"Where Does CAM Stand?," Feature Article, January 2007

Feature Focus: Manufacturing

where does CAM stand?

Researchers are pushing to get more manufacturing assistance from the computer.

by Jean Thilmany, Associate Editor

You've just wrapped up your part design and you've certainly earned a break. But don't shut down your computer just yet. You may not be through.

Depending on which computer-aided design system you use, you might need to translate your design data into an open format like STEP or IGES before sending it on to the computer-aided manufacturing system. There's a good chance that your company's CAM system might not speak your CAD system's native language. Even when CAD and CAM are compatible and you get to skip this step, you're likely to get some questions from the manufacturing floor as engineers there determine how to best transfer information from CAM to the machine tools that will manufacture the part.

There's even a possibility that you've designed your part in such a way that it can't actually be manufactured—or, at least, manufactured easily or at low cost. Your CAD system hadn't alerted you to that, had it?

These are some of the issues in CAD and CAM compatibility today, according to Satyandra Gupta, a mechanical engineering professor at the University of Maryland in College Park. Gupta's university profile lists his research interests as "geometric reasoning algorithms for computer aided manufacturing; integrated product/process design decision models."

A Minnesota company has nearly automated its mold making. Software designs the mold automatically and also automatically commands milling machines.

CAD and CAM can often communicate back and forth only in an open format. Then CAD info must be interpreted by the manufacturing engineers who move information from CAM to the numerically controlled machines. Interpreting CAM information in order to program those tools requires a human operator who has to know quite a bit about machining in order to select the best cutters, tools, machining processes, and tool paths, Gupta said.

Streamlining CAD and CAM handoff has long been the dream of engineering organizations that face handoff issues every day, he added. Many manufacturers would love to put an end to the cumbersome interpretation process.

"But you just can't get the human out of the loop," Gupta said.

"What designers ideally want to get to is a system where, after finishing a design, they could press a button on the computer and fabrication could automatically begin," Gupta said. "The idea is: They should be able to go from CAD to CAM with all the decisions about starting production made automatically."

In other words, press a button on your CAD system, wait a bit, walk down to the manufacturing floor, and get your part. CAM handoff is automatic and CAM's codes can be fed automatically to numerically controlled machines in the manner most appropriate to create the prescribed part.

In his Maryland lab, Gupta and his team work out algorithms that would power CAD software toward such an end. Leaders of at least one company, meanwhile, say they're already there, but they had to write the software on their own.

Cut Out the Middle

The company, Protomold Co. Inc., ties CAD directly with CAM, to do away with requiring a human in the loop. It makes plastic injection-molded parts from customers' CAD models. Protomold, just west of Minneapolis in Maple Plain, Minn., boasts that it can immediately offer a manufacturing quote based on a CAD model submitted via its Web site, and deliver parts to your door within three days.

"That was unheard of even three years ago," Gupta said.
How can Protomold quote so quickly and deliver so fast? By doing away with the middleman—in this case, the manufacturing engineer in front of the CAM software.

The Protomold story began in the middle 1990s with Larry Lukis, the founder of LaserMaster, which made computer printers and desktop publishing systems.

"He discovered he could get printed circuit board prototypes made overnight, but that little plastic parts would take a month or two to make and cost $20,000 because of the time and the hassle to make the plastics mold," said Brad Cleveland, Protomold's chief executive officer.

To make a plastic part, you must first make a part mold and then essentially press the plastic into the mold, eventually breaking it open to reveal the completed part. Molds can be as complicated or simple as the part they'll eventually create, but they're very often expensive to make, Cleveland said.

Lukis, who fostered a love of software programming, knew what to do. According to Protomold, Lukis and more than a dozen colleagues went to work writing software that would smooth the hassle and allow for the automatic programming of manufacturing machines from a CAD file.

Lukis founded Protomold eight years ago. Customers load their CAD parts on the company's Web site and the site quickly returns a quote. No human is involved even in that process. Lukis's software generates quotes automatically. If the part can't be easily manufactured, the software gives feedback on how to best change the CAD model for manufacture—again without anyone overseeing the feedback process.

The Protomold software, all of which Lukis wrote, runs on parallel processors—several computer clusters, themselves comprising dozens of linked computers.

When a customer orders a part, the company's software designs the mold automatically based on the CAD file and also automatically generates the commands—the tool paths—for the milling machines that will create the mold components. Protomold runs dozens of milling machines that make components of aluminum. Employees assemble the components into the actual molds in a factory next door, where parts are molded.

"We can do this in a few days largely because of the software that automates most of the process." Cleveland said.

