The headline is so common: "Design Engineers Face Increased Time-To-Market Pressures." Whenever I see such a headline, I smile. Is there ever enough time, budget, or staff to develop a new design and meet all of the project's expectations? Designers are constantly looking for ways to accelerate the product development process. But today's designer faces new challenges due to the increasing convergence of computing, communications, and control.
After talking with engineers and managers in the communications, automotive, and other industries, I found that there's a common need to reduce the bottlenecks in the design cycle. It's no surprise that one of the biggest anxieties is that increasingly unfamiliar technologies are converging to make products more complex. What's interesting is that the biggest bottleneck is no longer in the design simulation phase.
With the power of modern computing environments and sophisticated EDA software, today's multifunction products can be designed and simulated reasonably well. Increasing product complexity has moved the biggest "design" challenges to the subsequent parts of the design cycle: system integration and design validation. For example, consider the automotive electronics engineer. A simple AM/FM/cassette car radio from a decade ago has now become a full automotive telematics subsystem that combines audio, video, navigation, diagnostics interfaces, and wireless communications functions in a single device. The designer integrating these diverse technologies into a functioning prototype faces an extremely demanding task. The engineer has to assemble a bewildering array of audio, video, and data sources and acquisition tools to adequately validate system performance.
Here's another surprise: Although software simulation methodologies have evolved to take advantage of increased computing power, much of the validation phase is still done manually. At a recent meeting with a mixed-signal IC group at one of the world's leading semiconductor companies, the design team took me through its design and characterization facilities. The team was proud of its million-dollar EDA software toolset and the five equally expensive validation stations used to characterize its complex mixed-signal products. All of the systems had world-class computing environments and the latest network connections. I was obviously impressed.
When asked how the performance of the team's latest chip compared with its expectations from simulation, one designer presented paper printouts of the simulation runs and test results. This was a bit puzzling. Here were two multimillion dollar systems for simulation and characterization, producing gigabytes of data, yet the performance of a complex communications IC was being assessed by looking at two pieces of paper. Even with the latest network connections, the two toolsets didn't communicate. How much time and error could have been saved if the simulation and test data were simply in compatible formats and toolsets were integrated at a software level? Visual inspection could easily have missed a small crosstalk problem, causing the chip to have 1 dB more noise than expected—a critical loss for a wireless communications application.
Cooperation between toolsets is common in the EDA world but missing when we enter prototyping and characterization. At a recent conference, Stanford's dean of engineering, James Plummer, made an interesting forecast that the revolutions of the future would come from working on the interstitials, the spaces "in between" traditional disciplines. Applying this view to the design engineer, further streamlining of the design process requires us not only to work on better tools but also on connecting the simulation and validation halves of the design cycle. The exploding complexity facing engineers is a call to action to the tool community to define a common set of open interface standards that can provide cooperation across today's simulation and test tools. It probably is also a call to action to make sure we educate tomorrow's engineers with hands-on experiential learning that combines multiple engineering disciplines.