Hardware
TesterMagnetometer Tester
Test your phone magnetometer sensor: live microtesla X/Y/Z readings, check against Earth's 25-65 µT field, spot hard-iron interference, plus CSV/JSON export.
Have feedback? Report bugs, suggest features, or share your thoughts — we read them allAbout Magnetometer Tester
Magnetometer Tester visualizes the Earth's magnetic field in real time. Monitor X, Y, and Z axis readings, view the combined field strength on a live chart, and track session min/max/average to judge a sensor against the expected 25-65 µT range. The compass needle shows a magnetic, tilt-sensitive heading (accurate only when the device is flat, uncalibrated and not corrected for declination), so treat it as a quick indicator rather than a precise navigation compass. Export the full reading history to CSV or JSON for QA and repair records.
- Press Start testing and allow magnetometer access when prompted by the browser (HTTPS required).
- Hold your device flat to see the baseline magnetic field strength and compass heading.
- Move the device near magnets or metal objects to observe spikes on the realtime chart.
- Watch the session min / max / average to compare the total field against the 25-65 µT Earth-field range.
- Use Calibrate compass to zero the heading, then Export CSV or JSON to archive the test, and Reset to start fresh.
Frequently Asked Questions
A magnetometer measures the strength and direction of the local magnetic field around your device along three orthogonal axes (X, Y, Z). The unit is microteslas (μT) in SI or milligauss (mG) in CGS — 1 μT = 10 mG. Earth's magnetic field at the surface ranges from about 25 μT near the equator to 65 μT near the poles, with a horizontal component of 20–40 μT that gives compasses their direction. Phones use a tiny three-axis Hall-effect or magnetoresistive (AMR/GMR) chip — like the AKM AK0991x or ST LIS3MDL — that resolves field components down to ~0.15 μT. Beyond compass apps, the magnetometer enables AR orientation, metal detection, and tools like this tester.
Magnetometer faults are subtle and usually only show up when compass apps point in the wrong direction or when AR experiences yaw silently. Causes include hard-iron contamination from a phone case with magnets (MagSafe rings, kickstands), proximity to laptop speakers, recalibration corruption after a firmware update, and physical damage to the sensor itself. Testing reveals if all three axes respond, if the total field magnitude matches the expected 25–65 μT range for your location (look up your value at NOAA's magnetic field calculator), and if rotating the phone produces smooth field readings. Indoor steel-frame buildings often distort readings by 20–100 μT, which this tester also helps visualize.
Earth's surface field magnitude varies from 25 μT near the magnetic equator to about 65 μT near the magnetic poles. At a typical mid-latitude location like New York, Madrid, or Tokyo, expect roughly 45–55 μT total magnitude with a horizontal component of 20–25 μT. Indoors, steel structures, electric appliances, and concrete rebar can push readings well outside this range — values of 100–500 μT are common near a fridge or speaker. Strong neodymium magnets can saturate the sensor at 1000+ μT. The reference value for your exact GPS coordinates can be looked up using the World Magnetic Model (WMM2025), updated every five years by NOAA/BGS, and is what calibration apps compare your reading against.
Two effects distort compass headings: hard-iron and soft-iron interference. Hard-iron interference comes from permanent magnets near the sensor (case magnets, MagSafe rings, even some screen protectors) and adds a constant offset to the X/Y/Z readings — your data points in 3D form a sphere displaced from origin. Soft-iron interference comes from ferromagnetic materials (steel beams, refrigerator doors, car frames) that distort the local field and turn that sphere into an ellipsoid. Calibration apps ask you to rotate the phone in a figure-eight pattern to map out the displacement and stretch, then subtract them in software. Re-do calibration whenever you change phone cases, and always test outdoors away from steel buildings for a true reading.
Hard-iron offset shifts each axis by a constant — if a small magnet sits 1 cm from the sensor, you might see +30 μT on X regardless of orientation. The fix is bias estimation: collect samples while rotating the phone in all directions, compute the center of the resulting point cloud, and subtract those values from every future reading. Soft-iron distortion scales the readings differently along different axes, deforming the calibration sphere into an ellipsoid. The fix is a 3×3 transformation matrix derived by fitting an ellipsoid to the data and inverting it. Modern phones run this calibration continuously in the background, but rapid environmental changes (entering a car, picking up a magnetic case) require a fresh figure-eight motion to re-anchor the offsets.
