Reads your flight stick, throttle and rudder pedals at once and checks every axis, the hat switch and every button.
Connect each device by USB, open this page in Chrome, Edge or Firefox, then move a stick or press a button on each one. Windows can also list them under “Set up USB game controllers” if a device does not appear.
This is a controller. The gamepad tester and the stick drift test are made for it.
This looks like a racing wheel. The steering wheel tester reads its rotation and pedals.
One direction of stick movement, reported from -1 to 1. Each stick has two: left/right and up/down.
A small area around the stick centre that the game ignores, so a resting stick does not move the camera. Too small and drift leaks through; too large and slow aiming disappears.
Browsers only hand a device to a page after it has been used. Move the stick or press a button with the page open; each device on the desk appears as its own card.
As little as hides the rest offset and the jitter. The bench measures both for every centring axis while you let go, and gives the smallest deadzone that covers them.
Move an axis slowly end to end and read its resolution. The browser scales every axis to the same range, but the steps between values give the real count away.
A flight desk is several USB devices — the stick, the throttle, the pedals — and the browser lists each one with its axes and buttons but no names for them. The bench names the device from its USB ids, gives each axis a role from where it rests (a stick axis centres, a throttle sits at one end, a hat parks outside the range), then measures it while you fly it about: how far it reaches, where it settles when you let go, how much it wanders at rest, how fine its steps are and whether it jumps. It also traces the stick, lights the hat switch’s eight directions and counts every button press, catching a switch that bounces.
Measurement boundary: The browser sees the axes after the device’s own firmware and driver: software curves or deadzones set in the maker’s app are already applied, and a driver that smooths the signal hides its true resolution, in which case the bench says “at least”. Browsers list only as many axes and buttons per device as they support, so a few very large throttles show fewer buttons here than in the maker’s software. Force feedback is not tested.
How far an axis travels. A centring axis should reach both ends; a throttle or pedal should travel end to end. Under 95% means part of the range is lost — recalibrate.
Where a stick axis settles when you let go. A few tenths of a percent is normal; several percent off centre means the aircraft drifts unless a deadzone hides it.
How much an axis wanders while untouched. Hall-effect sensors barely move; a worn potentiometer can wander by a percent or more, which a game reads as a twitchy aircraft.
How many steps an axis has, shown in bits. 10 bits is 1,024 steps, 16 bits is 65,536; more steps mean finer control near the centre.
A value that jumps and comes straight back is a dirty or worn sensor. A button that registers twice in a few milliseconds is a worn switch.
Some throttles and HOTAS bases enumerate as separate devices for the base and the grip; that is normal. If one is missing, check Windows “Set up USB game controllers” and try another USB port, ideally not a hub.
Some throttles have a centre detent or report a split range. The bench names the axis from where it rests; the travel and reach it measures are still right.
Clean or replace the potentiometer if the maker supports it, or set a deadzone of the size the bench suggests in the game or the maker’s software. Hall-effect upgrades remove jitter for good.
Browsers cap the axes and buttons they list for one device. The maker’s software can test the rest, or map extra buttons onto a second, virtual device.
Open this page in Chrome, Edge or Firefox, move your stick, throttle and pedals so each appears, then move every axis end to end, push the hat and press every button. Each device gets its own report.
Yes. The bench reads it like any other flight device: the stick axes, the twist, the throttle on its base and the rocker, the hat and every button, and names it from its USB ids.
Move the stick around, then let go and keep your hands off for a second. The rest value of each axis is how far it settles from centre; the bench flags anything over 3% and suggests a deadzone.
Yes. Pedals are their own USB device; the bench reads the rudder axis and both toe brakes and checks that each travels end to end.
Calibration tells Windows the real ends and centre of each axis. If an axis reaches only part of its range here, recalibrating in “Set up USB game controllers” or the maker’s software usually restores it.
Move from observation to a narrower diagnostic without losing context.
Turn the wheel and press the pedals: a guided check names every axis, then measures centre, real rotation, resolution, pedal range and spikes, and says what to change.
02Watch every button, stick and trigger light up on a drawn PS5, Xbox or Switch pad, then test vibration, report rate and stick circularity — nothing to install, nothing uploaded.
03Measure the observed center offset, maximum movement and stability of both analog sticks with a repeatable browser test.
04 DesktopRead the raw HID reports of a DualSense or DualShock 4: stick centres to the byte, drift percentage, gyro bias, report rate, battery and touchpad, read-only over WebHID.
I answer from this site only — the 28 browser tests, the controls library for 46 games, the sensitivity converters and the two games. I can tell you which page to open, how a tool works and what its result means.
I will not guess. If something is not on the site, I will say so.
Answers come from this site's pages.