we talked about how buses and protocols work in industrial automation. This time we will focus on modern working solutions: we will see what protocols are used in systems around the world. Consider the technologies of German companies Beckhoff and Siemens, Austrian B&R, American Rockwell Automation and Russian Fastwel. And also we will study universal solutions that are not tied to a specific manufacturer, such as EtherCAT and CAN.
At the end of the article there will be a comparison table with the characteristics of the EtherCAT, POWERLINK, PROFINET, EtherNet/IP and ModbusTCP protocols.
We did not include the PRP, HSR, OPC UA and others protocols in the review, because there are already excellent articles on them on HabrΓ© by our fellow engineers who are developing industrial automation systems. For example, ΠΈ .
First, let's define the terminology: Industrial Ethernet = industrial network, Fieldbus = field bus. In Russian industrial automation, confusion occurs in terms of field bus and industrial network of the lower level. Often these terms are combined into a single vague concept of "lower level", which is referred to as both a field bus and a low-level bus, although it may not be a bus at all.
Why is that?Such confusion is most likely due to the fact that in many modern controllers the connection of I / O modules is often implemented using a backplane or a physical bus. That is, some bus contacts and connectors are used to combine several modules into a single node. But such nodes, in turn, can be interconnected both by an industrial network and a field bus. In Western terminology, there is a clear division: a network is a network, a bus is a bus. The first is designated by the term Industrial Ethernet, the second by Fieldbus. The article proposes to use the term "industrial network" and the term "field bus" for these concepts, respectively.
EtherCAT industrial network standard developed by Beckhoff
The EtherCAT protocol and industrial network is perhaps one of the fastest methods of data transmission in automation systems today. The EtherCAT network is successfully used in distributed automation systems, where interacting nodes are separated over a long distance.
The EtherCAT protocol uses standard Ethernet frames to transmit its telegrams, therefore it remains compatible with any standard Ethernet equipment and, in fact, data reception and transmission can be organized on any Ethernet controller, with the appropriate software.
Beckhoff controller with set of I/O modules. Source:
The protocol specification is open and available, but only within the development association - EtherCAT Technology Group.
Here's how EtherCAT works (the spectacle is mesmerizing, like a Zuma Inca game):
The high exchange rate in this protocol - and we can talk about units of microseconds - is realized due to the fact that the developers refused to exchange using telegrams sent directly to a specific device. Instead, one telegram is sent to the EtherCAT network, addressed to all devices at the same time, each of the slave nodes for collecting and transmitting information (they are often called OCO - communication device with an object) takes from it "on the fly" the data that was intended for it, and inserts to the telegram data, which he is ready to provide for exchange. After that, the telegram is sent to the next slave node, where the same operation takes place. Having passed all the USO, the telegram is returned to the main controller, which, based on the data received from the slave devices, implements the control logic, again interacting via the telegram with the slave nodes that issue a control signal to the equipment.
An EtherCAT network can have any topology, but in essence it will always be a ring due to the use of full duplex mode and two Ethernet connectors. Thus, the telegram will always be transmitted sequentially to each device on the bus.
Schematic representation of an Ethercat network with multiple nodes. Source:
By the way, the EtherCAT specification does not contain restrictions on the 100Base-TX physical layer, so the implementation of the protocol is possible on the basis of gigabit and optical lines.
Open industrial networks and PROFIBUS/NET standards from Siemens
The German concern Siemens has long been known for its programmable logic controllers (PLCs), which are used around the world.
Data exchange between the nodes of an automated system controlled by Siemens equipment is implemented both via a field bus, which is called PROFIBUS, and in an industrial PROFINET network.
The PROFIBUS bus uses a special two-wire cable with DB-9 connectors. Siemens has it in purple, but we have met others in practice :). For communication of several nodes, the connector can connect two cables. It also has a switch for the terminating resistor. The terminal resistor must be turned on at the end devices of the network, thus it is reported that this is the first or last device, and after it there is nothing, only darkness and emptiness (all rs485s work this way). If you turn on the resistor on the intermediate connector, then the section following it will be disconnected.
PROFIBUS cable with connectors. Source:
The PROFINET network uses a twisted-pair analog, usually with RJ-45 connectors, the cable is colored green. If the PROFIBUS topology is a bus, then the PROFINET network topology can be anything: even a ring, even a star, even a tree, even all together.
