine the benefits of straightforward
volume licensing with a high degree of flexibility for the required TwinCAT
3 license configuration. The disadvantages of a volume license, such as fixed

42
 license configuration and the fixation to customer-specific hardware, can be
circumvented. A TwinCAT 3 license dongle offers greater flexibility with
regard to the control computer (e.g. during service), since the TwinCAT
3 license is no longer tied to a specific IPC so it can be exchanged for
multiple IPCs, which is very convenient within a lab environment as the
same license file can be used in multiple setups (not simultaneously). The
only requirement is that the TwinCAT 3 hardware platform level must
match the corresponding TwinCAT 3 license.

TwinCAT standard licenses (as the TE1400) are always tied to a specific
hardware – generally a TwinCAT dongle (EL6070 (EtherCAT Terminal) or
C9900-L100 (USB stick)). In principle it is also possible to tie a TwinCAT
3 license to a specific Beckhoff IPC. However, this has the severe disadvantage
that, if an IPC is replaced, the TwinCAT 3 licenses are no longer valid for the
new IPC. If, on the other hand, the TwinCAT 3 licenses are tied to a TwinCAT3
license dongle, the IPC can easily be replaced.
Among all the different licensing options Beckhoff offers, we believe the USB
dongle with memory function is the most convenient within a lab and fast pro-
totyping environment due to its high flexibility and intrinsic advantages. In the
following sections we describe the required procedure to save a standard license
(TE14000) into a license dongle C9900-L100 (USB stick).

License Key USB Stick C9900-L100

The USB dongle stick provided by Beckhoff does not include any stored license.
Therefore it is necessary to include the required licenses on it. The following
steps summarize the procedure that has to be carried out in order to store a
license file into the USB dongle stick to further use it in a TwinCAT project:

1. Insert the USB drive into the USB slot of your computer. Note that in
the current TwinCAT version the TwinCAT3 USB dongle is not detected
automatically. It must be initialized (set as license device) once in the
project. The TwinCAT3 USB dongle must therefore always be located
precisely at the USB slot which was configured for this project (For a new
project you can, of course, select a different position for the TwinCAT3
USB dongle.)

2. In the TwinCAT System Manager, scan for new devices to check whether
the system finds the C9900-L100 license dongle, such as in figure 50. For
TwinCAT 3 volume license dongles please note that a special, custom ESI
file must be used for the volume license dongle. In case the system does
not recognise the dongle, the specific ESI file can be downloaded from the
Beckhoff Information System website [1]. The downloaded file should then
be stored in: TwinCAT\3.x\Config\Io\USB

43
Figure 50: Scan for new devices to check whether the system finds the C9900-
L100 license dongle.

3. In the project click on the subitem “SYSTEM”, then double-click on “Li-

cense”, such as in figure 51.

4. In the window that opens select the tab “License Device”2

5. In the hardware section select “Dongle (EtherCAT terminal EL6070, USB)”.

2
In Windows 10, there is a slightly different procedure to get the Dongle working. First,
select “Add” which is in the “Order Information (Runtime)” tab. This will open a sublevel
under the “Licence” tab, in which we can add the actual dongle.

44
Figure 51: (1) Scan for new devices to check whether the system finds the C9900-
L100 license dongle. (2) Click on the subitem “SYSTEM”. (3) Select the tab
“License Device”. (4) Select “Dongle (EtherCAT terminal EL6070, USB)”. (5)
Click on “Search”.

6. Click on “Search” to find TwinCAT3 license dongles at the USB ports or

in the connected EtherCAT network.

7. In the search results select the TwinCAT3 USB license dongle C9900-L100,
see figure 52.

Figure 52: (6) Select the TwinCAT3 USB license dongle C9900-L100. (7) Click
on “OK”.

8. Then click on “OK”. In the “Order Information” tab, the C9900-L100 can
then be selected as hardware basis for the licensing procedure.

9. Click on the “Order Information” tab at the top, figure 53.

10. Under “System Id” open the selection box by clicking on the down arrow.

45
Figure 53: (8) Click on the “Order Information” tab. (9) Under “System Id”
open the selection box. (10) Select “Dongle Hardware Id”.

