So, I spent last week sitting in an old New Hampshire mill building, knitting like a fiend. And the only reason that fact is relevant is that the existence of that mill gives me an excuse to talk about tailoring technology to fit the application. And we all know that the application decides what you measure, how you measure it, and just how complicated your solution is going to be.
The mill building in question is in Harrisville, NH, and forms part of the Harrisville Historic District, a well-preserved New England mill village. They still spin woolen yarn in Harrisville, incidentally, making them one of the handful of US-based mills that still do so. The mill is where it is because that's where the Nubanusit Brook is, a narrow waterway that incorporates a series of sharp drops along its 15-mile path from Nelson down to Peterborough. Because hydropower is all about the height of the water column, these steep drops in the brook made perfect spots to place water-powered mills and, if you follow the brook you will find either existing mill buildings or mill foundations at these spots. Because the brook is narrow, the mills themselves were narrow (longer crank shafts would twist and break). And because this section of the state has lots of granite and clay, the mill owners could create and maintain reservoirs of various sizes, controlled by dams, to make sure that the mills had a regular water supply with which to work. At the height of production in the mid 1800s, woolen cloth woven in Harrisville could be transported via train to Boston and daily packet to New York City to arrive within 24 hours. Did I mention that Harrisville is tiny and tucked away into the woods?
In the case of the New England mills, the application—whether it was sawing lumber, milling flour, spinning yarn, or weaving cloth—needed a power source. In my neck of the woods, the best available power source was water. The constraints of that waterpower decided the location, size, and internal layout of the water-powered mills. Someone specifying sensors for use in a modern sawmill is going to deal with different constraints than someone specifying sensors that will be built into a toaster, or implanted into a person, or used in a nuclear plant. An industrial vibration sensor for condition-based monitoring in a manufacturing plant is going to look very different to an accelerometer intended for a smartphone. Both are accelerometers, both may use the same sensing technology, but their specifications, packaging, and outputs are not the same because the demands of their particular applications are so different. (I love talking with application engineers because I always, always, learn something new and wonderful when I do. Sometimes it's a sensor application that's completely new to me, sometimes it's just a really clever use of an existing technology, but it's always educational.)
It's also not as if this state of affairs is static; we're always working to improve existing sensor systems as well as adding new applications, figuring out how to tailor the technology to better fit our specific needs. The devil is still in the details, whether you're designing a woolen mill or a car's stability control system.