An accelerometer is a sensor that measures acceleration, the rate at which an object speeds up, slows down, or changes direction. You interact with accelerometers dozens of times a day without realizing it. They rotate your phone screen when you tilt it sideways, count your steps, deploy airbags in a crash, and detect when an elderly person falls. At their core, these sensors convert physical motion into an electrical signal that a computer can read and act on.
How an Accelerometer Works
The most common type today is the MEMS (micro-electromechanical systems) accelerometer, a tiny chip small enough to fit inside a smartphone. Inside the chip sits a small weighted structure called a proof mass, suspended by microscopic springs and attached to a frame. When the device accelerates, the proof mass lags behind due to inertia, shifting slightly relative to the frame. That tiny shift is what the sensor measures.
To detect that movement, the chip uses pairs of capacitor plates: one set attached to the proof mass and another fixed to the frame. When the proof mass is stationary, the gaps between the plates are equal, and the electrical capacitance on both sides matches. When acceleration pushes the proof mass to one side, one gap shrinks while the other widens, creating a measurable difference in capacitance. For small displacements, this difference is directly proportional to how much the mass moved.
The sensor converts that capacitance difference into a voltage. A circuit drives the fixed plates with high-frequency electrical signals, and the resulting output voltage changes in proportion to acceleration. When there’s no acceleration, the voltage output is zero. When the sensor accelerates in one direction, the voltage goes positive; reverse the direction, and it goes negative. The final output is a clean electrical signal whose strength and sign tell you both how strong the acceleration is and which way it’s pointing.
Newton’s second law ties it all together. The displacement of the proof mass depends on the spring stiffness and the mass itself, so the chip’s designers can calibrate the relationship between voltage output and acceleration with high precision.
Units and Key Specifications
Acceleration is measured in “g,” where 1g equals the pull of Earth’s gravity (9.8 m/s²). A phone sitting on a table reads 1g pointing downward. A roller coaster might hit 3 to 4g. A car crash can exceed 100g.
The sensitivity of an accelerometer describes how much electrical output you get per unit of acceleration, typically rated in millivolts per g. A higher sensitivity means the sensor can detect smaller movements, but it also means the sensor saturates at lower forces. This is why consumer devices and crash-test rigs use very different sensors. A phone accelerometer might cover a range of ±16g, while an industrial shock sensor handles thousands of g.
Bandwidth, measured in hertz, defines how fast the sensor can respond. A sensor with a bandwidth of 500 Hz can track vibrations happening up to 500 times per second. Consumer devices need only modest bandwidth for screen rotation and step counting, but industrial and crash-testing sensors need much wider bandwidth to capture rapid, high-frequency events.
Three Main Types
Not all accelerometers work the same way. The three most common types each have distinct strengths.
- Capacitive MEMS: The type described above, found in nearly every smartphone, tablet, and wearable. These are inexpensive, tiny, and excellent at measuring low-frequency motion like tilt, orientation, and walking cadence. They’re the reason accelerometers became ubiquitous in consumer electronics.
- Piezoelectric: These use a crystal material that generates an electrical charge when squeezed by acceleration. They’re the most widely used type for professional vibration testing because of their wide frequency response, low noise, and durability. Charge-mode versions can survive extreme temperatures ranging from -200°C to 400°C, making them suitable for jet engines and even nuclear environments.
- Piezoresistive: Instead of generating a charge, these sensors change their electrical resistance under strain. Their standout feature is very wide bandwidth, which lets them capture extremely short, violent events. Automotive crash testing and weapons testing are their primary domain. They also measure down to zero hertz, meaning they can track steady-state forces and calculate velocity or displacement from the data.
How Your Phone Uses One
A smartphone accelerometer is a 3-axis MEMS sensor, meaning it measures acceleration independently along three perpendicular directions: forward/backward, left/right, and up/down. This is what allows your phone to know its orientation in space at any moment. When you rotate your phone from portrait to landscape, the sensor detects that the direction of gravitational pull has shifted relative to the device, and the operating system rotates the display accordingly.
