Selecting the Right Industrial Network

The vast number of available networks makes it a daunting task to pick one (or several) to best meet the needs of a particular application. This has probably hindered the use of networks in controls applications. The intent here is to discuss the process of picking a network best suited to an application, and not to compare the specific capability of different networks.

By Gary C. Peterson, Alvey Systems Inc. January 1, 1998

KEY WORDS

Networks and communication

Fieldbus

Device-level networks

Sensor/actuator-level networks

System integrators

Sidebars: Industrial Network Checklist Digital Networks: More Information at Lower Cost

The vast number of available networks makes it a daunting task to pick one (or several) to best meet the needs of a particular application. This has probably hindered the use of networks in controls applications. The intent here is to discuss the process of picking a network best suited to an application, and not to compare the specific capability of different networks.

Networks for controls are reaching a level of maturity where the benefits are now well defined, and the devices that attach to them are readily available and reliable. They are a viable input/output (I/O) solution for almost every application and should be considered for new and updated controls systems.

Networks for controls are usually classified into three categories—sensor networks, device networks, and field networks. There are overlaps in the capabilities between sensor and device networks, and device and field networks.

Sensor, device, or field?

Each category can be characterized by the way it communicates data.

Sensor networks communicate bits of I/O data associated with the status of a sensor, usually restricted to on/off status. This is usually limited to one or two bits per node; these networks cannot transmit words of data.

Device networks communicate bits of I/O data associated with status of a device (sensor or actuator). This can include diagnostic bits as well as status bits. Device networks are also capable of sending words of data in limited lengths per transmission. Data strings of more than eight or nine words can be segmented over several transmissions.

Field networks are characterized by long streams of data. Status bits of devices (if available) are embedded in words which have to be decoded to extract bit information.

Discrete, process control

When selecting a network, first identify the type of control application. A lot of overlap exists in the capabilities of different networks. Discrete control is best done and most cost effective with sensor or device networks. Process control applications tend to require longer data streams and the higher bandwidths characteristic of high-end device networks and field networks.

Open vs. proprietary

Open networks allow any manufacturer to develop interfaces for its devices. The “open” advantage is that the user has a choice among different manufacturers of the same type of device.

The disadvantage is that it falls to the controls integrator to ensure that different devices from various manufacturers all work together on the specific network. This brings up the issues of:

Conformance does a particular device communicate properly on the network?

Compliance: does a device work with all of the other devices on a network? and

Interoperability: can a device be replaced with the same type of device by another manufacturer?

Using devices that have been tested and certified to work on a particular network is an advantage for the controls integrator and end-user.

Processor, scanner interface

Networks are just I/O subsystems for the controls system. The choice of processor—PLC, PC, or other computer—will narrow the available choices of networks within a category. Scanner cards are the interface between the network and the processor. They are not all the same and should be examined for performance, even within a manufacturer’s product line.

Depending upon other factors below, part of the evaluation process may include an examination of alternative processors to control an application.

What devices are available?

The single most important factor is the type of devices required on the network for the specific application or range of applications. Additional products that can communicate on specific networks are becoming available daily. This can differentiate between a sensor and device network. Start with a list of devices required for the application.

For example, does the application require simple sensors, like photo eyes and proximity sensors, or does it require variable-frequency drives, operator interface terminals, barcode scanners, and other higher level devices, as well. This will drive the decision between a sensor network and a device network.

Diagnostics

Conventionally wired sensors and devices have little or no capability to communicate diagnostic information to the processor of a control system. Typically, if available, a single diagnostic bit is wired back, separately, to an input to indicate the status of a device.

Device-network and field-network devices have a microprocessor for communications, which can also be used to provide status and parameter data as part of the message to the processor. The information can then be used by the processor to do predictive failure analysis on the devices.

Photo eyes can have their excess gain—a measure of signal strength—compared to the optimal value, and thereby detect misalignment and dirt buildup on the lens. The same analysis can be done for motor currents, which can show increased loads from normal. This information can be used to direct maintenance people to take action at a convenient time, avoiding costly downtime. Payback for this capability is enormous and may not be considered in the cost justification for using this technology. If this capability is of benefit, it is a driver towards device networks and field networks.

Performance

When considering the different networks within a category, performance is an issue. The determining factor may be more than just raw speed.

For example, a network running at 500 kHz using a master/slave polled messaging scheme will have slower response than a network running at 125 kHz using a change of state messaging scheme. Another factor to be considered is the speed of the processor. If the cycle time of the processor is greater than the update time of the network, improving the latter’s update time will not improve the system throughput time.

