Controller testing methodology
How to test controller stick drift without misreading center noise.
A useful result is a pattern, not one decimal value. Use the same controller, connection and browser, collect an untouched sample, move through the full range, then verify that the stick returns to a repeatable center.
What the browser can measure
The browser reads axes, buttons, timestamps and mapping information through the Gamepad API. Each analog stick is normally exposed as a horizontal and vertical axis. Combining those axes gives a radial distance from center, which is easier to compare than either axis alone. The W3C Gamepad specification defines the browser interface, but the operating system, driver and controller still affect what reaches it.
Run a repeatable resting sample
Put the controller on a stable surface, wake it with a button, release both sticks and reset the session. Use the stick drift test to collect a ten-second Controller Health Report without touching either stick. Repeat the sample after rotating each stick around its full edge and releasing it naturally. A consistent directional offset matters more than a brief isolated spike.
Average, P95 and peak drift
Average drift describes the typical resting offset over the sample. P95 is the level that 95% of samples stay at or below, so it is less sensitive to one unusual spike. Peak drift is the largest observed value and helps reveal intermittent movement, but it should not set the deadzone by itself. Compare all three values across repeated tests.
Why a stick rarely stays at exactly zero
Analog sensors, springs, mechanical tolerances, calibration and signal conversion all introduce small variation. A reading slightly above zero is not automatically a defect. The meaningful question is whether the stick repeatedly escapes the threshold used by the game, favors one direction or returns to a different center after movement.
Choose the smallest stable deadzone
A deadzone hides central input; it does not repair a sensor. Start slightly above the repeatable P95 value, then use the controller deadzone test and verify the result inside the game. A larger deadzone can stop unwanted movement but also removes fine aim near the center. Different games may apply different response curves and thresholds.
Compare USB and Bluetooth deliberately
Test the connection you normally use first. If the result is unstable, repeat with a direct USB connection while keeping the browser and computer unchanged. USB and Bluetooth are separate test conditions: radio interference may affect delivery consistency, while a fixed directional center offset that survives both connections is stronger evidence of a controller-side issue.
What the result cannot prove
A browser test shows the signal exposed by the Gamepad API. It cannot identify a worn potentiometer, contamination, damaged spring, firmware fault or incorrect calibration with certainty. It also cannot prove raw USB polling rate or end-to-end game latency. Use repeated browser results as diagnostic evidence, then follow official calibration, warranty and repair guidance.
Published · 5 August 2026