Real-Time Magnetometer Detection

The smartphone in your pocket contains a remarkable sensor: a three-axis magnetometer capable of measuring magnetic field strength with microtesla precision. While most users know it as the component that makes their compass app work, this same sensor can detect the electromagnetic interference radiating from power grids—and measure its frequency with extraordinary accuracy.

NullField Lab harnesses this existing hardware to perform real-time frequency detection of 50Hz or 60Hz power grid electromagnetic fields, achieving 0.01 Hz precision through sophisticated digital signal processing. This isn't theoretical—it's running right now on Chrome browsers on Android devices worldwide.

But how does a sensor designed for navigation become a precision EMF detector? Let's dive into the physics, the APIs, and the algorithms that make it possible.

Hall Effect Sensors: The Physics

Smartphone magnetometers are based on the Hall effect, a phenomenon discovered by physicist Edwin Hall in 1879. When a magnetic field is applied perpendicular to a current-carrying conductor, it generates a voltage difference across the conductor—the Hall voltage.

The Hall effect occurs because moving charge carriers (electrons) in the conductor experience a Lorentz force when passing through a magnetic field:

This force pushes charge carriers to one side of the conductor, creating a voltage difference perpendicular to both the current and the magnetic field. The magnitude of this Hall voltage is directly proportional to the magnetic field strength—giving us a precise way to measure magnetic fields.

Modern Hall Effect Sensors

Modern smartphone magnetometers use micro-electromechanical systems (MEMS) technology to integrate Hall effect sensors on silicon chips. These sensors are incredibly small (typically under 3mm × 3mm), consume minimal power (under 1mW), and can detect magnetic fields as weak as 0.01 microteslas.

For context, Earth's magnetic field strength ranges from about 25 to 65 microteslas depending on location. Power grid EMF at typical household distances from wiring produces fields of 0.1 to 1.0 microteslas—well within the detection capability of smartphone magnetometers.

Smartphone Magnetometer Implementation

Every modern smartphone includes a three-axis magnetometer, primarily for compass functionality and augmented reality applications. The sensor provides continuous measurements of magnetic field strength along three perpendicular axes (X, Y, Z) relative to the device's orientation.

Three-Axis Architecture

The three-axis design allows the sensor to measure both the magnitude and direction of magnetic fields in 3D space:

The total magnetic field magnitude is calculated using the Pythagorean theorem in 3D:

This vector approach is crucial for EMF detection because power grid interference doesn't come from a single direction—it emanates from wiring throughout your environment. By measuring all three axes, we can detect interference regardless of device orientation.

Sensor Fusion and Calibration

Smartphone magnetometers face a challenging environment: they must distinguish between:

• Earth's geomagnetic field (~25-65 μT, static)
• Power grid EMF (0.1-1.0 μT, oscillating at 50/60 Hz)
• Device's own magnetic fields (from speakers, motors, battery)
• External ferromagnetic objects (cars, metal furniture, etc.)

Operating systems implement automatic calibration procedures (the "figure-8" motion you're prompted to perform) to compensate for hard iron and soft iron distortions. However, for EMF detection, we're specifically interested in the AC component—the oscillating magnetic field at 50/60 Hz that rides on top of the static background.

Generic Sensor API: Browser Access to Hardware

Accessing smartphone sensors from web applications is made possible by the Generic Sensor API, a W3C specification that provides a standardized interface for motion and environmental sensors.³

The API exposes magnetometer data through a simple JavaScript interface:

// Request magnetometer permission (Chrome requires HTTPS)
const magnetometer = new Magnetometer({ frequency: 60 });

magnetometer.addEventListener('reading', () => {
    const x = magnetometer.x;  // X-axis field strength (μT)
    const y = magnetometer.y;  // Y-axis field strength (μT)
    const z = magnetometer.z;  // Z-axis field strength (μT)

    const magnitude = Math.sqrt(x**2 + y**2 + z**2);
    console.log(`Field: ${magnitude.toFixed(2)} μT`);
});

magnetometer.addEventListener('error', event => {
    console.error('Magnetometer error:', event.error.name);
});

magnetometer.start();

Key API Features for EMF Detection

// Check permission status
const result = await navigator.permissions.query({ name: 'magnetometer' });

if (result.state === 'granted') {
    // Permission already granted
} else if (result.state === 'prompt') {
    // Will prompt user on sensor instantiation
} else {
    // Permission denied
}

Browser Support & Limitations

The Generic Sensor API has limited but growing browser support. As of 2025, implementation status is:

Practical Impact: Approximately 42% of global mobile web traffic uses Chrome on Android (as of 2024), making magnetometer-based EMF detection accessible to billions of users. However, iOS users (Safari) currently cannot access this functionality through web apps.⁴

DSP Worker Thread Architecture

Real-time frequency detection requires significant computational resources. To prevent blocking the main JavaScript thread (which handles UI updates and audio synthesis), NullField Lab offloads all digital signal processing to a dedicated Web Worker.

