3D Tuning Fork Beats: Why Two Almost-Equal Tones Pulse

A Blender-built acoustics bench: two tuning forks on resonance boxes, a microphone on a rail and a live sound field. Beats from exact wave superposition.

Switch to 2D

Presets

Playback

Parameters

Camera

Layers

Quality

Model check

The science

What is modelled

Two tuning forks on resonance boxes sound at slightly different frequencies. A microphone on a rail picks up the sum of the two sound waves. Because the tones drift in and out of step, their sum swells and fades: these are beats. The coloured sheet on the table shows the loudness (or the instantaneous pressure) of the combined sound field, and the lamp tower follows the loudness at the microphone.

Equations

  • Pressure at the microphone: p(t) = a1 cos(ω1t − k1r1) + a2 cos(ω2t − k2r2), with k = ω/c and spherical spreading ai = Air0/ri.
  • Equal amplitudes: cosω1t + cosω2t = 2 cos((ω1−ω2)t/2) cos((ω1+ω2)t/2): a tone at the mean frequency whose loudness is multiplied by a slow cosine, so the beat frequency is fbeat = |f1 − f2|.
  • General envelope: E(t) = √(a1² + a2² + 2a1a2 cos(Δωt + Δφ)), ranging from |a1−a2| to a1+a2. The path difference to the two forks sets Δφ, which is why the loudest and quietest spots form hyperbola-shaped bands on the table.
  • Speed of sound: c = √(γRT/M) ≈ 331.3 + 0.606 T m/s (T in °C), wavelength λ = c/f.
  • Detuning with a clip: extra mass Δm on the tines lowers the frequency, f ≈ f0√(meff/(meff+Δm)).

Checks run in the page

  • A brute-force synthesis of both tones at the microphone (no beat formula) is searched for loudness minima and compared with |f1−f2|.
  • The product-to-sum identity is verified at thousands of time points for the current frequencies.
  • The envelope formula is compared with the measured peak of the real signal, and the exact speed of sound with the linear fit over 0 to 40 °C.

Honest simplifications

Each fork is a point source radiating a steady sine wave in free space: no room echoes, no directivity of the resonance box, no non-linear distortion and no frequency drift as the forks cool. Real tines move well under a millimetre; their swing is exaggerated on screen (a layer switch turns that off). The picture runs in slow motion (the factor is a slider) because 440 vibrations per second cannot be followed by eye; the real-sound button plays the true frequencies in real time. Real forks ring down over seconds; the ring-down option uses a simple exponential envelope. The frequency of the added-mass formula uses an assumed effective tine mass of 14 g.

What to try

  • Set the detuning to 0, then nudge it to 1 Hz: the beats become so slow that the pulses take a second each.
  • Press Play real sound with a detuning of 4 Hz and count the pulses against the beat frequency readout.
  • Drag the microphone along the rail: at some places the two waves arrive out of step and the beats get shallow, at others they swell to full depth.
  • Make the loudness unequal: the beats no longer reach silence, the modulation depth readout drops.
  • Raise the detuning to 28 Hz: the beats merge into a rough tone.
Loading model...
Loading Blender model...
Two tones and their sum at the microphone (real time)
Loudness envelope over several beats
Loudness range along the microphone rail
Made with 🟠 Blender · Three.js