This document discusses use of an outside specialist for networking technology: How OEMs can focus on their core competencies; get high performance, functional, and conformance tested products in a short period of time; and bring a product to market much faster. It describes how an OEM can use a specialist as a resource for unusual or difficult applications.

Open Networks for New Designs

DeviceNet, Profibus DP, and Ethernet are rapidly becoming the basis of 85% of new installations and designs; the other 15% is distributed among less common networks such as Interbus, ControlNet, CANopen and AS-I. Foundation Fieldbus H1 and HSE are rapidly gaining ground in the process industry, as is Profibus PA. H1 and PA will be the dominant technologies for Intrinsically Safe environments for the foreseeable future.

Legacy Networks

In terms of existing installed base, here's what is being used most today:

• Allen-Bradley remote I/O


• Allen-Bradley Data Highway, DH+


• Modicon Modbus Plus


• GE Genius I/O


• HART

...and legions of serial protocols specific to various PLCs, drives, DCS systems, and I/O platforms.These networks are still important, but new installations and new business opportunities for these networks are declining.An in-depth discussion of the barriers to industrial networking, and Synergetic's vision to overcome them, is found in the white paper Crossing The Chasm: Overcoming the obstacles of industrial communication.

Which network do I do first?

Murphy's law dictates that as soon as eight months of DeviceNet development is complete, the first major customer is going to ask for Profibus. Lost opportunities are inevitable. The quickest answer to the "which Fieldbus" question is: implement Modbus RTU/ASCII, because Modbus is everywhere. In reality, it's probably the most supported of all industrial networks. Modbus on industrial equipment is like cassette players in cars. Its performance isn't great, and nobody's bragging about it, but it's out there. The advantage of Modbus is one can get from Modbus to just about any other network, using a gateway. Gateways are expensive and a bit "kludgey" but they can get customers out of a jam.

What does it take to implement a hardware design?

Cookbook designs are almost always available for Fieldbus interfaces. The hardware is the easy part. Regardless of which network is involved, the following components are needed:

• A microprocessor


• RAM & ROM/FLASH


• A communications ASIC (each particular network you're using has its own)


• Optical isolators / transceivers


• LEDs, connectors, DIP switches etc. depending on the network

A typical bill of materials cost for the above components can range from $20 (for an absolutely minimalist DeviceNet/CAN interface) to as much as $100 (for ControlNet, which has the expensive Rockwell ASIC).

What does it take to implement a firmware design?

Firmware must be written to manage communication between the host and the network. While the ASIC handle low-level portions of the protocol and message arbitration, most of the network services and protocol features must be handled by the firmware.The code must:

• Manage communication services (i.e., poll / broadcast / event driven messages / fragmented messages / configuration data)


• Properly pass data between the network and the product


• Handle asynchronous configuration / parameter data sent over the network


• Ensure that messages are always handled with 100% reliability


• Manage the communication controller.

Code complexity varies dramatically depending on how many services and messaging types are supported.Development time is 2-4 man months per Fieldbus for very simple implementations; 6 man months or more is typical. For quantities of fewer than several hundred units of each network, the development cost is easily 2-4 times the hardware cost!

Co-processor or single processor?

The existing host processor may be used instead of having a separate, standalone processor. However, one must make absolutely sure that the processor has enough extra bandwidth to properly handle all network messages under all conditions.In communication designs 16-bit processors are most common. In terms of general product design, 8-bit microprocessors are usually used for sensors; 16-bit microprocessors are used for drives, controllers, and operator interfaces; 32-bit processors are used for routers and very complex controllers.If the processor is already near its limits, then a second processor is a requirement. An industrial communication stack typically consumes 25-35% of a processor's power.

Native communication design, a gateway or plug-in?

Volume, time to market, and price determine the answer to this. A gateway is a "black box" solution that is expensive but quickly implemented. Usually these are used to solve an urgent problem on a specific installation. Plug-in modules are ideal for small to medium volume (i.e. hundreds of units per year) and can be quickly implemented. A chip-based solution may require the most up-front integration, but for high volumes the cost savings are dramatic.

A serial to Fieldbus gateway can provide an instant connectivity solution while implementing a more permanent one. Many times the customer requires that a barcode reader, drive, temperature controller, etc. connect to a Fieldbus network. But all that is available on the device is an RS232 or 485 port. Is there a solution?

Synergetic's Protocol Converters are "black boxes" which can quickly connect a serial device -- RS232, 422 or 485 -- to a Fieldbus network like DeviceNet, ProfiBus, InterBus, or CANopen. Alternately, a compact "Fieldbus cable" is available which has a Fieldbus connection on one side and a serial connection on the other.

