Fieldbus - Conflicting "Standards" Emerge,
By : Jim Pinto,
IntroductionThe major “buzz” in the industrial automation business is networking communications on the factory floor and in the process environment: connecting sensors to actuators and everything else in between - programmable controllers (PLCs), I/O, computers, data acquisition, alarms, displays, recorders, controllers.
Everyone in the industrial automation seems to be waiting for “fieldbus” - the bi-directional communications protocol used to connect field-level instruments and controls. The IEC/ISA SP-50 international fieldbus specifications are still under development, with strong demands from major users and support from most of the instrument vendors. The users expect interoperability as the primary benefit of standardization, but that remains elusive as vendors push to promote their proprietary differentiation. Even when the fieldbus specification is complete and products are available, this will not eliminate the need for other industrial networks for a variety of reasons - cost, speed, complexity, compatibility with already-installed devices and future expansion requirements. The requirements for “fieldbus” simply highlights the advantages and benefits of using industrial bus networks.
The Choice Between the BusesA wide variety of networks is available in the industrial automation environment and confusion arises because their capabilities overlap. The figure here shows a simple overview and comparison between the major industrial networks available today.
At the lowest level are the “sensor” networks which were originally designed primarily for digital (on/off) interface. These are fast and effective, but with only limited applications beyond relatively simple machine-control. ASI (actuator/sensor interface) is still popular in Europe, while Seriplex is a US development..
At the next level in the hierarchy are the “device” busses - which provide analog and digital support for more complex instruments and products. DeviceNet interfaces very well with programmable controllers; however, it is still only a local, device-level network, operating typically with up to 64 points within about 1,000 feet. Profibus DP and Interbus are somewhat comparable to DeviceNet and are excellent, general-purpose device-level networks. LONWorks operates over greater distances and is currently the only practical peer-to-peer network, extendible to many thousands of points - though it is still comparably slow and more complex.
Profibus PA provides improved performance at the “fieldbus” level, for instruments and controls, replacing the features and functions which are provided by HART (originally developed for transmitter calibration and diagnostics). The primary differences are in speed, complexity and distance. LONWorks also extends into this realm. Fieldbus SP-50 overlaps somewhat with Profibus-PA and Profibus-FM, and this is the primary area of contention in the ongoing IEC/ISA SP-50 committee discussions.
At the next level we have the “control” networks - which include ControlNet (developed by Allen-Bradley and utilized by Honeywell), clearly overlapping with some of the functionality and performance intended to be provided by Profibus-FMS and Fieldbus SP-50.
At the highest level is information and internet connectivity and EtherNet TCP/IP is clearly the standard. During the past year EtherNet is gaining wide acceptance, even at the lower levels, because its performance is well established at the higher end, while commercial proliferation and plummeting prices have extended usage at all levels.
The overlap between the various bus-standards is evident and the details are still subject to dispute. The combination of commercial viability and practical time-to-market far outweighs theoretical considerations. The choice is dictated by point-count, physical distances, speed of response, hardware configuration, software requirements, cost and - perhaps most important - interface with other equipment and systems. Typically, a choice of one particular bus causes proliferation until limitations (speed, point-count) are exceeded.
Competing StandardsIn the industrial automation business the end-users have a clear and pressing need for “inter-operability” and this results in a push for “standards”. However, because of the tremendous breadth and scope of applications and because the specifications and capabilities of each network overlap considerably, there are several competing alternatives. New technology developments will inevitably increase the overlap and competing claims. Emerging standards are not necessarily the technical best, but rather those that are established by de facto market acceptance. Standards are typically promoted commercially by one company or organization, which takes responsibility for the open specifications and for timely implementation of extensions and modifications which are demanded by the market.
DeviceNet is based on CAN - the Control Area Network chip originally developed by Bosch to eliminate wire-harnesses in automobiles. DeviceNet was originally developed and introduced by Allen-Bradley (Rockwell Automation). To encourage industry-wide support for an open standard, responsibility for DeviceNet was transferred to ODVA - the Open DeviceNet Vendors Association - which is now responsible for coordinating specification changes and helping to assure interoperability between products supplied by multiple vendors. Many other major industrial automation vendors are now members of ODVA and are contributing strongly to its development and proliferation. Other industrial network "standards" are also based on CAN - CAN-Open in Europe and SDS by Honeywell Microswitch - but these have not attained a significant level of worldwide acceptance.
