Your controller is probably not broken. In most cases, a controller that fails on a PC is being claimed by the wrong software layer, reporting under an input protocol the game does not read, or dropping off a USB bus that selectively suspends it. Three subsystems must agree before a button press reaches a game: the operating system's driver stack, the game's input application programming interface (API), and the controller's own connection state. When any one disagrees, you get a dead stick, doubled inputs, or a device that works in one launcher and not another. This guide explains each layer, the numbers behind common failure rates, and the exact tests that isolate the fault.

The driver stack and how a controller identifies itself

Windows does not know what a "controller" is. It knows what a device reports over USB or Bluetooth descriptors. When you plug in a pad, the USB host reads a set of descriptors that declare the device class, endpoints, and report format. Windows then matches a driver: either the class-compliant HID driver (hidusb.sys plus the HID class driver) or a vendor-supplied driver for devices that use proprietary report formats.

This matters because the same physical controller can enumerate in different modes. An Xbox controller exposes XInput, an API layered over HID that normalizes report formats so games can read a fixed 20-byte state structure. A DualSense in default mode exposes a generic HID device with a different report layout. A Switch Pro Controller exposes yet another. The driver stack converts raw reports into a device path, and the game queries that path through DirectInput, XInput, or a newer API. If any link in that chain fails, the controller appears in Device Manager but produces no input in the game.

The most common single cause is a driver that reports the device as "Unknown device" or with a yellow warning icon. That indicates the descriptor was read but the driver binding failed, often after a Windows update replaced a vendor driver with a generic one. Reinstalling the vendor driver, or removing the device and rescanning, restores the binding in the majority of these cases.

Protocol mismatch: XInput, DirectInput, and per-game API support

The second most common cause is a protocol the game does not read. Games do not read controllers directly; they read an input API. Older titles built on DirectInput see generic HID devices. Newer titles built on XInput see only devices that expose the XInput interface. A game built exclusively on XInput will not see a DirectInput-only controller, and a game built on DirectInput may see an XInput controller as a generic HID device with an unfamiliar mapping.

The result is a predictable pattern: the controller works in one game and not another, or works in Steam but not in a standalone launcher. Steam Input adds a translation layer that presents any detected controller as an XInput device to the game, which is why "works in Steam, dead elsewhere" is a common report. Third-party tools such as DS4Windows and reWASD do the same translation outside Steam. The fix is not to install more layers but to identify which API the game expects and which interface your controller currently exposes.

Symptom Likely layer First test
Appears in Device Manager, no input anywhere Driver binding Check for warning icon; reinstall vendor driver
Works in Steam, not in a standalone game API translation Enable or disable Steam Input for that game
Works wired, not wirelessly Connection state Re-pair; check Bluetooth adapter
Input doubled or drifting Report format or calibration Test in joy.cpl; recalibrate
Disconnects after seconds or minutes Power management Disable USB selective suspend
Works in one game, dead in another Per-game API support Check game's controller support notes

Quantified failure patterns and time courses

Controller failures are not random. They cluster around specific events: driver updates, OS updates, Bluetooth re-pairing, and physical wear. A first-generation analog stick uses potentiometers that degrade with use; drift typically emerges after hundreds of hours of play, and the threshold at which it becomes noticeable is roughly 10 to 15 percent of full deflection at rest. Hall-effect sticks replace potentiometers with magnetic sensing and do not develop this wear pattern, which is why they are marketed as drift-resistant.

Disconnects follow a different time course. A controller that drops after 10 to 30 seconds of idle is usually being suspended by USB selective suspend or by the Bluetooth adapter's power-saving policy. A controller that drops after 5 to 15 minutes of active play is more often a battery or interference issue. A controller that drops only during specific games is often a conflict with another input device or a background overlay.

Time to failure Typical cause Diagnostic step
Immediate, never detected Cable, port, or pairing Try a different cable and port
10–30 seconds idle USB selective suspend Disable in Power Options
5–15 minutes active Battery or 2.4 GHz interference Test wired; move adapter
After OS or driver update Driver binding replaced Roll back or reinstall driver
After hundreds of hours Stick wear (potentiometer) Test deadzone in joy.cpl
Only in one game API or overlay conflict Disable overlays; check API

The counterintuitive detail: a controller that tests perfectly in the Windows game controllers panel (joy.cpl) can still fail in a game, because joy.cpl reads the raw HID report while the game reads through a different API. Passing the panel test proves the driver layer works. It does not prove the game's API layer works.

