← Back to the chapter · Next detail →
Technical detail · Sensor Selection: Evidence Before Preference
Power and timing in the assembled system
Power claims are a common failure point. Oversampling means taking or combining multiple internal readings for one reported measurement; it can affect both measurement quality and energy use. A regulator’s quiescent current is the small current it uses to operate even when the load draws little power. Temperature derating means allowing for reduced battery capacity or performance at the temperatures where the product must work. A BME280-style environmental node may look excellent from sleep-current rows, but the battery-life claim also depends on measurement mode, oversampling, warm-up, I2C/SPI transaction time, pullup leakage, regulator quiescent current, radio transmit bursts, temperature derating, and whether firmware really enters sleep. A motion node has the same pattern around output data rate, FIFO watermark, interrupt threshold, host wake time, and false-wake rate.
Timing claims fail in similar ways. An accelerometer may support a data rate that looks adequate, but the product still depends on filtering motion above the usable sampling band so it does not appear as misleading lower-frequency motion (anti-alias filtering), FIFO depth, interrupt latency, MCU service time, timestamp alignment, and whether radio or flash operations block sample handling. A fixture is the mounting rig used for a bench test; its resonance is a frequency at which that rig can amplify particular vibrations. Noise density describes sensor noise per frequency bandwidth. A vibration-monitoring design using an ADXL355-class low-noise accelerometer therefore needs fixture resonance checks and real sampling traces, not only a low noise-density row.