The molding influence.
Developing products without thinking about their markets is a classic way of going broke. The principle sounds self-evident, but any number of influences, from natural, human assumption to executive ego, can make people lose sight of it.
Here is a worst-case scenario. Testers, code-writers, and everybody else agree that a new program is ready to run. Programmers have chased out all the bugs. The graphics are sharp, and so is the price. The package carries more bells and whistles than two fire departments. Then throw in one hitch no one counted on. That is, nobody looks at this whiz-bang tool and says, "Hey, I can use that."
That is what beta testing of programs is for: to avoid debacles and misses. It puts early versions of a program into the hands of the engineers the product is designed to serve. They work with it and can turn up glitches in the programming.
But according to software developers, the most important lessons from these field tests are answers to marketing questions: how easy a product is to use, whether it has the right features, whether tools are presented in logical order, whether it is useful.
When Unigraphics Solutions Inc. of St. Louis was creating a new application, MoldWizard, what developers wanted to learn from the beta test program was how effectively they had incorporated the knowledge of mold design into the software. The program sought to speed and simplify the design of plastic injection molds. A lot was riding on the product, because the company plans a series of software tools called Process Wizards.
Depending on the complexity of a mold and its eventual use, the design process can require as many as 50 different steps, including tasks such as importing and cleaning up the CAD model of the part, adjusting its size for shrinkage, separating the core and cavity, generating mold bases, and adding sliders, inserts, and other standard components. Design can take weeks. Creation of a finished mold, from design through building and final testing, can require four to six months.
Tom Hecht, a senior mold designer at Phillips Plastics Corp., which was one of about 80 beta test sites worldwide for MoldWizard, became involved because he had been talking to local sales representatives about developing specialized mold design software. It was through them that he learned that Unigrapics was recruiting companies like Phillips Plastics for beta testing. The company, in Hudson, Wis., is a custom injection molder of plastic and metal.
When he received the beta version of the software, Hecht put it to work in making molds for customers.
"I used it on any file I could get to turn into a solid, and that was the majority" he said. "I did about a half-dozen tools with it."
Hecht later spent three days sharing his observations with the developers at Unigraphics' development center in Cypress, Calif.
PUTTING BETA TESTING IN FOCUS
Tech Mold Inc., a designer and builder of injection molds in Tempe, Ariz., was already in the process of developing its own mold design software when it heard about MoldWizard. According to John Teenstra, CAD/CAM systems manager, Tech Mold learned of the beta test program during a focus group session at a Unigraphics users group meeting.
He was among the mold designers who went to meet the developers in California. When he received the beta version of the program, he loaded it into the company's computer, but Teenstra's agenda differed from Hecht's.
"We allowed numerous people to look at it," Teenstra said. "We compared it with our own piece of software, which we were writing here."
Tech Mold's own program dealt with the design of mold base assemblies. A number of suppliers sell standard bases, but Teenstra pointed out that by designing custom bases a company has greater flexibility in the size and shape of its molds. Tech Mold wanted to automate this routine part of its mold design work.
"We had finished the mold base portion of our program and were planning to start on the core and cavity when we heard about MoldWizard," Teenstra said.
Beta testers communicated with the developers over the World Wide Web. Unigraphics set up a password-protected website where the companies taking part in the test downloaded the program. Later, notified by e-mail, they came back and downloaded each revision to the software. When the testers found problems with the program or wanted to request enhancements, they submitted the information over the Web.
A news group at the website let the test users communicate directly with each other. When beta testers had questions about how to use the program, they posted them in the news group and other testers would respond.
According to Mike Rebrukh, the Unigraphics Solutions marketing product manager who oversaw the development of the product, "We were trying to build in as much process knowledge as possible so the software could automate the mold design process. We thought we'd done a good job, but we couldn't be sure until the people with the actual knowledge, the real-world mold designers, started using the program."
The communication among testers and back to the developers led to a number of changes in the product.
Beta testers suggested that the developers strengthen the software's ability to create the parting surface, which is the split between the core and cavity blocks of the mold. For a complex shape, the line can travel over an intricate path following faces and edges. The user must manually extend each face and edge of the parting surface and trim it to an adjacent surface.
According to Hecht at Phillips Plastics, "In the first version, the tool was kind of simplistic and didn't always result in the best parting surface in terms of machining. I wanted them to make the results a lot more machine-friendly so we'd have less work down the road."
He suggested adding some of the additional surface modeling functionality that was already available in the Unigraphics surfacing package.
Another beta tester, Minco Tool & Mold in Dayton, Ohio, found that the initial beta version of the program couldn't handle a complex mold with 4,000 faces. According to Jim Sutton, product design manager at Minco, the company passed its observations along to the developers, who came back with a version that could do more.
Minco tested a later version by giving it a 20-megabyte file containing more than 5,000 faces. The software split the core and cavity inserts in four hours. Sutton estimated that it would have taken designers 30 hours to split that part using a manual process.
"There's no one button in MoldWizard that says, 'Create the parting surface,' "Rebrukh said. "But what used to be a very laborious process goes a lot faster."
Testers at Minco also noticed that they weren't able to adjust the tolerance level used by the program when sewing mismatched surfaces from imported CAD models. "The program would sew all the sheets together, but the tolerance was locked in," Sutton said. "We asked the developers to make tolerance user-definable so that if we got a part from a system where the gaps are larger, we could loosen up the tolerance"
The tolerance adjustment was added to a subsequent version of MoldWizard.
Minco, whose business includes product design in addition to designing and building injection molds, set aside a team to test the program and, according to Sutton, "We have used it for actual work." Minco is using the program on a limited basis at present and intends eventually to build a process around the software.
Another significant change that came from the beta test feedback involved the way the software represents mold bases. The program included a library of standard mold bases to spare users the task of modeling them. The parts in the database were parametrically defined solid models, but what the user saw when he browsed the library were 2-D section views with dimensions.
Some beta testers complained that these views didn't look like the parts they saw in their catalogs. So the developers revised the entire library, scanning in pages from the manufacturers' catalogs so that the current version of the program presents the image of a mold base on the screen as it appears in a catalog.
Teenstra of Tech Mold and his colleagues suggested adding the ability to use predefined core and cavity solids instead of creating new ones each time. Teenstra pointed out that one threaded screw cap, for instance, doesn't differ that much from any other except in its proportions, which in a CAD file can be altered by increasing or reducing dimensions.
Tech Mold has not made a final decision about adopting MoldWizard, but according to Teenstra is "leaning toward yes."
Teenstra acknowledged that the program simplifies the design process. "We have developed and refined our mold design process over the last 27 years," he said. "MoldWizard follows the same procedures. But it can do in three steps what would take 50 steps to do without automation."
Beta testing takes time, and designers have little of that to spare, but many testers reported that the biggest benefit was the ability to influence the development of the product.
Frequently, developers polled beta testers to set up priorities of changes. "We asked them to vote on which ones they felt were most important," Rebrukh said. "Then we tackled them in the order that would benefit the majority of users as soon as possible. As a result, MoldWizard is one of Unigraphics' most successful new products, with customers experiencing an average of 10 to 1 productivity in mold tooling design"
Unigraphics introduced the commercial version of MoldWizard on July 1. According to the company, most of the suggestions from the beta program have been incorporated into the code.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||beta testing to develop computer aided design systems for the moldings industry|
|Comment:||Beta testing programs considerably enhance engineering software development as it places initial versions of computer programs into the hands of the users they are designed to serve.|
|Date:||Oct 1, 1999|
|Previous Article:||CAD meets CAE.|
|Next Article:||Smooth body.|