🔬 The Science Behind Ghost Hunting Equipment

Understanding how paranormal investigation tools actually work - no mysteries, just engineering and psychology

← Back to Home

🗣️ How the Ovilus Actually Works

The Ovilus is marketed as a device that "converts environmental energy into words" for spirit communication. But what's really happening inside?

The Algorithm (Simplified)

// Commercial Ovilus Algorithm (our best reverse-engineering) 1. Read environmental sensors: - Temperature: 22.5°C - Humidity: 45% - Pressure: 1013 hPa - Magnetic field: 512 (analog reading) 2. Combine into a hash value: hash = (temp * 1000 + humidity * 100 + pressure + mag_field) hash = (22500 + 4500 + 1013 + 512) = 28,525 3. Map hash to word index: word_index = hash % WORD_COUNT word_index = 28,525 % 512 = 397 4. Look up word in database: word = dictionary[397] // e.g., "DANGER" 5. Display and/or speak the word

The Math: How Many Unique Combinations?

Sensor Resolution: • Temperature (BME280): 0.01°C precision Range: 15°C to 35°C = 2,000 unique values • Humidity (BME280): 0.01% precision Range: 0% to 100% = 10,000 unique values • Pressure (BME280): 0.01 hPa precision Range: 950 to 1050 hPa = 10,000 unique values • Magnetic (Hall sensor): 12-bit ADC Range: 0 to 4095 = 4,096 unique values Total Possible Combinations: 2,000 × 10,000 × 10,000 × 4,096 = 819,200,000,000,000 That's 819 TRILLION unique environmental states!

⚠️ The Psychology of Small Word Databases

Why Commercial Devices Use Only 512-2048 Words

It's not a technical limitation - it's psychological manipulation.

Commercial Ovilus devices deliberately use small word databases to create artificial repetition:

  • 512 words = 1.6 trillion sensor combinations per word
  • 2,048 words = 400 billion sensor combinations per word
  • 10,000 words = 82 billion sensor combinations per word

What this means in practice:

With a 512-word database, if the room gets slightly warmer (22.5°C → 22.6°C), the device might still output the same word because billions of different sensor states map to the same word.

Why they do this:

  • Pattern recognition bias: "It said DANGER three times - the spirit is warning us!"
  • Confirmation bias: Repeated words feel more significant than random variety
  • Illusion of communication: "It's responding to my questions with the same word!"
  • Proprietary mystique: Small database = "carefully curated spirit vocabulary"

The truth: More words = better statistical distribution = more variety = less "spooky" repetition = less interesting for paranormal TV shows.

Word Database Size Comparison

Word Count Sensor States Per Word Behavior Use Case
100 words 8.2 trillion Extremely repetitive - same words over and over Beginner demo/testing
512 words 1.6 trillion High repetition - "classic" Ovilus behavior Mimics commercial Ovilus I/II
2,048 words 400 billion Moderate repetition - feels "responsive" Mimics commercial Ovilus III/IV
10,000 words 82 billion Low repetition - good statistical distribution Scientific investigation, better variety
50,000+ words 16 billion Minimal repetition - nearly unique per state Maximum entropy, data analysis

OpenParanormal Advantage: You Choose!

Our open-source firmware lets you configure word database size:

  • Classic Mode (512 words): Hardcoded in firmware - mimics commercial behavior for pattern-seeking investigators
  • Extended Mode (2,048 words): Hardcoded in firmware - balanced repetition vs. variety
  • Dictionary Mode (10,000+ words): Load from microSD card - scientific investigation with proper statistical distribution
  • Custom Mode: Create your own word lists - investigation-specific vocabularies (e.g., historical names, location-specific terms)

Why this matters: Transparency lets you understand bias in your data. If you know the device repeats words due to algorithm design (not spirits), you can make informed interpretations.

📡 EMF Meters: What They Actually Detect

EMF (Electromagnetic Field) meters measure magnetic fields in the environment. Commercial paranormal EMF meters work identically to electrician's EMF meters - the difference is just marketing.

What Creates EMF in Your Environment

  • AC Power Lines: 60Hz (US) or 50Hz (Europe) magnetic fields from household wiring
  • Electrical Devices: Transformers, motors, chargers, appliances
  • Radio Transmitters: Cell towers, WiFi routers, radio stations
  • Natural Sources: Earth's magnetic field (~0.5 mG), lightning, solar activity
  • Faulty Wiring: Creates higher-than-normal EMF spikes

The "EMF = Ghosts" Hypothesis

Popular theory: Spirits manipulate electromagnetic fields to manifest or communicate.

