Newton, MA—Here's a quick test: In the next 60 seconds, write down every product category you can think of that doesn't involve electronics.
Ballpoint pens? Wrong. Some have lights in them now.
The tires on your car? Sorry. Michelin just introduced a tire with an embedded sensor that monitors tire pressure.
Materials? What do you think gives smart materials their brains?
Even the venerable Swiss Army Knife now has a clock and altimeter.
Electronics are hot, and they are occupying an increasing amount of space in engineers' minds even as they save space in some products. The reason: Electronics can often present engineers with interesting design problems, including the need for thermal management. And more and more, engineers are using CAD and finite element analysis software to solve those problems.
You don't have to look at end products like computers to find the issues that electronics present to engineers. They are often in the manufacturing equipment and systems that produce those products. Phoenix-based Cottonwood Technologies Inc. found that out while testing integrated circuits.
Cooling can be a major challenge in a test environment. IC packages with a low thermal resistance require a large amount of heat generation to achieve good temperature resolution when measuring thermal performance. Cottonwood hired another Phoenix company, Phoenix Analysis and Design Technologies (PADT) to figure out how to beat the heat.
"Our task was to cool the chip during testing," says J. Luis Rosales, PADT's senior mechanical and analytical engineer. The chip, or die, sits on a spreader, which in turn sits on a heat sink. Air cooling, it appears, would have been the natural choice—except that there wasn't enough room. "The footprint of the die was less than a square inch," Rosales says. As a result, water cooling appeared to be the best option.
The strategy involved cooling water flowing through channels in the heat sink that are about 0.01-inch wide. But how many fins should they use and how tall should they be?
To find out, Rosales and his team ran a combination of 2D thermal analyses and computational fluid dynamics (CFD) analyses of different design configurations using ANSYS software (www.ansys.com).
"The package was originally expected to reach 600C (over 1,100F)," says Rosales. "It would have burned up." 2D analyses showed him a design configuration that would get it down to 115 to 120C. Next, Rosales ran a series of 3D analyses with ANSYS' Flowtran software. The more realistic 3D runs produced a design for 121C. Then, by replacing some copper parts with aluminum, Rosales found the design that would get the level to 114C.
"Without the software, we could only have given Cottonwood a best guess about the ideal design configuration," Rosales says.
Analysis resonates with design
At Nicolet Instrument Corp. (Madison, WI), engineers had a different problem. The company manufactures infrared spectrometers used in chemical analyses and polymer manufacturing. Nicolet mounts its optical components to a heavy aluminum casting that acts as a base for the spectrometer. Engineers redesigned the base, but needed to make sure that the new casting had no resonances in the frequency range that would be transmitted adversely to the optical detector.
They knew the existing casting didn't exhibit an adverse frequency response, so they wanted to copy the frequency response of the old version to the new design.
Mechanical Engineer George Skupniewicz and his team used SolidWorks (www.solidworks.com) to model the old casting, then ran the design through COSMOS/Works to find the theoretical natural frequencies. Then, they did physical tests to check the accuracy of the analysis results. "We were within 5%, so we knew our model represented the old casting fairly accurately," says Skupniewicz.
Next, they designed the new casting in SolidWorks and ran another COSMOS analysis. "We went through several iterations until the frequency matched the original design," Skupniewicz reports.
The CAD and FEA software both showed the engineering team they could thin out the ribs and add stiffening ribs to the casting itself. When they eventually made the first casting, they were again within 5% of the original casting. "Without the software," Skupniewicz says, "we could have found problems with the resonant frequencies late in the design process and added several weeks to the design cycle." He adds that the software was so easy to use that they allowed a college student to perform the analyses.
Time was also critical for All Wet Technologies in its efforts to prove that its semiconductor wafer-handling equipment could speed up wet wafer processing. "We had to prove that the maximum stress everywhere in the part was less than the yield point, so there was no wear-out mechanism," says Arthur Keigler, CEO. He also wanted to optimize the design for different applications. The company accomplished these goals with the help of MSC.Nastran for Windows.
By using Design of Experiments technologies with a Concept Selection matrix to drive the design iterations for different applications, Keigler shortened the time required to adapt the initial design to each application. The Concept Selection matrix has products requirements and issues in raws, and different design versions in columns. Pluses and minuses are listed for each design version. Engineers create new design versions by keeping the pluses and eliminating the minuses.
"Building and testing a virtual prototype lets you understand the dynamics, instead of just running off parts until they break and wondering why they are breaking," Keigler says. The software enabled him to finish the design in a week, rather than having to build physical prototypes for each application and testing each. All Wet invested $25K in the software and saved between $25K and $30K per design adaptation by eliminating all but the final prototype, Keigler says. "The impact of the software was to make us more daring," he says.
And daring is the name of the game in breakthrough design, in electronics as well as other applications.