Finding the right fit
Why industrial Ethernet? The use of Ethernet in industrial applications is growing and changing rapidly. There is a growing need to overcome the barriers to manufacturing efficiency: The demand for industrial Ethernet is rapidly growing due to business and technical reasons. The costs associated with Ethernet technology have plummeted in the past five years, making it more affordable to embed E...
Why industrial Ethernet?
The use of Ethernet in industrial applications is growing and changing rapidly. There is a growing need to overcome the barriers to manufacturing efficiency:
Rapidly increasing need for plant floor data
Major network-related redesign costs and reprogramming delays for manufacturing expansion or system changes
Increasing costs and errors associated with maintaining a growing plethora of different network types.
Because the costs of Ethernet technology have dropped in the past five years, it is now affordable to embed Ethernet in highly distributed basic plant floor devices.
Industrial Ethernet networks can have many more devices than fieldbus networks.
Cabling systems for industrial Ethernet have special considerations.
In Part 2 of this series, Larry Komarek discusses component selection for industrial Ethernet networks. He explains architecture requirements, plant-floor maintainability and environmental considerations.
Larry Komarek has more than 25 years experience in specifying and developing automation products. He can be reached at (717) 944-1300, ext. 3625 or at firstname.lastname@example.org .Article edited by Jack Smith, Senior Editor, Plant Engineering magazine, (630) 288-8783, email@example.com .
The demand for industrial Ethernet is rapidly growing due to business and technical reasons. The costs associated with Ethernet technology have plummeted in the past five years, making it more affordable to embed Ethernet in highly distributed basic plant floor devices. The introduction of industrial Ethernet connection devices (called infrastructure devices) provides the size, reliability and maintainability needed for installing and supporting from hundreds to thousands of Ethernet control devices in a plant (Fig. 1). The use of Ethernet in automation control applications provides several key advantages over traditional approaches.
Data capacity and performance
The major fieldbus networks continue to evolve with new functions to meet greater varieties of applications. Like fieldbus networks, Ethernet is an evolving standard. Since the original IEEE 802.3 specification, Ethernet has been expanded to include new media types such as wireless, and increasing performance such as expanding data rates (network speed) from 10 Mbps to 100 Mbps to 1 Gbps to 10 Gbps.
It has also expanded in functional areas such as auto-negotiation, redundancy and message-filtering protocols that reduce network traffic and increase throughput. Ethernet also provides a bigger "pipe" through which large amounts of data can be transmitted quickly (Fig. 2). Most fieldbus networks have data rates of 128 Kbps to 512 Kbps or, in some cases, 12 Mbps. Industrial Ethernet applications operate in the range of 10 Mbps to 100 Mbps.
Simplifying data transfers
No CEO wants to fire up a Web browser to view the status of 14,000 little sensors or motor starters operating in the plant. However, moving information through or between the plant's many control systems requires significantly lower effort with Ethernet vs. the traditional array of individual fieldbus networks or company-specific networks.
With Ethernet, the IP addressing range allows thousands of interconnected devices on a network. Major fieldbus networks have a limited address range of connected devices: a maximum of 32, 64, 128 or 256 devices, depending on the network type. Most systems never get near the maximum device limit because of the slow communications update times.
Control systems overcome this by connecting individual networks together, where each network ends in a PC or PLC. To transfer data between networks requires programming. This means if a new vision inspection camera is added and data has to pass from it to the PLC and through three separate PC/network systems to the production line QC-monitoring PC, all four devices must be programmed to receive the data and transmit it down the line (Fig. 3).
With Ethernet's large device capacity (thousands of devices vs. 64) and data capacity (100,000 kbps speed vs. 512 kbps), all the devices can be connected on one network. In this case, the data can pass directly between the vision camera and the QC computer without reprogramming all the intermediate PCs due to the traditional "multiple independent networks approach" (Figure 4). Compared to fieldbus networks, Ethernet networks can simultaneously transmit different protocols on the same wire. This allows different vendor devices with different protocols to connect to the same wire, which eliminates running many different network cables.
