Article

Coordinate measuring machine (CMM) software is experiencing more change now than it has at any other time in recent memory. In fact, in the world of CMM software, the next three years will be the most interesting we've seen, because change is happening so rapidly.

 One key factor driving this change is a strong, almost urgent, need to share proprietary part programs among the top CMM suppliers. This need is a result of CMMs having assumed their rightful place on the factory floor as true manufacturing machines.

 The second factor driving this change is the need for more accurate and repeatable computer-aided design (CAD) data. This may not be immediately urgent, but will become so in the next few years. To some extent, this trend is attributable to the role of the CMM as a manufacturing machine, but it also relates to trends and shifts within the CAD software industry.

 But let's put this into context.

 Within three years, the Dimensional Measuring Interface Standard (DMIS) will be fully accepted and employed throughout the industry, and everyone will be enjoying the benefits of a standard interface and a standard set of data formats. DMIS, developed by the Consortium for Advanced Manufacturing International (Bedford, Texas), has been approved by the American National Standards Institute as a protocol for efficient bidirectional communication of inspection data between CAD systems and CMMs. DMIS is a standard, vendor-neutral language for part inspection programming, similar to the Automatic Programmed Tool language for numerically controlled machine tools.

 However, like many other advances in computer-based technologies, DMIS looks forward, not backward. Although manufacturers will share its benefits in the future, there are still tens of thousands of inspection routines for thousands of critical parts that exist in proprietary, nonsharable formats. But they need to be shared now.

CMMs as manufacturing machines

 This urgent need for information sharing and routine sharing is the result of one trend during the last five years: the essential shift in the role of the CMM. Five years ago, CMMs were sold mostly as sophisticated go/no-go gages. Today, the CMM is a manufacturing tool, used less often to measure products at the end of production, with the objective of scrapping the bad and shipping the good, and more often to monitor processes, with the objective of taming manufacturing noise to keep all parts good by means of a continuous flow of CMM data.

 As the CMM's role has shifted from policeman to key player in closed-loop control, manufacturers have come to rely on the CMM as one more programmable machine in their arsenal of quality manufacturing tools. At the same time, in arenas from automotive to appliance to consumer electronics production, manufacturing has become global. Manufacturers now ask to see (and to measure) the same parts and products flowing from manufacturing lines worldwide.

 From this global perspective, manufacturing has to be consistent from site to site, despite differences in available production machinery, materials and the labor pool. The need to derive homogeneity from a variety of heterogeneous manufacturing environments has become particularly acute in this era of worldwide competition.

 Global manufacturers might have CMMs from two, three, four or more suppliers, all deployed in measuring parts and components for a single product line. These manufacturers have no desire to replicate the same data in three or four totally different proprietary formats, but they do have a great desire to continue to employ machines from different vendors. These machines haven't reached the end of their economic lives, and more important, they have demonstrated their usefulness. Simply replacing them with new, DMIS-based equipment is not feasible in the short-term.

Competitors talk among themselves

 The result is a clamor for intermachine, intervendor data sharing. The most forward-thinking CMM suppliers have begun to address this need, maximizing their existing equipment by creating a data interchange between their systems and those of their competitors. To accomplish this, three steps are required: de-emphasis of proprietary formats, open sharing of data formatting and significant allocation of software resources.

De-emphasis of proprietary formats

 The first step is by far the easiest of the three because most major CMM suppliers are already moving rapidly away from one-of-a-kind, proprietary software formats. The move is largely customer-driven. Many customers, after evaluating offerings from different suppliers, have reported that the software is pretty much on a par, with differences only in user interfaces and a few details. More and more, the choices are predicated on critical machine features or special capabilities for specific applications, along with the software maker's demonstrated level of commitment to open software standards.

 Given this sales reality, few suppliers have continued applying resources to proprietary software development, which clearly will give very little return in the future. For some time now, they have been thinking instead in terms of open standards. The groundwork is done, shaving literally years from the development effort.

