Why does my fan emit a high-pitched whine, and how do I fix it?

A high-pitched whine that tracks fan speed usually has one of three sources: PWM switching in the audible band (below ~18 kHz), current ripple interacting with the motor's torque structure, or a duty region where the control loop and the motor's dynamics interact. Each has a distinct fingerprint and a cheap test — find the source before buying silence with a bigger enclosure.

Fingerprint the whine

PWM whine is a pure, steady tone at (or tied to) the switching frequency whose pitch stays constant as duty changes — audible drive frequencies are the giveaway, typically the host is running below 18 kHz. Ripple-related whine modulates with speed and changes character with duty. Loop-interaction whine appears only in certain duty windows and often coincides with visible duty or tach jitter on a log. A phone spectrum app separates these in minutes: constant-pitch = PWM; speed-tracking = ripple; window-limited = loop.

Fixes in cost order

(1) Raise the PWM frequency to 25 kHz if the fan supports the window — firmware only. (2) Clean the supply: ripple and shared-rail noise couple into the drive; a local bulk capacitor near the fan header is pennies. (3) Rate-limit duty slew to a few percent per second to remove stepping artefacts. (4) Re-map the control table to avoid parked duties where the tone excites structure. (5) Only then consider hardware: a fan with different commutation design, or acoustic damping at the mount — both are custom requests we handle routinely.

• Verify the fix at the customer's ear position, not beside the fan.
• Re-check that the fix survives minimum rail voltage and cold start.
• Document the final PWM frequency and slew rate in the interface spec.