SciSpinner Max — g-level analysis

Zero-g offset by sphere fit · offset-corrected time-averaged ⟨g⟩ · sphere coverage. Everything runs in this browser; your files are not uploaded anywhere.

Drop the run's files here (imu.csv, experiment.json, sensors.csv), or click to choose them.
Needs imu.csv. Optional: experiment.json (speeds), sensors.csv (temperature). Dropping the whole folder works where the browser allows it; its timelapse/ frames are never read.

Frame rpm Core rpm Used for the kinematic prediction (filled from experiment.json when present).

Run

Samples

—

Duration

—

Core temperature

—

Sphere coverage

—

Sensor zero-g offset (sphere fit)

Offset vector

—

g, sensor axes (x, y, z)

Offset magnitude

—

Sphere radius

—

Offset stability

—

spread of fitted centre across 30-min blocks

Time-averaged ⟨g⟩

Vector mean over the window, then magnitude (as the instrument computes it). Ranges are over all window positions after the first hour.

Window	Raw (as displayed)	Offset removed	Kinematic model, same speeds

Time-averaged ⟨g⟩ — raw vs offset-corrected

15-minute and 1-hour trailing windows, log scale. Dashed line: 10⁻³ g.

Raw sensorOffset removedKinematic model

Direction of gravity in the sample frame

Orthographic views of the unit sphere; every point is one sample (offset-corrected).

Method

Each raw sample is a_meas = a_true + b, where a_true is gravity in the rotating sensor frame (|a_true| = 1 g) and b is the accelerometer's fixed zero-g offset. During a run a_true sweeps the unit sphere, so the raw samples lie on a sphere of radius ≈ 1 g centred at b. The centre is found by linear least squares (|a|² = 2 b·a + (r² − |b|²)); an axis-aligned ellipsoid fit gives per-axis scale factors. b is subtracted from every sample before averaging. ⟨g⟩ is the magnitude of the vector mean over one-second block means (identical to the instrument's calculation). The kinematic model is Clary et al. (2022) Eq. 1 evaluated at the run's speeds with no sensor. Coverage counts 288 equal-area cells (12 bands in sin(lat) × 24 in longitude) visited by the corrected, normalised samples.