Universal rules for devices
Ethernet has been used to link higher-level PLCs, computers, CNCs and DCS systems for more than 15 years. All major automation vendors offer Ethernet-based products. Today, plants may have a variety of networks to maintain. In addition to proprietary legacy remote I/O and data networks, different fieldbus networks may be used on different lines or manufacturing areas of the plants. Each fieldbus network has different installation and layout guidelines, and different diagnostic LEDs, flashing patterns, color changes, etc.
Typically, this results in certain maintenance people supporting network type "A" while others support network type "B." If someone is out sick or on vacation, downtime can be longer if the "expert" is gone. With Ethernet, all the network diagnostic LED indicators and layout/wiring guidelines are the same regardless of which vendors and which protocols are used.
Given its advantages, it is not surprising that the use of Ethernet in industrial applications is growing rapidly. However, like most things in life, there is no free lunch. The application of industrial Ethernet is different from applying fieldbus systems. These differences have a major impact on the layout, cabling, and maintenance procedures used to install, startup and maintain an automation system.
Fieldbus and industrial Ethernet: the differences
Network layout and device connections
Fieldbus systems are laid out in a trunk-and-drop manner where there is one main network cable and the devices connect to it via "T" or daisy chain connections. Industrial Ethernet systems are laid out in a point-to-point, or "star" scheme of cabling, where one cable connects to each device. These individual connections are made through infrastructure components such as Ethernet switches (Fig. 5).
With the star approach, multiple infrastructure components are interconnected and each component connects in a star topology to groups of individual control devices. While this provides much greater flexibility over the trunk-and-drop systems, it requires Ethernet switches or hubs instead of only the connectors used in fieldbus systems.
Ethernet has the advantage of allowing devices with different data rates (a 10 Mbps-rated device vs. a 100 Mbps-rated device) to be mixed in the same system and even connected to the same Ethernet switch. This allows existing installations or systems with a vendor's first-generation Ethernet devices to be easily expanded to use faster products. This is an advantage over fieldbus systems, where the slowest device determines the data rate setting of the whole network.
Maintenance implications of layout and connections
Ethernet infrastructure devices are added electrical devices that do require maintenance. Some are "connect-and-go" types of devices (unmanaged switches) that require little-to-no training to install and maintain. Others have built in web pages with "fill-in-the-blank" configuration screens similar to other small intelligent automation devices such as weigh scales, barcode readers, nano PLCs, etc.
To communicate with controllers such as PLCs or PCs, networked devices require an address on the network. In both proprietary networks and in fieldbus systems, this has normally been done using rotary or dual-inline-package (DIP) switches. As mentioned earlier, Ethernet has the ability to connect thousands of devices instead of tens of devices on typical fieldbus systems. This advantage of industrial Ethernet makes the task of setting addresses with rotary or DIP switches impractical.
Industrial Ethernet devices are addressed by the same Internet Protocol (IP) addressing used in your home or office computer's web browser. With this addressing scheme — four sets of three-digit numbers separated by periods (for example, 192.168.125.211), industrial products would need 12 sets of DIP switches — clearly not practical. Instead, industrial Ethernet devices are addressed using laptops with addressing software.
The area of addressing is not only different between fieldbus and industrial Ethernet devices, there is also a difference between how Ethernet-connected office computers and Ethernet-connected industrial devices are typically addressed. In an office environment, the IT department has a computer or server that contains IP addresses for all the computers in the facility. Whenever you turn on your PC, it asks the server to give it an IP address. The IT server picks one out of the list and addresses the office PC with it. Each day you power up, your computer gets a different IP address. The protocol used to address your office computer is called Dynamic Host Configuration Protocol (DHCP).
In the industrial control world, it would be an interesting but highly frustrating world if a PLC's I/O drops would change addresses each time they powered up. Another protocol called Boot Protocol (Boot P) is typically used to manually assign a single fixed address to an Ethernet device. Addressing software that uses the Boot P protocol is called a "Boot P server" (it "serves up" the address to the device using the Boot P protocol). Many automation companies provide Boot P server software in no-charge demo packages.
