1 April 2007
Fieldbus: Where do we stand?
There are two choices in network selection: Use your favored supplier and their network, or use a supplier who supports your chosen network. All networks are adequate.
- There is an overlap between factory automation fieldbuses and sensor networks.
- There is much debate on the virtues of field control vs. distributed control, most of it means nothing if the system is well designed.
- Many control loops are far too complex for field control.
By Richard Caro
When fieldbuses work in the factory, the purpose has been to reduce installation cost by moving the I/O interface from the programmable logic controller (PLC) to a remote I/O unit mounted close to the machine on the factory floor.
Factory automation fieldbuses are fast and deterministic. (Deterministic means the maximum worst-case time to obtain data across the fieldbus is accurately predictable and is not subject to chance.)
In the past, there was a distinction between fieldbuses and device buses, a name, which went to bus technology, intended for discrete automation. However, as a byproduct of the Fieldbus Wars of the 1990s, several device buses (Profibus and Interbus) were included in the IEC 61158 fieldbus standard.
Since most of the former device buses are capable of transporting scalar data, the distinction is no longer relevant. All of the network technology formerly called device bus is included in this article as a fieldbus.
All of the fieldbuses used for factory automation allow the remote device or I/O unit to be intelligent and execute software. All of these fieldbuses also work to interconnect PLCs and PCs into a network to share information.
However, there is no specific software to allow control logic to bridge across the network. This means logic in one PLC cannot link directly with logic in another PLC on the same network.
It is possible to accomplish these same goals, obtain the status of an I/O point, or an intermediate variable across the network to use in logic, but the timing of network access is not in synchronization with the timing of PLC logic and is usually not deterministic.
Fieldbuses used for process control came about explicitly to link smart field instruments to each other and to higher-level control systems.
While the speed requirements for process control are much less than for factory automation or material handling, within that speed the requirements are no less for determinism. In fact, process control adds an additional requirement for very tight time synchronization.
Additionally, the amount of data exchanged for process control is far larger than for factory automation or material handling, and almost always involves several floating-point numbers as well as several discrete status bits.
For many applications, process control fieldbuses are also required to conduct electrical power to field instruments over the same wires used to communicate data. In keeping with modern safety requirements in the chemical and petroleum industries, process control fieldbuses must prevent ignition of flammable gases by conforming to the requirements for intrinsic safety.
Factory automation fieldbus
The original application for factory automation fieldbuses was to connect remote I/O termination units to the PLC.
The original remote I/O unit was a small equipment rack to contain the PLC's I/O cards, but modern remote I/O (or block I/O) often appears to be an electrical termination block for connecting sensors and actuators. The signal-processing intelligence of the fieldbus is contained in the block.
There is an overlap between factory automation fieldbuses and sensor networks. Sensor networks cannot perform signal processing or simple control operations in the field but require the PLC to perform all logic.
The processors used to implement fieldbuses are perfectly capable of performing signal processing for discrete I/O and even simple logical controls, but control operations rarely take place at the fieldbus level.
It is possible to install intelligence into binary 2-state sensors and actuators with fieldbus communications. Two of the CAN-based fieldbus protocols, DeviceNet and SDS, have defined a number of functions only possible when there is a microprocessor operating in the field device with a direct connection to the sensor or actuator.
High-speed filtering and signal-processing functions supported by discrete automation fieldbuses becomes very practical with the close connection to the sensor, whereas remote connection to a PLC would require impractical high-speed polling.
Process control fieldbus
Typical process control operations are slower than those for factory automation applications allowing slower network speeds, but often requiring larger data transfers.
While factory automation networks must be fast, there are few demands for tight time synchronization between network nodes.
Process control, while not as demanding on network speed, demands tight time synchronization between nodes forming a cascade control loop.
Foundation fieldbus enables time-critical closed loop control data to move back and forth on its fieldbus network. The architecture of Foundation fieldbus is "field control," or the seamless construction of a control loop consisting of a highly interconnected set of function blocks to perform the closed loop control function with, or without, the participation of a host controller. Those who favor it consider field control to be more reliable, more accurate, less expensive, and more responsive than distributed control.
