Completing the industrial Ethernet connection

Specifications, industrial environment and network structure are factors to consider when selecting devices for an industrial Ethernet infrastructure. However, getting data from device to device is another important factor. Ethernet wiring types, categories, cabling and connection options are also important components within the industrial Ethernet infrastructure.

By Larry Komarek, Phoenix Contact, Inc. November 1, 2005

Specifications, industrial environment and network structure are factors to consider when selecting devices for an industrial Ethernet infrastructure. However, getting data from device to device is another important factor. Ethernet wiring types, categories, cabling and connection options are also important components within the industrial Ethernet infrastructure.

Cable and connector issues

There are several naming conventions used in Ethernet. A “T” or “TX” refers to twisted pair cabling. “FX” refers to glass fiber optic cabling. For example, a 100-Base-FX specification means 100 Mbps (Fast Ethernet) over fiber cable. A “10-Base-T” means 10 Mbps over twisted pair copper cable. Copper cable is used for convenience; fiber optic cable is used for distance and noise immunity. The number of available cabling options is growing. Each has its benefits and drawbacks (See “Alternatives for 10/100 Mbps Ethernet cable”).

Copper cable

Copper cable length is limited to 100 meters between any two devices regardless of data rates. The higher the copper cable category rating number is, the higher the frequency it is able to handle. Unshielded cables are popular in the US; shielded cables are popular in Europe. When using 100 Mbps Ethernet in areas of high electrical noise, shielded cables can provide added noise immunity.

For industrial applications, Category 5 (CAT 5) is the most widely used. CAT 5 cables have two twisted pairs, are low cost and handle 10/100 Mbps data rates. CAT 5E (E = enhanced) cable contains four twisted pairs and can handle data rates up to 1 Gbps, but can be more expensive than CAT 5, depending on the vendor. CAT 6 cables also handle 1 Gbps, but have greater electrical noise rejection capability than CAT 5E, and are more expensive than CAT 5E. They can also handle 10 Gbps signals. Though the CAT 5E and CAT 6 cables are more expensive, costs have been dropping and many companies are installing it now to avoid rewiring in the future.

Virtually all industrial devices have 10/100 Mbps connections. The automation industry is in transition from 10 Mbps to 100 Mbps-based systems. However, due to 10-times greater noise sensitivity and increased cost factors, gigabit Ethernet installations are rarely used in industrial automation systems. When they are, they are used to create a main Ethernet “backbone.”

When selecting Ethernet cables for applications near cutting fluids, special cable coatings are required. Use cables with thicker outer coverings between panels. Thinner-walled cables are easier to bend for in-cabinet wiring.

second numeral indicates protection from ingress of liquids. The “5” means jetting water, any direction; “6” means powerful jetting water, any direction; and “7” means temporary immersion in water.

The RJ45 connector is adequate for most industrial applications. But in applications near moving machinery, RJ45 connectors with locking hoods should be used to provide added strain relief and shock/vibration immunity. There are also many control panel bulkhead connection options available for connecting programming panel and device cables to enclosures without needing to open the cabinet. Water-tight IP65/67 connectors are available for connecting Ethernet devices in modular machine mount or wash down applications (Fig. 1).

Fiber optic cable

Although copper cables are the most popular, fiber optic cables are necessary for longer distance runs and noise immunity in high (electrical) noise areas. Multimode glass fiber is the most widely used. At 100 Mbps, it allows distances of 2,000 meters or more between Ethernet devices. The new generations of terminations have reduced the complexity of terminating multimode glass fiber cables experienced in the past.

Older multimode installations may use ST style of connectors that were popular with 10 Mbps Ethernet systems. Newer systems use SC style connectors recommend for 100 Mbps installations.

When upgrading existing fiber optic systems with new equipment, also check that the wavelength of the light sources are compatible. Today’s 100-FX systems are based on 1,300-nanometer light. Older systems may be based on 850-nanometer light. When adding new switches to existing installations, check switch data sheets for the frequencies/wavelengths that the fiber interfaces support.

Unmanaged switches may have one or two fiber optic ports. These are used to connect the switch to the main network (backbone) or main control panel. A single port is used when the control panel is physically isolated and is connected through one long fiber cable. Dual fiber port connections are typically used to daisy-chain multiple switches. This saves fiber cable costs by eliminating individual long “home run” cables back from each switch to a central control cabinet switch.

For very long distances, single-mode telecommunications-grade fiber cables allow distances greater than 15 kilometers between Ethernet devices. Single-mode cables/connectors can be 4 times more expensive than multimode cabling, and require precise terminations by well-trained personnel. Using pre-terminated cables or professional installers can prevent reliability problems or system startup delays.

