Gyroscope Test Guide: Check Your Phone's Rotation Sensor

Tilt your phone and the screen rotates. Turn it in a game and the view follows. Hold it up for an augmented-reality view and digital objects stay pinned to the real world as you move. Behind all of that sits the gyroscope, a tiny sensor that measures rotation: how fast the device is turning, and around which axis. The Gyroscope test reads that sensor in your browser and shows its values as you move the device, so you can see whether rotation is being detected at all, whether it is smooth or jumpy, and whether it drifts when the phone is still. It is a behaviour check, not a repair tool, and knowing what it can and cannot show is the key to using it well.

A gyroscope is often confused with an accelerometer, and the distinction matters for diagnosis. An accelerometer measures linear acceleration, including gravity, which is why it can tell which way is down and therefore how the phone is tilted. A gyroscope measures angular velocity, how quickly the phone is rotating around its three axes: pitch, roll, and yaw. Games and panorama photos need the gyroscope because they care about rotation itself, not just tilt. Modern phones combine both sensors, plus a magnetometer, in software to estimate orientation, which is why a faulty gyroscope does not always break screen rotation: the accelerometer may still handle simple portrait-to-landscape changes while rotation-heavy apps fail. If screen rotation works but a game or panorama app behaves wildly, the gyroscope is a prime suspect.

How to run the test

Open the test on the phone itself, not on a desktop, since a desktop has no such sensor to read. Grant motion permission if the browser asks: on some phones, especially newer iPhones, motion sensors are blocked until the user explicitly allows them, and a test that shows nothing may simply be waiting for that permission rather than reporting a dead sensor. Hold the phone still first and watch the readings: they should sit near zero, with only small flickers. Then rotate the phone slowly around one axis at a time, first left and right, then forward and back, then twisting like a steering wheel, and watch whether the values rise smoothly and fall back as you stop. Then rotate it quickly and check whether the values keep up or lag and jump. Finally, set it down flat and still for thirty seconds and watch for drift: a sensor whose resting value creeps steadily in one direction is drifting, which makes virtual objects slowly slide even when you hold still.

Worked example

Consider a sample test session, with values clearly labelled as an illustration. At rest, the three axis readings hover between minus 0.02 and plus 0.02 radians per second, which is normal noise. You rotate the phone slowly through a quarter turn, 90 degrees, over 2 seconds. A quarter turn is about 1.57 radians, since 90 divided by 180 times 3.1416 is roughly 1.57, and over 2 seconds the average rotation rate is 1.57 divided by 2, which is about 0.79 radians per second. The test shows a smooth rise to around 0.8 radians per second on one axis and a smooth fall back to near zero as you stop, which matches the expected value closely. Now imagine the faulty case: the same rotation shows values jumping between 0.1 and 2.5 with no smooth curve, and after you stop, the reading settles at 0.35 instead of near zero and stays there. The first case is a healthy sensor; the second shows noise plus a stuck offset, and no app setting will fix that pattern. The arithmetic gives you a yardstick: a slow, known rotation should produce a modest, steady reading, not spikes or a permanent offset.

Mistakes to avoid

The first mistake is testing on a device that cannot report the sensor, such as a desktop or a browser that blocks motion access, and concluding the phone is broken. Test on the phone, in a browser that supports motion sensors, with permission granted. The second is testing inside a case with a magnetic clasp or near a strong magnet, which can disturb the companion sensors and confuse the orientation estimate. The third is judging the sensor from one app: a single misbehaving game may have its own bug, so confirm with the dedicated test before blaming hardware. The fourth is confusing drift with damage: a small amount of drift is normal and software compensates for it, but steady, large drift that grows over a minute is a real fault. The fifth is attempting hardware fixes, such as opening the phone, on the basis of a browser test alone; a browser test is evidence for a service visit, not a repair manual.

Putting it into practice

If the test shows smooth rotation and a quiet rest state, the sensor is almost certainly fine, and any app problem lies in that app's settings, its permissions, or its calibration screen. Many games and camera apps offer their own calibration or recentring step, which is worth trying before anything else. If the test shows spikes, a stuck offset, or no response at all after permissions are granted, try a restart, then test again, then test in a second browser to rule out a browser fault. If the fault persists across browsers after a restart, note exactly what you saw, including the axis affected and the resting offset, and take that description to support or a repair service. For related device checks, the Microphone test and Webcam test cover other sensors the same way, and the companion guide on Desktop Computer Test covers broader device health. A sensor test is most useful when it turns "my game feels wrong" into a specific, repeatable observation.

Key points

  • A gyroscope measures rotation, while an accelerometer measures linear movement and tilt
  • A working gyroscope shows smooth, continuous rotation values as you turn the device
  • Calibration problems and app permissions can mimic a dead sensor
  • A browser test shows sensor behaviour, it cannot repair hardware or replace a service check

In short: hold still, rotate slowly on one axis at a time, and compare what you see with what a known rotation should produce. Smooth values that return to near zero mean a healthy sensor; spikes, stuck offsets, or silence after permission mean the problem is real and worth taking further.