This article examines the growing usefulness of computer-aided engineering (CAE) programs for the design of electronics packaging. CAE combined with expert interpretation, can help manufacturers quickly satisfy demand for increasingly small yet reliable products. Currently, three classes of software specifically support electronics packaging design: integrated computer-aided design (CAD) CAE software, general-purpose CAE software, and specialty CAE software solutions. The integrated CAE software emphasizes automatic creation and updating of finite-element-analysis (FEA) models based on CAD geometry. The effectiveness of this associativity between CAD and FEA depends on the product behavior to be evaluated and the quality of implementation. CAE simulation can calculate the maximum acceptable loads on pins, as well as the vibration characteristics of components. Solids-based CAD helps detect interference problems across components, as in these exploded views of a disc drive and a headset. While several general-purpose CAE suppliers permit engineers to build customized environments for automating model creation, specialty suppliers such as Pacific Numerix deliver the specific automated capabilities and connector libraries.
Consumers would no doubt relish wafer-thin wireless telephones, or computers with the dimensions of a business card—convenient products that can tucked away until needed. Of course, products with small dimensions must still satisfy basic ergonomic requirements, such as dialing a number, entering data, or producing high-quality sound or visual output. Current market trends suggest that the ideal consumer product would serve its function, yet not exist.
This appetite for increasingly small yet reliable products creates challenging functional electronics design requirements. The simultaneous need to be first to market fosters severe product development pressures, especially given the already accelerated pace of technology innovation in the electronics industry. Manufacturers struggle to design reliable electronics products in a limited time. A judicious application of computer-aided engineering (CAE) capabilities, combined with expert interpretation, can help manufacturers conquer this formidable marketplace.
To grasp the importance of CAE for electronics packaging, consider the current position of the electronics industry. Growing at more than 30 percent annually, the semiconductor marketplace promises to reach $335 billion by the year 2000, dominated by personal computing, wireless communications, automotive electronics, smartcards, and energy/environmental applications. This rapidly expanding market holds the promise of potential-ly huge financial rewards for companies that are able to satisfy consumer demands.
Currently, success in the electronics industry depends on reaching the market first with innovative new products that are smaller, cheaper, and more ergonomic, yet more reliable. These modern manufacturing parameters in turn lead to the creation of an increasingly complex, multidisciplinary design environment. To master the stringent requirements of smaller product design, manufacturers must break down the traditional divide between electronics design and mechanical design. And, as the price of consumer electronics continues to plummet, companies face severe pressure to minimize costs while maintaining margins.
Product design now determines 75 percent-of manufacturing costs. Roughly two-thirds of design costs derive from decisions made early in the design process. Additionally, companies pay dearly for product and preproduction prototype failures. According to data collected from a broad range of manufacturers over the past 15 years, design errors trapped early in product development clearly cost less and take less time to fix. The pie chart on page 66 shows the relative cost of fixing failures during the various phases of the product life cycle.
The cost of product fixes increases more than threefold during the preproduction stage (after the first prototype), compared with fixes before the first prototype. This ratio increases to more than 7:1 for errors that are detected during production.
The relative cost of fixing a product released to the field can be catastrophic— more than 20:1 when compared with pre-prototype fixes. Such product recalls or replacements can also exact a heavy price in customer loyalty. In addition, consider the negative impact of time lost making fixes—notably, the adverse effect on potential sales.
Companies, therefore, benefit tremendously from errors caught at the prototype or preproduction stages. Early error detection means faster resolution at a lower cost and a greater chance of product success. Therefore, increasing numbers of manufacturers now turn to solids- based computer-aided design (CAD) and CAE to address escalating electronic packaging challenges.
The Role of CAE
Although CAE alone delivers lower-fidelity results than prototyping, when used by experts it enables the most cost-effective assessment of many design alternatives. CAE provides enough fidelity to understand the impact of design choices, and so contributes to the deeper understanding of product behavior and problem detection before the commitment of production resources. Expertise retains a significant role in CAE modeling, because ability and experience are required to capture the functional behavior of components with abstract models. Engineers must then interpret results based upon the assumptions of the models.