11. Select “Dongle Hardware Id”.

12. The system ID required for licensing is now provided by the selected Twin-
CAT3 license dongle, see figure 54.

13. Enter the license ID (the corresponding order number provided by Beck-
hoff) in the field “License Id”. The lower section of the “Order Informa-
tion” tab shows a list of licenses, which TwinCAT has determined auto-
matically based on your project. If licenses were added manually, these are
also listed here. In the “Manage Licenses” tab you can select / add licenses
manually. Here we can manually add the TE1400 license. Double-check
that all listed licenses are also included in the order with the specified num-
ber – otherwise the Beckhoff license server would issue an error message! If
licenses are listed, which are NOT included in the order with this number,
these should be deselected in the “Manage Licenses” tab.

14. The “License Request file” can then be generated via the “Generate File”
button (14).

46
Figure 54: (12) The system ID required for licensing is now provided by the
selected TwinCAT3 license dongle. (13) Enter the license ID. (14) Generate the
“License Request file”.

A window opens, in which you can specify where the License Request file is
to be stored. (We recommend accepting the default settings.)

47
 Figure 55: Specify where the License Request file is to be stored.

Clicking on “Save” triggers a prompt as to whether the License Request file is

to be sent to the Beckhoff license server (prerequisites: an email program must
be installed on your computer, and your computer must be connected to the
Internet).

Figure 56: Send the License Request file

Click on “Yes” to send the License Request file right away. If no email pro-
gram is installed on your computer, or if you have no Internet access, answer the
question with “No” and copy the License Request file to a data storage device
(e.g. USB stick). Send the file from a computer with Internet access and email
program to the Beckhoff license server (tclicense@beckhoff.com). The Beckhoff
license server returns the License Response file to the same email address from
which the License Request file was sent. When the server receives the email with
your license request, it compares the request with the order number and returns

48
an email (to the same email address, from which the License Request file was
sent) with a “License Response file”.3 The contents of the License Response files
must not be changed, otherwise the license file becomes invalid. The next step
is to save the activated License Response file into the USB dongle. The memory
function requires TwinCAT 3.1 build 4018.26 or higher (Engineering and Run-
time). This version initially offers the main basic functions for uploading and
downloading of the License Response files. The TwinCAT 3 license dongles are
configured in the “License Device” tab of the TwinCAT license manager:

Figure 57: TwinCAT 3 license dongles are configured in the “License Device”
tab.

In order to save a license into the USB dongle, click on “Download license to
device”. A window opens in which you can select the previously obtained License
response file. This way, a copy of the requested license is permanently stored into
your USB dongle. TwinCAT 3 does not work directly with the files on the dongle;
it always loads a “working copy” of the files onto the hard disk of the IPCs. It is
therefore important to download the license files once manually or automatically
on each startup of the TwinCAT 3 Runtime for each project. To do it manually,
click on “Cache License data”. It Copies all files of the TwinCAT3 license
dongle to the hard disk of the IPC in directory “TwinCAT\3.x\Target\Licenses”.
3
The only difference between the License Response file and the License Request file is a
signature, which documents the validity of the license file.

49
However we recommend to click on “Cache or check License Response files during
startup”. When this checkbox is ticked, TwinCAT 3 checks automatically on
each start of the runtime whether the files on the hard disk match those in
the dongle memory. If the dongle contains newer or additional files, they are
automatically downloaded to the hard disk of the IPC (“cached”). Once the
project is run, you should be able to see which licenses are active (or valid)
within your project in the “project licenses” tab. If you followed all the previous
steps, you should be able to see your new TE1400 license as “valid”. If not, you
can press the bottom “Reload Info”. It reads the data of the selected dongle again
and update the display of the dongle parameters. If no valid data are displayed
under “Dongle Status”, the data can be re-read via this button. Finally, if you
want to delete the stored licenses in your dongle, click on “Clear License Store”