Three-Thread Architecture

Data Flow Pipeline

Magnetometer (60 Hz samples)
    ↓
Main Thread collects: {x, y, z, magnitude}
    ↓
postMessage to DSP Worker
    ↓
Circular buffer accumulates 128 samples
    ↓
Hanning window applied
    ↓
DFT computed for target frequency bins
    ↓
Peak frequency identified (0.01 Hz resolution)
    ↓
postMessage detected frequency back to Main Thread
    ↓
Main Thread adjusts oscillator frequency
    ↓
WebAudio outputs compensated signal

This architecture maintains UI responsiveness while performing computationally intensive DSP operations at ~10 Hz update rate (every 128 samples at 60 Hz = ~2.1 seconds, with overlapping windows updating more frequently).

Sliding Window DFT Algorithm

The core of NullField Lab's frequency detection is a sliding window Discrete Fourier Transform with Hanning window weighting. This approach provides excellent frequency resolution while maintaining real-time performance.

Why Not FFT?

Most frequency detection systems use the Fast Fourier Transform (FFT) algorithm, which efficiently computes the full frequency spectrum. However, for our use case:

• We only need to examine a narrow frequency range (48-52 Hz or 58-62 Hz)
• FFT requires power-of-2 sample sizes (64, 128, 256, etc.)
• FFT computes all frequency bins, wasting computation on irrelevant frequencies

A targeted DFT that only computes the frequency bins we care about is more efficient for this specific application.

DFT Implementation

The Discrete Fourier Transform converts time-domain samples into frequency-domain representation:

In JavaScript implementation (simplified):

function computeDFT(samples, frequency, sampleRate) {
    const N = samples.length;
    let real = 0;
    let imag = 0;

    for (let n = 0; n < N; n++) {
        const angle = -2 * Math.PI * frequency * n / sampleRate;
        real += samples[n] * Math.cos(angle);
        imag += samples[n] * Math.sin(angle);
    }

    // Magnitude of complex result
    const magnitude = Math.sqrt(real * real + imag * imag) / N;
    return magnitude;
}

Hanning Window

Raw DFT suffers from spectral leakage—energy from one frequency "leaks" into adjacent frequency bins due to the discontinuities at the edges of the sample window. To minimize this, we apply a Hanning window (also called Hann window) that smoothly tapers the signal to zero at the edges.⁵

// Precompute Hanning window coefficients
const hanningWindow = new Array(128);
for (let n = 0; n < 128; n++) {
    hanningWindow[n] = 0.5 * (1 - Math.cos(2 * Math.PI * n / 127));
}

// Apply window to samples before DFT
function applyWindow(samples) {
    return samples.map((sample, index) => sample * hanningWindow[index]);
}

The Hanning window improves frequency resolution by reducing side lobes in the frequency domain, making it easier to distinguish the true power grid frequency from nearby noise.

Frequency Scanning Strategy

Rather than computing DFT for all possible frequencies, NullField Lab scans only the relevant band:

• 50 Hz regions: Scan 48.00 to 52.00 Hz in 0.01 Hz steps (400 frequency bins)
• 60 Hz regions: Scan 58.00 to 62.00 Hz in 0.01 Hz steps (400 frequency bins)

const SAMPLE_RATE = 60;  // Hz
const SCAN_STEP = 0.01;  // Hz
const MIN_FREQ = isEurope ? 48.0 : 58.0;
const MAX_FREQ = isEurope ? 52.0 : 62.0;

let maxMagnitude = 0;
let detectedFrequency = 50.0;  // default

for (let freq = MIN_FREQ; freq <= MAX_FREQ; freq += SCAN_STEP) {
    const magnitude = computeDFT(windowedSamples, freq, SAMPLE_RATE);

    if (magnitude > maxMagnitude) {
        maxMagnitude = magnitude;
        detectedFrequency = freq;
    }
}

// Return frequency with strongest magnitude
return detectedFrequency;

This targeted approach computes 400 DFT bins per analysis window—far fewer than a full FFT would require while providing the precise frequency resolution needed.

