Capture
Watch
Position — orientation of the watch, regardless of mic placement
Six-position testing follows COSC / Master Chronometer convention. Log each position separately to characterise the movement's positional variation.
Microphone & sensitivity
Tips for best accuracy
- A quiet room is essential — the tick is on the order of −40 dBFS.
- Hold the watch case firmly against the microphone (a phone's bottom mic works surprisingly well; a contact / piezo pickup is best).
- Let the watch settle for ~10 s after handling, then capture for 30 – 60 s.
- Test each position separately and log it — most movements vary 5–15 s/day between positions.
- If the trace is noisy: narrow the band-pass, raise sensitivity σ, or move closer to the mic.
- Auto-detect snaps to the nearest standard BPH; set it manually if your watch is exotic.
Long-running session (days)
For multi-day testing, keep the device plugged in, enable Wake Lock, and set Auto-log to 5–15 min. The lifetime metrics below use every detected beat since "Start" — they grow more precise the longer you run.
Lifetime — whole session, since "Start"
Session log
Beta workbench experimental
Extra views a Witschi-class machine shows beyond the three headline numbers. These are derived live from the same beat stream — handy, occasionally finicky, and meant for tinkering. Flip the switch to wake them up.
Sub-pulse scope one beat, magnified
The escapement fires three micro-impulses per tick. Amplitude comes from the unlock→impulse gap Δt. If the green marker keeps vanishing, the unlock pulse is below the noise floor — press the watch closer or add gain.
Positional spread from the session log
Log a reading in each of the six positions; the rosette plots rate per position and the spread (max−min) — the single most telling measure of a movement's regulation.
Isochronism rate vs amplitude
As the mainspring unwinds, amplitude falls. A flat line means the watch keeps time regardless of amplitude — the hallmark of a good hairspring. Builds up as you log readings over a wind-down.
Tools calibration & audio
Trust your amplitude on one reference position? Enter its true amplitude and snap the lift angle so the measurement matches — every other position then reads correctly.
The device's audio clock has its own quartz tolerance (typically <10 ppm ≈ <0.9 s/day). Measure a reference timepiece of known rate, enter that rate, and snap — the offset is stored on this device and applied to every rate reading. Set it back to 0 with Reset.
Re-synthesises the detected tick & tock as two soft clicks at the measured cadence and offset. A perfectly regulated watch sounds like an even gallop; beat error makes it limp.