Siemens controller with connected PROFINET cable. Source: w3.siemens.com
There are several communication protocols on the PROFIBUS bus and on the PROFINET network.
For PROFIBUS:
- PROFIBUS DP - the implementation of this protocol implies communication with remote slave devices, in the case of PROFINET, this protocol corresponds to the PROFINET IO protocol.
- PROFIBUS PA - is essentially the same as PROFIBUS DP, only used for explosion-proof versions of data and power transmission (an analogue of PROFIBUS DP with different physical properties). For PROFINET, an explosion-proof protocol similar to PROFIBUS does not yet exist.
- PROFIBUS FMS - designed to communicate with systems from other manufacturers that cannot use PROFIBUS DP. The analogue of PROFIBUS FMS in the PROFINET network is the PROFINET CBA protocol.
For PROFINET:
- PROFINETIO;
- PROFINET CBA.
The PROFINET IO protocol is divided into several classes:
- PROFINET NRT (non-real time) - used in applications where timing is not critical. It uses the Ethernet TCP/IP data transfer protocol as well as UDP/IP.
- PROFINET RT (real time) - I/O communication is implemented using Ethernet frames, but diagnostic and communication data is still transmitted via UDP/IP.
- PROFINET IRT (Isochronous Real Time) - This protocol was developed specifically for motion control applications and includes an isochronous data transfer phase.
As for the implementation of the PROFINET IRT hard real-time protocol, two exchange channels are allocated in it for communication with remote devices: isochronous and asynchronous. An isochronous channel with a time-fixed exchange cycle length uses clock synchronization and transmits time-critical data; second-level telegrams are used for transmission. The transmission duration in the isochronous channel does not exceed 1 millisecond.
In an asynchronous channel, so-called real-time data is transmitted, which is also addressed by means of a MAC address. Additionally, various diagnostic and auxiliary information is transmitted over TCP/IP. Neither real-time data, much less other information, of course, can interrupt the isochronous cycle.
An extended set of PROFINET IO functions is far from needed for every industrial automation system, so this protocol is scaled for a specific project, taking into account compliance classes or application classes (conformance classes): Π‘Π‘-A, CC-B, CC-CC. Compliance classes allow you to select field devices and trunk components with the minimum required functionality.
Source:
The second exchange protocol in the PROFINET network - PROFINET CBA - is used to organize industrial communication between equipment from different manufacturers. The basic production unit in CBA systems is a certain entity called a component. This component is usually a collection of mechanical, electrical and electronic parts of a device or installation, as well as the associated application software. For each component, a software module is selected that contains a complete description of the interface of this component according to the requirements of the PROFINET standard. After that, these software modules are used to exchange data with devices.
B&R Ethernet POWERLINK protocol
The Powerlink protocol was developed by the Austrian company B&R in the early 2000s. This is another implementation of a real-time protocol on top of the Ethernet standard. The protocol specification is available and freely distributed.
Powerlink technology uses a so-called mixed polling mechanism, when all interaction between devices is divided into several phases. Particularly critical data is transmitted in the isochronous phase of the exchange, for which the required response time is configured, the rest of the data will be transmitted as far as possible in the asynchronous phase.
B&R controller with set of I/O modules. Source: br-automation.com
Initially, the protocol was implemented on top of the 100Base-TX physical layer, but a gigabit implementation was later developed.
The Powerlink protocol uses an exchange scheduling mechanism. A certain token or control message is sent to the network, with the help of which it is determined which of the devices currently has permission to exchange data. Only one device can access the exchange at a time.
Schematic representation of a POWERLINK Ethernet network with multiple nodes.
In the isochronous phase, the polling controller sequentially sends a request to each node from which it is necessary to receive critical data.
The isochronous phase is performed, as already mentioned, with a configurable cycle time. In the asynchronous phase of the exchange, the IP protocol stack is used, the controller requests non-critical data from all nodes, which send a response as they gain access to transmission to the network. The ratio of time between the isochronous and asynchronous phases can be adjusted manually.
Rockwell Automation Ethernet/IP
The EtherNet/IP protocol was developed with the active participation of the American company Rockwell Automation in 2000. It uses the TCP and UDP IP stack and extends it for industrial automation applications. The second part of the name, contrary to popular belief, does not mean Internet Protocol, but Industrial Protocol. UDP IP uses the Common Interface Protocol (CIP) communication stack, which is also used on ControlNet / DeviceNet networks and is implemented over TCP/IP.
