posted by Bryon Moyer
TI caught my eye when they released a microcontroller that they said was “haptics-enabled.” A few seconds of thought convinced me that this concept needed some unpacking.
Haptics is all about devices providing feedback through some kind of touch mechanism. It could be as passive as raised bumps telling you that your fingers are in the right place, or it could be through vibrations or other active events that you can feel. It’s a hot topic, one we’ll probably be seeing much more of.
But… TI’s new MSP430TCH5E microcontroller is… a microcontroller. How can that generate haptic feedback? Does it have a specific hardware module for driving a specific vibratory engine? Seems unlikely, since haptics has lots of ways of being implemented; there’s no “mainstream” mechanism that’s suitable for hardening. Is there?
The release does talk of software libraries and SDKs. Could this be just about software? But… if so, why is it unique to this microcontroller?
I checked in with them, and the details of whatever the answer is are confidential; they’re not saying. But it does have to do with protecting IP. So my take on it is that this is a microcontroller/software bundle that includes haptic libraries. And you can’t use those libraries on other microcontrollers. Why not? Not sure… it could be the license: to get this you most likely have to promise to play by their rules. And if the solution is worth it, most upstanding businesses are not willing to risk legal hassles by playing games trying to port to another processor.
But it may also be that there’s some kind of hardware lock – something specifically put in place that the libraries interrogate to ensure that they’re running on a designated platform. Since, as far as I know, this specific microcontroller isn’t available without the haptics library, that may be the case. (It would be an easy design strategy to have a basic platform that simply has an ID that can be changed with one mask to make the device “unique.”)
I don’t know if this is what they did, but it would certainly be doable, and would add some practical teeth to the license. And if the low-level code is in machine language, it would be really hard to hack.
You can read more about what you can do with this in their announcement. And if you have any other clues about what's going on, please post in the comments.
posted by Bryon Moyer
Last month STMicroelectronics announced a new accelerometer “engineered to withstand stresses of modern mobile life.” They see those stresses arising from increasingly thinner phones and the mechanical and thermal challenges they cause. They called out board bending as a particular challenge to the mechanical integrity of the works inside the accelerometer package.
So how do you improve the mechanical structure of the accelerometer to do this? First, it helps to realize that there are two structures in ST’s accelerometers. One operates in-plane and provides both x and y acceleration information. A separate structure is used for the out-of-plane z axis acceleration. On older models, these two structures were set side by side.
To illustrate how things might be improved, they made reference to stability in an airplane, even though the comparison can’t be taken too literally. If you want the smoothest ride in the plane, you sit in the middle, between the wings. Especially to the extent that the middle has the least stress and that stresses radiate out from that, there’s more disturbance (bumpiness) at the extremes – the wingtips and nose and tail – than in the middle.
It turns out that the z-axis accelerometer is the most sensitive, so improving it was a goal. So they moved it to the middle of the die layout rather than having it off on one side. And where would the x/y structure go if the z structure is hogging the middle? Symmetry is achieved by splitting the x/y structure and putting one half on either side of the center z structure. The two halves become the “wings.”
The other improvement was to double the number of anchoring points on the z structure from 2 to 4. This reduced the stresses on those points, making them less subject to failure.
You can find more details on the performance of this acceleromete
posted by Bryon Moyer
The sensor market is highly fragmented. Many sensor companies are good at one or a few sensors based on super-secret ways they have of building them. When combining with other sensors for things like IMUs, they may actually bring in partners for sensors they don’t make themselves, and they may use sensor fusion software from yet another company.
Bosch Sensortec has been taking a different tack, however. As one of the two really big sensor guys, it’s taking the “you don’t need anyone else” approach, doing a wide variety of its own sensors and writing its own fusion code. And, for those that don’t want to customize and optimize the code, they then combine them into complete units that abstract the lower-level stuff away. They refer to these as application-specific sensor nodes (ASSNs).
The first of these was their orientation sensor. And they’ve now announced another one: an environmental sensor, the BME280. It combines pressure, humidity, and temperature sensors. Yes, it pretty much looks like a weather station in 2.5x2.5 mm2 package. Although they say that it has application for fitness as well. And, in fact, they’re touting the pressure sensor as providing an altitude response for indoor navigation (even thought that would seem to be a better fit for the orientation sensor). Its accuracy for that is ±1 m, enough to discriminate floors in a building.
One thing they’re particularly proud of is the humidity sensor. While typical versions take 5-10 seconds to register a value, they can tap the humidity in less than 1 second. They do this using a polymer that absorbs water; when it does, it changes its dielectric constant. Apparently the variables in this physical structure – the choice of polymer, mode of access, layer stack, and thickness – all affect the diffusion time both in and out. (Yeah, after you’re done measuring the humidity right now, you need that water to leave if the outside humidity drops.)
Because it’s in a ported package (the outside air has to get inside to be sampled), they also see this as lending itself well to future gas sensors. So there may be more to come here.
You can read more about the BME280 in their announcement.