An accelerometer is a sensor that measures acceleration — the rate of change of velocity. It detects how fast something is speeding up, slowing down, or changing direction.

Here's the thing: it also feels gravity. When the sensor sits still on your desk, it reads 1 g pointing down. That's not a bug — it's how the sensor knows which way is "up." Your phone uses this to rotate the screen when you tilt it.

In short: an accelerometer converts mechanical motion into an electrical signal you can read with a microcontroller.

The Basic Idea

Picture a tiny weight attached to springs. When you move the sensor, the weight resists the motion due to inertia — it lags behind. That lag creates a displacement or force you can measure.

Newton's Second Law says force equals mass times acceleration. The sensor knows the mass. It measures the force. Division gives you acceleration. Simple in theory, tricky in practice.

MEMS Accelerometers (The Ones You'll Actually Use)

Over 90% of accelerometers sold today use MEMS technology — tiny mechanical structures etched into silicon chips.

Here's what happens inside:

A microscopic silicon block (the proof mass) hangs from flexible beams. It forms one plate of a capacitor. Fixed electrodes form the other plates. When acceleration hits, the proof mass shifts position. That changes the capacitance. On-chip circuitry converts that capacitance change into a voltage or digital number.

Why MEMS won the market:

Cheap — Built on standard semiconductor wafers. Costs 
0.50to
0.50to5 in volume.
Tiny — A complete 3-axis sensor fits in a 2mm × 2mm package.
Low power — Some draw less current than a digital watch.
Easy to integrate — Slap a gyroscope on the same die and you've got an IMU.

Piezoelectric Accelerometers

These show up in industrial vibration monitoring. A piezoelectric crystal generates electrical charge when mechanically stressed.

Key limitation: no DC response. It can't measure static tilt or gravity — only dynamic vibration. But it handles frequencies up to 30 kHz, far beyond MEMS capabilities. Costs 

50to 50to500+.

Piezoresistive Accelerometers

Strain gauges bonded to a flexure change resistance as they stretch. Measures both static and dynamic acceleration. Common in automotive crash sensors. Mid-range pricing at 
10to100.

By Sensing Technology

By Axis Count

Single-axis — One direction only. Rare in practice.
Tri-axis (3-axis) — Measures X, Y, and Z simultaneously. The default choice for most projects.
6-axis IMU — 3-axis accelerometer plus 3-axis gyroscope. For orientation tracking.
9-axis IMU — Adds a magnetometer for compass heading. Full navigation capability.

By Quality Grade

Measurement Range

How hard can you push it before the output clips? Expressed in g (1 g = 9.8 m/s²).

• ±2 g — Tilt sensing, screen rotation, step counting

• ±4–8 g — Wearables, general motion tracking

• ±16–24 g — Sports equipment, impact detection

• ±50–500 g — Crash sensors, shock testing

• ±2,000 g+ — Ballistics, explosives research

Pick a range that covers your worst case plus 30% margin. Too narrow and you clip during impacts. Too wide and you waste resolution.

Sensitivity

Output per unit of acceleration. Analog sensors use mV/g. Digital ones use LSB/g.

Higher sensitivity resolves smaller accelerations — good for seismic work. The trade-off: high-sensitivity parts usually have narrower ranges.

Resolution

The smallest acceleration change you can trust. Limited by electronic noise.

Screen rotation needs ~50 mg resolution. Inertial navigation demands ~0.1 mg. Most projects fall somewhere in between.

Bandwidth

How fast can the signal change before the sensor misses it?

Consumer MEMS handles up to ~1 kHz. Industrial piezoelectric units hit 30 kHz. Wider bandwidth means more noise — narrow it to your signal of interest for cleaner data.

Zero-g Offset

What does it read when perfectly still? Should be zero (or 1 g on the vertical axis). Never is, exactly.

Consumer parts wander ±50 mg. Automotive-grade holds ±20 mg. Industrial stays within ±5 mg. Temperature drift matters too — a sensor calibrated at 25°C might read 20 mg off at 85°C.

People mix these up constantly.

An accelerometer measures linear acceleration — gravity counts. It tells you which way is down and how fast you're moving in a straight line.

A gyroscope measures rotation speed. It tells you how fast you're spinning around each axis.

An IMU combines both (and sometimes a magnetometer) to track full 3D orientation and position.

Why you need both: An accelerometer in a moving car can't tell if the 1 g reading is gravity pulling down or the car accelerating forward. The gyroscope sorts that out. That's why phones use both — and why navigation apps get confused when you spin in place.

What does "g" mean in accelerometer specs?

1 g equals standard Earth gravity: 9.8 m/s². A ±2 g sensor handles -19.6 m/s² to +19.6 m/s².

Why does my accelerometer read 1 g when sitting still?

Gravity never stops pulling. Sit still and the sensor feels 9.8 m/s² straight down. Turn it on its side and that same 1 g shows up on the X or Y axis. Your phone uses this to figure out which way is up.

Can an accelerometer measure speed or distance?

Technically yes, practically no. You'd integrate acceleration over time to get velocity, then integrate again for position. Every tiny error compounds. After ten seconds you're off by meters. After a minute you're in the wrong zip code. Use an IMU instead.

What's the best accelerometer for Arduino?

The ADXL345 if you only need acceleration. The MPU-6050 if you want orientation tracking too. Both have massive community support and libraries everywhere.

Can I use a MEMS accelerometer for vibration analysis?

Works for low-frequency vibration — building sway, human motion. For high-speed rotating machinery, use a piezoelectric sensor. The bandwidth and noise performance are in a different league.