There is much debate on the virtues of field control vs. distributed control, most of it meaning nothing if the system is well designed. Most DCSs have implemented control with all cascades limited to a single controller, meaning cascade setpoints do not connect to the control level network. Most DCSs allow many controllers to exist and share information on the control level network as long as it is not time-critical closed loop control information.
Many of the benefits of field control rest on "better control," which is not often discussed in textbooks on process control. Better control means the normal oscillation of a closed control loop about the setpoint is smaller. Generally, removal of deadtime and hysteresis from the control loop will result in control with fewer and smaller deviations from setpoint-if the control loop is correctly tuned.
Conversely, proponents of distributed control, as in the classical DCS, have their own set of valid claims as well. Actually, there are two parts to the control story: signal processing and closed loop control. Both groups favor signal processing in the field device closest to the actual sensor. Signal processing consists of conversion of the raw data from the sensor to engineering units after removing the effects of noise and thermal drift. This could happen in the controller, but more and higher frequency data would be required. Doing the signal processing close to the sensor eliminates the need to send high frequency volumes of data. Additionally, alarm and limit testing can also take place in the field device in either architecture.
Distributed control proponents claim it is less costly and simpler to do the control function in a reliable device specially built for this purpose and share the cost among many control loops. With the cost of control room electronics constantly in decline, there is some merit to this claim. Simplicity of doing all control processing in a few multifunction controllers is a given; it is far simpler than field control. Additionally, there are many control loops that are far too complex for field control.
For most processes, a combination of field control and distributed control is probably the most economic, responsive, accurate, and reliable. Foundation fieldbus enables this type of mix for all DCS architectures. Profibus-PA enables only field signal processing, but does not enable field control. WorldFIP shares a common architecture with Foundation Fieldbus, but few choose to use it in process control applications. HART does not permit field control, but is fully capable of field signal processing, although is not often used for this purpose.
The networks we use
InTech's fieldbus survey showed responders' propensity to use (in reverse alphabetical order) Profibus, Modbus, HART, Foundation fieldbus, and DeviceNet. Responders leveraging Ethernet-based networks largely listed Modbus/TCP and EtherNet/IP over ProfiNet and FF-HSE. See the numbers on page 40.
DeviceNet is one of the specific implementations of CAN protocol in which a more complete physical layer and application layer has been specified.
The DeviceNet network is a fieldbus network that provides connections between simple industrial devices, such as sensors and actuators, and higher-level devices, such as PLCs and PCs.
DeviceNet offers master/slave and peer-to-peer capabilities in a flexible, open network with devices from a number of vendors.
Allen-Bradley launched DeviceNet in 1994. The ODVA (www.odva.org) now stewards the technology and its standards.
DeviceNet is a fieldbus and not a sensor network limited to connect discrete sensors and actuators. Variable-speed drives, bar-code readers, and other devices work with the DeviceNet network.
However, DeviceNet is usually promoted as a low-cost communications link to connect industrial I/O devices to a controller and to eliminate expensive direct-wired I/O.
EtherNet/IP came about through the combined efforts of ODVA, ControlNet International, and Rockwell Automation. Earlier efforts by Rockwell to encapsulate ControlNet protocol on Ethernet/TCP/IP was the basis for this development.
EtherNet/IP (Ethernet industrial protocol) implements data transfer using common industrial protocol (CIP) at the application layer. It operates on commercial Ethernet but spawned one of the ODVA special-interest groups to investigate alternative physical wiring and connectors more suitable to industrial automation.
The standard proscribes the use of Category 5E, Category 6, and Category 7 cable, bulkhead connectors with an RJ-45 form factor, and a round M-12 (12 mm) 4-pin connector.
The objectives of EtherNet/IP are to provide a full industrial-grade data communications service, using as much commercial off-the-shelf Ethernet hardware and cabling as possible.
