Arduino Theremin Design: Ultrasonic vs IR Sensor Concepts

The original theremin, invented by Léon Theremin in 1920, is one of the first fully electronic musical instruments. It relies on analog heterodyne beat frequency oscillators (BFO) and radio frequency (RF) interference to detect hand proximity. However, building an Arduino theremin in 2026 discards volatile analog RF circuits entirely. Instead, modern makers leverage digital time-of-flight (ToF) sensors, infrared reflectance, and Direct Digital Synthesis (DDS) to achieve superior stability and programmable scales.

This concept guide breaks down the physics, sensor selection, and audio generation architectures required to design a professional-grade digital theremin using modern microcontrollers.

The Core Concept: Analog RF vs. Digital Sensor Mapping

In a classic analog theremin, the player's hand acts as a grounded plate in a capacitor. Moving the hand changes the capacitance, which shifts the frequency of an RF oscillator. The difference between this variable oscillator and a fixed reference oscillator creates an audible beat frequency.

An Arduino-based theremin replaces this continuous analog physics with a discrete digital pipeline:

1. Sensing: A digital or analog sensor measures the physical distance between the instrument and the player's hand.
2. Processing: The MCU reads the sensor data, applies smoothing filters (like a Kalman or moving average filter), and maps the distance to a target frequency.
3. Synthesis: The MCU generates the audio waveform, either via Pulse Width Modulation (PWM), an R-2R resistor ladder DAC, or an I2S digital-to-analog converter.

Expert Insight: The primary challenge in digital theremin design is not generating the sound, but eliminating sensor jitter. A 1mm jitter at the sensor level can translate to a 5Hz frequency wobble at higher pitches, which the human ear perceives as an unpleasant, out-of-tune warble.

Sensor Selection Matrix: Ultrasonic vs. IR vs. ToF

Choosing the right distance sensor dictates the playability of your instrument. Below is a comparison of the three most common sensor architectures used in MCU theremin builds.

Edge Cases and Failure Modes

While the HC-SR04 is cheap and ubiquitous, it suffers from multipath interference. Sound waves bounce off the player's arm, torso, and nearby walls, causing phantom echoes that result in sudden pitch drops. Furthermore, ultrasonic sensors struggle with soft, sound-absorbing clothing.

The Sharp IR sensor avoids acoustic echoes but is highly susceptible to ambient light saturation. If you perform on a stage with 5000K LED lighting or direct sunlight, the IR receiver will saturate, causing the distance reading to flatline or spike erratically.

For professional builds, the VL53L1X Time-of-Flight laser sensor is the superior choice. It uses a 940nm VCSEL laser and a SPAD (Single Photon Avalanche Diode) array, making it largely immune to ambient acoustic noise and offering millimeter precision. However, it requires I2C communication, which introduces slight bus-latency overhead compared to the analog read of the Sharp sensor.

Audio Generation: Moving Beyond 8-Bit PWM Buzz

Beginners often use the Arduino tone() function to generate sound. This function manipulates hardware timers to output a 50% duty cycle square wave. While functional, a square wave is rich in odd harmonics and sounds like a harsh, retro buzzer—nothing like the smooth, eerie sine wave of a real theremin.

To generate true sine waves, you must implement Direct Digital Synthesis (DDS). DDS involves stepping through a pre-calculated 256-byte sine wave lookup table stored in the MCU's flash memory at a precise sample rate (e.g., 20kHz to 44.1kHz).

Architecture 1: ATmega328P and Timer1 Interrupts

On legacy 8-bit boards like the Arduino Uno, you can manipulate Timer1 to trigger an interrupt at your desired sample rate. Inside the Interrupt Service Routine (ISR), you fetch the next byte from the sine table and write it to an 8-bit R-2R resistor ladder DAC connected to PORTD. As detailed in the Secrets of Arduino PWM documentation, manipulating the TCCR1A and OCR1A registers allows for high-frequency PCM audio output, though it is limited to 8-bit resolution, which introduces noticeable quantization noise (hiss) during quiet passages.

Architecture 2: ESP32 and I2S DACs (The 2026 Standard)

Modern designs favor the ESP32-S3 due to its native I2S (Inter-IC Sound) peripheral. Instead of hacking timers, the ESP32 streams 16-bit or 24-bit audio samples directly to an external I2S DAC amplifier, such as the Adafruit MAX98357A. This offloads the timing-critical waveform generation to dedicated hardware, yielding CD-quality, noise-free sine waves while leaving the main CPU cores free to handle complex sensor filtering and wireless MIDI transmission.

The Psychoacoustics of Pitch Mapping

The most common mistake in Arduino theremin code is using a linear mapping function. Human pitch perception is logarithmic, not linear. An octave represents a doubling of frequency (e.g., A4 is 440Hz, A5 is 880Hz).

If you use the standard Arduino map() function to map 0-50cm directly to 220Hz-880Hz, the lower octave (220Hz to 440Hz) will occupy the first 33cm of physical space, while the upper octave (440Hz to 880Hz) will be squished into the remaining 17cm. This makes the higher notes nearly impossible to play in tune.

The Logarithmic Mapping Solution

To achieve an even, playable physical scale, you must map the linear distance to MIDI note numbers (which are linear in pitch steps), and then convert the MIDI note to frequency using the equal temperament formula:

f = 440 * 2^((n - 69) / 12)

Where n is the MIDI note number. In C++, this translates to:

float distance = readSensor(); // e.g., 0.0 to 50.0 cm
int midiNote = map(distance, 0, 50, 93, 45); // Maps to MIDI C6 down to A2
float frequency = 440.0 * pow(2.0, (midiNote - 69) / 12.0);

This mathematical approach ensures that every physical centimeter of hand movement corresponds to a consistent musical interval, drastically improving the instrument's playability.

Adding Expression: The Volume Antenna

A true theremin features two antennas: one for pitch (vertical) and one for volume (horizontal). For the volume antenna, a secondary VL53L0X (shorter range ToF sensor) can be mounted horizontally.

Rather than mapping distance directly to amplitude (which causes abrupt audio cutoffs), map the distance to an exponential volume envelope. Furthermore, implement a 'mute threshold' at the extreme end of the sensor's range. When the player touches the volume antenna, the distance reads near-zero, and the MCU should apply a rapid 10ms fade-out envelope to prevent harsh digital clicking (pop noise) when the audio stream halts.

Summary of Best Practices

• Sensor Choice: Use ToF laser sensors (VL53L1X) over ultrasonic to avoid acoustic multipath jitter.
• Filtering: Apply a Kalman filter or a weighted moving average to sensor readings to smooth out micro-tremors in the player's hand.
• Audio Engine: Abandon tone(). Use an ESP32 with an I2S DAC for 16-bit DDS sine wave generation.
• Mapping: Always use logarithmic math to map physical distance to musical frequencies, respecting human psychoacoustics.