Scientific reality: No controlled experiments have demonstrated paranormal EMF manipulation. All documented EMF spikes have prosaic explanations:

  • Wiring in walls (especially old buildings)
  • Electrical panel boxes
  • Appliances turning on/off automatically
  • Cell phone signals (investigator's own phone!)
  • Fluorescent lights (even when "off" can leak EMF)

Our approach: Log ALL EMF data with timestamps, correlate with environmental factors, look for patterns - not cherry-pick "anomalies."

Commercial K2 Meter vs. DIY EMF Meter

✅ OpenParanormal EMF Meter

  • Logs data to SD card with timestamps
  • Graphical display shows trends
  • Calibrated readings in milligauss (mG)
  • Costs $15 to build
  • Open-source = understand the algorithm

⚠️ Commercial K2 Meter

  • No data logging - just LEDs
  • No calibration - arbitrary LED thresholds
  • Costs $50-70
  • Black box design - can't verify accuracy
  • Optimized for dramatic TV reactions

📻 Spirit Box: Radio Static or Communication?

A Spirit Box rapidly scans AM/FM radio frequencies, creating white noise from fragments of radio stations. The theory: spirits manipulate this radio static to form words.

How It Works

Spirit Box Algorithm: 1. Tune radio to starting frequency (e.g., 87.5 MHz FM) 2. Wait X milliseconds (sweep speed, e.g., 50ms) 3. Increment frequency by 0.1 MHz 4. Repeat from 87.5 to 108.0 MHz 5. Loop forever What you hear: • Brief fragments of radio stations (music, talk, ads) • Static (white noise between stations) • Occasional clear words (from actual broadcasts) • "Pareidolia effect" - brain pattern-matches noise into words

The Psychology: Audio Pareidolia

Pareidolia is the tendency to perceive meaningful patterns in random stimuli. With audio:

  • Your brain is designed to extract speech from noise (evolutionary advantage)
  • Rapid radio scanning creates random audio fragments
  • Your brain fills in gaps, creating "words" from static
  • Confirmation bias: you hear what you expect/want to hear

Test: Use a Spirit Box in different languages/countries. "Spirits" will speak the local language of radio broadcasts - not Latin or ancient languages.

OpenParanormal Advantage: Recording & Analysis

Our Spirit Box can record sessions to SD card, allowing:

  • Playback at different speeds (slow down to analyze fragments)
  • Spectral analysis (identify if "words" are from radio stations)
  • Blind listening tests (remove investigator bias)
  • Correlation with other sensor data (EMF spikes, temperature)

💡 Novel Device Ideas: Inventing New Tools

Beyond copying commercial equipment, OpenParanormal can pioneer new investigation methods based on sensor technology and data science:

1. Environmental Change Correlator

Concept: Instead of measuring individual sensors, detect simultaneous changes across multiple sensors.

Hardware:

  • BME280 (temp/humidity/pressure)
  • EMF sensor (magnetic field)
  • Microphone (audio level)
  • PIR motion sensor
  • All logged with microsecond timestamps

Algorithm:

// Detect correlated anomalies if (temp_delta > threshold AND emf_delta > threshold AND audio_spike detected AND all within 100ms window) { FLAG AS ANOMALY Log all sensor values + timestamp }

Why this is better: Filters out individual false positives. Correlated multi-sensor events are statistically rarer and harder to explain via environmental factors.

2. Entropy Analyzer

Concept: Measure "randomness" in environmental data vs. expected baseline.

Theory: If paranormal phenomena exist, they might introduce non-random patterns (or extra randomness) into sensor data.

Hardware:

  • High-speed ADC sampling ambient EM noise
  • Temperature sensor array (multiple points)
  • Accelerometer (vibration/movement)

Algorithm:

// Statistical entropy calculation 1. Establish baseline entropy over 60 seconds 2. Continuously measure entropy in rolling 5-second windows 3. Calculate z-score (how many standard deviations from baseline) 4. Alert on significant deviations (z > 3.0 or z < -3.0) 5. Log all raw data for post-analysis

Scientific value: Quantifiable metric. Can be tested in controlled vs. "haunted" environments.

3. Infrasound Detector

Concept: Measure sound below human hearing range (< 20 Hz).