50
6 Experimental implementations
6.1 Frameless Motor
In this section a small experimental setup equipped with EtherCAT technology
is presented. The setup is run from a Dell Latitude E6440 laptop on which Twin-
CAT is installed as well as MATLAB 2015b. The main hardware components of
this setup are a Kollmorgen RBE series frameless motor, and an EPOS3 motor
controller. The motor of the setup is depicted in figure 58. The main idea of this
setup is to implement a very simple impedance controller in Simulink and to run
this Simulink model realtime from a Windows PC with the hardware connected.
The simple impedance control law that is implemented has the following
equation:
T = Kp (x − xref ) + Kv (ẋ − ẋref ) (1)
where Kp is a proportional position gain, Kv is a proportional velocity gain, x
is the motor axle position and ẋ is the motor axle velocity. To obtain a spring
like behaviour in the motor the reference position xref can be set to 0 and Kp
now functions as the spring constant of the spring that the motor emulates. In
order to obtain a stable system we of course need a small damper in the form of
a small value for Kv and ẋref = 0 but also a high cycle time to allow the system
to respond to disturbances.
The equation can easily be converted to a Simulink model. The model can
be compiled with MATLAB/Simulink coder and afterwards be imported as a
TcCom object in TwinCAT. The model can now be run in realtime on the
dedicated cores in the computer and still be controlled from Simulink running
in external mode on the windows cores of the computer.
The results of a simple experiment with the frameless motor are depicted in
the graph of figure 59. The motor emulates a spring and as can be seen from
the graph, different values for a spring stiffness were easily implemented and
emulated in the motor. Graphs are easily obtained and stored in MATLAB for
further data processing.

51
Figure 58: A Kollmorgen RBE type 12 pole frameless motor as used in the
simple test setup. The aluminum casing was made to fit the motor and allow
for testing to obtain the motor performance characteristics. The sensors in the
motor include hall sensors for the commutation of the motor and an optical
incremental encoder for the position of the motor. A small ring is clamped to
the main axle to be able to turn the motor axle manually.

52
 0.5

0.4

0.3

0.2

0.1

-0.1

-0.2

-0.3

-0.4

-0.5
-150 -100 -50 0 50 100 150

Figure 59: A graph of different springs emulated by the frameless motor with on
the x axis the deviation angle of the motor axis in degrees and on the y axis the
torque in Nm. Different spring different values, including negative ones, for the
spring stiffness were programmed, as can be seen from the graph.

6.2 Cyberlegs Prosthesis

As a case for the described technology this section presents the Cyberlegs an-
kle knee prosthesis [9]. The Cyberlegs prosthesis is equipped with a custom
developed EtherCAT slave controller (ExotronicsTM board) that is compatible
with Maxon Escon power modules. These power the drivetrain of the prosthesis
which includes the ankle motor, the knee motor and the weight acceptance motor
which is used to provide an energy efficient way of remaining upright. Secondly,
the controller was programmed in Simulink. The Simulink control model can
easily be compiled with MATLAB/Simulink Coder after which it can be run as
a module on the TwinCAT target created on the researcher’s Windows laptop.
After the prototyping of the controller, the Simulink model can be compiled in C
code, and added to the code containing the SOEM master software. The result
will be a transfemoral prosthesis without a laptop in the backpack but only with
a single power cable and some batteries in the backpack. The full realtime logic
and controller is implemented and runs on the prosthesis itself.

53
Figure 60: The Cyberlegs ankle-knee prosthesis. This prosthesis has four Ex-
otronics boards installed to drive the motors. Two of these motor drivers are
labeled 4011 and 4012 and can be seen in the picture. Another observation that
must be made is the white Ethernet cable with the yellow plug and the black
power cable running to the prosthesis. These two cables are the only two cables
required for prototyping control architectures with Simulink and TwinCAT from
a Windows laptop.

54
6.3 Dual Motor Setup
Another setup where TwinCAT was applied is shown in Figure 61. The setup
is used to test a dual-motor architecture. Two DC motors, a 90W Maxon RE35
motor with a 43:1 reduction and a 150W Maxon RE40 motor with a 15:1 reduc-
tion, are coupled to a single output (an inertial load) by means of a planetary
differential. Furthermore, both motors are equipped with controllable brakes. If
engaged, the brakes deliver a static torque, and the motor can be shut down.
This way, energy losses in the drivetrain are avoided. The dual-motor architec-
ture creates a redundant kinematic degree of freedom. It allows for distributing
the power requirements over both motors in the most energy-efficient way. With
a proper control, the system can be made more energy-efficient than a single
motor [10]. As low-level controllers, two Maxon MAXPOS 50/5 drives are used.
They communicate with the TwinCAT target through the MAXPOS’s Ether-
CAT connection. The controllable brakes are normally engaged, and can be
disengaged by applying a 24V voltage. A transistor is used to switch between
the on and off state; the transistors are controlled by means of the digital out-
puts on the MAXPOS drives. These can be accessed on the TwinCAT target,
just like the readings from the encoders on the Maxon motors. Finally, a torque
sensor and an encoder are used to determine the output speed and torque. The
signals are acquired by Beckhoff modules (EL5101 and EL3102). The Beckhoff
EK1100 coupler transmits the values to the TwinCAT target through Ether-
CAT. A crucial aspect is the synchronization of the low-level motor controllers
and the controllable brakes. Just like in the previous examples, the controller is
written in Simulink and converted to C code by MATLAB/Simulink Coder. Just
like in the previous examples, this code is run on the TwinCAT created on the
researcher’s Windows PC. The controller runs at a rate of 1 kHz, and succeeds in
simultaneously controlling the speed of both motors in real-time using feedback
from the encoder on the output.

