Related: Automotive Production
Not so long ago, the applications of coordinate measuring machines (CMMs) and vision-based gaging systems for production-line operation was characterized by an either-or proposition. That is, either a plant would use a vision system or they would install CMMs. But now, according to Walter Pettigrew, vice president, LK, Inc. (Brighton, MI), that has changed. Each type of equipment has found its own niche, and so the proposition is one that calls for the installation of equipment that is most apt for the particular application. And it is likely that the either-or is now "both." LK, incidentally, is a CMM manufacturer and its equipment is used by a variety of OEMs as well as suppliers.
In fact, Pettigrew points out that there is a great deal of interest in metrology among the supply base. The reason: In order to meet the standards of their customers—customers who aren't in the least bit interested in or inclined to do incoming inspection—the suppliers are installing CMMs. And in Pettigrew's view, the suppliers aren't necessarily going for the low-priced bench-top CMMs. Rather, they are acquiring machines that are just like those in use by their customers. Presumably, that's a good way to make sure the bases are covered should a problem arise. What's more, Pettigrew notes that some operations—like GM Truck & Bus—actually hold CMM user group meetings, during which the customer explains its expectations to its suppliers.
According to Pettigrew, the right place for CMMs is more often in instances where there is an off-line situation, for such things as original prove out, auditing and trend-analysis. The CMMs are flexible and provide geometric and relational information, but the tradeoff is that they are, compared to dedicated, in-process vision-based gages, slow. However, he points out that the information from these in-process gages is typically less than that which is provided by a CMM, and that in the case of a design modification to the part being measured, the changeover of the dedicated gage is more-time consuming than is the case with a CMM. So the scenario, whether this is in a powertrain application or sheet metal area, the vision should be on-line and the CMM off-line—but nearby. That is, LK, like all of the major CMM manufacturers, are building equipment that can deal with ambient shop conditions. The reason: The need for quick response. If there is, say, a tool break on a machining line, having the CMM in close proximity is beneficial to getting back to producing quality parts more quickly than would be the case if, as was typical in the past, the CMM was housed in a lab or separate quality department. And although there is, Pettigrew admits, a trade-off in terms of absolute accuracy vis-à-vis a machine built for being used in an isolated environment and one built to be out in the midst of the action, he notes that the shop floor CMMs tend to be up to the tasks: "CMMs that are accurate within 2 microns are being used to measure sheet metal with a 2-mm accuracy."
As for CMMs having the wherewithal to measure it all, Pettigrew remarks, "CMMs performing 100% inspection on a head line? It's not going to happen. It would cost more than the head line."
Speaking to the issue of who is specifying and using the CMMs, Pettigrew says that nowadays it tends to be the production engineers rather than the quality engineers who held sway in the past. He explains that the production engineers have the interest in making sure that the process is running and under control, so they're the ones who are interested in the metrology equipment.
The Software Issue
A key issue related to CMMs—one that seems to be a key issue related to virtually all types of manufacturing equipment nowadays—is software. "With the software," Pettigrew says, "we are at a point where we were with the machines about eight years ago." He explains that some people would like to simply pop a CAD model on the screen of the CMM control and voila!, a measurement program would result. There are at least a couple of difficulties related to this at the present time, he notes. One is that not all CAD systems "read" the same, so their needs to be some tailoring. Second, unlike in machining applications, where, if the program generated from the CAD model isn't right, things can be "tweaked," that's not possible with an inspection program since the CAD model is the source of the tolerances. If the model isn't right, neither is the inspection program.
The biggest software change on the immediate horizon, he suggests, is a migration to ANSI/CAM-I 101-1995, DMIS 3.0. Or, in plainer English, the Dimensional Measuring Interface Standard that permits bi-directional communication of inspection data between computer systems and inspection equipment. The development of DMIS goes back to 1983, under the lead of the Consortium for Advanced Manufacturing-International (CAM-I). There was a slight hiccup in implementation in that a CMM vendor had managed to patent the previous release (DMIS 2.1). DMIS is a "neutral format." This means that the information exchange between, say, CAD systems and CMMs will be one in which all those complying will "read" the same. (This is not a hand-in-glove simple thing, it should be noted. There is a lot of give-and-take between the CAD system and the CMM controller.)
All of the major CMM vendors in the U.S. participated in the development of DMIS. All offer DMIS as well as their own native, or proprietary, software. LK, for example, offers a Windows-based program called Visual CMES. But Pettigrew says that the company is going to be making a major effort to promote a DMIS-based package. One of the benefits of using DMIS is that there are third-party software developers that can write packages that will work with DMIS, so the capabilities of the entire system—the CMM and the computer—can be enhanced. This, like a more general move toward DMIS among the vendors, should take two to three years.
