01 Browser input diagnostic
Test your mouse scroll wheel
Scroll in the test area below and this page reports exactly what your browser received — event counts, direction, and delta units. Free, private, no account.
Mouse wheel test
Not tested yet01Input type
Choose “Touchpad / Free-spin” for touchpads or free-spinning wheels — coasting and inertia are normal there and the wording stays cautious.
02Test direction
03Test duration
A longer duration provides a larger observation sample. It does not make the result inherently accurate.
04Active test area
Not tested yet — press Start Test, then scroll here
waiting for input
05Live result
Live graph: no input yet. Each tick is one received wheel event; tick height reflects delta size.
Run dashboard — browser-observed values
Every value below describes events this browser delivered. Units are shown beside each metric; mixed delta units are never summed in the analysis.
| Metric | Value | Unit |
|---|---|---|
| Duration (configured) | — | s |
| Sample count (events) | — | events |
| Event rate | — | events/s |
| Latest deltaY | — | delta units |
| Accumulated vertical input | — | delta units |
| Positive / negative peaks (deltaY) | — | delta units |
| DeltaMode mix (pixel / line / page) | — | events |
| Latest deltaX | — | delta units |
| Accumulated horizontal input | — | delta units |
| Expected-direction changes | — | count |
| Possible opposite-direction events | — | events |
Observation: Result
Sample quality
Test validity
Timing profile
Timing and magnitude profile of received events — descriptive only. Event gaps are measured from browser delivery times, not hardware polling.
View event table
View run summary
Run A / Run B
Use it in three steps
- Choose a mode. Pick vertical or horizontal scrolling, the standard wheel or touchpad setting, and a duration. Fifteen seconds suits a quick check; longer presets give a larger sample.
- Collect a controlled sample. Focus the pad, make three slow calibration scrolls, then keep one steady speed and direction until the timer ends.
- Interpret and compare. Read the observation, quality, and validity cards, then save the run and repeat it after any single change to compare like with like.
What was observed — and what was not measured
Each completed run observes the wheel events your browser delivered while the pad was focused: timestamps, deltaX, deltaY, and deltaMode. Nothing is captured before Start, after Stop, or while the pad is unfocused.
Not measured: USB polling rate, sensor scan rate, physical wheel notches, encoder condition, or whether anything was repaired. The page cannot see missed physical movement — only the events the browser chose to deliver.
Metrics guide
| Metric | What it means | What it cannot prove |
|---|---|---|
| Sample count | Wheel events the browser delivered during the run. | Not the number of physical wheel notches. |
| Event rate | Events per second of measured wall time. | Not hardware polling rate. |
| Latest deltaY / deltaX | The vertical / sideways size of the most recent event, in delta units. | Not scroll distance on screen. |
| Accumulated input | Signed sum of deltas on each axis. | Not distance; opposing inputs cancel out. |
| Peaks | Largest upward and downward single deltas. | A single tall peak is usually acceleration, not damage. |
| DeltaMode mix | How many events arrived as pixels, lines, or pages. | Units are never summed or converted. |
| Direction changes | Times the primary-axis sign flipped between consecutive events. | Cannot distinguish intent from hardware. |
| Opposite-direction events | Events against the requested (or dominant sideways) direction, with a rate. | A cause — smoothing and coasting are common. |
Methodology in brief
The tool listens for standard WheelEvent objects (see MDN's WheelEvent reference). Each event contributes deltaX, deltaY, deltaMode (0 = pixels, 1 = lines, 2 = pages), and a receipt timestamp. Ctrl+wheel zoom gestures are set aside. For horizontal runs the dominant sideways direction decides, and minority-side events count as opposite.
Sample-quality logic: fewer than 8 events is too few to interpret; a stopped run is a partial sample; any focus loss, hidden tab, or blur pauses timing and is counted as an interruption. Classification algorithm version: st-wheel-1. The full method, synthetic examples, and change log are on the How the Scroll Test Works page.
Troubleshoot by symptom
- No input: click inside the pad so it holds focus, keep this tab visible, and scroll with the wheel — not the page scrollbar. Still nothing? Try another browser, then another USB port.
- Intermittent input: check for interruptions in the validity card. Hidden tabs and lost focus pause the timer by design.
- Unexpected reversal: repeat the identical run. If the minority rate repeats across browsers, work through the jumpy-wheel decision path, one change at a time.
- Horizontal noise: many vertical wheels emit no deltaX at all — that is normal. Touchpad swipes coast after the fingers lift.
- Touchpad or free-spin inertia: use the Touchpad / Free-spin setting and expect trailing events; compare longer runs so the sample outlasts the coasting tail.
Scroll test FAQ
Press Start Test below, focus the test area, and scroll steadily in the requested direction for the full duration. The mouse scroll test counts the wheel events your browser actually received — no install, no account.
Opposite-direction input can come from touchpad inertia, free-spin coasting, OS or browser smoothing, a mid-run correction you made, or a worn encoder. One run cannot identify the cause — repeat the same scroll wheel test twice and see whether the pattern repeats.
Run the 15-second test below in each direction you use. If events arrive steadily with the expected direction and units, the browser path from wheel to page is working. If nothing arrives, check focus, try another browser, then another USB port.
It means one received wheel event pointed against the requested (or dominant sideways) direction. The result lists the count and rate. It is an observation, not a diagnosis — smoothing and coasting produce these routinely.
No. This scroll wheel tester reports browser-observed input and names non-hardware causes beside every result. Only a repeating pattern across browsers and machines makes hardware suspect, and even then this page offers no warranty verdict.
Yes — choose Touchpad / Free-spin before starting. Coasting after the fingers lift is normal there, so the wording stays cautious and trailing events are expected.
Yes. Select the Horizontal direction here or run the horizontal scroll test. Many vertical wheels report no sideways input at all, which is normal for that hardware.
It counts browser-delivered wheel events during a chosen period and reports events per second plus the median gap between events. It does not measure hardware polling rate and never claims to.
No. There is no backend. Events, saved runs, and preferences live only in your browser's local storage. Exported JSON files go only where you save them, and reports never appear in the URL.
Save the first run as Run A on the compare page, change exactly one condition (USB port, browser, extension state, OS setting, or cleaning step), then record Run B with identical settings. Compare two browser-observed runs locally. A difference describes input — it does not prove a repair.
Privacy and sources
Your scrolling never leaves this device. There is no server, no account, and no analytics payload carrying wheel input. Saved runs live in local storage until you delete them; exports are files you keep. Full details: Privacy.
Sources: MDN — WheelEvent · Google — SEO starter guide