ProfiBus was originally developed by Siemens in Europe and has a significant group of allied vendors. PTO - the Profibus Trade Organization - includes both vendors and end-users dedicated to proliferation of this fairly widely used industrial network. Profibus includes three flavors - -DP at the lowest level, -PA and -FMS at higher levels of speed and complexity.
Interbus was originally developed by by Phoenix Contact and ASI (Actuator Sensor Interface) by Siemens and others, both with significant multi-vendor support and involvement. MODBUS was originally developed by Modicon, which is now a part of Groupe Schneider. Other “open” standards - such as Echelon’s LONWorks - are operated primarily by one company, with specifications published for any interested users and implementers.
Perhaps most important, the sheer volume of use of EtherNet in the commercial/office environment during the last few years has quickly made it a viable candidate for industrial environment networking, with comparable cost even at device-level. A new Industrial EtherNet Association has been formed recently (May '99) dedicated to proliferation of EtherNet in industrial applications. This group of users and vendors will focus on EtherNet adaptations and extensions suited to factory automation and process control, while still retaining interoperability and internet connectivity. Industrial features include real-time response, industrial connectors, etc.
Usage & ProliferationProfibus had made progress to the extent that more than a half-million nodes have been installed (mainly in Europe), with some installations arriving in the US and other parts of the world. Echelon claims to have installed “several million” LON nodes, though it must be noted that a large portion of these are in the building-automation markets and embedded in special non-inter-operable products. ODVA reports an installed base of approximately a half-million nodes for DeviceNet, though primarily in the US. ASI and Interbus are estimated in the same region, though primarily in Europe.
Ethernet usage is in the hundreds of millions, and it is anticipated that the sheer breadth of application will inevitably cause encroachment into the industrial device and fieldbus markets - to compete with, if not replace, the primary industrial network contenders. Recognizing this inevitability, most industrial networks are expanding their scope to include at least some level of EtherNet and/or TCP/IP compatability.
This important fact must be noted - no single fieldbus product will emerge as the “universal standard” in the industrial automation environment. It is far more likely that several standards will emerge at the various levels of performance and complexity, in different applications and environments.
InteroperabilitySince “inter-operability” between equipment from multiple vendors is a major factor for many users, all the major “open-bus” vendor organizations offer independently “certified” inter-operability - typically through independent third-party test establishments. This directly address the key point - whether or not I/O from vendor A will operate with sensors from vendor B and controllers from vendor C. Typically, the products are certified to work together. However, there may still be difference in speed of operation and other technical features, which still provide proprietary differentiation.
Silicon Chips - Availability & Cost
The price of silicon chips is dependent on size (chip "real-estate") and sheer unit volume. The industrial automation business is extremely fragmented, with very few products that literally sell in the millions. Silicon chips used in industrial controls typically cannot support volumes beyond a few hundreds of thousands, and hence silicon designed specifically for industrial products cannot achieve a pricing level that compares with consumer products. Therefore, the best alternative for designers of industrial controls is to “piggyback” on chips that are already utilized in high volume in other markets.
Echelons LONworks technology was initially a contender for low-level industrial networking, though it has made more headway in building-automation markets rather than industrial applications. The LON chip is manufactured by Toshiba and is available in high quantity at a price of about $5. CAN-chips are manufactured by Motorola, Intel, and several others, with high volume in other applications already yielding a cost of about $3 in reasonable quantities. Profibus chips are available primarily under license from Siemens and similarly, ControlNet chips are available only from Rockwell(Allen-Bradley).
At the Fieldbus level, both SP-50 and Foundation Fieldbus (the vendor sponsored fieldbus alternative that is being utilized by several major vendors while SP-50 is still being debated), as well as WorldFIP (the French-developed fieldbus alternative) have silicon available, though chip-prices and complexity of implementation are still indicative of early stages of development.
Recognizing that “higher-level” Fieldbus is significantly more complex, with the added burden of the politics of committee-cooperation, it is unlikely that final silicon will be available at anywhere near the cost which justifies general application in “low level” industrial networking. This limitation also inhibits increase of volume. In any case, under the best of circumstances, total worldwide unit volume is unlikely to exceed a few million, which is a relatively small quantity in the silicon production environment. Also, the chip (at least initially) will be relatively large in size, which keeps the price relatively high (perhaps $25). For these reasons, Fieldbus is still relegated to amorphous “higher level” applications, the realm of proprietary vendor/user partnerships rather than multi-vendor interoperability.