The biggest bang for the invested buck in DMIS will be the ability to have platform independence, thereby allowing the implementation of a variety of CMMs. That is, presently user companies have part measurement programs created in a language like, say, CMES. These programs can represent a huge investment. So there is little—if any—inclination for those users to switch brands. DMIS can allow the selection of specific machines for the task. Pettigrew thinks this may result in the niching of CMMs, as well as a situation where the CMM software and the CMM machinery, now offered as a package, will separate.
Only time will tell.
Tips for CMM Implementation in 3D Sheet Metal Inspection
Here are some tips from the people at Brown & Sharpe Manufacturing Company (North Kingstown, RI)— whose products not only include those with the Brown & Sharpe name, but coordinate measuring machines (CMMs) and related equipment and software from DEA and Leitz, as well—for using CMMs on car body production lines:
•Equipment selection. There are basically two types from which to choose: measuring robots and high-speed measuring machines.
1. Measuring robots. Generally these are horizontal-arm CMMs. The reason: high dynamics, good accuracy and long measuring strokes. They can be installed in-line with production equipment as they are designed to perform under typical shop-floor conditions. The machines can be fitted with continuous or indexable wrists.
2. High-speed measuring machines. These are typically vertical arm machines more well suited for measuring sheet metal panels, small sub-assemblies, glass, internal plastic panels, bumpers, exhaust pipes, and other 2-1/2-D items. These machines resemble conventional CMMs, but are ruggedized for factory-floor conditions.
- Control systems. Motion, data processing, analysis of results, and integration with other equipment are all handled by the control system. This means that the control needs to be multi-tasking. As it is likely to be installed on the factory floor next to the measuring system, it needs to be protected.
- Software.There are dedicated packages that can handle sheet metal measuring routines for automatic inspection and for presenting the results. The operator enters nominal data. The system then generates the part program with the necessary positioning and probing instructions. During operation, the program employs automatic self-adaptive search routines and skip cycles in case the parts are (a) misaligned or (b) not there. Off-line part programming is recommended so that the measuring machine is not diverted from its real task. Some CAD/CAM software vendors are offering packages written in the DMIS format that can accelerate the programming. Beyond the measuring, the software used should be able to statistically evaluate the data in real time to avoid the production of out-of-tolerance parts.
- Sensors. Two-axis, servo-driven wrists that can orient the probe—be it tactile and/or non-contact—in any attitude following 3D trajectories are key. Servo wrists can be fitted with extension bars for inside car body inspection with no loss of accuracy. Software compensation routines are available to ensure probe positioning accuracy without calibration.
- Flexible fixturing. Thin-walled parts need to be held in their nominal geometry during inspection. Dedicated fixtures are a possibility. But they tend to be complicated to design, time-consuming to manufacture and maintain, and costly. An alternative is the flexible fixture. DEA has developed one that employs modular support and reference devices that are actually manipulated by the measuring robot. These fixtures are re-usable, thereby minimizing additional costs as new parts are introduced.
Better Contour Checking
Welles Manufacturing (Northvale, NJ) produces engine components—including shafts, followers, and rocker arms—for construction, marine, and stationary power applications, for both OEM and aftermarket customers. The factory manufactures approximately 150,000 parts per week; there are 2,000 part varieties. One thing all have in common: they must have a surface finish spec in the 8 to 18 microinch range. What makes matters somewhat trickier is that the surfaces checked are often curved, not flat.
Observes Ed Larre, QA manager and training director at Welles, "There aren't a whole lot of surface testing machines out there with proven ability to handle rounds and curves."
The company had been using a manual, portable surface testing machine produced by Mitutoyo, MTI Corp. (Aurora, IL) for checking parts; it did the job, but it wasn't optimized for curved-surface measuring. Because it was a manual machine, whoever did the measurement routine had to manually record the results. As Welles was preparing itself for ISO 9000 certification, and as many of its customers were, too, Larre and his colleagues reassessed what they were doing vis-à-vis measuring.
So, in the fall of 1996, the manual machine was replaced with a Mitutoyo Surftest SV514 surface tester. The system, in addition to the measuring unit, includes an IBM PC, Windows-based surface roughness software, and a printer. The stylus type measuring unit features digital filters and round surface compensation functions—it is built to handle curves. It measures in all the common ISO, DIN and ANSI modes.
Observes Larre, "Now I'm sure we can prove every surface roughness value we record on any curved surface. I couldn't say that in the past. Confidence levels on any surface roughness measurement are well within all customers' specs and ISO 9000 standards, regardless of the curvature."
And automatic printing eliminates any manual recording errors.