Open sharing of data formatting

 The second step is perhaps the toughest. Layer after layer of software evolution has occurred in any given set of machines. In the absence of standards, the software teams developed each layer from their own viewpoint, using their own terminology and focusing on that layer's set of specific problems. To move from one machine to another, we need to reconcile those differences.

 Moving from one set of proprietary software to another involves much more than simple data translation, because even the definition of the data itself is often different from one CMM supplier to the next. In any case, data translation brings up its own set of issues, including problems of exact translation. What's required is interoperability, and that's a major challenge.

 The outlook isn't as gloomy as this challenge suggests. Immediate, open sharing of data among differing software is possible, but suppliers will have to take the unprecedented step of working together to meet the needs of a given customer. Fortunately, once they begin working together, they will find that each specific situation presents a fairly constrained problem set: One major customer will have one specific set of parts and geometries to be handled cooperatively. During the next few months, expect to see major CMM suppliers cooperating to ensure that their key customers' machines have all of the accuracy, repeatability and speed needed.

Allocation of resources

 The third step is actually part of yet another trend: greatly increased emphasis on customer service on the part of CMM suppliers. This will become paramount as controls, machines and software continue to evolve toward open-standard platforms. Most major CMM suppliers are aware that once openness is in place, a neutral, open standard will inevitably occur. CMM suppliers will no longer be selling machines, but solutions.

Buying best-of-breed

 To see how best-of-breed works, we can trace the evolution into the near future. Once the DMIS standard allows any part program to run on any CMM, customers will be able to mix and match best-of-breed components into systems that have been optimized for their needs. Customers will buy the machines, controllers, software and even probes wherever they want. Each CMM supplier will find a niche and build its business around that specialty.

 From the customer's point of view, more specialization calls for greatly increased customer service from the CMM-manufacturing community. Although five years ago the primary need was largely for a machine representative, the need will increasingly be for a consultant. To be assured of the best possible setup, the CMM buyer will have to work with someone who knows the breadth and scope of the market, the ins and outs of specific applications, the range of roles CMMs can play in manufacturing and product quality, the services available across the market, and, of course, the software.

No longer offering just machines

 Only those CMM suppliers that can provide top-of-the-line machines, services and software and are willing to allocate the resources for this kind of consultative selling will rise to the top. Five years ago, if you had asked CMM suppliers what they did for a living, they would have confidently told you that they sell machines. Today, CMM suppliers offer not only machines but also services and software. Even the CMMs are now sold less as machines and more as just one piece of a comprehensive solution for the worldwide production needs of a specific customer's product line.

 As indicated earlier, within the body of production, CMM suppliers are offering the arm of precision management. This means they're not really selling metrology; rather, they're providing a path to consistent, drift-free manufacturing. CMM suppliers are selling production capability. Measurement plays a part, but measurement and software are not going to be chief factors of competitive advantage in the future. Instead, the capabilities for technical expertise, customer consultation, deployment and after-sales support will be the major differentiators among CMM suppliers.

Needed: accurate CAD models

 The trend that is turning CMMs into production machines is demanding not only vendor-independent interoperability but also better and more consistent CAD data. Today, designs go from art to part rapidly, and they must also be able to go from art to measurement rapidly as well. The easier it is to send data directly from CAD to CMM, the more urgent the need becomes for dependable geometric information.

 Standardizing mathematically derived data is not as easy as it sounds. There is considerable geometric variability between core CAD engines. A CAD engine (or "kernel") is a collection of algorithms for expressing a given geometric shape. For example, when a user draws a circle in a CAD system, the CAD software conveys information about the radius and position of the points along the circle, mathematically defining the circle.

 For more complex shapes, such as Bezier curves, many more parameters define the shape, and the CAD engine uses a long chain of sophisticated mathematical formulas to derive it. Highly sophisticated shapes might require millions of operations that mathematically resolve approximations down to representations that are as exact as possible. As with many branches of applied mathematics, the more sophisticated the need, the greater the likelihood that several formulas could be employed for each geometric shape. A standard engine ensures that only one traceable and consistent set of formulas is used each time.

