热度 22
2012-4-7 10:59
2294 次阅读|
0 个评论
I get tired of hearing from people who have little or no idea of what it takes to bring an innovative idea or product concept to market reality—and I refer mostly to those facile, glib, smooth-talking pundits and politicians out there. Somehow, they are always alluding to some great idea—"magic bullet" and "silver bullet" seems to be the phrase in vogue—which will cut through the clutter and take us directly to a quick solution to some unpleasant problem, and with just a few small details to clean up along the way. While there are many times when a very good or brilliant idea is a real game-changer, making that idea into a workable solution is rarely easy or straightforward as they make it sound. I was reminded of the reality when I met at the recent Design West event with Raman Sharma of Energy Micro AS, about their extremely low-power EFM32 Gecko microcontroller. Besides the energy-usage data he showed me, what caught my attention was his clear exposition of the multiple tactics used to squeeze and squeeze the power/energy consumption further down. (I am sure that other vendors have used some or many of these ideas, but it was instructive to see them all called out in one clear listing by Raman.) In addition to an inherently low-power semiconductor process—that's a "given" for sure—the device uses a combination of topological and operational techniques; sometimes these are called "tricks" but I think that term is dismissive and demeaning. These included: - Low active-power consumption, of course - Minimizing time in any active mode - Very fast wake-up time - Ultra-low standby current - Autonomous peripheral (timers, ADC) operation, to avoid the need for the CPU intervention - Direct signaling between peripherals, again without the CPU - Multiple energy-usage modes - Low power I/O peripherals - ADC and associated support elements can function while CPU is sleeping - Development tools which allow the designer to observe and thus fine-tune energy use It's an old story, but often glossed over: even genuine magic-bullet ideas need fierce, relentless attention to detail, plus working though all the first-, second-, and even third-order problems, to be successful. Consider the steps between the Bell Labs germanium point-contact transistor of 1948 and eventual unbelievably high-volume, pennies-each discrete component we now have, or the first IC and how it became what it is today. Or, for a more bounded lesson, you can carefully study Jim Williams' 1976 EDN article on the high-accuracy scale he built for the MIT Nutrition Lab, "This 30-ppm proves that analog designs aren't dead yet") where he systematically identified, understood, and minimized the many subtle error sources. Have you ever been caught in the "here's the magic bullet—the rest should be no big deal" trap? Was it naively set by you? Or for you, but by others?ÿ