Protomold employees monitor the process to ensure that it runs smoothly, he said. The company designs dozens of molds per day, he added. The software can now make much more complex molds than when Lukis first fired it up.

"There'll always be limitations, typically on size and complexity," Cleveland said. "If it's too big, we tell that to the customer. If it's too complex, we might recommend they change little areas. So we give them back helpful advice."

In fact, a handful of engineering professors have begun to ask students to submit their designs to the company's site for a practical critique of manufacturability, Cleveland said. Feedback like that is vital to fledgling engineers, Gupta added.

Easy Overrides

So why can't more companies follow the Protomold example? Perhaps because Lukis wrote his software exactly for one company's needs—Protomold's.

In addition to being too expensive for vendors to sell, the automatic CAD-CAM handoff research now in the pipeline at Gupta's university might be too generic for the many manufacturers that tune their manufacturing lines to their own needs and appreciate CAM software that can keep up. The advanced CAM software would automatically generate tool paths using settings that might not work as standard at every engineering company. Many companies customize their CAM software to accommodate special glitches in their manufacturing equipment, Gupta said.

"Right now, we're working on generic software where you can press a button and it spits out an answer for a generic machine." he said. "But you might not do things that generic way.

Protomold employees assemble mold parts after they've been created. They also oversee the largely automated mold-creation process to ensure that it runs smoothly.

"Let's say your software says you need to clamp this part in a vise, but you don't have a vise mounted. You want to use a tool clamp," Gupta said. "You have to manually override that. Or let's say you're getting a CAM system, but you're going to set milling speeds higher than it specifies."

Again, a manual override is in order.

Often companies will call in CAM consultants to customize software, or they'll try to do it themselves, although the customization is often over the heads of all but the most advanced software users. This, of course, takes time and money.

"Right now, it's unbelievably expensive to customize those things," Gupta said.

CAM systems of the future should include easy workarounds that any company could use to customize the software. And Gupta hopes to eventually have a hand with development.

Gupta estimates that his algorithms will be worked out in five to six years. CAD vendors will then be able to incorporate them in their software, he said. The software in this scenario would alert engineers to potential manufacturing problems that a part faces. An alert would give engineers the option to redesign for easier manufacturability. This type of feedback would resemble the analysis-on-the-fly information now returned to engineers via desktop analysis software, Gupta said.

"CAD systems today do include some manufacturing expertise, like they tell you a section thickness is too large. They can point out obvious problems," Gupta said. "But as problems get more subtle, you don't get good manufacturability feedback."

Still in the Loop

Like other computer-aided engineering applications, manufacturing software is being pushed forward, although innovation and research is mainly the purview of academics like Gupta. Whether CAD and CAM vendors can make a business case for implementing them remains to be seen, as closely linking CAD and CAM could make software more expensive, Gupta said. If it can be commercialized, a streamlined CAD and CAM handoff in the mechanical engineering realm might one day come to mirror some of the technologies already available today.

For example, take laser cutting, an industrial manufacturing method that uses a laser to shape and cut materials, like carbon steel or stainless. Laser cutters take input directly from a CAD drawing to produce forms of great complexity. Although the laser cutters mainly slice through flat sheet material as well as through structural and piping materials, some types of lasers cut parts that have been preformed by casting or machining methods, Gupta said.

For a technique that marries CAD and CAM closer to home, consider rapid prototyping. When it first came on the scene some 20 years ago, the ability to print a digital design in three dimensions seemed like a technology straight from the pages of science fiction. The shape stored in your computer assembled itself from a claylike material right before your eyes. You could hold that design in your hand—often that same day.

Rapid prototyping as it stands today reflects where researchers like Gupta hope to see the CAM industry eventually evolve for mechanical engineers. To make a rapid prototype of your CAD part, you press a button and the system loads instructions into the rapid prototyping machine that automatically begins building the part one layer at a time. Three-dimensional printing, a form of rapid prototyping on a smaller scale using more compact technology, allows you to print single parts within moments. And you barely have to leave your desk. The printer can sit nearby.

Two decades after the introduction of rapid prototyping, some manufacturers now turn to the technology to make production-quality parts in small numbers. Because pieces are still made in small batches and production per piece can be relatively expensive, the process doesn't support large-scale production. It's still cheaper for manufacturers to use injection molding to make plastic parts, for instance, and thus Gupta continues to work in the CAD-CAM field.

So, while there's been talk "almost forever" about getting the human out of the CAD-CAM loop, Gupta said, it looks like the scenario might no longer be that far-fetched.