What is being measured
A thumbstick reports two axes, each between −1 and +1, and at rest both should read 0. Drift is the failure of that promise: the stick reports movement nobody asked for. The headline figure here is the largest magnitude the pair reached during the run — √(x² + y²) — because that is the quantity a dead zone compares against, not the individual axes. A stick resting at x = 0.04, y = 0.04 has a magnitude of about 0.057 and will escape a 0.05 dead zone even though neither axis alone reaches it.
The per-axis maxima are reported underneath because they localise the fault. Drift confined to one axis usually points at one potentiometer or one Hall sensor; drift on both tends to be a centring or calibration problem. Neither is a diagnosis this page can make for you — it reports what the axes did.
The zero-sample trap
This is the part most drift-testing pages leave out, and it invalidates a great many confident results. Dead zones stack. Your controller's firmware may apply one, the driver may apply another, and only what survives both ever reaches the browser. A stick with a real resting offset of 0.04 sitting behind an upstream dead zone of 0.08 will report a perfectly flat 0.000 here, forever, on every browser.
So the tool reports the share of samples that were exactly zero, and if the entire run was zero it marks the result inconclusive, never clean. An all-zero run proves that nothing escaped the dead zones already in the chain. It does not prove the stick is centred, and this site will not tell you it does.
PadTest's drift bands
A raw number like 0.037 is hard to think about, so the tool attaches a word to it. Those words come from the scale below. This scale is a definition invented for this site. No standards body, manufacturer or repair guide set these boundaries — we did, so that the same reading always gets the same label. They are not a warranty threshold, not a fault diagnosis, and not a measurement of any population of controllers.
| Band | Offset magnitude | What it describes |
|---|---|---|
| Reads exactly zero | exactly 0.000 | Every sample was 0.000. Either the stick is genuinely still, or a dead zone applied by the driver, firmware or OS is hiding movement before the browser can see it. This test cannot tell those apart — see the note under the result. |
| Quiet | above 0.000 – 0.020 | A little jitter is reaching the browser, but it is smaller than the dead zone most games apply. Nothing here would move a camera. |
| Restless | above 0.020 – 0.050 | Visible in a readout, still under a typical game dead zone. Worth writing down so you can compare the same stick in a month. |
| Wandering | above 0.050 – 0.100 | Large enough to escape a small dead zone. This is the range where people start reporting slow camera creep in games that let you set the dead zone low. |
| Runaway | above 0.100 | A tenth of full deflection with nobody touching the stick. Games that expose a dead-zone slider can usually mask it; the underlying movement is still there. |
Does your drift actually matter? Check it against a dead zone
Drift only becomes a symptom when it escapes the dead zone applied on top of it. This table is the whole question in one grid: down the side, a resting offset; across the top, a circular dead zone. A tick means that dead zone hides that offset completely.
| Resting offset | d = 0.05 | d = 0.08 | d = 0.10 | d = 0.15 | d = 0.20 | d = 0.25 | d = 0.30 | Smallest that hides it |
|---|---|---|---|---|---|---|---|---|
| 0.010 | hidden | hidden | hidden | hidden | hidden | hidden | hidden | 0.05 |
| 0.020 | hidden | hidden | hidden | hidden | hidden | hidden | hidden | 0.05 |
| 0.035 | hidden | hidden | hidden | hidden | hidden | hidden | hidden | 0.05 |
| 0.050 | escapes | hidden | hidden | hidden | hidden | hidden | hidden | 0.08 |
| 0.080 | escapes | escapes | hidden | hidden | hidden | hidden | hidden | 0.10 |
| 0.120 | escapes | escapes | escapes | hidden | hidden | hidden | hidden | 0.15 |
| 0.200 | escapes | escapes | escapes | escapes | escapes | hidden | hidden | 0.25 |
Read your own measured offset off the tool, find the nearest row, and you know which games will show it. A stick measuring 0.035 is invisible to anything using 0.05 or more, and visible to a game that lets you drop the dead zone to 0.02 — which is exactly why players who lower their dead zone for responsiveness are the ones who notice drift first. The dead zone test lets you drag that threshold across your own live input and watch the transition happen.
How to run a test that is worth comparing
- Place the controller on a flat, still surface. A lap, a cushion or a hand adds movement you will later mistake for drift.
- Disconnect anything else that could nudge it — a charging cable under tension will lean a lightweight pad.
- Start the run and take your hands off entirely, including off the desk if it wobbles.
- Record all four numbers: max, mean, per-axis maxima, and the zero-sample share. The last one is what makes the others meaningful.
- Repeat after any change — recalibration, cleaning, a firmware update — under identical conditions. A single measurement is a data point; two under the same conditions are evidence.
If you are testing before buying second-hand, run this alongside the full gamepad tester so you cover buttons and triggers in the same sitting, and check the analog triggers separately on the analog trigger test — a trigger that no longer returns to zero is the same class of fault as a drifting stick and is missed by every button test.
What this page deliberately does not tell you
It publishes no drift figures of its own for any controller model. We have no controllers to measure, and a table of plausible-looking numbers for popular pads would be invention dressed as data. Any page ranking controllers by drift owes you a method, a sample size and a date; in their absence, the only number worth trusting about your controller is the one your controller just produced.