Currently, three classes of software specifically support electronics packaging design: integrated computer-aided design (CAD) CAE software, general-purpose CAE software, and specialty CAE software solutions. Corporations across all industries have adopted integrated CAD-CAE software as the linchpin of their strategies to use solids-based technology to drive design efforts. It provides capabilities for the most general assessment of a product’s engineering behavior and properties, detecting the most fundamental problems.
This class of software excels in its ability to quickly calculate the physical properties of components, assemblies, and products, such as weight, moment of inertia, and center of gravity. These measurements can be critical to the understanding of both ergonomics and performance. For example, the weight and center of gravity can partially reflect the ergonomics and utility of wireless telecommunications equipment, while the knowledge of inertial properties may be important for quick predictions on the performance of military electronics in the field. In addition, CAD software enables the detection of gross component interference problems that prohibit proper component assembly.
The integrated CAE software emphasizes automatic creation and updating of finite-element-analysis (FEA) models based on CAD geometry. The effectiveness of this associativity between CAD and FEA depends on the product behavior to be evaluated and the quality of implementation. Leading suppliers include Dassault Systèmes (sold by IBM), Parametric Technology Corp., Unigraphics Solutions Inc., and SDRC.
General-purpose CAE software suppliers focus on using the breadth, depth, and reliability of their finite-element simulation capabilities to address a broad range of engineering problems, with options for automating and customizing the user environment. Supporting analysis capabilities include structural static with geometric nonlinearities, plasticity, creep, fracture, and fatigue; vibration analysis; linear and nonlinear transient dynamic analysis; heat transfer analysis with conduction, convection, and radiation for steady state and transient conditions; computational fluid dynamics; and electromagnetic analysis.
The MacNeal-Schwendler Corp. of Los Angeles is the largest and best-known general-purpose CAE provider. Other leading vendors include Ansys Inc., Fluent Inc., Hib-bit-Karlsson, Sorensen, and MARC Analysis Research Corp.
Mechanical engineers working in the electronics industries report that once they have mastered a given general-purpose CAE program, it becomes very difficult to change products. The competitive pressures on the product development front simply do not accommodate the typical learning curve for this genre of software. Adopters should, therefore, carefully consider their selection of a general-purpose CAE supplier. Potential buyers should evaluate staying power, reputation, breadth and depth of finite-element- analysis capabilities, and quality of support.
Specialty CAE solutions offer specific expertise to address a limited set of highly focused, key problems with the highest reliability available today. Electronic Design Validation System (EDVS) from Pacific Numerix is the most notable example of specialty CAE electronics packaging software. This software offers the most fluid environment for building and evaluating electronics-specific simulation models. Arguably, general-purpose software can address some of the problems that EDVS targets. However, EDVS automates much of the work that must be performed manually or programmed with general-purpose CAE software. No other software matches EDVS for other types of simulation, such as Printed Circuit Board (PCB) manufacturing, including the solder reflow process and the optimization of component placement, considering thermal and vibration requirements.
Other small specialty suppliers, not focused on electronics-based applications, offer specific capabilities unmatched by general-purpose suppliers. For example, Centric Engineering offers superior expertise for coupled fluid-thermal analysis; this is useful for component cooling studies. Engineering Software Research and Development (ESRD) offers expertise in evaluation of stress concentrations, important for evaluating connectors and mechanical failure at multimaterial interfaces such as those encountered in multichip modules.
Mechanical Design Challenges
Shrinking functional and spatial requirements often force designers to introduce ever-larger numbers of components into increasingly small spaces. The positioning and grouping of components must still satisfy operational needs, that is, functional dependencies of neighboring components and effective heat removal, while minimizing wasted space. Here, designers and engineers have adopted solids-based integrated CAD-CAE software to evaluate spatial considerations, weight, and other engineering properties. They apply electrical and mechanical CAE simulation software to predict functional performance of the overall system. Specialty software can optimize component placement.
Consumer and military electronics can encounter a variety of harsh environmental conditions, including shock, vibration, excessive heat and humidity, and harsh chemical environments. Designers must ensure that mechanical packaging protects against these conditions. General-purpose and specialty CAE software enables engineers to evaluate a host of mechanical and thermal conditions.