SoftwareAny networking system is intrinsically a tight combination of both hardware and software - firmware within the products and also external software, typically interfacing with a PLC (programmable logic controller) or industrial PC. Software provides one of the most important benefits of networking - remote configuration of the network and the individual modules; diagnostics (modules working and connected correctly to the other modules); prioritizing network access (some I/O may require more frequent access; monitoring (display), recording (storage of parameters over time) and HMI (human-machine-interface - a combination of all the above requirements).
Who Will Benefits From Standards ?A "standard" benefits end users - not vendors. Indeed, it is a reasonable and justified objective for any vendor to provide "differentiation", to offer products that produce better profit-margins. Expecting vendors to support a standard that eliminates their advantage is futile. The vendors all support "interoperability" within the standard, and the associations sponsor "interchangeability" between products from different vendors through independent testing. However, the incompatibility between vendor and user objectives will inevitably prolong the confusion until one - or more likely a "few" - networks attain de-facto leadership. Until that leadership is clearly established in the marketplace, conflicting claims will continue to cause confusion.
It should be recognized that a committee-generated "standard" is intrinsically a "bureaucratic" thing that obsoletes itself in the present-day fast-moving technological environment. The realities of a practical competitive environment demonstrate that no one is really willing to pay the price for a standard, when performance is available at a fraction of the cost.
What Are The Real Benefits ?Industrial networking for all factory automation and process control devices has significant benefits - reduced wiring, interoperability among multiple-vendors, enhanced field-level controls, simpler systems integration, easier troubleshooting and maintenance. A variety of bus networks will demonstrate these significant benefits over the next few years, paving the way for a complete re-organization of the industrial environment based on networking of sensors, controllers, computers, I/O, actuators and virtually everything in the industrial environment. Choices are dictated by clear benefits - in business that means productivity improvements and cost savings. Ultimately, industrial networks enable greater manufacturing flexibility, productivity, higher quality products and improved regulatory compliance.
PredictionsThe great debate on the IEC/SP50 Fieldbus standard will continue along the direction of its predecessor, MAP - the Manufacturing Automation Protocol. When it is finally complete, it will continue to attract lip-service and capitulation to demands by all major vendors. But, in parallel with their “interoperable” Fieldbus products, vendors will continue to offer proprietary extensions and performance improvements which will continue to defeat the underlying users need for “interchangeability”.
The reality of interoperability is a marketing dichotomy that will never really be resolved. Major vendors will show “interoperable” products, when demanded by major users, but will always inevitably retain the best performance for their own products when utilized as a family. In other words, a Honeywell valve may work with a Fisher/Rosemount system, but the best performance will be available only when that valve is used with a Honeywell system. And, of course, the reverse will be true. Some aspects of operation must, of necessity, remain “proprietary” - the fruits of “intellectual property”.
What will emerge is the use of “gateways” between buses and between proprietary products and open architectures. With the availability of cheap memory and processing power, it is far more practical to make a whole range of I/O products work with any bus through a gateway, without the need for converting the entire product line to operate with an “alien” bus and all its attendant hidden proprietary features. Gateways will theoretically appear to have intrinsic delays and performance limitations - but practically, can be made to operate with very little disadvantage.
In the meantime, it is likely that at least a few “de-facto” standards will emerge - CAN and LON at the lower end, which will eliminate HART, and various Sensor and Actuator bus products, and provide connections directly into the higher level networks like Ethernet and ARCnet which will provide enterprise-level communications. Fieldbus itself will remain a useful, more complex and costly, lesser used alternative, inhibited by its dependence on bureaucratic multi-vendor committee involvement.
Industrial networks bring lots of gains |
But that brings with it lots of games
The fieldbus wars cause a lot of fuss
The Users want an Open bus.
The Vendors simply can’t agree
You’ve gotta read between the lines
De facto standards win all debates
Return to Index of all JimPinto Writings
Return to JimPinto.com HomePage
If you have ideas or suggestions to improve this site, contact: firstname.lastname@example.org