Product development teams face increased pressures today. They must turn products out faster than ever before with fewer people and resources. But what happens to quality and reliability under such constraints? Developers often neglect them to meet schedules and reduce costs. The end result is that products frequently fail outgoing test, customers’ specifications or standards (such as ISO 9000), or field trials. Few things are more frustrating than a completed design that keeps coming back for rework, like the “Man Who Came to Dinner” (and stayed!). Fortunately, simple procedures can make such outcomes less likely without taking a lot of time or raising costs significantly. The key is to plan for quality and reliability from the beginning, and making them part and parcel of every aspect of the development cycle.
Quality and reliability are particularly difficult to maintain when projects must continually reduce material/component costs and simplify manufacturing processes. Historically, components had enough built-in margins to allow for simple testing to meet specifications. However, recent aggressive cost reduction efforts by manufacturers have often eaten away margins and increased component variability. This makes test-to-specification (TTS) a dangerous approach in today’s world. A much better idea is a well thought out test-to-fail approach (TTF). A TTF methodology includes selecting applicable parameters and driving a random group of samples to failure (usually 20-30 from different lots) by increasing their values. A failure distribution can predict the lifespan of the component in an application or provide comparisons with previous generations to understand the effects of changes in materials, designs, or manufacturing processes. A common example in use today is thermal cycling to failure of components soldered to circuit boards. The resulting Weibull failure distribution can predict lifespan in the end user environment. Another example is increasing temperature or voltage on a capacitor to create a failure distribution. TTF takes time, but developers can minimize the effects on their schedules by planning ahead and selecting parameters and magnitudes properly.
Developing a robust final qualification test plan followed by detailed analysis is a critical last step before product launch. Unforeseen interactions can occur and must be caught at this time. Testing the final product to specification may be appropriate. However, studies show that the product should be deconstructed and evaluated thoroughly after testing. For example, solder joints may be cracked although still functional. Such analysis may reveal weaknesses that could cause problems in incoming test at customer facilities or during field trials.
Developers cannot afford to forget about quality and reliability in the rush to production. Today’s tight schedules and cost and staffing constraints make sound engineering practices even more essential than before. Consistent planning for quality and reliability in today’s difficult environment will keep overall costs down, avoid expensive and time-consuming re-designs, and keep customers satisfied (and presumably ready to order more!).
Randy Schuller is Senior Technical Staff at DfR Solutions. You can reach him at firstname.lastname@example.org.
Most engineers and management will agree that critical decisions regarding design and reliability should be based on robust analyses and data. Despite this concurrence, product teams within the technology space continue to use outmoded approaches for reliability assurance and risk mitigation. This includes empirical handbook predictions, like MIL-HDBK-217 and equivalent, arbitrary derating rules, reliability by similarity, prior experience, or even no reliability assessment until physical test.