Tilt sensors, accelerometers, gyroscopes, IMUs, and attitude sensors are often mixed together as if they all measure “angle”.
That is too rough.
Many tilt sensors are based on accelerometers, but a tilt sensor does not necessarily contain a gyroscope. A 6-axis IMU usually means a three-axis accelerometer plus a three-axis gyroscope. That is not the same thing as every tilt sensor.
The first model is: in static or low-dynamic scenes, tilt can be calculated from the gravity vector projected onto the sensor axes. In dynamic scenes, linear acceleration and vibration pollute the accelerometer, so gyroscopes, filtering, and fusion become necessary.
Static:
3-axis accelerometer -> gravity components -> roll / pitch
Dynamic:
accelerometer + gyroscope -> fusion algorithm -> more stable attitude
What Tilt Really Means
Tilt is the relationship between the device coordinate system and gravity.
When a sensor is fixed to a device, gravity projects onto the sensor X/Y/Z axes. When the device tilts, those projections change. From those components, roll and pitch can be estimated.
Gravity direction is fixed
Device coordinate system rotates with the device
Gravity components on X/Y/Z change
-> Tilt can be inferred
The boundary is important: an accelerometer does not directly measure angle. It measures acceleration. Angle is calculated later.
Why an Accelerometer Can Estimate Static Tilt
When the device is still, the accelerometer mainly senses gravity:
ax, ay, az
If the sensor lies flat, one axis may be near 1 g and the others near 0 g. When it tilts, gravity is distributed across axes.
Common formulas look like:
roll = atan2(ay, az)
pitch = atan2(-ax, sqrt(ay^2 + az^2))
Sign conventions depend on axis definitions and mounting direction, but the idea is the same: estimate gravity direction from acceleration, then infer tilt.
This works well for installation angle, slow platform tilt, boxes, engineering machinery at low speed, and solar panel tilt monitoring.
When Accelerometers Mislead
An accelerometer measures acceleration, not only gravity. When the device is moving, vibrating, or being hit, it sees:
measured_acceleration = gravity + linear_motion_acceleration + vibration + noise
If the algorithm treats all of that as gravity, it converts motion into false tilt.
Typical cases:
- Vehicle acceleration or braking looks like pitch change
- Motors, fans, and pumps make tilt readings shake
- Moving, bumping, or dropping the device creates jumps
- Robots and drones have strong linear acceleration
- Flexible mounting mixes structural vibration with real tilt
Static tilt can be simple. Dynamic attitude cannot be solved by accelerometer-only math.
The same MEMS accelerometer is also often used in IoT devices for vibration, shock, movement, and low-power wakeup. That is a dynamic-acceleration application. It can show that a device moved or changed state, but it does not directly diagnose a specific fault.
A Gyroscope Measures Angular Rate, Not Angle
A gyroscope measures angular velocity:
gx, gy, gz -> angular rate around X/Y/Z
To get angle, the rate must be integrated:
angle = angle + gyro_rate * dt
Gyros respond quickly to rotation and are not directly fooled by linear acceleration like accelerometers. But they drift. Any small bias integrates into a growing angle error.
MEMS gyros also have temperature drift. Temperature changes can affect zero bias, scale factor, and noise. Bias drift is especially troublesome because it accumulates after integration.
Common handling includes:
- Zero-bias calibration at startup
- Bias-versus-temperature compensation
- Online correction using accelerometer, magnetometer, vision, or other references
- Managing thermal environment
Why a 6-Axis IMU Needs Fusion
A 6-axis IMU contains:
- 3-axis accelerometer
- 3-axis gyroscope
It does not automatically output true attitude. The useful result comes from fusion.
The core idea:
Use gyroscope for short-term fast motion.
Use accelerometer to pull long-term attitude back to gravity.
A simple complementary filter can be written as:
angle = alpha * (angle + gyro_rate * dt) + (1 - alpha) * accel_angle
This is not every product’s exact algorithm, but it captures the tradeoff: gyros are fast but drift; accelerometers are stable long-term but polluted by motion.
Why Balance Vehicles and Quadcopters Need IMUs
A two-wheel balancing vehicle needs body pitch and falling rate in real time so the motor can correct it. Acceleration, braking, and road vibration pollute accelerometer readings, so gyro feedback is essential.
A quadcopter is even more dynamic. It estimates roll, pitch, yaw, and angular rate continuously to control motor speeds. The control loop relies heavily on gyros; accelerometers provide low-frequency attitude reference; magnetometers, barometers, GPS, or vision may support heading, altitude, and position.
Accelerometer can estimate static tilt.
Dynamic control needs IMU, fusion, and a controller.
Why 9-Axis IMUs Add a Magnetometer
Some IMUs add a three-axis magnetometer:
3-axis accelerometer + 3-axis gyroscope + 3-axis magnetometer
The magnetometer provides a long-term geomagnetic reference for heading, or yaw.
But magnetometers are easily disturbed indoors and inside devices. Motors, currents, magnets, steel structures, screws, and wiring can change the local field. A magnetometer is not a magic north reference.
This also explains why accelerometers can estimate roll and pitch, but cannot reliably know rotation around gravity. Yaw needs another reference.
A Tilt Sensor Does Not Always Need a Gyroscope
If the goal is static or slow tilt, a three-axis accelerometer with filtering, temperature compensation, and calibration may be enough.
Typical concerns are:
- Static accuracy
- Resolution
- Zero stability
- Temperature drift
- Response time
- Anti-vibration filtering
- Mounting calibration
A gyroscope adds cost, power consumption, calibration complexity, and drift handling. It is valuable when the scene is dynamic, not just because “more axes” sounds better.
Mounting and Zero Point Often Matter More Than the Chip
Tilt measurement depends heavily on installation.
Common problems include:
- Sensor coordinate axes not aligned with device axes
- PCB mounted at a slight angle inside the housing
- Housing assembly adding extra tilt
- Mounting surface not level
- Zero calibration done before final assembly
- Vibration source coupled directly into the sensor
For many products, final installed calibration matters more than the nominal accelerometer specification.
Filtering Changes Output Meaning
Filtering can make tilt readings stable, but it changes response.
Low-pass filters reduce vibration but add delay. Median filters suppress spikes but can hide fast changes. Long averaging windows make the display smooth but slow.
The design must choose between:
- Stability
- Response speed
- Vibration rejection
- Latency
Equipment installation monitoring may prefer stability. Vehicles, robots, and moving platforms may need faster response and lower phase delay.
Engineering Takeaway
A tilt sensor is not always a 6-axis IMU.
If the task is static or slow tilt, a three-axis accelerometer with filtering, temperature compensation, and calibration may be enough.
If the task is dynamic attitude, such as vehicles, robots, drones, handheld motion, or rapid rotation, a gyro and fusion algorithm are valuable.
Before choosing a sensor, ask:
- Static tilt or dynamic attitude?
- Is there acceleration, vibration, or shock?
- Do you need yaw, or only roll and pitch?
- Can you calibrate after final mounting?
- How much latency can filtering introduce?
The key sentence is:
Accelerometers measure acceleration and infer gravity direction.
Gyroscopes measure angular rate and drift after integration.
Stable dynamic attitude comes from fusion and valid assumptions.