It is important to understand the performance requirements of the system, and match the processor and network to them. The Nyquist Criterion, which states that in order to capture an event you must sample at twice the rate of the fastest event, applies to networks as well as any controls application. This will define the performance requirements of an application.

Determinism, design

Determinism is the predictability of network performance. This is very important in any control system. This is characteristic of most networks, but should be examined as part of the selection process.

To make the task of designing a network easier, some manufacturers have developed design tools to assist in the process. Most networks bus power along with the signal to the devices. The number and type of the devices, the distance between devices, and network topology will determine the highest operating speed of the network, and the number and placement of power supplies required by the devices.

Design tools assist in this task, and the availability of these tools should be another factor in the decision process.

Device configuration

Devices have to be configured to establish the address of the device on the network, and set performance parameters. Two common methods are used: hardware switches and software tools. Software tools also allow the setting of performance parameters in the devices.

The option for hardware or software configurable devices is not specific to the type or brand of network, in most cases. This can also be a determining factor in the selection of specific devices for a given network.

Network diagnostic tools

The various networks and devices are designed to be “plug and play.” The reality is that sometimes things that are supposed to plug in and work do not live up to that ideal. Determining the type and availability of network diagnostic tools is another factor to be considered when choosing a network.

Sensor, device, and field networks are cost effective as an I/O solution, when compared to the installed cost of conventional I/O systems. At least one network suits every controls application. The process of defining the application will lead to several viable alternatives.

Any perceived risks associated with this “new” technology have been diminished to the point of not being a factor. It is time to move controls networks from the leading edge into the mainstream of controls technology.

Alvey Systems Inc. (St. Louis, MO), established 1911, provides turnkey material handling systems for beverage and other consumer goods operations, employing its own brand line of package conveyors, high-speed sorters, pallet and unit handling conveyors, automatic palletizers and depalletizers, gantry robotics, and related software and controls. Alvey is one of six companies operating under the umbrella of Pinnacle Automation, which provides material handling equipment, software and systems for distribution, logistics, and manufacturing operations.

For more information visit www.controleng.com/info .

Author Information

Gary C. Peterson is manager of Controls Engineering for Alvey Systems Inc.

Industrial Network Checklist

What “level” of communications is required?

Sensor-level;

Device-level; or

In the field.

What’s the application?

Discrete control?

Process control?

Batch control?

An integrated application?

Are you seeking an open or proprietary network?

Open: generally more manufacturers, but may be additional integration issues.

Proprietary: one vendor is usually responsible.

Will the network need to work with another network at a higher or lower level?

Does the network have devices available to suit the application?

Would your application benefit from diagnostic information?

With the number of required connections, does the network provide adequate sampling speed?

Does the network protocol and powering method support the application’s topology and distance between devices?

Are there design tools to help with configuration?

How deterministic (predictable) does the network need to be?

Do you want devices addresses to be:

Hardware configurable?

Software configurable?

Are all, or a majority, of available devices certified or somehow guaranteed to be plug-and-play?

How complex is the project/will you need help?

Do-it-yourself; or

Hiring a system integrator.

Sources: Alvey Systems Inc.; Control Engineering

Digital Networks: More Information at Lower Cost

Why should users consider digital networks?

Standard digital networks offer many benefits to users of industrial automation systems and devices. These benefits accrue throughout the life cycle of a plant and its control systems—including planning, installation, operation, maintenance, and expansion.

The most-often cited benefit of a digital network is cost reduction in wiring and installation, with users of newer generation digital networks routinely citing savings of 50% and higher on wiring, panels, and junction boxes.

In the big picture, however, the most important benefit of digital networks is increased information availability. Today’s demanding business and regulatory environment requires companies to gather more information about their processes and the instrumentation connected to the processes.

Instrumentation is becoming more intelligent as embedded processing capabilities continue to get less costly, more powerful, and lower in power consumption. A digital network makes it possible to capitalize on the power of these intelligent devices.

More reliable, higher accuracy measurements can be communicated. Devices can perform and communicate multiple measurements, requiring fewer instruments and lowering equipment costs. System commissioning time is reduced because configuration data are readily visible from the control room. Calibration and diagnostic information can be remotely accessed, alleviating the need for technicians to go to the field for routine maintenance checking.

With some networks, deterministic control can actually be distributed in an interoperable way, using intelligent devices as open controllers.

In short, interoperable digital networks make it possible for users to really know what’s happening in their control systems with a higher degree of reliability and accuracy. These digital networks will dramatically change the types of information available and the way information is used. The ultimate benefit will be greatly enhanced ability for users to meet the real business needs of higher quality, increased productivity, and lower costs.

For more information, visit