Why Thermal MEMS?
The reasons might resonate with you.
“Capacitive Versus Thermal MEMS for High-Vibration Applications” is the title of a recently issued white paper from sensor maker MEMSIC, Inc. Its author, James Fennelly, business development manager — automotive and industrial sensing, at the company, spoke with EECatalog recently.
EECatalog: Why did you write this white paper [Capacitive Versus Thermal MEMS for High-Vibration Applications]?
James Fennelly, MEMSIC, Inc.: To address what I saw as a perception in the market that all the MEMS were capacitive-based and that they all had very similar characteristics. I found myself in numerous customer engagements explaining that [thermal MEMS] is a different technology. With the white paper I wanted to reach the broader community of engineers working with accelerometers and looking for ways to solve problems.
EECatalog: In the white paper you note a number of design variables to consider. What are your recommendations for how to weigh one against the other?
Fennelly, MEMSIC, Inc.: I think the quickest way to get to a decision as to which technology to choose as the lead technology is to think about three things in the application. First, assess the vibration and mechanical shock environment by making some wideband measurements using an instrument-grade accelerometer that has wide bandwidth, so you can make some high-speed measurements. Perform an FFT [Fast Fourier Transform] on those measurements to see the spectral content of the vibration. [Doing so] can quickly determine the answer to the question, “Is a capacitive accelerometer going to work in that environment, or will I have to take some big steps to try to make it work in that environment?”
Second, determine what bandwidth is needed in the application, meaning, what responsiveness is needed by the accelerometer for the application itself? If the acceleration to be measured is less than 50Hz, then thermal MEMS is likely going to be the best choice, or at least it is not going to be disqualified as the best choice.
On the other hand, if the bandwidth of the application needs to be greater than 50Hz, you are going to have to try deal with the vibration environment using capacitive-based MEMS. As a rule of thumb, the low pass filter response of our thermal MEMS device is going to make it harder to respond to accelerations that are greater than around 50Hz.
Third, consider the power budget that is available for the application. Although power would not be the gating decision for a device that can be recharged, a thermal MEMS device is always going to require more power than a capacitive-based accelerometer. If, for example, the design is intended to run off a coin battery, last 10 years, and be used to monitor something where it is not accessible, thermal MEMS is probably not the best choice. An example would be putting an accelerometer into a piece of equipment to monitor a residential gas line for gas flow and/or for an event that could cause a gas leak.
So if the bandwidth is less than 50Hz and you are not power-budget constrained, thermal MEMS is a very good choice to look at as a lead candidate. In addition, if there is a lot of vibration or mechanical shock in the application, then thermal MEMS is likely the best choice to pick as your lead design consideration. I found that in many disciplined engineering environments engineers will often look at multiple types of sensors, do some early prototyping work with two or three and then cull to a lead candidate.
EECatalog: Can you cite a situation where these considerations came into play?
Fennelly, MEMSIC, Inc.: Yes. Digital SLR cameras need orientation detection for re-orienting the screen. You wouldn’t think digital SLR camera use is a high vibration environment or a particularly harsh environment, and our customer had gone ahead and designed in a capacitive-based accelerometer. What they found, when they introduced the camera to the market, is that the lens action, the shutter that opens and closes with a “click click” was causing problems for that capacitive-based accelerometer. It was sending it temporarily into resonance and screwing up the orientation because it was like an impulse event—one fairly large in magnitude but very, very short in duration. If you look at the spectral content of that by doing an FFT, such an event actually has theoretically infinite frequency content. Some of that content turned out to be at the resonant frequency of the accelerometer in this SLR camera application, so we ended up winning back that design because our part doesn’t have that problem.
EECatalog: What are three pieces of information the reader of the white paper should take away, even if that reader takes away nothing else?
Fennelly, MEMSIC, Inc.: One, thermal MEMS have a distinct advantage in applications where vibration and mechanical shock are present because of the transducer’s inherent low pass frequency response. That is the key takeaway.
Second, thermal MEMS are ideally suited for lower g applications—less than 10g’s—because they have no mechanical resonance.
Third, thermal MEMS are extremely repeatable and reliable because they take a “monolithic design and no moving parts” approach to sensing acceleration.
Conversely, if your application needs to measure high g or high vibration, then this isn’t the right technology. If what you are trying to measure is the actual vibration, and that vibration has higher than 50Hz frequency content, then this is not the right technology.
For example, one of the first major adoptions of accelerometers in the marketplace, the first really big volume driver, was crash sensing [for automotive], and it is still a huge volume driver for the market.
Satellite sensors sit on the periphery of the car, and they detect a crash. There is also a safing sensor, which is more centrally located, that confirms the crash happened. Algorithms determine when to deploy the airbag, That application is high g, high bandwidth, and that is where capacitive accelerometers excel—that is the perfect type of application for them, because they can have stiff mechanical structures, and they have wide bandwidth and they need the bandwidth in order to do the calculations to deploy the airbag at the right time.
Conversely, in that same automobile, there are accelerometers that are used more centrally for things like electronic stability control and for rollover detection. Those applications are low g applications. Typically a car [excluding race cars] can only generate 1g of acceleration—maybe slightly more, but that is really pushing it, even under heavy braking conditions or cornering. So that is a low g application in the same automobile, and the thermal MEMS is great for those types of applications.