How MEMS Get Educated


Does anybody not have a smartphone these days? Actually, some people don’t. I ran into a guy in an airport the other day using the old standby flip phone with what looked like a mono screen. Imagine! He’s probably better for it, not feeling the need to constantly fill every idle second fiddling with what’s really an electronic pacifier. Not for nothing were they called Crackberries.

How long has this smartphone obsession been going on and why? Scrub your mental tape back to 2005—a decade ago—and you wouldn’t have seen many people riding the bus obsessively stroking their phones. But three years later, there we were—a nation of phonebots and there are ever more of us. The why relates partially to Apple’s iPhone, which was by no means the first of a class of products we’ve come to call smartphones. Apple just cleverly morphed a niche/geek market into a mass market thanks to a prescient leveraging of a simple, robust operating system, new display technology, small capable processors and a clever little thing called a MEMS device—micro electrical mechanical system. These are basically miniature motion sensors we often think of as “solid state” gyros, but aren’t exactly that, to be perfectly accurate.

The iPhone’s signature screen-flipping capability wouldn’t be possible without MEMS technology, nor would apps that act as crude compasses, g-meters and AHRS utilities. (Emphasis here on crude; like the dancing ant, that an iPhone can do such things well is less amazing than that it can do them at all.)

Products you use routinely have MEMS devices. Your car has them for the airbags; for on-the-fly suspension changes and ride stability; motorcycles have them for traction control and ABS; game controllers have them to sense the player’s hand position and movement; cameras use them to stabilize image recording; your coffee maker probably has one. Since these things are now made in the millions, the market is competitive and prices are at the commodity level.

The impact on aviation is most visible in EFIS displays and the AHRS or ADAHRS that drive them. Inexpensive MEMS-based rotational sensors and accelerometers have ushered in a cottage industry of AHRS products that display accurate attitude on either tablets or dedicated displays with remarkable accuracy and at a price most of us wouldn’t have thought possible a decade ago. An example? Just five years ago, a single-axis MEMS automotive gyro cost about $15; now a two-axis gyro is $1.80 and prices are still falling.

If it weren’t for FAA certification requirements, I think you’d see far more affordable EFIS displays than are out there now. Then again, there are a lot of them out there, many built by Seattle-based Dynon Avionics, which formed more than a decade ago for the very purpose of leveraging this kind of emerging technology into the experimental and light sport segment. Dynon has a complete range of affordable EFIS products, including comm radios, and this quite naturally spun off their D2 product, a compact, battery-operated EFIS intended as a backup or basic reference gyro in an airplane that might have no attitude instrumentation at all, like our Cub. I toured Dynon’s factory in Seattle to shoot a video on the D2 recently and got good a look at the innards of the device.

Not that there’s not much to see. The largest component is the color display and the rest is given over to a sizable Li-ion battery that will run the D2 for up to eight hours. The actual working electronics is a package smaller than a credit cardand containing the aforementioned MEMS chips, plus GPS and a processor to make it all work. Once they’re assembled, powered up and given some initial testing, the D2s are plunged into a freezer for cold soaking and later into an oven for a heat session. That this is necessary has less to do with your D2 working at McMurdo Station than it does with what economics gives, economics also taketh away.

The inexpensive MEMS that make something like the D2 saleable for about $1100 aren’t, shall we say, aerospace-quality devices; they’re automotive-quality products, so what they sense as motion drifts and requires massaging in the software. Note that the second M in MEMS is mechanical, which means that these chips sense motion in one of several ways, but all are based on the effect motion has on tiny vibrating parts etched into a silicon substrate. Depending on the sensor type, these can be nano tuning forks, oscillating wheels or nano pendulums. For rotation sensing, Coriolis effect acts on these small parts, creating a tiny voltage that can be interpolated to measure rate of movement.

But even the best automotive-grade MEMS sensors drift and are affected by temperature changes that might otherwise be interpreted as physical rate change. One of Dynon’s design engineers, Ian Jordan, told me that even from the same manufacturer and the same lot, each MEMS outputs differently in response to temperature, hence the cold and heat soaking to more or less sample each device’s personality. After the cold soak, the devices are placed on a mechanical turntable to expose them to known inputs. Each one gets the equivalent of its own little database of temperature-related output and that’swhat the system’s software uses to smooth out and correct for drift so the sensor’s understanding of what’s up and what’s down corresponds to reality consistently.

Tablets, which may or may not always have rotational sensors, generally don’t have thislevel of sensor output conditioning. Nor do they have the processing horsepower to deal with it, since a tablet’s hardware is hardly dedicated to attitude sensing, as an EFIS is.This is noticeable, too. Some tablets don’t do well in sensing rotational movement if they’re placed in an odd attitude.Side by side with the D2, the difference in the fluidity of motion sensing is noticeable. The D2 doesn’t have airdata—it’s an AHRS not an ADAHRS—so it uses just GPS for positional aiding. The algorithms that run these little gadgets can get complex, using GPS or airdata as another truthing source to compare with what the MEMS sensors think they’re detecting. Even without the air data aiding, the D2 doesn’t get airsick easily.I wrung out the previous version, the D1, to try to invoke its version of gimbal lock, but it always seems to reorient itself. I haven’t tried any aerobatics, but the D2 is good enough to have its own dedicated G-meter page for just that purpose.

MEMS devices seem to get cheaper and more capable every year as designers find more ways to use them. The really high-end MEMS are orders of magnitude more expensive than automotive-grade products and come with their own controllers to sort out errors and drift rates. But they’re still not up to the accuracy of fiber optic and ring laser gyros, whose drift—what engineers call angular random walk—is 1000 times less than high-quality MEMS.

Then again, that a $500 iPad can do what it does with 50-cent MEMS is remarkable and that the D2 significantly improves on that in a package you can fit into a shirt pocket is more remarkable yet. Kudos to Dynon for engineering such a thing and to Toyota, Ford, GM, Chrysler and Apple for making it possible at a price you can afford.