There are two philosophies of addressing industrial Ethernet devices. One is to address them one-by-one before connecting them to the network. Industrial devices can be addressed one-by-one by connecting the Ethernet port or connector of the device to the laptop or desktop computer containing the addressing software. Some devices have a built in serial RS-232 port and a built in addressing screen (accessed by using the "hyper terminal" function that is standard on most computers).
Hardware MAC addresses vs. IP addressing
The second philosophy of addressing is to connect them all to the network then address them. This is possible because each Ethernet device really has two addresses: the IP address that is used to communicate with all the control software/user interfaces, and a hardware address that is used internally between the device and Ethernet switches. This hardware address is burned into the device at the vendor's factory, is unique for each product and cannot be changed. This address is called a Media ACcess (MAC) address. This is a series of six two-digit addresses, separated by periods, usually on a label on the device (for example, 00.AO.45.00.83.B3).
To address devices on the network, first obtain from the controls or IT department the list of the MAC and IP addresses to which to set them. When you power up the industrial Ethernet devices, they will automatically transmit their MAC addresses to your addressing software. Using the "fill-in-the-blank" screen of your addressing tool, you simply select the Mac address of the device, and fill in a blank with the IP address and click "OK." Follow this procedure for each device until all the addressing is complete.
Setting data (baud) rates on Ethernet devices
Typical fieldbus devices also require that the user set both the address and the data rate using DIP switches. Newer generations of fieldbus devices have "auto-baud" capability, where the device determines to what baud rate to set itself. In 1995, the IEEE standardized on an "auto-negotiation" specification that defines the procedure that Ethernet devices are to go through to automatically determine the correct data rate (for example, 10 Mbps vs. 100 Mbps) and full duplex (simultaneously transmit and receive) vs. half duplex (transmit one direction at a time) settings required to communicate with each other.
Because today's devices auto-negotiate, data rate settings are not needed, except in two rare cases. One is when you desire added electrical noise immunity, which can be gained by forcing communications at the lower 10 Mbps data rate, and the other is where you are connecting an older device that may have been developed prior to or just after the auto-negotiation standard.
Maintenance implications of addressing
In order to install or replace existing industrial Ethernet devices, training with computer/laptop-based addressing software is necessary. Maintenance personnel can no longer address and install a system with only a screwdriver. With smaller systems, devices can be collected into one area such as a maintenance crib; physically connected to, addressed with, and disconnected from the computer one by one until the entire group is complete; then sent out to the plant floor to be physically installed in the various control cabinets.
As applications grow in complexity to many tens or hundreds of devices, you may have 50 identical products with different IP addresses to be installed across 20 different panels. In this case, the chance for putting the wrong IP addressed device in the wrong panel is higher. For large applications, it may be easier to first install them all, and then assign the IP addresses over the network based on a list of which devices (MAC addresses) are actually installed in which panels.
Greater use of fiber optic cabling
As stated earlier, one of the advantages of industrial Ethernet is the capability of faster data rates. Many of today's installations are based on first-generation 10 Mbps industrial Ethernet devices. New products have both 10 Mbps and 100 Mbps (called "Fast Ethernet") capability. However, as with all networks, faster data rates have shorter specified copper cable lengths. With Ethernet's high data rates, the maximum copper cable length is 100 meters. This is fixed regardless of the data rate. In the IEEE Ethernet specifications, it was decided to maintain the fixed 100-meter length, but require increasingly higher frequency carrying cables (at increasing costs) as the data rates increase.
Ethernet's 100-meter copper cable length specification is shorter than the 400-meter to 1600-meter or longer copper cable lengths possible with fieldbus systems. Because of this, fiber optic cables are required (not optional) for most long distance applications. The first step is selecting the right fiber cable based on differences in distance, cost and ease of termination. Plastic fiber cables are available that are optimized for creating short (50%%MDASSML%%100 meter) noise-immune custom cable lengths using standard wiring tools. Glass fiber cables cover greater distances (typically 2000 meters), and have become much easier to terminate than early connector generations.
Maintenance implications of fiber cabling
With the 100-meter copper cable length restriction of Ethernet, fiber optics will become required in more installations. The maintenance team will need to be trained on terminating fiber cables, or have preterminated cables purchased and available in the maintenance crib.