How do heart rate sensor in smartwatches work?

29 Tem 15:18

When I started testing the cardio function of my watch during my jogging sessions, the first thing I noticed was that the optical sensor relies on photoplethysmography (PPG): a small LED (usually green because blood absorbs this wavelength better) illuminates the skin, and a photodiode measures the reflected light. With each heartbeat, the blood volume under the skin increases, reducing reflection; the photodiode thus detects a sinusoidal variation corresponding to the heart rate. The watch's microcontroller converts these analog variations into high-frequency digital data (around 25–100 Hz) and then applies several processing steps: a band-pass filter to eliminate low-frequency noise (blood pressure) and high-frequency noise (light interference), followed by a peak detection algorithm (often a first derivative with an adaptive threshold) that identifies each contraction. During a hike in the mountains, I noticed that wrist movements introduced significant artifacts: vibrations from the LED and changes in angle altered the amount of light received. The watch then uses dynamic filtering: Kalman filters or weighted moving averages reduce these fluctuations, and in some models, an integrated acceleration sensor correlates detected peaks with movements, rejecting those that don’t coincide with a real pulse. Thanks to these steps—PPG acquisition, noise filtering, motion correction, and peak detection—the reading remains reliable even during intense activity. In practice, I found that wearing the watch slightly higher on the arm, where movement is minimal, improves signal stability, which aligns with what manufacturers recommend in their user guides.