This article focuses on the growing role of computer-aided engineering programs for the design of electronic packaging. While computer-aided engineering (CAD) clearly offers the potential to improve products and cut manufacturing costs and time, those who have adopted CAE are facing significant challenges in making it work. Historically, CAE has failed to deliver results fast enough to exert a major influence on design decisions. Instead, corporations have used finite-element results to validate previously determined designs. But because design changes later can cost more than correct up-front decisions, CAE simulation can offer cash benefits when it has a role in the initial stages of product design. Technical staff leaders must translate objectives into CAE work that is efficient and effective. The status of the measurements must always be available, in order to monitor progress.
While computer-aided engineering clearly offers the potential to improve products and cut manufacturing costs and time, those who have adopted CAE are facing significant challenges in making it work. Historically, CAE has failed to deliver results fast enough to exert a major influence on design decisions. Instead, corporations have used finite-element results to validate previously determined designs. But because design changes later can cost more than correct up-front decisions, CAE simulation can offer cash benefits when it has a role in the initial stages of product design.
In the past, CAE technology ran slowly, but over the past five years advances in geometric modeling, automatic meshing, visualization, and hardware performance have cut in half the time that it takes to perform finite-element analyses. Between 1991 and 1996, median time to define an initial model declined from 30 hours to 22; the median number of analyses for each model remained constant at three, but the median time required to review results and refine a model after each run shrank from 10 hours to three. Total median time for a simulation, which had been 60 hours in 1991, was down to 31 hours by 1996.
While a 50 percent reduction in turnaround time improves CAE efficiency, companies do not achieve quantum leaps in CAE effectiveness until they make radical changes in how they use the technology. Lucent Technologies made such a change when it needed to specify the manufacturing parameters for a polyethylene covered copper wire, a short segment of a power conductor for an electronic system. The design effort, led by Anthony Rizzo, a member of Lucent's technical staff, attached a stiff, braided cable to the top end of the conductor. The cable and other components beyond it applied force to the conductor, whose lower end was attached to a stiff metal support. The designers needed to learn how the buckling load of a short segment of the copper wire was affected by several variables, including length , position of the load relative to the end of the wire, and the extent of polyethylene coverage.
Nonlinear finite-element analysis simulating changes to design variables generated the pattern of the buckling load as the design variables changed. The information provided designers with the insight needed to design systems that would allow for any change in the weight of parts. The final design was successful in its first prototype. Fundamental to Lucent's strategy, finite-element simulation was part of the initial specification of design parameters. It was not simply a check on design decisions already made.
Today's parametric solids modeling software, coupled with FEA capabilities, provides an unprecedented opportunity to implement this CAE-driven design strategy. To succeed, however, designers, engineers, and manufacturers must agree on definitions for key design parameters.
This article is the second of a two-part series that examines the growing role of computer-aided engineering programs for the design of electronic packaging.
A Shortage of Experts
On average, one out of every 15 designers and engineers in electronics companies has experience with mechanical CAE simulation tools. Not even all of these experienced personnel have the requisite expertise to work effectively with the software.
This contrasts with the popular notion of designers applying CAE software as casual users to make design decisions. Casual users have never proved successful at applying this technology. While innovations over the past several years have improved the ability to reference CAD geometry to build finite-element models, engineering expertise still must translate the design challenge into the appropriate model. Interpreting results requires an equally high caliber of skill and experience. Finally, effective use of software products involves a substantial learning curve that designers can ill afford, given the demands on their time.
As the consensus of respondents to a 1996 survey by D.H. Brown Associates Inc. demonstrates, a collaboration between designers and CAB experts works best. CAB experts should participate directly in design efforts. They should establish precedents for conducting highly focused analyses, such as evaluating standard connectors, on a routine basis. By creating templates for analysis and providing guidelines for interpretation of results, they can embed their expertise into a customized environment. Then, the CAE experts can monitor analyses conducted by design engineers.
Leading electronics manufacturers report difficulty in obtaining reliable critical data, such as the material behavior of semiconductors, ceramics, and solder, or the power dissipation of components during operation. Other industries, such as aerospace and automotive, employ American Society for Testing and Materials standards to document behavior and to standardize procedures for materials testing. Most companies and suppliers in the electronics business consider materials information strategic and proprietary.
A cooperative effort to standardize common information would benefit the electronics industry as a whole. Manufacturers could establish comprehensive corporate materials databases. Suppliers could also participate in the documentation effort.
Mechanical engineers have reported difficulty in obtaining data on power dissipation during performance. Component manufacturers typically publish power-dissipation data for peak operating conditions. However, most components dissipate power at lower levels during ordinary operation. As electrical engineers typically have this information, they could share it with mechanical designers.
All industries face occasional or frequent failures of the standards for exchanging model geometry. The shortcomings of standards have discouraged the practice of using solids-based CAD geometry to create finite-element models. According to the most recent survey by D.H. Brown Associates of more than 100 companies' CAE practices, better than 50 percent of the simulations "involved geometry re-created directly from drawings. Recent developments in programming have increased the potential for referencing CAD data directly and building finite-element models from them, which cuts modeling time. Some interfaces also will update the finite-element models for CAB studies as changes are made in a product's CAD version.
Programs from major software suppliers do not support the multiphysics simulation required to solve several current design problems, such as solder reflow simulation or degradation of the electrical properties of PCBs and joints due to fracture and fatigue.
Also, they do not support hp-adaptive FEA for the most reliable evaluation of structurally and thermally induced stress concentrations. Manufacturers must weigh lost opportunities for improving design processes due to these and other technology gaps.
Based on the potential of the opportunities, manufacturers must demand development efforts as strategic partners with their software suppliers, or adopt the technology from specialty suppliers, paying particular attention to the feasibility of integrating the specialty applications with their other software.
After spending millions of dollars, some companies derive a few benefits from CAE, but see no significant effect on the bottom line. In such instances, managers typically lose patience with the technical staff, and the technologists resent business managers for their lack of understanding.
Management and technical staff must align business objectives with technical objectives. First, managers must communicate their objectives. Business and technical management must agree on measurements to gauge progress. A typical objective might be reduced time to market; possible indications are the length of design and manufacturing cycles. For quality in1provement, key measures might be the number of product failures or the mean time to failure, inventory required for repair, or the rate of customer returns.
Technical staff leaders must translate objectives into CAE work that is efficient and effective. The status of the measurements must always be available, in order to monitor progress-or to adjust strategy if improvements are not being made. There can be a downside to the free exchange of that kind of information, though. Sometimes staff may feel that others in the organization will use it toward political ends, so trust and personal integrity are fundamental to achieving the necessary level of cooperation.