55
Figure 61: Dual-motor actuator setup. The setup consists of two DC motors
coupled to the same output by means of a planetary differential. Holding brakes
on the motors allow switching off the motor, while the brake delivers a static
torque. The output is a simple flywheel. A speed controller for the setup, us-
ing information from the encoder on the output as feedback, is programmed in
Simulink and run on the TwinCAT target at a speed of 1 kHz.

The dual-motor actuator was incorporated into a Series Elastic Actuator instead
of a regular electric motor. The resulting Series Elastic Dual-Motor Actuator was
then implemented on the Marco Hopper II, a hopping robot consisting of a two-
link segmented leg with actuated knee [11]. The controller, again programmed in
Simulink and run on a TwinCAT target on a windows PC, generates repetitive
hopping patterns. Furthermore, it calculates the optimal way to distribute the
required output power over the two motors on-line, and sends the appropriate
speed commands to the low-level controllers.

56
Figure 62: Marco Hopper II setup. A Bowden cable is used to transmit the
motion from the Series Elastic Dual-Motor actuator (left) to the knee of the
hopping robot (right).

6.4 Test bench for robotic devices and actuators

This section is devoted to an EtherCAT based one degree of freedom test bench
for performing static as well as dynamic measurements on actuators or robotic
devices. The setup is run from a Lenovo ThinkPad W541 laptop on which Twin-
CAT and MATLAB 2015b are installed on Windows 10. As a specific illustration
of the use of this testbench, the total setup depicted in figure 63, consists of an
Omron R88M-1L75030C-S2 AC servo motor (load motor) powered by an Omron
R88D-1SN-10F-ECT EtherCAT 400V AC servo drive and equipped with a Wit-
tenstein NP 035S-MF2-40-0E1-1S planetary gearbox. From a Simulink model
running in realtime in TwinCAT, this load motor is commanded the ankle load
torque or the ankle angle of the AMP-Foot prosthesis connected at the output
of the test bench, and such mimicking the ankle torque caused by the amputee’s
weight. The prosthesis itself is driven by a maxon EC 4 pole motor in spin-
dle drive setup and powered by an Elmo DCWHI20/100EEH EtherCAT servo
drive. The ankle torque is measured by a DRFL-VI-500-w ETH messtechnik
torque transducer and since this device does not have any Ethernet ports, this
data is transferred to the EtherCAT network by the means of a 16 bit, 2-channel,
-10V/+10V analog input terminals, Beckhoff EL3102 module connected to an

57
EK1100 coupler. A 4-channel 5 V DC digital output terminal (EL2124 Beck-
hoff module) in combination with a KL9505 Beckhoff module providing 5V DC
power supply is used to control a small hobby servo motor that (un)locks the
clutch. Measured variables such as phase voltages, phase currents, ankle out-
put torque, encoder position and velocity of all servo drives are saved in the
MATLAB workspace for easy further data processing. This example illustrates
a simple and save way to perform dynamic measurements (to obtain e.g. elec-
trical energy consumption during different gait cycles) or testing novel control
strategies without the possibility of the amputee getting injured in the first trials.

Figure 63: One degree of freedom test bench for performing static as well as
dynamic measurements on actuators or robotic devices running TwinCAT

58
References
[1] “Beckhoff Information System.” https://infosys.beckhoff.com/index en.htm.

[2] “IEEE Standard for Ethernet,” IEEE Std 802.3-2015 (Revision of IEEE
Std 802.3-2012), pp. 1–4017, Mar. 2016.