Edge cases that mimic hardware failure

Several conditions produce symptoms indistinguishable from a dead controller but originate elsewhere. USB port power delivery is one: a port that supplies less than the standard 500 mA for USB 2.0 or 900 mA for USB 3.0 can cause intermittent disconnects, particularly with controllers that draw power for rumble motors or lighting. A powered USB hub resolves this.

Bluetooth interference is another. The 2.4 GHz band is shared with Wi-Fi, and a controller paired over Bluetooth can drop when the adapter is congested or when the controller is more than a few meters from the adapter with obstructions. Wired testing isolates this in under a minute.

Overlay software is a third. Steam, Discord, and GPU vendor overlays can capture input before the game sees it. A controller that works with overlays disabled but not enabled points to overlay capture, not hardware.

Red flag: the controller appears as "Unknown device" in Device Manager with a warning icon. Stop here and fix the driver before testing anything else. Red flag: the controller works wired but not wirelessly, and re-pairing fails repeatedly. The Bluetooth adapter or its driver is the likely fault, not the controller. Red flag: inputs are doubled or a stick drifts at rest in joy.cpl. This is a report-format or calibration issue, not a game bug. Red flag: disconnects begin immediately after a Windows update. Check for a replaced driver before replacing hardware.

Misconceptions that waste time and money

The most expensive misconception is that a controller that fails in one game is defective. Protocol mismatch explains the majority of single-game failures, and the fix costs nothing. Buying a replacement controller before testing the API layer is the most common wasted purchase in this category.

A second misconception is that third-party controllers are universally incompatible. Many third-party pads expose XInput correctly and work without translation. The variable is the firmware's reported interface, not the brand. A controller that reports XInput works in XInput games regardless of who made it.

A third misconception is that "reinstalling the game" fixes controller issues. It almost never does, because the fault is in the driver or API layer, not the game's files. Reinstalling the game resets settings but does not rebind the driver or change the API.

A professional caveat worth stating plainly: firmware updates can change a controller's reported interface. A pad that worked in XInput mode can switch to a different mode after a firmware update, which is why "it worked yesterday" is compatible with a protocol change.

Practical diagnostic sequence

Work from the bottom of the stack upward. Each step isolates one layer, so a failure at any step tells you where the problem lives.

Test in joy.cpl. Open the Windows game controllers panel and confirm the device appears and responds. If it does not appear, the driver layer is failing; reinstall the vendor driver and rescan. Test wired with a known-good cable and a rear motherboard port. If wireless fails but wired works, the fault is the Bluetooth adapter or pairing, not the controller. Disable USB selective suspend. In Power Options, set USB selective suspend to disabled, then retest idle disconnects. Identify the game's API. Check whether the game supports XInput, DirectInput, or both, then confirm your controller exposes the matching interface. Test with overlays disabled. Close Steam, Discord, and GPU overlays, then retest. Recalibrate or reset the controller. Use the vendor's calibration tool if drift or doubled inputs appear. Update firmware last. Apply a firmware update only after the software layers are ruled out, because updates can change the reported interface.

Thresholds to reconcile: a disconnect within 30 seconds of idle points to power management, not hardware. A disconnect after 5 to 15 minutes of active play points to battery or interference, not the driver. Drift that appears only after hundreds of hours points to stick wear, and a deadzone adjustment is a temporary workaround, not a repair. These thresholds overlap at the edges, so test in the order above rather than guessing.

Limits of what this can tell you

No diagnostic sequence can distinguish a failing USB controller chip from a failing cable without substitution testing, and no software test can detect a marginal solder joint that fails only under heat. When the sequence above clears every layer and the controller still fails, substitution is the only remaining test: a different controller on the same port, and the same controller on a different PC. That isolates the hardware in two steps.

The mechanisms here are stable and do not depend on current prices or firmware versions. Specific vendor tools, driver version numbers, and the exact features of any given controller change over time, so verify those against the manufacturer's current documentation rather than treating any version detail as permanent. The diagnostic logic, however, holds: driver layer, then protocol layer, then connection state, then hardware.