The EtherNet/IP specification is public and free. The topology of an Ethernet/IP network can be arbitrary and include a ring, star, tree, or bus.
In addition to the standard functions of the HTTP, FTP, SMTP protocols, EtherNet/IP implements the transfer of time-critical data between the polling controller and I/O devices. The transmission of non-time-critical data is provided by TCP packets, and the time-critical delivery of cyclic control data is via the UDP protocol.
For time synchronization in distributed systems, EtherNet/IP uses the CIPsync protocol, which is an extension of the CIP communication protocol.
Schematic representation of an Ethernet/IP network with multiple nodes and Modbus device connections. Source:
To simplify the setup of an EtherNet/IP network, most standard automation devices come with predefined configuration files.
Implementation of the FBUS protocol at Fastwel
We thought for a long time whether to include the Russian company Fastwel with its domestic implementation of the FBUS industrial protocol in this list, but then we decided to write a couple of paragraphs to better understand the realities of import substitution.
There are two physical implementations of FBUS. One of them is a bus in which the FBUS protocol runs on top of the RS485 standard. In addition, there is an implementation of FBUS in an industrial Ethernet network.
FBUS can hardly be called a high-speed protocol, the response time strongly depends on the number of I / O modules on the bus and on the exchange parameters, usually it ranges from 0,5-10 milliseconds. One FBUS slave can only contain 64 I/O modules. For a field bus, the cable length cannot exceed 1 meter, so it is not a question of distributed systems. Rather, it goes, but only when using the industrial FBUS network over TCP / IP, which means an increase in the polling time by several times. To connect the modules, bus extensions can be used, which makes it possible to conveniently arrange the modules in the automation cabinet.
Fastwel controller with connected I/O modules. Source:
Total: how all this is used in practice in process control systems
Naturally, the species diversity of modern industrial data transfer protocols is much greater than we have described in this article. Some are tied to a specific manufacturer, some, on the contrary, are universal. When developing automated process control systems (APCS), an engineer selects the optimal protocols, taking into account specific tasks and constraints (technical and budget).
If we talk about the prevalence of one or another exchange protocol, then we can give a diagram of the company HMS Networks AB, which illustrates the market share of various exchange technologies in industrial networks.
Source:
As you can see from the diagram, PRONET and PROFIBUS from Siemens are in the lead.
Interesting that 6 years ago .
The table below summarizes the data for the described exchange protocols. Some parameters, such as performance, are expressed in abstract terms: high / low. Numerical equivalents can be found in performance analysis articles.
|
|
EtherCAT
|
POWERLINK
|
PROFINET
|
Ethernet/IP
|
ModbusTCP
|
|
Physical layer
|
100/1000BASE-TX
|
100/1000BASE-TX
|
100/1000BASE-TX
|
100/1000BASE-TX
|
100/1000BASE-TX
|
|
Data transfer layer
|
Channel (Ethernet frames)
|
Channel (Ethernet frames)
|
Channel (Ethernet frames), Network/Transport (TCP/IP)
|
Network/Transport(TCP/IP)
|
Network/Transport(TCP/IP)
|
|
Real time support
|
Yes
|
Yes
|
Yes
|
Yes
|
No
|
|
Performance
|
High
|
High
|
IRT - high, RT - medium
|
Average
|
Low
|
|
Cable length between nodes
|
100m
|
100m/2km
|
100m
|
100m
|
100m
|
|
Transmission phases
|
No
|
Isochronous + asynchronous
|
IRT - isochronous + asynchronous, RT - asynchronous
|
No
|
No
|
|
Number of nodes
|
65535
|
240
|
TCP/IP Network Restriction
|
TCP/IP Network Restriction
|
TCP/IP Network Restriction
|
|
Collision resolution
|
Ring topology
|
Clock synchronization, transmission phases
|
Ring topology, transmission phases
|
Switches, star topology
|
Switches, star topology
|
|
Hot swap
|
No
|
Yes
|
Yes
|
Yes
|
Depending on the implementation
|
|
Cost of equipment
|
Low
|
Low
|
High
|
Average
|
Low
|
The fields of application of the described communication protocols, field buses and industrial networks are very diverse. Ranging from the chemical and automotive industries to aerospace technology and electronics. High-speed exchange protocols are in demand in real-time positioning systems for various devices and in robotics.
What protocols did you work with and where did you apply them? Share your experience in the comments. π
Source: habr.com