Plastic-clad glass (HCS) and new all-plastic POF fiber optic cables have recently emerged with 100 Mbps data-rate capability. They have shorter distance limits of 50—200 meters, but they are the most easily terminated. The POF connectors can be terminated reliably in 2—3 minutes without special tools. This makes them useful when routing within a cabinet around drives or starters; near welders, robots or large motors; or as a faster means of terminating custom cable lengths (vs. terminating RJ45 connectors). Plastic cables typically use SMA style connectors.

Modular Ethernet switches provide the ability to freely mix cable types: copper cables for in-cabinet cables, long distance glass fiber and nearby control panels with plastic fiber (Fig. 2). Some switches allow fiber connections to be accessed from the bottom, which allows you to mount the switch in shallow working-depth junction boxes without taking the extra space required to accommodate the fiber optic cable bend-radius. Exceeding the fiber optic cable bend radius specifications, which is typically about 2 inches, can crack or degrade the cable, leading to eventual failure.

Control device connections

nfrastructure. I/O systems allow real-time communication between PC/PLCs and sensor/actuator devices. Gateways translate the TCP/IP protocol to the fieldbus with which you wish to communicate.

Media converters, serial servers, I/O systems and gateways connect plant floor control devices to industrial Ethernet systems (Fig. 3).

Media converters

Virtually all industrial control devices today connect to Ethernet using copper cables with RJ45 connectors. However, media converters are used when fiber cable is needed to connect a device to the network. Media converters convert copper cable connections to one of the various fiber optic cable types. They are also useful for connecting devices to 850-nanometer legacy fiber optic systems or to new 1,300-nanometer fiber systems. Because the fiber optic ports on unmanaged switches are typically used to connect to other switches, media converters at the switch may be required when a third (or more) fiber connection is needed to go to an isolated device or cabinet. Media converters are available that support all glass and plastic fiber types.

Serial servers

Serial communications or device servers are used to connect existing RS-232 or RS-485 devices to Ethernet, replacing long serial cable runs. These servers insert serial transmissions inside the Ethernet packet and transmit them to either another serial server or a PC with software that strips out the information at the receiving end. These are intelligent devices that require IP addresses. They are configured through their web pages or serial ports.

I/O systems

Connecting discrete, analog, encoder and other control device types is accomplished via I/O stations. Ethernet I/O is available in form factors optimized for different applications. Modular Ethernet I/O is used in main control panels that have a greater quantity or variety of signal types to interface through. Block style I/O provides a fixed configuration of I/O in smaller sizes and at lower costs. These are optimized for small distributed I/O drops for multiple sections of a machine, or for monitoring systems. IP65/67 machine/process-mount I/O is also starting to emerge.

Other than the typical I/O system feature differences among vendors, there are several Ethernet-specific issues to consider. The main issue is deciding which real-time control protocol will be used. This is usually specified by the PLC or PC-based control system in use. Modbus/TCP, EtherNet/IP, Profinet and Foundation Fieldbus HSE for process control applications are the current major offerings. Some vendors standardize on one Ethernet protocol; others offer several for the same I/O form factors.

For monitoring applications, some I/O systems allow data and diagnostics to be viewed with Web browsers, reducing the need for higher-cost HMI/SCADA packages. There is a wide variability of diagnostics supported by available I/O systems. Most have Web-based diagnostics, link LEDs that show status (electrical and Ethernet connections OK or not OK), communication activity indication and any LEDs required by the protocol specification. Some may have numerical displays that can provide added fault-code diagnostic information without software. For example, a display could indicate that the device is waiting to be assigned an IP address.

A major option to check is fault response status. This feature determines whether the I/O system is to turn outputs off, leave in last operating state or go to a predefined state in the event communication is interrupted. In fieldbus systems, if a cable is cut, the physical connection is lost. The I/O station detects this and the outputs go into a predefined fault state (usually off).

With Ethernet, the connection between a PC/PLC and an I/O station might pass through many Ethernet switches. If a cable is cut between the middle two switches, it doesn’t affect the immediate connection between the I/O and the immediate switch to which it is connected. Without additional attention, the outputs would be held in the last state while the I/O waits for output update information that will never come.

Some protocols include a built-in timer. If an output command or a heartbeat signal is not received within a certain delay after the last update, the I/O assumes the communication is lost and goes into a fault state. Some protocols do not specify this function and it is up to the vendors to add the timer in the I/O device as a product feature. If one is using a new Ethernet I/O control protocol, ask the vendor how these types of faults are handled.


The major fieldbus protocols offer a gateway to the Ethernet implementation of their protocol. This allows existing fieldbus installations to be upgraded with Ethernet capability.


The new generations of Ethernet infrastructure components allow the cost effective implementation of distributed wiring approaches. Modular switches allow tailoring of port capacity and cable types while incorporating standards based redundancy and advanced message filtering functions. Device servers, media converters and I/O systems allow the types of control devices to be tied into a system in phases. Industrial infrastructure devices with easy to understand PLC-like Web pages and network management software allow networks that can be installed and maintained by plant floor personnel, yet are compatible with the IT world’s standards and support tools. Flexibility, performance and lower lifecycle costs are key elements for enterprise-wide industrial Ethernet integration and growth.

The Bottom Line…

  • Copper cable is used for convenience; fiber optic cable is used for distance and noise immunity.

  • When upgrading existing fiber optic systems with new equipment, check that the frequency of light is compatible.

  • Some I/O systems allow data and diagnostics to be viewed with Web browsers, reducing the need for higher-cost HMI/SCADA packages.

  • Larry Komarek can be reached at (717) 944-1300, ext. 3625 or at .

    • Alternatives for 10/100 Mbps Ethernet cable

      Cable type Distance/segment Application issues
      Twisted pair 100 meters In panel
      Lowest cost
      Plastic optical fiber 50 meters at Connect nearby electrically-noisy devices
      10/100 Mbps Easiest custom (fiber optic) termination
      Hybrid clad silica (HCS) fiber 200 meters at 10 Mbps Connect electrically-noisy devices
      100 meters at 100 Mbps Easy custom termination
      Multimode glass fiber 2 kilometers to Long distances and plant back-bones
      6.4 kilometers Moderate termination difficulty
      Single mode glass fiber 15 kilometers to Longest distance
      36 kilometers Highest termination difficulty
      Highest cost

      Third of three parts September — “Create the connection to industrial Ethernet”
      October — “Select the right industrial Ethernet device for the job”
      November —”Completing the industrial Ethernet connection”
      ABB: Technology competition claims “false” Wading into the ongoing debate over the value of FDT/DTM vs. EDDL for automation industry end users, an official for ABB Inc. said recent statements are misleading end users about the debate.
      “The authors of these statements want their audience to believe that FDT/DTM technology and EDD are competing, mutually exclusive technologies — either you support EDD or you support FDT/DTM,” said Mark Taft , senior vice president, systems marketing, ABB Inc., and a board member for the Fieldbus Foundation. “This assertion is simply false.”
      Noting that the company has supported users groups for FDT/DTM technology as well as EDDL cooperation to address interoperability issues, Taft said discussion of “Fieldbus Wars” is overblown. “A couple of vendors have chosen to withhold support for FDT/DTM from their customer-owners,” Taft said. “Automation system and field-device supplier members of the FDT Group have gone on record as supporting both FDT/DTM and EDDL for their customer-owners. The FDT Group has grown to nearly 40 member companies made up of automation and field device suppliers like ABB, Endress & Hauser, Honeywell, Invensys, Metso, Rockwell, Schneider, SMAR, and Yokogawa, and customer-owners like Saudi-Aramco and Shell .
      “We will support both EDDL and FDT/DTM as complementary technologies in our process instrument, control system and power technology products. We feel strongly that exclusion of either technology would compromise that goal,” said Taft.
      Standardized IEEE networks show top growth potential Led by 802.11b, standardized IEEE 802.11 networks are expected to account for the most significant gains in share among wireless industrial products in North America by 2007. According to a recent VDC study, IEEE 802.11a, b and g networks combined to account for 33.6% of the 2004 North American market for wireless products used in on-site industrial monitoring and control applications. VDC forecasts this share to increase to 41.4% in 2007.
      Proprietary protocols operating in the 900 MHz band accounted for the largest share of shipments in 2004, followed by IEEE 802.11b. The two networks combined to account for just over half of shipments and are expected to combine for a similar share in 2007, but with IEEE 802.11b gaining ground. Similar gains in share are also expected for IEEE 802.11a and IEEE 802.11g. Most of these gains will be at the expense of products using proprietary protocols in ISM bands.
      Reasons for this shift include:

      • Lower prices of IEEE 802.11 products compared to most proprietary networks

      • Standardization of office and plant floor networks

      • Multiple vendor options

      • High data throughput

      • Brand name familiarity from the IT department

        • Despite the advantages of IEEE 802.11, proprietary networks are often preferred in industrial applications where there is a need for longer transmission distances and bandwidth requirements are not high. There is also the perception that these products provide higher reliability and security.

Related Resources