Step counting is more complex. Your phone’s accelerometer picks up the repetitive, rhythmic impact pattern of walking. Algorithms break the raw data into short time windows and look for the dominant frequency of that repeating pattern, which corresponds to your walking cadence. The total step count is calculated by summing the estimated steps across all windows. This is trickier than it sounds because a phone can be in your pocket, hand, or bag, each producing different signal shapes. Modern open-source algorithms address this by working with data from various body locations, making the count more reliable regardless of where you carry your device.
Airbag Deployment in Cars
One of the most safety-critical uses of accelerometers is triggering airbags. Vehicles contain accelerometers that continuously monitor for sudden deceleration. When the sensor detects a spike exceeding a preset threshold, it wakes up the airbag processor, which begins analyzing the vehicle’s full deceleration profile in real time.
The system doesn’t just fire the airbag the instant it feels a jolt. It evaluates how severe the crash is and decides whether to deploy at all, when to deploy, and how many inflator stages to activate. In one study of GM vehicles, the first-stage driver airbag deployed with 50% probability at a speed change of about 9 mph, while both inflator stages deployed with 50% probability at around 26 mph. Deployment times were remarkably fast: the average was 7 milliseconds from the moment the processor woke up, with the median across real-world crashes at 15 milliseconds. Some edge cases saw deployment as low as 3 to 4 mph of speed change, while in other situations airbags didn’t deploy at speed changes above 26 mph, reflecting the sophistication of the algorithms involved.
Fall Detection and Health Monitoring
Wearable devices and medical alert systems use accelerometers to detect falls, particularly for elderly users. These systems typically place sensors at the waist, wrist, or head and continuously track three-axis acceleration data. A fall produces a distinctive signature: a sudden spike in acceleration (the impact), preceded by a brief period of near-weightlessness (free fall), followed by a shift to a horizontal orientation (lying on the ground).
Fall detection algorithms set acceleration thresholds based on measurements of normal activities like sitting, standing, and walking, then flag events that exceed those thresholds. They also analyze the body’s angle after the event, checking whether the person ended up in a prone or supine position through a transition that wasn’t deliberate. Combining acceleration magnitude, body angle, and post-event stillness reduces false alarms from activities like sitting down quickly or dropping the device.
Industrial Vibration Monitoring
Factories use accelerometers for predictive maintenance, mounting them on rotating machinery to detect early signs of mechanical failure before something breaks. A healthy bearing produces a smooth vibration signature. As a bearing degrades, it generates characteristic vibration frequencies tied to specific defect types, such as damage to the outer ring or inner ring of the bearing. These frequencies appear at predictable multiples or fractions of the machine’s rotation speed.
Low-sensitivity, wide-bandwidth accelerometers are chosen for this work because they can pick up the high-frequency vibrations that signal roller element defects and gear mesh faults. General vibration monitoring covers machinery running at speeds from 300 to 7,200 rpm. By tracking changes in a machine’s vibration profile over weeks or months, maintenance teams can schedule repairs before a catastrophic failure, avoiding unplanned downtime.
How Accelerometers Combine With Other Sensors
An accelerometer alone can measure linear motion along three axes (surge, sway, and heave), giving you three degrees of freedom. But it can’t directly measure rotation. That’s where a gyroscope comes in, measuring angular motion around those same three axes (pitch, roll, and yaw), adding another three degrees of freedom. The combination of an accelerometer and a gyroscope is called an inertial measurement unit, or IMU, providing six degrees of freedom.
Adding a magnetometer, which measures the Earth’s magnetic field to determine compass heading, creates what’s known as an attitude and heading reference system. This 9-degree-of-freedom setup is what powers navigation in drones, VR headsets, and robotics. Each sensor compensates for the others’ weaknesses: gyroscopes drift over time, accelerometers can’t distinguish tilt from lateral movement on their own, and magnetometers are easily thrown off by nearby metal. Fusing all three produces a stable, accurate picture of how an object is moving and oriented in space.