The serial device must communicate using the MODBUS or other common serial protocol. Using RS485, one can multi-drop up to 31 serial devices on the serial side. Devices are mapped into a single slave node on the Fieldbus network. The Protocol Converter comes with a configuration utility that establishes communication parameters and maps one side to the other.

A more permanent solution is to use a snap-in communication daughterboard, such as the Synergetic OEM COM module, for all the major Fieldbuses. This is appropriate for smaller volume applications. A chip-based solution makes sense for larger, higher volume applications.

What is the opportunity cost of engineering a Fieldbus in-house?

Time to market: A basic network implementation takes 3 months to complete the basic design, 3 months to finalize the design and obtain certification, and 3 months to document, "productize" it, and inform marketing and sales channels of the new capabilities. This is a total of 9 months of dedicated effort per network.

If projected sales for the new product are $500,000 for the first year and $1,000,000 per year after that, then the cost must be taken from the best year. So a 9-month delay represents a loss of $750,000 of sales. For multiple networks, the only way to keep this from literally dragging into years is to assign multiple engineers to communication projects.(This is one of the hidden reasons why factory networking has not become nearly as popular as many people thought it would five years ago: The cost for OEMs supporting multiple networks is too high.)

Obviously, this cost greatly exceeds the engineering cost itself. If a turnkey solution is available which reduces the design, certification, and productization time to 3 months, then this drives $500,000 to the top line.

Focus: Managers must decide where to direct their limited energy and resources. A drive manufacturer must devote the majority of his engineering talents to core drive technology. Assigning key engineers to communication projects drains resources away from core competencies, and forces the organization to develop entirely new skill sets. The loss of focus is likely to have long term detrimental effects.

Successful Product Introduction: The greatest risk in "Do It Yourself" (DIY) network implementation is this: Because network connectivity is usually added in response to a major customer's requirement, the vendor is "on the spot." The major customer is now a beta site for a brand new product. Major customers don't like being guinea pigs.

And once again, Murphy's law comes into play: even assuming that the product is finished on schedule (!), the customer will still find bugs that nobody else could find. The vendor risks losing one of his largest customers on the success of a project which is outside his core competency.

What is the strategic significance of Fieldbus projects?

Network based systems tend to be large, involve many nodes, many dollars, and be "high visibility." Vendors report that they represent only 5-10% of their customer base, but 20-25% of their business! The larger an installation, the more likely a Fieldbus will be the backbone of the system.

How much up-front cost vs. per-unit cost?

Click here to view chart. Based on 3 Fieldbus networks and simple "slave" implementation

Time to market cost based on unit price of $500 per device

Engineering cost based on $10,000 per man month

Notes:
1.Gateways are handy, but economical only for very small quantities.
2.Time to market opportunity cost is the greatest liability of DIY
3.The engineering costs of DIY simply reflect the high cost of reinventing the wheel
4.The chip-based solution offers a powerful additional advantage: running the entire host application on the EC1 chip. This is discussed in detail in the white paper Embedded Communication: Option or Standard Feature?

What does it take to obtain conformance testing & certification?

All of the major Fieldbus trade organizations have appointed testing laboratories which apply rigorous tests to new products. The Profibus Interface Center reports that the average new product passes certification after two to three tests; Open DeviceNet Vendor Association reports an average of 2.5 tests. Testing costs an average of $5000 per product, plus travel.

What is required for tech support?

If the network implementation is done properly, support staff need only understand the basics of installing Profibus, DeviceNet, or Ethernet to support the product. It is a good idea to appoint a staff member to take charge of this responsibility for the company. However, knowing that these installations can involve finger-pointing contests and subtle problems, it is also reasonable to consider service contracts with qualified outside firms.

How can finger-pointing contests be prevented?

Miscommunication, and the failure to establish clear expectations and lines of responsibility, are the number one cause of these problems. Finger-pointing contests can certainly be technical, but more often they are caused by fundamental management problems.

From a technical standpoint, a few basic practices have proven to be effective:

1. Vendors must clearly communicate what network services are supported -- and not supported -- and documentation must anticipate problems in advance.
2. Customers and vendors both greatly benefit from having adequate network troubleshooting tools.
3. Most network problems are related to wiring problems, not device flaws.One of the advantages of going to a specialist for network product design is that the specialist is available to assist with difficult application problems, and in most cases has seen that type of problem before. The vendor becomes your "internal support network."

