What chatter actually is
Inside every switch, two metal contacts close when you press. They do not close cleanly — they bounce, making and breaking contact several times over a few milliseconds. Keyboard firmware hides this with a debounce window: after the first contact it ignores further transitions for a few milliseconds. A worn, dirty or oxidised switch bounces longer than the firmware expects, and one physical press leaks through as two keystrokes. You type "the", the screen shows "tthe". That is chatter — a hardware aging symptom, most familiar on heavily-used mechanical switches and old membrane domes.
Reading the gap numbers
| Same-key gap | Most likely explanation | Action |
|---|---|---|
| Under ~30 ms | Contact bounce — no human intent repeats a full press this fast | Treat as confirmed chatter |
| ~30-80 ms | Almost always chatter; extreme gaming taps can occasionally land here | Repeat the test typing normally — if flags persist, it is the switch |
| ~80-150 ms | Grey zone: very fast intentional double-taps live here | Only suspicious if you were not double-tapping |
| Over ~150 ms | Normal deliberate repetition ("ll" in "hello") | Not chatter |
Worked example: confirming a bad switch
Suspected key: T. Procedure: type the sentence "the fast attack started at ten" three times at normal
speed, then tap T alone twenty times at a relaxed pace. Result: four red flags, all on
KeyT, with gaps of 22, 28, 31 and 26 ms; zero flags on any other key. Interpretation: a
relaxed single tap cannot produce a second press 22 ms later, so the switch is double-firing. The
flags arriving during both sentences and single taps rules out typing technique. This is RMA/switch-swap
territory, and the logged gaps are your evidence.
Worked example: a false alarm
A gamer sets the threshold to 200 ms "to be safe", spams their dodge key, and sees a wall of red flags with gaps of 150-190 ms. Nothing is wrong with the keyboard: those gaps are simply their real double-tap speed, and a 200 ms window classifies deliberate taps as faults. Dropping the threshold back to 80 ms — or 50 ms for twitch-input testing — makes the flags vanish. The threshold field exists precisely because "impossibly fast for a human" depends on what the human is doing.
Fix list, cheapest first
- Clean: unplug, remove the keycap, blast compressed air; for mechanical switches a drop of contact cleaner into the stem area can revive an oxidised leaf.
- Firmware debounce: on QMK/VIA-capable boards, raise the debounce setting a few ms and retest here.
- Software masking: duplicate-keystroke filters hide chatter system-wide at the cost of fast legit double-taps — a bridge, not a cure.
- Swap the switch: on hot-swap boards this is a two-minute fix; soldered boards need an iron.
- Replace: membrane and laptop keyboards are not switch-serviceable — see the laptop keyboard test for triaging those.
After any fix, rerun this page and type a few paragraphs: zero flags at 80 ms is the pass condition. If your problem is dropped keys in combos rather than doubled letters, that is a different fault — measure it with the key rollover test. A chattering spacebar shows up dramatically in the spacebar test as an impossible presses-per-second rate. Full key-by-key coverage lives on the main keyboard tester.