Basics of industrial Ethernet
Ethernet is not the internet; Ethernet is not a fieldbus; and Ethernet is not a control system. Ethernet is a networking standard. Period.
An Ethernet network can have up to 1024 nodes, hundreds of cables, and many combinations of network management devices. In order of complexity, these network management devices have evolved to include:
Hubs or repeaters
The difference among these devices is the protocol layer with which it is designed to operate (See "Ethernet, OSI, and TCP/IP").
The terms hub and repeater typically are interchangeable. A hub simply redistributes data; it does not interpret or sort the messages that pass through them. Hubs extend the length of the network by repeating the signal, connect LAN segments together to form a single network, and allow conversion between cable types such as from unshielded twisted-pair (UTP) to fiber optic.
Since hubs are designed to regenerate signals without filtering or directing them, they offer little or no network protection. Some simple hubs amount to little more than electrical buffers with some basic noise filtering thrown in. However, some have limited store-and-forward capability.
Hubs work at the physical layer only (layer 1). They are not assigned MAC or IP addresses. Although they connect Ethernet nodes, all of the nodes share the same bandwidth. This means that if more nodes are added to the network, they have to compete for finite bandwidth.
But the primary reason that hubs are not reliable on the plant floor is that they forward data packets to all nodes simultaneously, making data collisions inevitable. Because data collisions create network delays of undeterminable duration, hubs are useless for control applications, which require that data packet transmission time is &100 msec.
Bridges connect separate networks. They operate on the physical (layer 1) and data-link (layer 2) layers and manage data traffic between networks of the same type, such as between Ethernet networks. Accounting and plant engineering departments may be on separate networks. However, a bridge can connect them. Bridges also provide protocol conversion between related networks, such as between Modbus and Modbus+.
Bridges differ from hubs because they open and check the data packets they receive. Most bridges can learn node addresses, so that they allow only necessary data traffic making forwarding decisions based on MAC addresses. Another helpful function of bridges is the ability to work with error-detection schemes such as CRC to prevent ill-formed packets from passing to other networks, and passing only the healthy packets. Some intelligent bridges can learn which devices are connected to each side, and determine which messages to forward and which ones to block.
A router is similar to a bridge because it routes information. However, routers operate at the network layer (layer 3), make forwarding decisions based on the IP address instead of the MAC address, connect complex networks such as the internet and corporate LAN, and divide large networks into logical subnetworks, such as geographically separated plants, offices, or divisions.
Whereas bridges route information among addresses on a single network, routers contain information on both networks and individual addresses in a network by maintaining tables of IP addresses on each segment. Routers open data packets to determine initial and final destinations, and purpose. They can filter messages based on specific applications and users.
A bridge can control data packets within a factory network. However, routers can manage factory network data packets as well as determine which data packets should be sent to different networks — including those connected directly to the internet. They learn the most efficient paths for sending messages to their destinations. The internet itself is made up of routers that control traffic among its stations.
A switch is more complex than a hub. It provides full bandwidth and storage to each node or node group. A switch directs data packets to the appropriate node or port instead of broadcasting to every node simultaneously like hubs do. Typically, switches operate on layers 2 and 3.
Because switches are multiport devices with fast backplanes and inherent intelligence, they can act as both bridges and routers. They accept multiple concurrent data packets and feed them onto the high-speed Ethernet backbone or between ports.
Switches can convert 10-Mbps Ethernet and high-speed Ethernet. They divide a large network into many smaller networks and can dedicate one port per device.
A switch directs, or switches, messages between the input port and an output port. It can operate in half-duplex or full-duplex mode. In half-duplex mode, a port cannot receive and transmit at the same time. If it tries to do both, the received signal is detected as a collision. However, full-duplex mode allows the switch to transmit and receive at the same time — effectively doubling the network bandwidth.
Switches used in full-duplex mode are not subject to collisions, therefore offering deterministic operation for industrial networks. An industrial Ethernet switch increases network bandwidth and provides network determinism for industrial control applications, and provides the most cost effective solution for industrial environments (Fig. 1).
Primarily, the cable types used for Ethernet are coaxial, twisted pair, and fiber optic. In the early days of Ethernet, coaxial cable was used primarily. Since coax is a closed transmission line, a terminator must be used at the end of the line to prevent reflected waves, which could cause a host of problems on an Ethernet network. The system is called closed because the shield is supposed to keep energy trapped within the cable.
Originally, coax used for Ethernet had a 50-V impedance and used an N-type connector. Its large diameter made the cable stiff and unwieldy and perhaps is why it was called "Thicknet." In 1998, a newer version of Ethernet was developed that uses a thinner, more flexible 50-V cable — RG-58A/U, which uses a stranded center conductor, a single-layer braided or foil shield, and BNC-type connectors. This version became known as "Thinnet" or "cheapernet."
