A chip smaller than a fingernail measures acceleration, tilt, vibration, and shock with microvolt-level precision.
MEMS accelerometers do this by detecting the displacement of a microscopic proof mass etched from silicon, a structure you cannot see with the naked eye but that follows the same physics as any spring-mass system.
This article covers how MEMS accelerometers work, the key electrical parameters engineers need to select the right device, and the six most important industrial and consumer applications.
A MEMS accelerometer is a microelectromechanical sensor that converts physical acceleration into an electrical signal by measuring the displacement of a silicon proof mass suspended on microscopic flexure springs.
The same chip technology that fits in your smartphone gyroscope also monitors bearing health in industrial motors, detects earthquakes in seismographs, and triggers airbags in cars within milliseconds of a collision.
What Is a MEMS Accelerometer and Why It Matters
MEMS stands for Microelectromechanical Systems. It refers to devices that combine mechanical structures and electronic circuits on a single silicon chip using the same photolithography and etching processes used to make integrated circuits.
A MEMS accelerometer etches a tiny proof mass, typically between 0.1 and 10 micrograms, from a silicon wafer. The proof mass is connected to the chip frame by thin flexure beams that act as springs.
When the chip accelerates, the proof mass resists the motion due to inertia and deflects relative to the frame. That deflection, often measured in nanometres, is what the sensor converts into a voltage or digital signal.
Before MEMS technology, measuring acceleration required bulky piezoelectric or servo accelerometers costing hundreds of dollars each. Today a sensor with better performance costs under one dollar in volume.
This cost reduction is why accelerometers now appear in every smartphone, wearable device, industrial vibration monitor, and automotive safety system.
Unlike a piezoelectric vibration sensor, which only detects dynamic vibration and cannot measure static tilt or gravity, a MEMS capacitive accelerometer responds to both static and dynamic acceleration.
Static acceleration includes gravity and constant tilt. Dynamic acceleration covers shock and vibration. This dual capability makes it far more versatile for industrial condition monitoring.
MEMS Accelerometer Working Principle: Capacitive Sensing
Most industrial and consumer MEMS accelerometers use differential capacitive sensing. The proof mass forms the movable plate of two capacitors, one on each side, with fixed electrodes attached to the chip frame forming the other plates.
At rest, both capacitors are equal: C1 = C2. When acceleration displaces the proof mass by a distance x toward one electrode, the gap on that side shrinks and the gap on the other side grows.
This creates a capacitance imbalance: C1 increases, C2 decreases. The differential signal C1 minus C2 is proportional to the displacement and therefore proportional to the acceleration.
The capacitance change for a displacement x is given by the parallel plate formula. A typical device with a 1 microgram proof mass sees only 0.98 nanometres of displacement under 1g.
The resulting capacitance change is around 20 attofarads. Dedicated charge amplifier circuits on the same chip convert this tiny signal to a usable voltage with noise levels well below 1 millivolt.
The resonant frequency is important because the usable bandwidth must stay below about 30% of it to avoid amplitude errors and phase distortion.
A device with a 15 kHz resonant frequency reliably measures signals up to about 5 kHz, which covers all standard industrial vibration monitoring requirements.
Four Types of MEMS Accelerometers and Their Sensing Mechanisms
Capacitive MEMS Accelerometer
The most common type. Detects differential capacitance change between movable and fixed comb fingers as the proof mass displaces. Responds to both static and dynamic acceleration. Noise below 100 ug per root Hz. Used in smartphones, IoT, industrial vibration, and inertial navigation.
Piezoresistive MEMS Accelerometer
Uses strain gauges diffused into the flexure beams. Stress in the beam changes resistivity, detected by a Wheatstone bridge. Simpler to manufacture than capacitive but higher noise and temperature sensitivity. Common in low-cost, high-shock applications like drop testing and crash recorders.
Piezoelectric MEMS Accelerometer
A piezoelectric thin film on the flexure beam generates charge directly from strain, no separate sense circuit needed. Zero DC response, so it cannot measure tilt or gravity. Best for high-frequency shock, ultrasound, and impact detection above 1 kHz.
Thermal MEMS Accelerometer
No proof mass. A tiny heater creates a hot gas bubble; acceleration shifts the temperature distribution asymmetrically. Thermopile sensors detect the asymmetry. Extremely robust to shock and vibration, no mechanical fatigue. Lower bandwidth than capacitive types but nearly immune to mechanical damage from dropping.
