Microelectromechanical systems market began with pressure sensors, exploded with automotive airbags and is now headed into IoT and medical applications.
Microelectromechanical systems (MEMS) seems to be everywhere these days — in our iPhones (and Android phones), our fitness trackers, our smart home devices like Amazon Echo and Nest, our automotive safety systems, our drones and our VR headsets, in pretty much every electronic product that you can name.
But I remember the days when we first started MEMS at Analog Devices.
In fact, I was working there, with my late wife Terry Core and other MEMS pioneers like Richie Payne, Roger Howe, Steve Sherman, Bob Sulouff and many others. Sometimes looking back at the beginning of a thing is the best place to start when understanding the thing itself.
Why did we conceive of the first commercial MEMS device 25 years ago, how did we build it and where are we going with MEMS?
Back in the day
My first exposure to MEMS dates back to the early 1980s when companies like Foxboro Corporation were using MEMS pressure sensors for process control.
I was working at Analog Devices as a manufacturing device engineer tasked with taking new processes and making them manufacturable.
When I learned that Richie Payne, ADI’s wafer fab operations and development manager, wanted to make an accelerometer (Analog’s first stab into the world of MEMS), I recommended that my wife, Terry, join the team. Hiring Terry was good for more than reducing our commuting costs, she had previous research and commercial experience in MEMS.
Teaming with Roger Howe, who consulted with us from Berkeley — she and Roger ended up patenting many of the wafer processing concepts that enabled that first MEMS accelerometer. They went on to patent another 20 MEMS concepts, laying a foundation for ideas that are still in use today.
When other companies were still focusing on semiconductors, why did ADI branch out into this new world of MEMS?
Richie was the guy with the vision. He won funding from DARPA to work on an accelerometer for military applications — which happened to be directly in line with what he wanted to make for automobiles. Since Richie already had a number of solid state products in automotive, he could see that sensors were getting bigger in cars, and that the airbag was the obvious target.
Bumps in the road
Automotive manufacturers had actually experimented with the safety aspects of airbags in the late 1980s-early '90s. This started with GM making 10,000 Cadillacs equipped with airbags.
The first airbag sensor was a ball-in-tube device. With a de-acceleration, the ball, held by a magnet, would release, roll down the tube and close the sensing circuit by bridging two exposed conductors. Every time a car would hit a bump, there was a chance that the ball would come off the magnet.
To reduce the number of false deploys — which could happen with the right shock from a pothole — GM installed three of the ball-in-tube sensors in each car; two of the balls had to come off in order to indicate a crash. Needless to say this “voting” approach was not flawless.
And that wasn’t the only weakness of this system. There was no ability for “self-test” — a feature that is standard today. Because the ball-in-tube existed as an open circuit, it was impossible to test the full sensor system prior to a crash.
While this early experiment in airbag safety did demonstrate improved survivability in real-world crashes, it also had an unacceptable number of false airbag deploys. Clearly, we needed a better crash-sensing technology for airbags to become mainstream safety equipment in all autos.
Next page: Delco the early adopter