[3] M. Felser, “Real-Time Ethernet - Industry Prospective,” Proceedings of the

IEEE, vol. 93, pp. 1118–1129, June 2005.

[4] J. Jasperneite, M. Schumacher, and K. Weber, “Limits of increasing the

performance of industrial ethernet protocols,” in Emerging Technologies and
Factory Automation, 2007. ETFA. IEEE Conference on, pp. 17–24, IEEE,
2007.

[5] J.-D. Decotignie, “Ethernet-Based Real-Time and Industrial Communica-

tions,” Proceedings of the IEEE, vol. 93, pp. 1102–1117, June 2005.

[6] G. Prytz, “A performance analysis of EtherCAT and PROFINET IRT,” in

Emerging Technologies and Factory Automation, 2008. ETFA 2008. IEEE
International Conference on, pp. 408–415, IEEE, 2008.

[7] M. Knezic, B. Dokic, and Z. Ivanovic, “Topology aspects in Ether-

CAT networks,” in Power Electronics and Motion Control Conference
(EPE/PEMC), 2010 14th International, pp. T1–1, IEEE, 2010.

[8] G. Cena, I. C. Bertolotti, S. Scanzio, A. Valenzano, and C. Zunino, “Eval-

uation of EtherCAT Distributed Clock Performance,” IEEE Transactions
on Industrial Informatics, vol. 8, pp. 20–29, Feb. 2012.

[9] L. Flynn, J. Geeroms, R. Jimenez-Fabian, B. Vanderborght, N. Vitiello, and

D. Lefeber, “Ankle–knee prosthesis with active ankle and energy transfer:
Development of the cyberlegs alpha-prosthesis,” Robotics and Autonomous
Systems, vol. 73, pp. 4–15, 2015.

[10] T. Verstraten, R. Furnémont, P. Lopez-Garcia, D. Rodriguez-Cianca, H.-L.

Cao, B. Vanderborght, and D. Lefeber, “Modeling and design of an energy-
efficient dual-motor actuation unit with a planetary differential and holding
brakes,” Mechatronics (submitted Aug. 2017), no. 9, 2017.

[11] J. Oehlke, M. A. Sharbafi, P. Beckerle, and A. Seyfarth, “Template-based

hopping control of a bio-inspired segmented robotic leg,” in Biomedical
Robotics and Biomechatronics (BioRob), 2016 6th IEEE International Con-
ference on, pp. 35–40, IEEE, 2016.

[12] V. Grosu, C. Rodriguez Guerrero, B. Brackx, S. Grosu, B. Vander-

borght, and D. Lefeber, “Instrumenting complex exoskeletons for im-
proved human-robot interaction,” Instrumentation & Measurement Mag-
azine, IEEE, vol. 18, no. 5, pp. 5–10, 2015.

59
Appendix A
In this section more information is presented on some common ’slave devices’
that we use in the lab. The three devices that are briefly mentioned here are first
the Exotronix boards of Victor Grosu, secondly the Maxon EPOS3 controller and
lastly the Beckhoff EK1100 bus coupler.

Exotronics board

Figure 64: Exotronics Board produced by Victor Grosu [12]. This board can be
used in combination with a motor drive like for example an Maxon ESCON. The
board can read an incremental encoder, a magnetic (SPI) encoder, a loadcell and
other IO. The board can send a pwm setpoint to the motor controller.

60
EPOS3 motor controller

Figure 65: Maxon EPOS3 controller. The EPOS3 controller is a versatile motor
drive that can be configured for position, velocity and torque control. These are
no longer sold by Maxon. Alternatives are MAXPOS and EPOS4 drivers. These
drives ’speak’ EtherCAT and can easily be imported into TwinCAT. Extensive
documentation on the devices can be found at the specific page on the website
under downloads

61
Beckhoff EK1100 coupler

Figure 66: Beckhoff EK1100 EtherCAT coupler with IO modules. This is an

essential device in the vast majority of all test-setups. The EK1100 coupler
module can be coupled to a great number of E-bus modules. These modules can for
example be analog inputs, analog outputs, digital IO but also encoder interfaces
or motor terminals. The for example analog read values are directly published on
the EtherCAT network and can be seen under the EK1100 module in TwinCAT.

62