Key Electrical Parameters for MEMS Accelerometer Selection
| Parameter | Typical Range | What It Means | Selection Guidance |
|---|---|---|---|
| Measurement Range | ±2g to ±400g | Maximum acceleration the sensor can measure without clipping or damage | Choose 2× to 3× your expected peak acceleration to avoid saturation |
| Sensitivity | 0.3 mV/g to 1000 mV/g | Output voltage change per unit of acceleration. Higher sensitivity means better resolution at low acceleration. | High sensitivity (>100 mV/g) for low-g motion; lower sensitivity for shock and high-g measurement |
| Noise Density | 25 ug/√Hz to 1 mg/√Hz | RMS noise per unit bandwidth. Multiply by sqrt of your signal bandwidth to get total RMS noise. | For tilt sensing at DC: any value fine. For vibration at 1 kHz: use <100 ug/√Hz |
| Bandwidth | DC to 5500 Hz | Frequency range over which the sensitivity is flat within ±3 dB. Signals above this are attenuated. | Must exceed the highest vibration frequency you need to detect, with margin |
| Cross-Axis Sensitivity | 0.5% to 5% | Output on one axis due to acceleration on a perpendicular axis. Lower is better for 3D accuracy. | Below 2% for navigation; up to 5% acceptable for simple tilt or event detection |
| Supply Voltage | 1.7V to 5V | Operating voltage range. Lower voltage sensors suit battery-powered IoT devices. | Match to your microcontroller supply rail. 3.3V is most common in modern systems. |
| Interface | Analog / SPI / I2C | How data leaves the sensor. Digital interfaces include on-chip ADC; analog output requires external ADC. | I2C for simple low-speed systems; SPI for high-speed data; analog for legacy signal chains |
| Shock Survival | 500g to 10,000g | Maximum non-destructive shock the device can survive as a brief impulse (0.5 ms pulse). | Industrial and automotive applications need at least 2000g shock survival |
MEMS Accelerometer Noise and Resolution Calculator
MEMS Accelerometer Applications in Industry and Consumer Electronics
How to Select a MEMS Accelerometer for Your Application
✅ Right Selection Approach
- Start with the measurement range: set it 2× to 3× above your expected peak acceleration
- Calculate required noise floor: noise density × sqrt(bandwidth) must be well below your minimum signal
- Check bandwidth: must exceed the highest frequency component you need to capture
- Choose digital output (SPI/I2C) for modern microcontroller systems to avoid external ADC noise
- For battery-powered IoT devices, prioritize low supply current and standby modes
- Verify shock survival rating exceeds the worst-case handling or installation shock
❌ Common Selection Mistakes
- Choosing ±2g range for an application that sees 5g peaks, causing output clipping
- Ignoring noise density for low-frequency tilt applications where noise accumulates at DC
- Selecting a sensor with bandwidth below the vibration frequency of interest
- Using analog output with a long cable run in an electrically noisy environment
- Confusing sensitivity (mV/g) with resolution: high sensitivity does not mean low noise
- Installing on a resonant bracket that amplifies vibration beyond the sensor's range
MEMS Accelerometer vs Other Acceleration Sensors
| Sensor Type | DC Response | Frequency Range | Typical Noise | Cost | Best For |
|---|---|---|---|---|---|
| MEMS Capacitive | Yes | DC to 5 kHz | 25 to 500 ug/√Hz | Under $5 | IoT, industrial vibration, tilt, navigation |
| Piezoelectric (bulk) | No | 1 Hz to 30 kHz | 0.1 to 10 mg/√Hz | $20 to $500 | High-frequency shock, machine tool vibration |
| Piezoresistive MEMS | Yes | DC to 10 kHz | 0.5 to 5 mg/√Hz | Under $3 | Drop testing, low-cost shock detection |
| Servo (force balance) | Yes | DC to 300 Hz | Below 1 ug/√Hz | $500 to $5000 | Seismology, inertial navigation grade |
| Hall Effect Tilt Sensor | Yes | DC only | High (degrees) | Under $2 | Simple tilt switch, inclinometer replacement |
Watch: How MEMS Accelerometers Work
MEMS Accelerometer Questions Engineers Ask
External References
- MEMS Accelerometer Capacitive Sensing — Analog Devices
- How to Interpret Accelerometer Specifications — STMicroelectronics (PDF)
What We Learn Today
- A MEMS accelerometer measures linear acceleration by detecting nanometre-scale displacement of a silicon proof mass on microscopic flexure springs
- Capacitive MEMS accelerometers use differential capacitance change between movable and fixed comb electrodes, the most common type for industrial and consumer use
- Proof mass displacement at 1g is only about 1 nanometre; the resulting capacitance change is in the attofarad range, converted by on-chip charge amplifiers
- RMS noise = noise density (ug/√Hz) × sqrt(bandwidth in Hz); always check this against your minimum detectable signal
- Usable bandwidth must stay below 30% of the resonant frequency to avoid amplitude errors and phase distortion
- Capacitive MEMS responds to both static (tilt, gravity) and dynamic (vibration, shock) acceleration, unlike piezoelectric sensors which are AC-coupled only
- Key selection parameters: measurement range, noise density, bandwidth, interface type, and shock survival rating