The next Ethernet standard was Category 5 (Cat-5) unshielded twisted pair (UTP), which is still commonly used for 10BASE-T and 100BASE-T Ethernet – referring to IEEE cable nomenclature – in office environments (See "Regular, fast, and ultrafast Ethernet"). However, UTP typically is not suited for the plant floor in many applications because its pull strength is low (about 25 lb), it is not crush resistant (typical tie-wraps could damage it), its bend radius must be >1 in., and its maximum run length is 100 m.
About the time Cat-5 was standardized, Category 5e was introduced. Category 5e is known as enhanced Ethernet as well as Gigabit Ethernet. It works by dividing the signal among its 4 pr of cable, transmitting 250 Mbps on each pair, and allowing full-duplex operation.
In 2002, Category 5i was proposed as a Telecommunications Industry Association (TIA)/Electronic Industries Alliance (EIA) standard. The idea was to standardize a Cat-5 version for industry. However, the proposal never became a standard because it did not accommodate the IP67 requirement for Ethernet cabling in harsh environments. IP67 represents ingress protection from a 1.0-mm object and and temporary immersion in water.
Category 6 can carry higher frequencies than 5e. This property makes it superior for the industrial Ethernet, right? Not necessarily. Because Category 6 has such a high frequency-carrying characteristic, it may be more sensitive to electrical noise from motors, drives, arc welders, and other equipment found in most plants. Shielded twisted pair (STP) cable resists electrical noise and is preferable to unshielded twisted pair or UTP in typical plants.
Fiber optic cable uses glass strands to transmit signals on a carrier of light. It is the best transmission medium of the three types of cable because it can carry higher data rates over longer distances. Fiber optic cable is more expensive than coax, UPT or STP, but it circumvents electrical problems such as noise, harmonic distortion, and grounding — problems that are compounded in high-speed networks.
Ethernet standards specify the RJ-45 connector for 10BASE-T and 100BASE-T. However, as with Cat-5, standard RJ-45 connectors fall short of many plant floor applications because they are not rugged, waterproof, or vibration proof. However, sealed RJ-45 connectors with IP67 rating have emerged recently (Fig. 2). Several vendors offer this and the M-12 connector, which has an IP68 rating (continual immersion), and recently was adopted as a standard by Open DeviceNet Vendor Association (ODVA).
Not so fast
If you think that your corner office supply can outfit your factory floor, think again. Installing Ethernet in an industrial environment is not the same as hooking up an office LAN. To make Ethernet work on the plant floor, issues such as determinism, topology, equipment robustness, power, and redundancy must be considered.
Network configuration and speed affect determinism. Office environments can tolerate occasional data packet collisions, along with the associated error recovery delays. Industrial equipment — especially control systems — is not that forgiving. Depending on the network requirements, topologies can be configured to ensure deterministic operation. Switches, along with star and ring configurations virtually eliminate data packet collisions making the technology aspect of industrial Ethernet possible. Much more can be said about redundancy and topologies — so much more, in fact, that it will take a series of articles to do them justice.
Equipment used for office LANs is not robust enough for the plant floor in most industrial applications. However, vendors are rapidly introducing Ethernet products specifically for use in industrial settings. Compared to standard office-type Ethernet products, industrial Ethernet offerings have higher temperature ratings, higher tolerance to shock and vibration, resilient industrial cases, DIN rail mounting, industrial power input — with redundancy if necessary, and in some cases, hazardous location certification. Although not every application requires waterproof connections, sealed RJ-45 and M-12 connectors are available for those that do.
There may be light industrial applications where one could get by with commercial off-the-shelf (COTS) Ethernet products. But since industrial Ethernet products are available now, why take the chance?
PLANT ENGINEERING magazine extends its appreciation to Automation Direct; Comtrol Corp.; Hirschmann; N-TRON Corp.,; ODVA; Phoenix Contact Inc.; Rockwell Automation; and Schneider Electric for the use of their materials in the preparation of this article.
A protocol is a set of rules that determines how two devices should communicate. The tasks required of protocols include:
Error detection and correction
Routing messages through complex networks
Data encryption and security
Consistent signal levels among devices
Media access control (MAC) is a protocol operating at the data link layer used to manage a station’s access to the communication channel. A MAC address is a unique address assigned to a station interface, identifying that station on the network. With Ethernet, this is the unique 48-bit station address. MAC address is the same as the physical address.
Internet protocol (IP) is the network layer in the TCP/IP communications protocol. IP contains a network address and allows messages to be routed to a different network or subnet. IP does not ensure delivery of a complete message, but the TCP transport layer is used to provide that guarantee. An IP address is the address of a computer or device attached to a TCP/IP network. Every client and server station must have a unique IP address. Client workstations or devices have either a permanent address or one that is dynamically assigned to them.
