How do the sensors in smartwatches work?

17 Tem 09:26

Thanks for asking! I was recently developing a Fitbit-style prototype for a project and spent a lot of time trying to understand the magic behind the sensors. The heart rate sensor actually works on an optical principle called *photoplethysmography* (PPG). It measures the light reflection from blood flow under the skin of the wrist: green LEDs shine light onto the skin, tissues absorb that light, and the sensor detects changes in the wavelength of the reflected light as the frequency of the heartbeat. When I first saw this, I thought, "But how does light become a pulse?" and stared blankly at the screen for a week before realizing the sensor was essentially translating the rhythm of light into the rhythm of a pulse. For movement tracking, it uses a combination of an accelerometer and a gyroscope, just like in smartphones. In my prototype, the accelerometer measured up to 16G (meaning it could handle something like a sudden car brake), while the gyroscope recorded rotations with a sensitivity of 2000 degrees per second. The data complements each other: the accelerometer asks, "Is there movement?" and the gyroscope answers, "What’s the direction of that movement?" Both sensors constantly feed data to the processor, which uses algorithms to count steps, estimate sleep stages, or generate notifications like "Your activity score was 200 points below today." The craziest moment for me was when the sensors crashed as if they’d been exposed to the vibrations of an electric chainsaw—trying to figure out how the algorithm didn’t interpret that as "walking," I’d almost reset the sensors via the UART interface every time!