As all electronics components generate heat by dissipating energy, effective thermal management is fundamentally important. Most semiconductor devices are not rated for junction temperatures above 175°C, and their performance degrades rapidly with increased temperatures. Mechanical designers, therefore, must take heat removal into account. Certain operating environments—for example, under the hood of an automobile—pose particular risks to the integrity and reliability of electronics. Here again, general-purpose and specialty CAE software support a variety of thermal analysis capabilities, such as conduction, convection, and radiation, for a broad range of materials.
Manufacturing costs usually exceed design costs by ratios of more than 6:1. Design innovations that translate to shop floor assembly savings become critical in the extremely competitive electronics market, because they appear quickly on the bottom line. Solids-based CAD and a specialty electronics manufacturing simulation allow for realistic evaluation of shop floor assembly and PCB manufacturing, providing insight to possible improvements. To maintain customer loyalty, products must allow for easy removal and replacement of failed components in the field. Solids-based CAD can effectively evaluate serviceability. In addition, virtual reality software has been demonstrated as a potential method for field service training.
Most mechanical failures occur at mechanical connections between components, such as soldered joints and pins. General-purpose CAE simulation software and specialty CAE software provide tools for predicting probable locations of connector failures. These predictions help designers take proactive steps to guard against failure.
All electronics components have minimum power requirements for effective operation. The challenge of reliably predicting satisfactory power distribution crosses both the electrical and mechanical disciplines. Most of the materials used in electronics design, such as ceramics, plastics, semiconductor materials, and solder, exhibit complex temperature-sensitive behaviors. Creep, fatigue, and other related behaviors degrade mechanical and electrical material capabilities over time. Creep causes connectors to lose up to 50 percent of their contact force under high temperatures, and up to 20 percent under extremely low temperatures. Microcracks and other forms of mechanical degradation influence characteristics such as electrical resistance, which can further impact mechanical behaviors (including creep) during product operation. Although partially degraded materials can still support product operation, component life becomes difficult to predict. To date, no well-known commercial CAE software effectively addresses this challenge.
A few electronics-packaging requirements relate primarily to electrical design, but can have mechanical implications. Design for testability can affect the spatial layout of components and, therefore, the minimum packaging volume. Shielding from external magnetic radiation and interference, and prevention of harmful electromagnetic radiation leakage are other concerns. Selection and arrangement of materials and components can influence the minimum packaging volume as well as the heat transfer characteristics. These requirements cannot be addressed in an isolated fashion. For example, a design proposal that optimizes shop floor assembly might seriously compromise serviceability or other organizational requirements.
Roughly two-thirds of product design costs derive from decisions made early in the design process.
Common Mechanical Failures
Most electronics products fail due to mechanical, rather than electrical, problems. These mechanical failures typically manifest themselves as an electrical short, an electrical open or incomplete circuit, or a circuit that works intermittently.
Electrical shorts occur when two or more circuits that should be electrically insulated from each other create a conducting path. The most common cause of catastrophic failures, electrical shorts, sometimes spark fires, because of the excessive heat from the high currents they draw.
General-purpose and specialty mechanical CAE software are good at detecting shorts that result from mechanical stress. However, today’s software still requires excellent engineering judgment to reliably predict the potential sources of such shorts. No commercial software currently available completely models all the conditions that affect physical behavior. CAE contributes partial solutions for electrical solutions that provide insight when en the uncertainties and insufficient data on actual material properties.
Intermittent failures frequently appear while the product is operating, but disappear when testing off-line. Bent pins and connectors often cause mechanical intermittent failures. Under vibration and shock conditions, the low normal contact forces arising from a deformed shape can cause intermittents. For example, pins on a pin grid array are highly susceptible to damage during handling and shipping. To minimize the chance of damaging pins during shipment, manufacturers must pay special attention to packaging. In addition, the vibration characteristics of components with moving elements frequently influence mechanical intermittent failure.
CAE simulation can calculate the maximum acceptable loads on pins, as well as the vibration characteristics of components. However, the calculation of mechanical effects on electrical behavior, such as contact resistance, remains difficult if not impossible today.