The benefits are to obtain the speed of Ethernet at the lowest cost possible stemming from broad usage in the commercial market.
Foundation fieldbus was response to the needs for a two-way, all-digital, data-transmission network technology for use in process control.
From the beginning, it was to replace the 4 to 20-mA DC transmissions previously used for analog control instrumentation. It was also to use the same type of wire typically used for analog transmission, supply power to field instruments, and to fully conform to intrinsic safety requirements.
The ANSI/ISA 50.02 standard met all of these requirements and served as the basis for Foundation fieldbus.
The Fieldbus Foundation (www.fieldbus.org) came into existence in 1994, at the urging of many users, from two competing organizations, WorldFIP North America and Interoperable Systems Project.
The initial Foundation Fieldbus specification was for H1 for the targeted instrumentation connection 4-20 mA replacement application. H1 (Hunk 1) operates at 31,250 Kbps, a very low speed for a communications bus, but necessary because of the need to reject noise, deliver DC power, and provide intrinsic safety.
HART stands for highway addressable remote transducer.
It launched to allow digital field instruments to communicate data to host systems, while simultaneously transmitting on ANSI/ISA 50.1 standard 4-20 mA.
Since a digital processor is required in the transmitter, many parts of the instrument now process as and in the software, thus replacing more expensive hardware and analog circuitry.
This has lowered the selling price of HART (www.hartcomm2.org) transmitters to parity with pure analog electronics. Customers with plans for the future tend to use HART transmitters on all new processes and in retrofits of existing plants.
HART enables range changing. Modern instruments operate over much larger ranges than older analog instruments that tended to build the range into the design in order to achieve high accuracy.
Digital transmitters are able to use a portion of the range of a wide-range sensor and still maintain high accuracy. This allows the range to change through software as long as the narrower range is within the limits of the wide range.
Furthermore and as to any non-linearity due to the wide range? The software can compensate for that.
Modbus in all of its forms is the most popular control level bus.
Modicon originally created Modbus (www.modbus.org) for computers to gather information and control the operation of their PLCs.
The data organization of all PLCs is as a set of addressable registers organized into sets as I/O, control relay, analog inputs, analog outputs, and variables. The organization of PLC I/O is such that each digital discrete input device appears as a single bit in the I/O registers according to its location in the I/O hardware.
This usually means digital discrete output often mixes in some of the registers requiring a register mask to define the outputs.
Modbus commands provide ways to transfer the content of one or many registers from the PLC to the host device that may be a computer or may be another PLC.
The Modbus command set was so popular when it first came out in 1979, that other PLCs often copied it.
The idea behind Modbus, a command set operating on 16-bit registers, has found a home with all PLC suppliers and with ISO 9506 Manufacturing Message Specification.
Additionally, Modbus continues to be the most common protocol for use in Supervisory Control and Data Acquisition, SCADA. The same structure is at work in OPC/DA.
Modbus/TCP surfaced in 1998 and was declared "open."
The specification is on the Modicon/Schneider Web site, but ownership has moved to the independent Modbus association (www.modbus.org).
Modbus/TCP enables several improvements to Modbus communications. It lowers cost by using commercial Ethernet components, enables remote operation via the corporate LAN or the Internet, and increases operational speed to the LAN choices of 10/100/1000 Mbps of Ethernet.
It also exposes the PLC to the usual Internet security problems requiring the use of protection with known methods.
Profibus-DP: Although Profibus (www.profibus.com) started out as a standard communications link between PLCs and host systems such as HMI, the earlier Profibus-FMS was to be too slow to support HMI update.
When a standard connection with PLC remote termination units or remote multiplexers became a requirement, Profibus-DP came out to solve both problems. The high speed of Profibus-DP, up to 12 Mbps, became its most attractive asset.
This makes Profibus-DP both a control level bus and a fieldbus. Profibus International prefers the term Profibus rather than any of its modifiers such as FMS, DP, or PA, but industry continues to use these designations.
