Never miss a story from SoC Integration. Subscribe for in-depth analysis and articles.
Have you ever attended the DesignCon show? I attended the recent event for the first time and was surprised by what I saw: tons of high-bandwidth coax cables, circuit boards, connectors, and other hardware—all harnessed to very expensive scopes and channel analyzers displaying perfect eye diagrams. This was definitely a show for hardware engineers.
Walking around the show floor brought on an unexpected feeling of familiarity, something I had lost to some degree many years ago on a particular day. It was in 1998 when I first realized that a big segment of the chip design process was morphing into software engineering. It was bad enough that RTL had already replaced schematics, but now I was in a kickoff meeting for a project that was about to use a precursor to SystemVerilog for creating digital simulation test scenarios. The conversation was full of software terms like “objects”, “classes”, “methods”, and other phrases that seemed heretical to “real engineering”. In fact, back then, we didn’t even use the term software “engineer”. All that would quickly change.
But now, once more, DesignCon brought me back to the world of voltages, currents, and the most reviled, but necessary element, noise. Yes, noise is necessary to each of us, because without it, we wouldn’t have jobs. Noise is the enemy; it makes electronic design hard. But if it wasn’t hard, the whole process would be completely automated by now.
For me, the highlight of the show came when I wandered into a presentation describing an in-depth simulation experiment that modeled the noise being propagated from one, “aggressor”, via stack to another, “victim”, via stack on a printed circuit board. Now this was more like it—aggressors and victims in the battle for signal integrity. The presenters used hallowed terms like “resonant cavity”, “tank circuit”, “harmonics”, and other mystical phrases from the fine art of high-frequency analog design. They explained the reasons behind counterintuitive actions like adding multiple bypass capacitors to reduce inductance (Each capacitor creates a “lead” inductance due to its connection to the circuit board. Many lead inductances in parallel result in a lower total inductance than that caused by a single capacitor). Truth be told, I didn’t really understand this stuff back in school, so I went off in the digital direction. But at least now, decades down the road, I had remembered enough to appreciate the elegance of the work the presenters had done.
Leaving that presentation, I wandered back to the Cadence booth where tools embedding our Sigrity noise and power analysis technology were being demonstrated—tools that today’s engineers can appreciate.
What I thought would be a journey of a couple hundred yards around the show floor turned out to be a journey of 35 years around my career.