Avoiding Design Mistakes With CAE
Designers seek to minimize space between components in order to satisfy volume constraints for electronics packages. Smaller volume, in turn, affects thermal management, manufacturing, serviceability, and reliable operation of the components. Solids-based CAD enables “first-cut” detection of interference problems across components based on nominal dimensions. Through kinematic analysis and specialty robotic-simulation software, manufacturers can assess manufacturability and serviceability. The figure on page 71 depicts solid models of a disc drive and a headset. The exploded views reflect the challenge of assembly within volume constraints. By assigning material behavior to each component, users can quickly estimate physical properties with reasonable accuracy.
Component positioning and orientation also influence the quality of soldered connections during the solder reflow process. Connector problems, such as poor solder adhesion, cold joints, and shorts, may occur. Effective simulation of the reflow process requires modeling a complex set of physical effects, including heat transfer, structural characteristics, and the flow characteristics of solder. Commercially available specialty CAE software such as the EDVS SolderSIM module from Pacific Numerix captures this complex behavior.
LG Electronics of South Korea adopted Pacific Numerix’s soldering process simulation software to address difficulties arising from higher-density printed circuit boards. When LG Electronics first automated its soldering process with wave solder machines, the component placement and board temperatures affected the quality of soldered connections between components and the board. Engineers encountered a host of solder reflow problems, including DIP (Dual Inline Package) shorts, occurring when globs of solder ran into each other; poor adhesion between solder and pins due to thermal problems; component rotation, which furthers soldering problems; and small holes in the soldered joints. By applying SolderSIM to reposition components, LG Electronics reduced the number of soldering process defects on its electronics boards by 73 percent.
Electronics products must be resistant to damage from accidental drops and harmonic excitation. Mechanical impact fractures components and connectors. Over time, cyclic loads, such as harmonic vibration, cause failure from fatigue. More catastrophic cyclic damage occurs if the excitation frequency matches one of the resonant frequencies of the product.
Currently, manufacturers often successfully use finite- element-analysis (FEA) capabilities to capture the gross characteristics of system behavior from CAD and specialty CAE suppliers, and to make reasonable estimates of lower natural frequency modes. A more accurate evaluation of higher vibration modes, subsystem behavior, and impact analysis would require more detailed finite-element models. To date, general-purpose and specialty CAE software offer the most efficient and reliable means of performing this demanding work.
For example, the German television manufacturer Grundig adopted MacNeal-Schwendler’s MSC/DropTest to improve product reliability and competitiveness. Every year, customers return tens of thousands of televisions because of transportation damage. These returns cost television manufacturers millions of dollars in unnecessary expenses, as well as inflicting damage on their public images. Therefore, television manufacturers must design the casings’ packaging to minimize damage during shipment.
However, changing the design directly impacts manufacturing costs. Grundig reports that dies for the front and back of the television casing cost approximately $300,000 apiece. The average cost to make a single change to the die typically runs 10 percent of the original manufacturing cost, or $30,000. Grundig would save significant time and money if casings were designed correctly before reaching the prototype phase—slashing costs on testing iterations and the resulting die changes. The MSC/DropTest simulation provided Grundig with the insight the company needed to improve its design.
Inadequate cooling and thermal management occur when engineers fail to understand the ambient thermal operating environment, miscalculate heat generated by components, or underdesign a product for heat removal. Material aging accelerates when materials are exposed to higher-than-expected thermally induced loads, resulting in fatigue and creep in plastic components and solder joints, stress relaxation of contact and housing material, and diffusion of base materials. Worse yet, junction temperatures exceeding 175°C can cause the total failure of semiconductors.
Inherently complex, the effective simulation of thermal cooling involves reliably modeling heat generation, convection, conduction, radiation, and fluid flow. Most general-purpose FEA software packages enable the modeling of heat generation and conduction reasonably well. While engineers can usually identify the locations of hot spots reliably, predictions of the actual temperatures often remain challenging.