Ethernet, OSI, and TCP/IP
Ethernet is the most common networking standard, because it supports many protocols, and its cost is relatively low. However, Ethernet is not a complete network on its own; it needs upper protocols.
Ethernet was developed by Intel, Digital, and Xerox (DIX) in 1979. However, in 1978, the International Standards Organization (ISO) had already established the Open Systems Interconnect Reference Model (OSI/RM), which divides network functions into seven layers. The OSI model is a template used for comparing network communications standards.
Layer Name Description Types 1 Physical Ethernet layers Transforms data into bits that are sent across the physical media. Media layers 2 Data link Determines access to the network media in terms of frames. Its MAC sublayer is responsible for physical addressing. 3 Network TCP/IP layers Determines data paths within the network. 4 Transport Provides end-to-end, reliable connections, often in terms of segments. Also includes error checking. 5 Session Allows users to establish connections using intelligently chosen names in packet. Also opens and closes communication paths. Host layers 6 Presentation Negotiates data exchange formats, also in terms of packets. Building blocks of data encryption. 7 Application Provides the interface between the user application and the network through messages.
Ethernet uses Transmission Control Protocol/Internet Protocol (TCP/IP) to provide layers of the OSI model. Most networks do not use all layers. For example, Ethernet and RS-232 are physical layers; layer 1 only for RS-232; layers 1 and 2 for Ethernet. TCP/IP is a protocol, not a network, and uses layers 3 and 4 regardless of whether layers 1 and 2 are a telephone line, wireless connection, or 10BASE-T Ethernet cable.
Data move from layer to layer within the seven layers of the OSI model. Ethernet supports the physical and data link layers. With (TCP/IP) as its protocol, it supports all seven layers of the OSI model. Several types of Ethernet cables support the physical layer.
IEEE cable nomenclature:
Customarily, IEEE naming practices can be confusing and overwhelming. However, insight into the rationale behind names chosen for Ethernet standards can be gained by examining what each part symbolizes.
The first term is numerical (10, 100, 1000) and indicates the transmission speed in megabits per second (Mbits/sec or Mbps). The second term indicates transmission type: BASE = baseband; BROAD = broadband. The last number indicates segment length. A 5 means a 500-m segment length from the original backbone.
In recent IEEE standards, letters replace numbers. For example, in 10BASE-T, the T means unshielded twisted-pair cables; in 100BASE-T4, the T4 indicates four unshielded twisted pairs.
Regular, fast, and ultrafast Ethernet
Ethernet supports several types of cables — each intended for different purposes. These cable types include:
10BASE-2 (Thinnet) — supports network speeds of 10Mbps. 10BASE-2 uses RG-58 coaxial cable to transmit baseband signals on 200-meter segments. Total network length can be 925 meters.
10BASE-5 (Thicknet) — supports network speeds of up to 10Mbps and uses RG-8 or RG-11 coaxial cable to transmit baseband signals in 500-meter segments. Total network length can be 2500 meters with up to 300 nodes. Today, 10BASE-5 is rarely used.
10BASE-T (twisted-pair Ethernet) — supports network speeds of 100 Mbps. The most widely used Ethernet cabling to date, 10BASE-T uses two pairs of 22 or 26-AWG UTP cable to transmit baseband signals on maximum 100-meter segments — one pair to transmit and the other pair to receive. 10BASE-T permits a maximum of 1024 segments and 1024 nodes. There is no limit on network length.
100BASE-T (fast Ethernet) — is 10BASE-T with the original Ethernet media access controller (MAC) operating at 10 times the speed.
100BASE-TX — uses two pairs of Cat-5 UTP or Type 1 STP cable. It is popular for horizontal connections
100BASE-T4 — uses four pairs of Category 3 or better cable. Use of 100BASE-T4 is not common
1000BASE-TX — is also known as Gigabit Ethernet over UTP. It uses all four pairs to transmit in both directions and can transmit and receive simultaneously. Gigabit Ethernet is also called ultrafast Ethernet because it is 10 times faster than fast Ethernet.
10BASE-FL (fiber optic link) — replaces the 1987 fiber optic inter-repeater link (FOIRL) specification and is backwards-compatible with existing FOIRL devices.
10BASE-FP and 10BASE-FB — are dead. P = passive; B = backbone.
100BASE-FX — uses two strands of multimode fiber. It is popular for vertical or backbone connections
Ethernet cabling tip:
Install strain relief at the ends of cables to prevent stress on connectors.
Ethernet cabling tip:
Use color-coded cable markings to identify network sections and branches.
Ethernet cabling tip:
Good cable minimizes errors and provides faster data throughput.