Network Hardware Making the Connection

Industrial networks demand rugged solutions for multiple environments; commercial networks are generally lower-cost but hardware may not suit rugged appliclations. Adapting commercial networks, such as Ethernet, is increasingly popular among users and manufacturers.Considerations in making network connections include topology, type of hardware needed (see related text on gateways, bridges...

By Mark T. Hoske, Control Engineering March 1, 1999

KEY WORDS

Networks and communications

Fieldbus

Device-level networks

Sensor/actuator-level networks

Sidebars: When Things Go Wrong Building the industrial highway Installation recommendations before power-up

Industrial networks demand rugged solutions for multiple environments; commercial networks are generally lower-cost but hardware may not suit rugged appliclations. Adapting commercial networks, such as Ethernet, is increasingly popular among users and manufacturers.

Considerations in making network connections include topology, type of hardware needed (see related text on gateways, bridges, routers, switches), media and device information, and the level or levels at which networks can serve.

Ethernet emerges

InterlinkBT (Plymouth, Minn.), a joint venture between Banner Engineering and Turck Inc., specializes in fieldbus-related products. Bob Svacina, marketing director, says Ethernet is the biggest emerging trend, but there’s uncertainty. “With a PC or PLC host, it works OK, but device manufacturers are having trouble at the slave level because Ethernet communication chews up a huge amount of real estate on the circuit board, pushing the cost up. To make Ethernet deterministic requires smart switching hubs, which is expensive, perhaps up to $100 per drop. In the office, if a print file messes up, you just send it again, but if a robot goes crazy, a lot of damage can be done.”

Hirschmann-USA (Pine Brook, N.J.), which says it’s first with a line of industrially hardened Ethernet hardware, has switches, hubs, fiber-optic rail transceivers, and other products. People there say they’re just in time, as major PLC manufacturers offer ports to Ethernet.

In October 1998, Hirsch-mann unveiled its Industrial Rail Line Family, a hardened line of Ethernet products, says John McGilvreay, Hirschmann network systems division sales manager for North America. (Parent company is based in Nuertingen, Germany.)

Mr. McGilvreay says Hirschmann’s line, unlike commercial products, is DIN-rail mounted, 24 V dc power (rather than 110 V), redundant, and offers link recovery in less than 300 milliseconds. Designed for industrial automation, the products take the heat at 60 °C, without fans. “Customers tell us commercial Ethernet hardware can’t last three months in hot, humid, industrial applications.”

Mr. McGilvreay says, “With full duplex operation, prioritization and wire speed switching of up to 1 gigabit/sec, Ethernet has become a deterministic Ethernet service.

Proper network design can go a long way toward improving performance, including, for critical applications, a fault-tolerant Ethernet network design. Equipment can’t make up for poor design, says Mr. McGilvreay. “You can poorly design an Ethernet network, just like any other network.”

Customers want an open system standard, he says. “We’re going to see more Ethernet TCP/IP—it lends itself to remote control and management through the Internet. Ethernet also has scalability from 10 megabit, to 100 megabit, to gigabit Ethernet. As demand in-creases, prices will fall, like the commercial

market, where an Ethernet networking card is available for $20 and a five-port hub is under $40.” Issues manufacturers need to address include standardizing connectors, network security via firewalls, and establishing a standard application layer, he suggests.

Gateways translate

Beyond whether to offer Ethernet or not, device manufacturers have been faced with deciding which industrial networks to support.

SST Division of Woodhead Canada (Waterloo, Ontario) helps by offering the UCS (Universal Communication System), to offload the communications burden from device manufacturers. Using SST’s UCS removes the risk of choosing, developing, and maintaining the network interface, allowing device manufacturers to focus on core competencies.

Corina Silberman, product marketing manager, SST Gateway Technologies, adds, “With UCS, when connecting to current and future modules, no code change is required to the host product. This makes UCS unique; other products require code change to the host product. The most viable previous alternatives included integrating one network at a time to a device using a protocol stack.” Another SST product, X-Link, provides gateway translations from one network to another, such as between Rockwell Automation/Allen-Bradley Data Highway Plus and Groupe Schneider’s Modicon Modbus Plus (photo).

