Debounce is the short window after a press during which firmware ignores the switch to hide contact bounce. Set it too long and fast double-taps vanish. Lock a key by pressing it, then double-tap and triple-tap it as fast as you physically can. The page records every press-to-press interval, finds the fastest one your keyboard accepted, flags intervals that look like a swallowed tap, and estimates the window. Run the keyboard latency test to see your result in seconds, right in your browser.
First key you press is locked in
-- ms press → press
Key --
Tap the key you want to test.
Press-to-press intervals (ms, 2 ms buckets; red zone = likely swallowed taps)
Keep double-tapping. Ten fast intervals are needed before an estimate is shown.
0 / 10 fast intervals
Values are common defaults, not guarantees; check your own board's documentation or configurator.
| Firmware / board type | Typical debounce | Note |
|---|
Your keyboard debounce test starts with one key and a few light taps: press a single letter about fifteen times, and the page flags rapid double keydowns that land too close together. Those flags show where stray double letters come from in your typing and gaming. This chatter check also works as a quick input test, with results in seconds.
The keyboard debounce test reads each key press as the browser receives it, so there is nothing to install on your keyboard or computer, and every input is logged for the session. Follow these steps: The rapid trigger test runs entirely in your browser, so nothing you press is stored or sent anywhere.
Each keydown that follows the previous one within about 25 ms is a bounce event. In a sample run on the J key, 15 taps produced a 9 ms and a 17 ms retap and a median hold of 84 ms: two flagged taps, or 13% of the run. The tool shows real-time events from the HID report, and the key press duration of each tap sits beside it. An 84 ms hold is normal; a gap that short between two presses is not. Run it on each suspect key, since a keyboard fault usually hits one key, and compare the keyboard against another computer.
Keyboard debounce is the filter that hides the flicker of a closing contact: the electrical signal jumps on and off for a moment. Mechanical switches do this, so the keyboard firmware waits out the flicker before reporting one press. If that wait is too short, you get duplicate characters and double characters inside a word, and the gap is only a few milliseconds long. Checking keyboard polling stability? The tool runs in your browser and takes about a minute.
The keyboard debounce time is the filter window. The scan rate is how often the key matrix is checked, a frequency in Hz. The polling rate, or report rate, is how often the keyboard reports to the PC. Together with the connection type they set your response time and system latency, and the keyboard latency you finally feel at the keyboard. Never confuse the three: a high polling rate cannot repair a short debounce window.
$$\text{Scan rate (Hz)} = \frac{1000}{\text{shortest press (ms)}}$$A shortest press of 2 ms gives 500 Hz. Higher scan rate means lower latency and a faster response time.
Characters you never pressed point to ghosting or limited rollover, so use a chord test for that. A repeated letter from one press is the opposite problem, and it causes missed taps and repeats in fast bursts. Keep the two apart before you replace a keyboard or buy a new keyboard.
Dana is deciding whether to return a keyboard after two weeks, because the E key keeps repeating a letter, as in "eexample". The decision hinges on one number: is this a worn switch or a setting? Dana taps E 20 times and the test flags 5 rapid double keydowns, so 5 of 20 taps, a 25.0% repeat rate. The shortest press reads 1.6 ms, which the calculator turns into 1000 ÷ 1.6 = 625 Hz.
Next come the latency inputs: a 1.00 ms polling interval, a 5 ms debounce window read from the keyboard's settings app, and 4 ms of system overhead. The latency figure this calculator produces is 5 + 1 + 4 = 10.0 ms, the bottom of the 10 to 15 ms band. That rules out a slow connection: the response is fine, the repeats are the fault.
The cheap experiment is to widen the debounce window to 8 ms in those settings, then rerun the same 20 taps:
| Debounce window | Flagged taps | Repeat rate | Total latency |
|---|---|---|---|
| 5 ms | 5 of 20 | 25.0% | 10.0 ms |
| 8 ms | 1 of 20 | 5.0% | 13.0 ms |
The number to watch is the trade: three extra points cut repeats by four fifths, and 13.0 ms is still under the 15 ms line where I would start suspecting the connection. I cross-checked the latency by hand and got the same 13.0 ms.
One caveat matters here. A longer debounce window also swallows genuine fast double taps, so a rhythm-game player could lose real presses before the band changes. The remaining flagged tap is the clue: a worn contact keeps repeating however wide the window gets, so Dana re-runs the test on a second key. If E alone still misbehaves at 8 ms, the return is justified.
A keyboard latency test adds the delay from filter, polling and system: 6 of debounce, 2 of polling and 4 of system gives 12 of latency in total. Lower input lag improves speed and performance, because every press reaches the screen sooner. An accurate reading needs repeated runs, so average at least ten presses against a benchmark rather than trusting one number.
| Latency (ms) | Responsiveness |
|---|---|
| Under 5 | Excellent for competitive play |
| 5 to 10 | Good for most players |
| 10 to 15 | Fine for typing |
| Over 15 | Check for lag |
Compare your own response time and latency against this table. A latency of 12 fits the third row: fine for typing, slightly slow for shooters. Latency that changes from run to run points to system load, while a steady high value points to the keyboard or the connection.
In FPS titles and rhythm games, a gaming keyboard under 5 of latency suits competitive gaming and esports. Slower keyboards are fine for everyday gaming, where a little extra lag goes unnoticed.
If a keyboard delay test or an input lag test looks poor, work through the cheap fixes first. Typists who see repeats should run a typing test afterwards, because the software and operating system can hide a real fault. Retest after each change so you know which one worked, and note the latency each time.
A wired connection over USB usually beats Bluetooth for latency. Compare a wired keyboard with a wireless keyboard, and try a 2.4GHz dongle before blaming your technique. Plug straight into the PC, not a hub, and keep the receiver close. If repeated letters appear only over a dongle, test the same key with a cable to separate interference from a worn switch.
A keyboard scan rate test in a browser measures what the software receives, not the hardware inside the keyboard, so treat its scan rate and polling rate as estimates. It cannot show factory debounce settings, and system latency adds noise. Use a dedicated tool for exact latency, and pair this test with the polling lab.
It watches one key for rapid double keydowns, the software-visible sign of switch bounce or chatter. It does not read the factory debounce setting inside the keyboard.
A single press that types two letters, or a test that flags two keydowns on the same key within about 25 ms, points to chatter. Repeat the test on several keys and on another computer.
Debounce time is the filter window that hides contact flicker. Scan rate is how often the keyboard checks its key matrix. Polling rate is how often it reports to the computer.
Under 5 ms is excellent for competitive play, 5 to 10 ms is good for most players, 10 to 15 ms is fine for typing, and over 15 ms is worth investigating.
Update keyboard drivers and firmware, turn off Filter Keys and Sticky Keys, compare a wired connection with Bluetooth, and retest after each change.
Holding a key triggers operating system auto-repeat, which is not switch chatter, so held keys are excluded from the bounce count.
It measures events the browser receives, not the keyboard's internal timing, so treat scan rate and latency as estimates and use hardware tools for exact figures.