If engineers have accurate heat transfer coefficients, CAD-integrated and general-purpose FEA software can create effective convection analysis. However, these coefficients—which depend on airflow, heat transfer, and geometry characteristics—can be complex to determine, and may require supporting fluid-dynamics computations. Specialty CAE software proves most effective for evaluating complex forced and natural convection simulation.
Because radiation is modeled as nonlinear behavior, it requires a higher level of expertise with the software. In addition, establishing reliable calculations for all required shape factors can be a daunting process. However, engineers can operate effectively with general-purpose or specialty FEA software, if they understand the specific idiosyncrasies of performing nonlinear analysis with their selected product.
Thermal expansion mismatch represents another major source of failure. This occurs when two bonded materials expand at different rates because of temperature changes. These differing expansion rates generate high stresses at the interface between the materials. Frequently, failures due to fatigue and thermal cracking result. In most cases, mismatches typically occur at interconnection points such as soldered joints and wire bonds.
Any quality thermal-mechanical CAE capability enables the quick evaluation of potential problem spots.
However, the thoroughness of the evaluation depends on the detail added to the finite- element model. General-purpose and specialty CAE software offer the most efficient software algorithms, superior model management, and comprehensive material modeling capabilities.
Undetected stress concentrations and inadequate stress relief cause connector failures by permitting highly localized loads that exceed the limits of connector materials. For example, abrupt changes in cross-section transitions can create localized stress concentrations, resulting in fracture. Connectors subjected to bending fracture most frequently. Without stress relief, crimped and soldered wires subject to tension will often fail.
Although standard design practice should avoid such errors, designers can easily overlook some requirements for the multiple connectors present in an electronic product. FEA can detect potential problems. However, engineers must carefully evaluate the magnitude of highly localized stress concentrations. A significant concern is that the dimensions of elements will match characteristic dimensions of the connector material anomalies at the microscopic level, invalidating commonly used stress-based material failure criteria. Therefore, failure prediction should be based on criteria less sensitive to mesh characteristics, such as strain energy or generalized stress intensity factors.
To adequately detect trouble spots and stress relief requirements, manufacturers should establish a systematic approach to uncovering trouble spots. One solution would be to deploy a library of parametric finite-element models representing the set of commonly used connectors. Engineers could then select components from the library, saving model creation time. HP-adaptive finite-element analysis from specialty suppliers such as ESRD would most effectively analyze the joints— given the characteristics of the meshes and highly localized stress concentrations.
Overly flexible PCBs or printed circuit carriers may bend under the weight of their components, producing high stresses at soldered joints that, in turn, create cracks. Excessive bending and torsion of the boards may also occur during manufacturing. And, as the density of components increases on new products, the increased density of pins on a pin grid array causes each pin to become more fragile. Assembly or maintenance, therefore, commonly bends pins and contacts. Linear and nonlinear structural finite-element software available from general-purpose and specialty suppliers provides powerful capabilities for evaluating excessive bending and torsion. Ideally, components and features can be applied from solid models to calculate physical properties and to serve as reference geometry for creating simulation models.
By generating stress concentrations at connection locations, overly stiff connectors can accelerate fatigue failure under dynamic conditions. To evaluate structural compliance, engineers can first use random vibration analysis on boards or subsystems, determining potential problem domains. Once they have identified the domains, engineers are then able to conduct local detailed analysis of connectors.
Currently, the leading commercial suppliers of general-purpose finite-element software support structural compliance evaluation. However, it remains challenging and time-consuming to create and update models that match both the component and lead attachment patterns, as well as the irregular edges of boards, holes, and cutouts. Specialty software offers automated capabilities to generate the finite-element models of boards based on component types and their locations. Access to libraries of finite- element models for commonly used connectors further increases productivity. While several general-purpose CAE suppliers permit engineers to build customized environments for automating model creation, specialty suppliers such as Pacific Numerix deliver the specific automated capabilities and connector libraries.
Component damage can result from selecting a PCB or printed wiring board size so that one of its mechanical resonant frequencies nearly matches one of the electrically induced frequencies to which it will be subjected. Although PCBs are normally large enough to avoid this problem, engineers can check for potential design failures through finite-element evaluation with any reputable finite-element software.