Synergetic Micro Systems Inc. (Downers Grove, Ill.) makes networking life easier via software drivers and fieldbus interface cards for multiple networks. Its set of Universal Software Drivers allows an integrator to build a system around one network, then transfer the investment to the next without making any software changes, according to company president, Michael B. Justice. Synergetic technology partner HMS Fieldbus Systems (Halmstad, Sweden) offers the AnyBus module, an interchangeble OEM fieldbus module for multiple networking applications.

Mr. Justice says, “Synergetic has defined industrial Ethernet as three things: an industrial DB9 connector (following the Siemens and Hirschmann specifications), a product or board that will survive in an industrial environment, and a long-term reliable supplier. These items along with with standard Microsoft Windows driver compatiblity give the industrial user a reliable product at a reasonable cost from a constant source.”

Other networking products are becoming more industry specific, such as use of Teflon and stainless steel materials for semiconductor and pharmaceutical industries. Another example is Turck Inc.’s recent MSHA (Mine Safety and Health Administration, U.S. Dept. of Labor) approval for more than 50 industrial-grade cable/connector components (photo). Similarly, Daniel Woodhead Co.’s Brad Harrison (Downers Grove, Ill.) unit has widened its range of sizes and breadth of quick-disconnect networking products.

Network topologies

Topologies, the patterns of interconnections among nodes, also can influence a network’s cost and performance. Topologies include bus (linear, branching, and tree), ring, and a combination of bus and ring through gateways, routers, and bridges. Star topology, often used in Ethernet, has a central node connecting to individual nodes. Mesh topology connects each node with every other node. For more about topologies, see Control Engineering , March 1998, p. 198, “Back to Basics” column, “How to Untangle Network Topologies.”

Network component manufacturers also can help confirm if the pattern of interconnection among nodes will allow the physical network operate as needed.

The number of devices permitted varies, as does distance and network cycle time. Up-to-date information is available from various network organizations, as well as from hardware vendors and system integrators. Major automation vendors often sell or have partners provide hardware for multiple networks. Most belong to multiple network organizations, which often have membership exceeding 100.

Connect with specific networks at:

ARCNet, www.arcnet.com ;

AS-Interface (AS-i), www.as-interface.com ;

CANopen (CAN-based bus): www.can-cia.de ;

ControlNet, www.controlnet.org ;

DeviceNet (CAN-based bus), www.odva.org ;

Fieldbus Foundation, www.fieldbus.org ;

Gigabit Ethernet, www.gigabit-ethernet.org ;

HART, www.ccsi.com/hart ;

Interbus, www.interbusclub.com ;

LonWorks, www.lonmark.org ;

Modbus, www.modicon.com/openmbus ;

Profibus-DP, -FMS, -PA, www.profibus.com ;

SERCOS, www.sercos.com ;

Seriplex, www.seriplex.org ;

Smart Distributed System (SDS, based on CAN), www.honeywell.com/sensing/prodinfo/sds ; and

WorldFIP, www.worldfip.org .

When Things Go Wrong

First question when things go wrongwith a DeviceNet network should always be: What has changed?

If nodes have been added or replaced, wiring changed, or device configured, go to where the work was done and search for the problem.

If no problem is found, see if the problem is in the physical media, node communication fault, or in network power distribution.

In general, check physical media and node configuration before checking network power distribution or isolating node communication faults.

—DeviceNet Troubleshooting Guide, 1998 InterlinkBT

Building the industrial highway

Bridges, switches, routers, and repeaters comprise the roadbed upon which industrial network protocols run.

Looking at each device as it relates to the protocol’s use of the seven-layer International Standards Organization/open systems interconnection (ISO/OSI) model is a great way of understanding these connection devices. To see how each relate to the model, see the network hardware diagram.

Switches are becoming the most widely used interconnection devices, as they assist in maximizing network throughput between PCs and PLCs. Routers are a key item in connecting different networks or for connecting the plant automation network to the Internet. As bridges don’t allow multiple stations to connect together simultaneously (as do switches), they are becoming less popular.