How much do communication engineers cost?

Engineers with communication and networking expertise are quite simply the most sought after in the technical world. It is well known that information technology specialists and network administrators are extremely well paid these days. Some highly skilled SAP specialists earn in excess of $1000 per hour.

Given the enormous growing demand for PDAs, handhelds, Internet appliances, and wireless devices, engineers who specialize in embedded networking are in just as much demand as IT personnel. A young engineer who can write firmware and software applications for networking earns anywhere from $55,000 to $85,000 per year, and is inclined to expect additional benefits and incentives. Obviously the fully burdened cost of such an employee easily exceeds $100,000 annually.

Firms who employ such engineers are advised protect their employees from recruiters, because it is certain that they will be calling. Firms who assign a core product design engineer to a communications project may in fact be grooming him or her for a more lucrative job in another company.

Which naturally leads to the question: What are the implications of losing a prized employee with communications expertise? Certainly countless projects have stalled for this very reason. The engineer leaves, and no one else is able to effectively pick up where he or she left off, or the staff are unable to support and maintain a recently released product.This is not an issue to be taken lightly, and it certainly illustrates the value of outsourcing a solution to an established communication specialist.

What is involved in writing the documentation?

Documentation is easy to write when core functionality and network connectivity are cleanly integrated. This requires broad planning at the outset, and a firm definition of how product functions will relate to network communication.The greatest value in a Fieldbus is the rich information that would otherwise be impractical to obtain in a hardwired system. With a network, all variables can be transmitted over the same wire. Diagnostic functions, and the ability to read and write all parameters at will, potentially add enormous value to the process.The device manufacturer must research what standard profiles exist for that type of device, and formulate an approach that takes advantage of as many commonalities as possible between the various Fieldbuses.

How does one deal with the evolving specs of Fieldbuses?

Specs are updated frequently. Any particular change will not usually affect most products. However, it is important to stay abreast of new developments. Changes which do affect a product will involve new communications firmware and recertification of the product. Open DeviceNet Vendor Association posts one or two changes to the DeviceNet spec per year.One of the advantages of a modular design is that it draws a clear line between sections of a product, meaning that changes to the product do not always require recertification.

Is it necessary to attend trade organization and SIG meetings?

Membership in the trade organizations is useful, and can become a marketing strategy component. However, this is optional for vendors who purchase a turnkey solution. Also, special interest groups (SIGs) within a trade organization may address important topics specific to an industry niche.

What's the difference between master vs. slave functionality?

A master can be considered five times as complex as a slave, and five times more expensive to implement. The most complex aspect is the configuration tools, which must establish all network device parameters and pass all relevant process data to the control program. This piece is easily overlooked but is without doubt a major undertaking.

To be a "master" is to take responsibility for most of the compatibility issues of an entire control system and to assure interoperability with devices from 100+ vendors. It is rarely economical to develop a master from scratch. EC1 offers conformance tested firmware and a variety of excellent choices for configuration tools.

What about "minority" protocols -- i.e. CANopen, LonWorks, ControlNet?

Occasional opportunities will arise for networks like these. They represent perhaps 15% of the nodes, but at times can be significant. Synergetic's OEM modules address these, whereas the EC1 chip only addresses CANopen, DeviceNet, and other CAN protocols; Profibus, Ethernet, and serial based protocols.

What expertise is available?

Dedicated specialists are on hand to help with product design. They represent a "safety net" for complex tech support problems, and can lend valuable advice on secondary issues like device profiles.

How fast can an outsourced network solution be brought to market?

Synergetic has customers who have implemented working solutions in a matter of weeks. The technical aspects are relatively straightforward. At that point, time to market is a question of how quickly management, sales and marketing can realize the benefits of this.

How can a communications partner help resolve "finger-pointing contests"?

From time to time, desperate situations arise where the best possible resource is the backing of a communication specialist who can resolve the problem quickly. This is part of the service which should be available from a communications partner, and Synergetic is committed to providing this.

How can a "generic" communication partner understand the subtleties of a specialized OEM control product?

It is easy for designers to feel that their product is so unique that no one can design acommunication interface better than they can.This is actually true -- which is why Synergetic has defined a universal communication interface that is the same for all supported networks. The universal interface makes all network services available so that the OEM can use his expertise to assign those services to device data and parameters.The advantage is that the Synergetic universal interface is much simpler to use than the network protocol itself.A communication partner can also offer insight into "best practices" and help the OEM understand how his customers will be using the network, and what limitations he should be prepared to deal with.