DevicedialHow bad is my stick drift?
The Devicedial stick drift test scores what a stick reports while nobody is touching it, from 0 to 100: under 20 is normal, 21 to 50 is mild drift, and over 50 is serious. The score is the largest resting deflection either axis reaches during the sample, as a share of full travel โ the Gamepad API hands the browser a value from โ1 to 1 per axis, and 100 would mean a stick reading fully pushed while it sits still.
The reason 20 is the line is that games never use the raw value. Microsoft’s XInput header recommends ignoring anything below 7,849 of 32,767 on the left stick, which is 24% of full travel, and 8,689 โ 26.5% โ on the right. Drift under that threshold is swallowed by the deadzone and you will never feel it. Above it, the character walks on its own, and the number tells you whether cleaning is worth trying or the module is finished.
What is the drift score measuring?
One thing only: how far a stick says it has moved when nobody is moving it.
The Gamepad API gives a web page each axis as a number between โ1 and 1, where 0 is centre. Put the controller down, take your hands off it, and a healthy stick sits within a few hundredths of zero while a worn one parks somewhere out along an axis. The stick drift test samples both sticks for a few seconds, takes the largest absolute deflection it sees, and multiplies by 100. A stick resting at 0.15 scores 15.
Nothing is smoothed and no deadzone is applied, which is deliberate. Every other place you could test a controller โ a game, a console menu, most desktop utilities โ has already thrown away small values before you get to see them.
Where do the thresholds come from?
From what games ignore. XInput, the interface most PC titles use for controllers, ships recommended deadzone constants in its header: 7,849 for the left stick and 8,689 for the right, against a full-scale value of 32,767. That is 24% and 26.5% of full travel discarded before the game reacts at all.
So the bands follow from the deadzone rather than from opinion:
- Under 20 โ inside every normal deadzone. Measurable, invisible in play, no action needed.
- 21 to 50 โ beyond a tight deadzone and closing on the standard one. This is where a menu cursor starts creeping and where careful aim gets harder without an obvious cause.
- Over 50 โ more than half of full travel with your hands off the pad. Nothing hides this; the character walks by itself.
The bands are ours, and they are named so they can be quoted. What they are anchored to is not: 24% is a published figure from Microsoft’s own header file.
Is this drift, or is it the game’s deadzone?
They are opposite halves of the same picture, and confusing them is the single most common mistake in stick-drift threads.
Drift is the controller reporting movement that did not happen. It is a property of the hardware, and it exists whether or not anything is listening.
A deadzone is software choosing to ignore small values so that a slightly imperfect stick still behaves. It is a property of the game, and different titles pick very different sizes โ which is why one game shows your drift and another does not.
That is why “it only happens in one game” is not evidence that the controller is fine. It is evidence that one game has a smaller deadzone than the others. Measure outside the game, with a test that applies no deadzone, and you get a number that does not depend on which title you loaded.
How should I run the test?
Flat, untouched, more than once.
- Put the controller on a level surface โ not on your lap, not in your hands.
- Take your fingers off entirely, triggers included.
- Run the stick drift test and note the score for each stick.
- Wait a few minutes and run it twice more.
A score that is the same every time points to a calibration offset: the electrical centre has moved, and the platform’s own controller calibration or a firmware update may re-centre it. A score that wanders between runs, or climbs as the pad warms up, points at wear in the sensor itself. If the sticks are fine but buttons or triggers are not responding, that is a different test โ the gamepad tester shows every button and axis reacting live.
What actually fixes it?
In ascending order of risk. Recalibrate first, through the console or platform settings rather than through third-party software. Update the controller’s firmware, because manufacturers do ship centre-point fixes. Blow compressed air around the base of the stick, where dust and skin debris collect in exactly the place that matters.
After that it becomes surgery: contact cleaner into a potentiometer, or replacing the stick module outright. Both work and both go wrong. If the pad is still in warranty, the sensible move is to record the drift score before you open anything โ a measured number and a repeatable procedure is a far better support ticket than “it drifts sometimes”.
While you are checking gear, drift is rarely the only thing that has aged. A mouse of the same vintage often has a switch starting to bounce, which the double click test finds in ten seconds and the guide to double clicking explains.
Frequently asked questions
What drift score is still fine?
Anything under 20. A perfectly centred stick reads 0, and a few points of resting deflection is normal for an analogue sensor โ no potentiometer or Hall sensor returns to exactly the same electrical centre every time. Between 21 and 50 you have real drift that a game with a small deadzone will show. Above 50 the stick is reporting more than half of full travel while untouched, which no standard deadzone will hide.
How do I test drift when the game has its own deadzone?
Test outside the game. A game applies a deadzone before you ever see the value, so it is measuring the stick and then hiding the result โ which is exactly why in-game testing tells you so little. A browser test reads the Gamepad API directly and applies no deadzone at all, so it reports the raw resting deflection. Run it with the controller flat on a desk and your hands off it entirely, including the triggers.
Why does the test show drift when the game feels fine?
Because the game is throwing your drift away. With XInput’s recommended left-stick deadzone at 24% of travel, a resting deflection of 15 is invisible in anything that follows the guidance. That is not a false reading โ it is a genuine measurement of a stick that has started to wander, and it is useful precisely because it warns you before a title with a tighter deadzone makes it obvious.
Is drift a calibration problem or a hardware fault?
Both exist and they behave differently. A calibration offset is steady: the stick rests at the same wrong place every time, and recalibrating or a firmware update can re-centre it. Wear is unsteady: the score moves between runs, changes after you flick the stick, or drifts in one direction as the pad warms up. Run the test three times a few minutes apart โ a constant number points at calibration, a wandering one at the sensor.
What can I try before replacing the controller?
In order of risk: recalibrate through the console or platform’s own controller settings; update the controller firmware; blow compressed air around the base of the stick where dust collects; and only then consider opening it. Contact cleaner sprayed into a potentiometer is a common repair and a common way to make things worse. If the pad is in warranty, measure the drift first, note the score, and let the manufacturer handle it.
How the test works
Sources: