IEEE Robotics & Automation Magazine - March 2018 - 23

accompanies this article in IEEE Xplore. Additionally, the
tutorial contains successful implementations on several systems, such as the Cybernetic Lower-Limb Cognitive OrthoProsthesis (CYBERLEGS) Beta-Prosthesis (Figure 1).
EtherCAT in Layman's Terms
The functionality of the EtherCAT protocol's general concepts
and origins of industrial communication are briefly recalled,
and the general working principles that apply specifically to
EtherCAT are discussed here with a certain level of abstraction, as the goal of this tutorial is only to give the reader a
basic understanding of the system to successfully deploy the
protocol. Specific resources are conveyed to the reader interested in delving more deeply into the field.
EtherCAT Origins
Technologies such as programable microcontrollers and digital signal processors allowed for the replacement of purely
analog control loops with digital controllers, such as programmable logic controllers (PLCs). This led to the core of
industrial networking, termed fieldbus protocols. Many fieldbuses were developed in parallel, and, essentially, every company designed its own protocols, which naturally led to
confusion for consumers and developers [11]. As a reaction to
this, there was the fieldbus war, a web of company politics and
marketing interests that, in 2000, led to the establishment of
an international fieldbus standardization, the International
Electrotechnical Commission Standard 61158 [12]-[15]. In
the meantime, Ethernet protocols were already well established in the office world. The increased data rates of newer
Ethernet standards (for example, 802.3 u fast Ethernet) made
it easier to create real-time Ethernet protocols, as the transmission and retransmission times are significantly shorter.
While fieldbus systems such as Profibus, Serial Real-Time
Communication System (SERCOS), CANs, and many others
have allowed for distributed industrial automation systems,
the performance of these protocols was considered to be too
limited when compared to the available performance (mainly
in terms of data throughput) of nonreal-time networks such
as Ethernet [16]. In addition, Ethernet bandwidth enables bus
cycle times in the microsecond range instead of the millisecond range, which, together with the superior performance of
modern personal-computer-based control systems, allows
one to close the control loops over the fieldbus that previously
had to be closed locally in the peripheral systems [17]. Finally,
the large amount of research that has gone into developing
Ethernet as a standard, as well as inexpensive and readily
available hardware, and the Internet of Things [18], which
enables the connection of almost anything with everything,
anywhere, illustrates the wish to develop RTE communication protocols. Nonetheless, the many challenges that presented themselves as the requirements between those two
domains-the office world and industry-are very different,
as discussed in [11] and [19]. The two main differences in
requirements are related to communication time determinism and precise clock synchronization.

First, time determinism is the requirement for the transmission times of data packets to be known. This means that
the latency of a signal must be bounded and have a low variance. The variance of the response time of a signal is often
referred to as jitter. A low jitter is required because the variance in time has a negative effect on control loops (the derivative and integral portions of a control loop are affected by
time variation). The second difference is 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 nonnegligible 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, the former of which is believed to offer the
best performance in terms of communication efficiency and
short cycle times [8].
Working Principle
Today, there are more than 25 different RTE solutions on the
market, and they are offered by diverse manufacturers and the
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, and so on. Primarily, EtherCAT is industrial
Ethernet and utilizes standard frames and the physical layer as
defined in the Ethernet IEEE Standard 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 allow effective data rates of 100 Mb/s [17]. Furthermore, the EtherCAT protocol employs the master/slave principle to control access to the medium. The master node
(typically the control system) sends the Ethernet frames to the
slave nodes (such as sensors and actuators), which can extract
data from and insert data into these frames. These process
data (of all the network devices) are carried together in one

Figure 1. The CYBERLEGs Beta-Prosthesis, which actuates two
degrees of freedom (the knee and ankle), is controlled via
an EtherCAT protocol. The prosthesis integrates series elastic
actuators controlled by custom-made electronics boards.

March 2018

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2018