The World Magnetic Model (WMM2025, currently valid through 2029) is a mathematical description of Earth's main magnetic field, updated every five years by NOAA's National Centers for Environmental Information and the British Geological Survey. It tells you the expected field strength, declination (the angle between magnetic north and true north), and inclination at any latitude/longitude/altitude on Earth. Phones use the WMM to convert raw magnetometer readings into true compass headings — without it, a compass pointing to magnetic north would be off by up to 30° in some regions. Declination drifts ~0.1° per year, which is why the model needs periodic updates. You can look up your local declination at the NOAA Magnetic Field Calculator.
Two APIs are involved. The legacy DeviceOrientationEvent fires alpha values that are fused with magnetometer data to produce a compass heading in degrees — this works on iOS Safari (with permission) and Android Chromium. The newer Generic Sensor API exposes Magnetometer and AbsoluteOrientationSensor classes that return raw μT values per axis along with a calibration accuracy indicator (LOW/MEDIUM/HIGH). Generic Sensor is currently supported on Chromium-based browsers behind a flag and via the Sensor permission. On iOS, raw magnetometer access is not exposed to the web — only the fused compass heading from DeviceOrientationEvent.webkitCompassHeading. This tool detects available APIs and falls back gracefully.
Consumer three-axis magnetometers (AKM AK0991x, ST LIS3MDL, Memsic MMC5983) are characterized by full-scale range (±1300 μT to ±4900 μT typical), resolution (0.15–0.6 μT/LSB), noise density (0.3–1.5 μT RMS), and zero-field offset stability over temperature. ISO 12000 covers magnetic field measurement procedures, and IEEE 1451.4 standardizes the smart-sensor interface. The reference field model is the WMM2025 maintained by NOAA/BGS, against which all consumer compasses are calibrated. For comparison, scientific-grade fluxgate magnetometers used by geophysicists reach 0.01 nT resolution — about 15,000 times more sensitive than a phone — but cost thousands of dollars. For consumer use, phone magnetometers easily meet the IATA aviation compass accuracy spec of ±5° heading.
Raw three-axis microtesla values come from the Generic Sensor API's Magnetometer class, which is exposed on Android Chromium (Chrome, Edge, Brave, Samsung Internet) — there you see live X, Y, Z and total field in µT. iOS Safari does NOT expose the raw magnetometer to the web; it only provides a fused compass heading via DeviceOrientationEvent.webkitCompassHeading, so on iPhone/iPad this tool runs in heading-only fallback mode (no µT axes). All sensor access requires a secure context: the page must be served over HTTPS (or localhost), otherwise the browser blocks it with a SecurityError. The first time you press Start, the browser shows a permission prompt (and on iOS 13+ a motion-and-orientation prompt) that you must allow; if you deny it or are on plain HTTP, the tool reports the exact reason instead of failing silently.
Yes, it is fully private. Every reading is processed on-device inside your browser: the X/Y/Z values, magnitude, compass heading, statistics and the recorded session log all live in page memory only. Nothing is uploaded to any server, and there is no account, tracking pixel, or telemetry on the sensor stream. When you press Export, the CSV or JSON file is generated locally with a Blob and downloaded directly to your device — it never passes through the network. Closing or reloading the tab discards the in-memory session, so export first if you need to keep the record.
Each axis is the magnetic field component along one device direction in microteslas (µT): X is across the screen (device's right is positive), Y runs up the screen toward the top edge, and Z points out of the screen. Individual axes swing positive or negative as you rotate the phone — that is normal. The number that matters for a health check is the total field magnitude, sqrt(X²+Y²+Z²), shown in the Magnetic field strength badge. Lay the phone flat, away from metal and magnets, and it should settle near your local Earth-field value (roughly 25–65 µT, about 45–55 µT at mid-latitudes). A healthy sensor changes all three axes smoothly as you turn the device while the total magnitude stays roughly constant. A flat-lined axis, a magnitude stuck far outside 25–65 µT, or wild jumps with no movement point to a hard-iron offset (a magnet/case nearby) or a faulty sensor.
During a run the tool records every sample with a timestamp and tracks the min, max, and average of the total field magnitude plus the sample count. To grade a sensor: hold the device still, away from metal, for a few seconds and read the average — for a pass it should sit inside the 25–65 µT Earth-field band (compare against your exact location using NOAA's WMM calculator), with min and max staying close together (a tight spread means low noise). A wide min–max gap on a motionless device indicates noise or interference; an average pinned well above 65 µT with the phone still suggests a hard-iron offset from a magnetic case. Press Export CSV (timestamp_iso, x_uT, y_uT, z_uT, magnitude_uT, heading_deg) or Export JSON to download the full reading history for repair tickets and before/after comparisons. Export stays available after you Stop so you can archive a finished test, and is disabled only when no data has been captured.
SENSORS13
WUTOOLS