Why this matters:

  • Infrasound (especially 18-19 Hz) can cause feelings of unease, dread, visual distortions
  • Sources: HVAC systems, wind, traffic, geological activity
  • Many "haunted" feelings may be infrasound exposure

Hardware:

  • Low-frequency microphone or custom resonant chamber
  • ESP32 with FFT analysis
  • Display shows 0-20 Hz spectrum in real-time

Scientific value: Distinguish psychological effects (infrasound-induced) from potential paranormal phenomena.

4. Baseline Environment Mapper

Concept: Create a "normal" baseline map of a location before investigation.

Process:

  1. Mapping phase (daytime): Walk through location with device, logging EMF, temp, humidity, light levels every meter
  2. Analysis: Generate heatmap of "normal" readings
  3. Investigation phase (nighttime): Device alerts only when readings deviate significantly from baseline map

Hardware:

  • GPS module or manual room selection
  • Multi-sensor array (EMF, temp, humidity, light)
  • MicroSD for storing baseline profiles
  • Visual/audio alert when anomaly detected

Why this is revolutionary: Eliminates false positives from environmental factors (e.g., "high EMF" is just a wall outlet). Only true changes trigger alerts.

5. The "Control Device" - Placebo Detector

Concept: A fake device that looks identical to real equipment but generates random data.

Scientific purpose:

  • Blind studies: Give investigators 2 devices (1 real, 1 fake), don't tell them which is which
  • If both devices get "equal" paranormal hits → investigator bias confirmed
  • If real device significantly outperforms fake → possible genuine signal

Hardware: Identical case to real EMF/Ovilus, but outputs random values

Ethical note: Only for controlled experiments, with informed consent afterward

🔬 Bringing Science to Paranormal Investigation

The Problem with Traditional Ghost Hunting

  • Cherry-picking data: Only reporting "hits," ignoring misses
  • Confirmation bias: Interpreting ambiguous data as supporting pre-existing beliefs
  • Lack of controls: Not testing equipment in non-"haunted" locations
  • Black-box equipment: Not understanding how devices work
  • No data logging: Relying on memory and subjective interpretation

The OpenParanormal Scientific Approach

Core Principles

  1. Log everything: All sensor data, all the time, with timestamps
  2. Establish baselines: Measure "normal" conditions before claiming anomalies
  3. Use controls: Test equipment in known non-paranormal environments
  4. Understand your tools: Open-source = you know exactly how devices work
  5. Blind analysis: Have someone unfamiliar with the investigation review data
  6. Quantify uncertainty: Use statistics, not just "we felt something"
  7. Replicate findings: Anomalies should repeat under similar conditions
  8. Rule out mundane explanations first: Is it a draft? Faulty wiring? Radio interference?

Data Analysis Best Practices

Example Investigation Protocol: Phase 1: BASELINE (1 hour minimum) - Deploy all sensors in location - Log all data continuously - Calculate mean, standard deviation for each sensor - Identify environmental patterns (HVAC cycles, traffic noise, etc.) Phase 2: INVESTIGATION (2-4 hours) - Continue logging all sensors - Mark events manually (investigator notes, timestamps) - Do NOT review data during investigation (avoid bias) Phase 3: ANALYSIS (offline) - Load all sensor logs into analysis software - Flag readings > 2 standard deviations from baseline - Correlate multi-sensor anomalies (within ±1 second) - Compare "anomaly" frequency to control dataset - Statistical significance: p < 0.05 to claim finding Phase 4: REPLICATION - Return to location at different time/date - Repeat investigation with same equipment - Compare datasets - do anomalies occur in same locations? - Only report findings that replicate across sessions

Publishing Your Findings

OpenParanormal encourages investigators to share data openly:

  • Raw data dumps: Upload sensor logs to public repositories (GitHub, OSF)
  • Methodology transparency: Document equipment specs, calibration, analysis methods
  • Negative results count: "We found nothing anomalous" is valuable data
  • Peer review: Let other investigators analyze your data independently

🎯 The Mission: Elevate Paranormal Investigation

OpenParanormal exists to bring transparency, scientific rigor, and affordability to ghost hunting. We believe:

  • Understanding = Better Investigation: Knowing how equipment works helps you interpret data correctly
  • Science ≠ Skepticism: Scientific method doesn't assume paranormal phenomena don't exist - it provides tools to test hypotheses rigorously
  • Open Source = Truth: Commercial black-box devices hide their flaws. Open-source reveals both capabilities and limitations
  • Data > Feelings: Logged, timestamped data is more reliable than subjective experiences
  • Innovation Through Sharing: When investigators share designs and data, the entire field advances

Build. Investigate. Share. Advance the science.