Great topic, and I’ve spent way too many hours debugging this stuff on my Wooting and a couple of self-built magnetic boards. On calibration edge cases, the single biggest overlooked factor is physical switch wobble and magnet alignment rather than the sensor itself. When your keycap stem isn’t perfectly perpendicular to the plate, the magnet passes the Hall sensor at a slight angle, which skews the field readings across the travel curve. You’ll see asymmetric actuation points—say rapid trigger fires cleanly at 0.1mm down but the reset point drifts to 0.15mm on release—because the effective distance changes depending on direction of travel. The fix isn’t always software recalibration; sometimes it’s a loose switch housing or a deformed stabilizer wire. Also worth checking whether your board does per-switch calibration or global—if you’re using per-switch and one column reads consistently off, that’s almost always a weak magnet or a cold solder joint on the sensor, not a firmware issue.
On failure modes and debugging, start by ruling out thermal drift versus genuine sensor decay. Hall sensors shift with temperature (typically a few percent over normal room variance), so if your trigger points move predictably after 20 minutes of heavy use, that’s normal thermal behavior baked into the calibration curve—don’t chase
"Liquid metal applied on direct die cooler | 18°C temperature drop on load"