Many companies support Profibus communications for their products. Although it began as a German national standard, it has now been fully internationalized.
Among the benefits of Profibus are it is the factory communications standard for Siemens, one of the world's largest system integrators and the largest manufacturer of PLCs in the world. Integration with Siemens products becomes much easier when using Profibus.
Profibus-PA is a hybrid protocol using Profibus-DP command structures but the same physical layer as Foundation fieldbus H1. Profibus-PA is for use in traditional process control applications where delivery of DC power to the field instrument and support of intrinsic safety is necessary.
Unlike Foundation fieldbus H1, Profibus-PA is a master/slave network that is an extension of Profibus-DP.
Normally, the field instruments wire to a field junction box where they terminate in a Profibus DP/PA coupler. Profibus-DP serves as the higher-level control level fieldbus to connect PA segments to the control system master.
Field instrument power often comes from the junction box. Since intrinsic safety (IS) barriers do not exist for Profibus-DP, intrinsically, safety systems require the junction box with the DP/PA coupler to be in a safe area and the intrinsic safety barrier goes on each Profibus-PA segment.
ABOUT THE AUTHOR
Richard Caro (RCaro@CMC.us) has worked in industrial automation for almost 50 years and is the author of Automation Network Selection, ISA Press 2004. He is CEO of CMC Associates, a senior member of ISA, holds two patents, and has two chemical engineering degrees. He managed the ISA and IEC Fieldbus standards committees.
InTech Market Study: Fieldbus growing
Fieldbus continues to be an important element of the automation industry. Along those lines, InTech magazine conducted an InTech Market Study to find out fundamental trends and what users see as challenges and opportunities for the protocol.
During the survey it became clear manufacturers use the technology because when asked if their manufacturing facility uses fieldbus, 80% of respondents said yes.
One of the other areas of the study that also became obvious was manufacturers use more than one fieldbus protocol in their process. More than three times the amount of people responded to this question than any other question in the survey. Of the respondents that answered yes to which protocol or network do they use, the top five listed were Hart, with 17% of the responses; ModBus and Profibus, with 15% each, Foundation fieldbus, with 14%, and DeviceNet, with 12%. Other protocols receiving votes were DCS, AS-I, ControlNet, CAN, InterBus, CC-Link, and LonWorks.
Ethernet showing up
While using a bus is one thing, adding the protocol onto Ethernet is another. When asked if they used an Ethernet version of these buses, 62% of respondents said yes. The protocols they are using are, again accounting for multiple answers, EtherNet/IP with 37%; Modbus/TCP with 36%; ProfiNet with 14%, and FF-HSE with 11%, with other coming in at 2%.
Fieldbus will be coming to a plant near you as 65% of respondents said they have plans to implement the protocol this year, and 81% said they have plans to add it in the next three to five years.
While more manufacturers will be adding fieldbus, when they were asked if their company used an analog DCS, fieldbus, or fast Ethernet, 51% of those answering said they used a combination of those protocols. Of the individual items, Ethernet scored 20%, while DCS garnered 16%, fieldbus had 11%, and the remaining 2% was other.
In another area that shows the varied aspects of the automation industry, 44% of respondents said they practiced process control, while 24% said they worked in discrete control, 18% said batch, 12% said hybrid, and the remaining 2% was other. Manufacturers said they definitely benefited from distributed control, with 87% answering in the affirmative, while 84% said they gained from using smart instruments.
Is your fieldbus connected to the Internet? No was the resounding answer with 82% responding in the negative, while only 18% said yes.
With the Internet, though, comes a fear of cyber hackers launching into a system. Of those responding, 46% said they have a cyber security system in place, while 54% said no. Of those that have a cyber security plan in place, 78% said they were satisfied, while 22% said no.
Wireless is also making an appearance in the fieldbus area with 71% of respondents saying wireless will become a part of their fieldbus and control system solution in the future.