OSI model use

The simplest device is the repeater, which operates at the physical level of the OSI model and simply retransmits an incoming electrical signal. A repeater amplifies and retimes the signal received on one segment onto all other segments. Collisions, truncated packets, or electrical noise on one segment also are transmitted onto all other segments.

The main reason for using a repeater is to extend the segment beyond the recommended length. The number of repeaters is generally restricted to two but a maximum of four can be used. Using more repeaters can cause timing problems. A drawback with repeaters is that any activity on one segment of the network is repeated onto the other segments, thus increasing the overall traffic and reducing response time for individual stations on the network.

Bridges pass valid data

Bridges are used to connect two separate networks to form a logical network. A bridge allows only valid messages to pass to destination addresses on the other network. The bridge stores the frame from one network and examines its destination address to determine whether it should be forwarded over the bridge.

The bridge maintains records’ node addresses on both networks to which it is connected. However, the data link protocol must be identical on both sides of the bridge.

Bridges can be used to extend the length of a network (as with repeaters) but in addition they improve network performance. For example, if a network is exhibiting fairly slowresponse times, the nodes that mainly communicate with each other can be grouped together on one segment and the remaining nodes can be grouped together in another segment. The lower activity segment should see an excellent improvement in response times. Bridges should be designed so that 80% of the traffic is within the LAN and only 20% crosses the bridge.

Switches add bandwidth

Normally Ethernet uses a media shared by all the devices connected to it. The maximum transmission speed on the standard Ethernet network is 10 megabit/sec. However this speed must be divided by the number of devices that are attached to the LAN. Each attached device has an effective transmission speed of 10 megabit/sec divided by the number of stations. To increase the bandwidth available to all devices, an Ethernet switch can be used to connect individual LANs or stations together and they can then communicate at the maximum speed of 10 megabit/sec.

Routers maintain addresses

Routers transfer data between two networks that have the same network layer protocols, but not necessarily the same physical or data link protocols. (Network layer is the IP in the TCP/IP—Transmission Control Protocol/Internet Protocol—a commonly used combination protocol for communications via Ethernet.)

Routers maintain network tables and can route messages. Routers use the Network (IP) address—which contains routing information—to determine where the message should be sent. They maintain tables of the optimum path to reach a particular network and redirect the message to the next router along the path.

For more information from IDC, call 800/324-4244, or visit

Steve Mackay P.Eng MBA B.Sc(Elec.Eng) B.Sc(Hons) is technical director for Instrument Data Communications (IDC), global consulting firm and provider of engineering training workshops.

Installation recommendations before power-up

Vendors and associations for digital industrial networks provide substantial documentation to assist with network design, installation, and startup. Recommendations vary from network to network.

Honeywell’s Smart Distributed System provides advice in three stages: before power, when powering, and when excuting the control program. Here are some before-power recommendations (condensed from

Use an ohm-meter to search for short circuits and reversed wiring in communication and power lines.

With two 120 ohm terminating resistors on ends of the bus, resistance across bus positive and negative (with power off) should be 50-70 ohms.

Ensure cable and tee support eliminates cable connection stress. Bend radius for mini cable should exceed 8 in.; micro cable bend radius should exceed 3 in.

Verify topology, checking bus speed, and host baud rate, length of bus and branches, and device addressing and configuration.

Ensure number of devices do not exceed limit (64).

Verify tightness of screw-terminal connections, and that green ring lines on plug-in connections show (hand-tighten only).

solate wires a minimum of 3 in. from other wiring, especially when dealing with high voltage, switched load, or electric motor drives. Bus wiring should not be installed in parallel with power or other system wiring.

Verify that the shield drain-wire is intact, measures less than 1 ohm to earth ground, attaches at a single point to control earth-ground start point.

For pigtail, bulkhead, or hardwired connections, verify the shield drain wire is clipped flush where it exits the cable and that the black-white and brown-white pairs remain twisted and as short as possible.