 A little inaccuracy in a numerically controlled program is not the end of the world, because close measurement will reveal the discrepancy and the numerically controlled program can be touched up quickly. A little inaccuracy in a CMM machine, however, is a particularly nasty problem. After all, this is the unit on which we rely to find part discrepancies. The CMM can't do that if the geometry it's employing is flawed. There's no check and no balance.

 Where do these geometric shifts occur? Think of a major automotive manufacturer handing a CAD file to a first-tier supplier. That supplier may or may not have a standard CAD engine. And it may transmit the data to a second-tier supplier, which again might be working with a home-grown CAD engine. If the CAD algorithms don't follow a standard, there is no guarantee that consistently accurate information will be available all along the supply chain once the part designs have moved into production. If, on the other hand, each tier works with a known engine, a known standard, then there is a greater likelihood of successful transition from one site to the next.

Moving to standards

 Fortunately, the CAD world is moving to proven engines. There seems to be a polarization occurring with Unigraphics' Parasolid engine and Spatial Technology's ACIS. For example, LK Metrology Systems Inc. has opted for ACIS, knowing that it can use that kernel for consistent and accurate results.

 Although that's fine for one company's set of machines, there remains the larger problem of geometric accuracy from CAD to CAD and CMM to CMM. What is required (and required soon) is an error-free way to validate all CAD data by a standardized test technology, a test algorithm.

 Prismatic features, including holes, diameters, radii, bosses and part envelopes, are reasonably safe from unwanted variability. Once an accurate datum is described, most of these simpler features are well within the capabilities of even the most problematic of one-off CAD engines.

 Unfortunately, complex surfaces, which are dependent on Bezier-splines (B-splines), nonuniform rational B-spline curves and even more sophisticated algorithms for CAD representation, are far more problematic. As the current trend toward "soft" or "sculpted" shapes has grown, the issue has become critical.

 If it were just the shell of a product depending on a sophisticated CAD representation, the problem would at least be controllable. The human eye can see minute flaws in extremely sophisticated shapes. But the trend is to pack these sculpted contours with more and more closely fitting mating assemblies, themselves designed and built within the nonprismatic world. The outer limits of products are reasonably fixed (how much larger can the largest car get?), but the number of inner assemblies continues to increase as product complexity and functionality increase. Control over the manufacturing and assembly of these spatial jigsaw puzzles is becoming more dependent on CMM capabilities. In these circumstances, unwanted geometric variability is far more than just a matter of a minor flaw in exterior beauty.

Standards are the solution

 The only way around unwanted, unsuspected geometric variability is by depending on a known, standard CAD engine. A manufacturer can know that both manufacturing and measurement are verifiable only with such an engine. We need to see this as a call to action: if manufacturers are to ensure the accuracy of every part and process, the CMM world needs a means for testing the accuracy of the CAD data it receives.

The culmination of a decade of trends

 Ten years ago, quality control personnel were likely to be found wrangling with those in manufacturing and engineering as barrels of scrap quietly built up in factory corners. Today, it's hard to find a walled-off quality control department among top manufacturers. The quality function has worked its way back directly into engineering and manufacturing, where it belongs.

 With that trend, the CMM has moved from its role as scrap-maker to one of process monitor. The more widely this function is deployed, the more insistent becomes the need for machine, part and control interoperability. Watch for this interoperability to undergo explosive advancement over the next few years, even among proprietary software. More important, watch for the continued adoption of standards, both within the CMM community and in the computer-aided design world.

About the author

 Walter Pettigrew, vice president of LK Metrology Systems Inc., has been with LK MSI for more than 13 years. He has 20 years' experience in the quality industry and has spoken nationally on multiple topics pertaining to the CMM industry. E-mail him at wpettigrew@qualitydigest.com .