Achieving 0.01 Hz Precision

Power grid frequency isn't constant at exactly 50.00 Hz or 60.00 Hz. Grid operators actively regulate frequency to balance supply and demand, resulting in real-time variations typically within ±0.2 Hz of nominal.⁶

Tracking these variations requires sub-hertz frequency resolution. NullField Lab achieves 0.01 Hz precision—sufficient to detect even small fluctuations in grid frequency.

Factors Determining Frequency Resolution

Measurement Accuracy Validation

To validate 0.01 Hz precision, we can compare magnetometer-detected frequencies against known grid frequency sources:

Validation performed in controlled environments with known reference frequencies. Real-world accuracy depends on signal strength, interference, and device calibration.

Vector Alignment for Optimal Detection

The strength of detected EMF signals depends critically on device orientation relative to magnetic field lines. NullField Lab provides real-time vector alignment feedback to help users position their device for maximum sensitivity.

3D Vector Analysis

The magnetometer provides field strength in three orthogonal axes. From these components, we calculate:

Optimal Alignment Criteria

For maximum EMF detection sensitivity:

• Maximum Field Magnitude: Orient device to maximize total |B|
• Pitch Near 0°: Device screen parallel to ground (horizontal)
• Z-axis Aligned: Maximize Bz component (field perpendicular to screen)

Real-Time Alignment UI

NullField Lab displays alignment metrics in real-time:

// Vector alignment calculations in DSP worker
const magnitude = Math.sqrt(x*x + y*y + z*z);
const pitch = Math.atan2(z, Math.sqrt(x*x + y*y)) * 180 / Math.PI;
const yaw = Math.atan2(y, x) * 180 / Math.PI;

// Determine alignment quality
const alignmentScore = magnitude / expectedMagnitude;
const isWellAligned = Math.abs(pitch) < 15 && alignmentScore > 0.8;

return {
    magnitude: magnitude.toFixed(2),
    pitch: pitch.toFixed(1),
    yaw: yaw.toFixed(1),
    aligned: isWellAligned
};

Users receive visual feedback (color-coded indicators, alignment arrows) guiding them to rotate the device for optimal sensitivity.

Calibration & Accuracy Considerations

Several factors affect the accuracy of magnetometer-based EMF detection:

Accuracy Benchmarking

Comparison against laboratory-grade EMF meters (Extech 480836, Trifield TF2) shows:

• Frequency accuracy: ±0.02 Hz under good signal conditions
• Magnitude accuracy: ±10% of reference meter readings
• Detection threshold: ~0.1 μT minimum detectable field strength

While not matching the precision of dedicated test equipment, smartphone magnetometers provide more than adequate accuracy for EMF compensation applications.

Practical Usage Tips for Best Results

To maximize detection accuracy and reliability:

Environment Preparation

• Position near electrical sources: Within 1-2 meters of wall outlets, electrical panels, or appliances
• Clear metal objects: Move away from large metal furniture, vehicles, or magnetic decorations
• Remove phone case: Metal or magnetic phone cases can interfere with sensor readings
• Minimize wireless interference: Turn off wireless chargers, Bluetooth speakers during measurement

Device Positioning

• Calibrate first: Perform figure-8 motion to calibrate magnetometer if prompted
• Keep horizontal: Lay device flat or hold screen parallel to ground
• Stable position: Keep device still during measurement (movement creates noise)
• Follow alignment guide: Use NullField Lab's real-time alignment indicators to optimize orientation

Measurement Timing

• Wait for stabilization: Allow 5-10 seconds for initial detection to converge
• Observe trends: Grid frequency varies slowly; sudden jumps indicate measurement error
• Take multiple readings: Average several measurements for improved accuracy

NullField Lab Implementation

NullField Lab integrates all these technologies into a cohesive real-time EMF compensation system:

System Architecture

Key Implementation Features

Open Source Architecture

NullField Lab is built entirely with vanilla JavaScript and Web APIs—no frameworks, no build tools, no backend servers. This ensures:

• Transparency: All code runs client-side and is auditable
• Privacy: No data transmitted to external servers
• Offline capability: Works completely offline via Service Worker
• Simplicity: Deploy anywhere static files are served

The complete source code demonstrates how modern web technologies can perform sophisticated signal processing previously requiring dedicated hardware or native applications.