Smarter Sensors Could Add Intelligence to Everyday Products

Electronic products ranging from mobile phones to fitness monitors may soon be smarter than ever. Thanks to sensors with onboard intelligence, those products will be able to "think" as never before, gathering data and then using it to reach conclusions - not just about the information gleaned from a single sensor - but about the world around them.

Click here for larger image.

For the electronics industry, the new breed of sensors represents a small revolution of sorts. Freescale Semiconductor, a component supplier that's just now bringing the technology to market, says that the micro-machined smart sensors could enable everyday products - such as automotive airbags or iPhones - to put their knowledge into context.

"More and more, it's not just about taking in the data from a single element, but about understanding everything around you," says Wayne Chavez, operations manager for consumer and industrial sensors at Freescale. "Whether it's altimetry, directional heading, or just the pace that you're walking, all these things are measurable. And all of them can be put into the context of your surroundings."

This "contextual" sensing represents a stark departure from the embedded computing that's gone on up to now. The reason: Most of today's sensors are dumb; they take an analog pressure or temperature reading and send it off to a microcontroller, which does the thinking.

Freescale believes there are profound advantages to be realized by changing that scenario. That's why the company has invested in the development of a two-level sensor - a MEMS sensor chip, such as an accelerometer, stacked atop an ASIC (application specific integrated circuit) incorporating a microcontroller. In this embodiment, the CMOS-based ASIC chip houses the brains, as well as an interface to an I²C communications databus. During operation, it takes the analog signal from the sensor chip above it, then using information that's available from the databus, begins to put the information into a larger context.

The advantages are not only that the "thinking" takes place on board the sensor, but that the sensor offloads computing chores from the main CPU and saves power in the bargain. The configuration is especially enticing to designers of smart phones, who now must integrate readings from GPS systems, pressure sensors, accelerometers and a half-dozen other sources.

"Today, all of this data comes into the phone and the question becomes, how do you manage it all?" Chavez asks. "You could manage it with your applications processor and be interrupted every time you receive a pressure sensor reading, magnetometer reading or gyroscope reading. Or you could put the intelligence on the sensor."

New Frontier of Applications
Freescale engineers cite the automotive airbag as a product that could benefit from contextual intelligence. Airbags, which do an exceptional job of being available when called upon, are nevertheless known to have rare inadvertent deployments - or worse, no deployment when needed. Most such situations occur when the integrity of a sensor signal is breeched, causing the controller to inappropriately disable or enable a bag.

But Freescale believes it can reduce or eliminate those situations by building self-diagnostics into the sensor. "We're reaching the point where we can check ourselves, more often and in real time, to better understand the integrity of a measurement," says Jim Grothe, marketing manager for Freescale's MEMS Automotive Business. With smart sensors, Grothe says, airbag modules would be unlikely to make wrong decisions based on a single-point, faulty sensor measurement.

Moreover, smart sensors could enable airbag systems to be more selective. By communicating with one another and understanding the nature of a particular crash, the airbag system might be more able to turn on and off the appropriate bags, thereby saving on replacement costs. "If you have a front-end crash, you might only want the front airbag to deploy," Grothe says. "If you have a rollover and there's no frontal impact, maybe you don't want that front bag to deploy."

Nowhere, however, are the advantages of smart sensors more evident than in the new breed of mobile phones. There, the proliferation of applications is growing so fast that many engineers are hard-pressed to keep up with them. Many phones now incorporate accelerometers for gesture recognition, magnetometers for heading information, GPS for location, gyroscopes for game play and even pressure sensors that serve as barometers, helping users transform their phones into mini-weather stations.

Smart sensors become especially important in such products because the applications are often interdependent. To accurately measure a magnetic field vector, for example, the phone must know its own orientation - that is, if the user is tilting it up or down, or holding it sideways.

"The three-axis accelerometer becomes critical for making other measurements," Chavez says. "No matter how you hold your handheld device, you have an orientation based on gravity, and that has to be used to auto-correct the measurement that describes your heading."

Click here for larger image.

Freescale says it's ready for such complex applications. In June, the company rolled out a MEMS-based accelerometer called the MMA9550L, which incorporates a ColdFire 32-bit processor, along with databus connectivity and power management in a 3 x 3 x 1 mm package. More recently, it has introduced a magnetometer-based sensor, along with software solutions for gesture recognition systems and pedometers, to its Xtrinsic line of sensors. Ultimately, the company's engineers say they will pack a gyroscope, pressure sensor, touch sensor and magnetic sensor in a 5 x 5 x 1 mm package, along with the accelerometer and processor.

The company's engineers believe the idea of a sensor with onboard intelligence will grow and creep into new applications, including monitors for fitness, home heath and onboard tire pressure, as well as e-book readers and tablet computers.

"There are still going to be some applications that will still use traditional sensors," Chavez says. "But for these sensors, the frontier is wide open to discovery."