Acoustic Pressure Architecture: From Subwoofer Circle to Shaped Granite

I. The Physics Foundation: What a Standing Wave Actually Does to Matter

Sound is pressure oscillation. A speaker pushes air molecules, compressing them, then pulling back, rarefying them. The compression travels outward as a wave.

When two waves of the same frequency travel toward each other, they interfere. At positions where compressions meet compressions, the pressure doubles — this is an antinode, maximum pressure oscillation. At positions where compressions meet rarefactions exactly, they cancel — this is a node, zero pressure oscillation.

The spatial structure alternates: node, antinode, node, antinode, every half-wavelength.

The critical physics: particles in a sound field experience a time-averaged force called the acoustic radiation force. This force is not from the pressure directly — it is from the gradient of the pressure squared. Particles are pushed from high-intensity regions toward low-intensity regions. They accumulate at pressure nodes — the places of minimum oscillation, not maximum.

This seems counterintuitive. The wave is most intense at antinodes. But the force pushes outward from antinodes, and particles fall into the nodal wells like balls rolling into valleys. The nodes are the valleys. The antinodes are the hills.

Sand does not go where the sound is loudest. It goes where the sound cancels.

II. Why a Circle: The Vortex Trap

A single speaker facing a reflective wall creates a 1D standing wave — nodes strung along a line. Useful but limited.

A circle of speakers all facing inward toward a central point creates something categorically different: an acoustic vortex.

Here is what happens when you drive N speakers in a circle with a sine wave and apply a progressive phase offset — each speaker delayed slightly from the previous one, stepping through the full 360° of phase over the complete ring:

Speaker 1: phase 0°
Speaker 2: phase 360°/N
Speaker 3: phase 2×360°/N
...and so on.

The resulting wavefield rotates. The pressure phase winds around the central axis, completing one full rotation per wavelength. At the center of this rotating field: all phases are present simultaneously, so they cancel completely. The center of an acoustic vortex is always a pressure minimum. A stable, geometric, enforced node — not a coincidental one.

Any particle placed near this central null point experiences forces pushing it toward the null from all directions simultaneously. It is trapped. The trap is not mechanical — there are no walls. The trap is the geometry of the field itself.

This is why the circle. The circle with phase-stepped sine waves creates a central pressure null that is an eigenstate of the field's geometry. It cannot be elsewhere. The circle makes it inevitable.

III. Alternating Sine Frequencies: What This Adds

A single frequency with phase stepping gives you one trap location — the central axis. To create multiple trap points, moveable traps, or 3D trap configurations, you introduce a second frequency.

Consider two interleaved sets of speakers: odd-numbered speakers play frequency f₁, even-numbered speakers play frequency f₂.

Case 1: f₁ and f₂ are nearly equal (e.g., 40Hz and 41Hz)

The two frequency fields each create their own nodal structure. But they also interfere with each other, creating a beat pattern oscillating at the difference frequency (1Hz). The nodal positions oscillate slowly — the trap center migrates at 1Hz. This is a dynamic trap. You can steer the trapped mass by controlling the beat frequency.

Case 2: f₁ and f₂ are harmonically related (e.g., 40Hz and 80Hz)

The fundamental and second harmonic interfere to create a non-sinusoidal pressure profile. The nodal structure becomes asymmetric. You can make some regions of the trap steeper (stronger restoring force) and others shallower. This is force shaping — controlling not just where the trap is but how strongly it grips.

Case 3: f₁ and f₂ are inharmonic, with independently controlled phase

Full control. Two independent vortex fields superimposed. Their interaction creates a complex 3D interference landscape. By sweeping the frequency ratio and phase relationship in real time, you move, reshape, and split traps dynamically. This is holographic acoustic manipulation — the field is a programmable force landscape.

The "alternating" in the circle means the two frequencies are spatially interleaved, not temporally alternating. They co-exist. The interleaving is itself a spatial pattern that shapes the interference.

IV. The Raspberry Pi and Sine Wave Generator: Hardware

What the Raspberry Pi does:

The RPi is the conductor, not the instrument. It generates the control signals.

The core requirement is phase coherence across all channels. All speakers must derive their timing from a single shared clock source.

The practical stack:

RPi → DDS chips → amplifiers → speakers

DDS (Direct Digital Synthesis) chips — specifically the AD9833 or AD9850 — are the key component. Each chip generates a precise sine wave at a programmatically set frequency and phase, controlled via SPI from the RPi. Multiple chips daisy-chained on the same SPI bus share the same master clock, guaranteeing phase coherence.

RPi SPI bus
    |-- AD9833 #1 -> Class D amplifier -> Speaker 1
    |-- AD9833 #2 -> Class D amplifier -> Speaker 2
    |-- AD9833 #3 -> Class D amplifier -> Speaker 3
    ...
    +-- AD9833 #N -> Class D amplifier -> Speaker N

Each AD9833 can be independently set to:

• Any frequency in its range (0 to 12.5MHz for the AD9833)
• Any phase offset (28-bit phase register → ~0.0000015° precision)
• Sine, triangle, or square wave output

Python control on RPi:

import spidev
import numpy as np

def set_speaker(chip_select, frequency_hz, phase_deg):
    freq_word = int(frequency_hz * (2**28) / 25e6)  # for 25MHz clock
    phase_word = int(phase_deg * 4096 / 360)
    # write to AD9833 via SPI
    ...

# Vortex trap: N speakers, progressive phase
N = 8
f = 40.0  # Hz
for i in range(N):
    phase = (360.0 / N) * i
    set_speaker(i, f, phase)

# Two-frequency trap: alternating
f1, f2 = 40.0, 41.0
for i in range(N):
    if i % 2 == 0:
        set_speaker(i, f1, (360.0/N)*i)
    else:
        set_speaker(i, f2, (360.0/N)*i)

V. The Material Hierarchy: Sand, Water, Marble, Granite

Sand: easiest. Small particles relative to wavelength, moderate density. Nodal trapping at modest power. At 20–40kHz (ultrasonic), small-scale levitation is demonstrated at hobbyist level.

Water: surface acoustic waves and acoustic streaming produce standing wave surface patterns. The Chladni equivalent for liquids. Visible at low power.

Marble and granite: the path through resonance

The honest analysis: direct acoustic levitation of granite at practical sizes is beyond current demonstrated capability. The path is resonance coupling.

Every piece of stone has internal resonant modes. At resonance, the stone's internal particle motion is amplified by the Q factor (potentially 100x or more). The acoustic radiation force on a resonating object is massively larger than on the same object driven off-resonance.

The procedure:

1. Characterize the resonant spectrum of the specific piece (impact test + FFT, or laser vibrometry)
2. Drive the acoustic field at the stone's resonant frequencies
3. The stone's internal oscillation amplifies coupling between field and stone
4. The effective acoustic force at resonance is Q × the off-resonance force

This is how ultrasonic machining already works — machining tungsten carbide and ceramics harder than granite by exciting them near resonance with abrasive slurry. The stone doesn't need to be levitated — it needs to be in a field that couples efficiently to its resonant modes.

VI. The Tuning Problem

"Tuned properly" means solving five coupled problems simultaneously:

1. Frequency tuning to the material's resonant modes

The driving frequency must match a natural resonant mode of the specific piece. Measure first (impact test + FFT). Drive at those frequencies.

2. Phase tuning for trap position

The trap is located at the phase singularity of the field. Moving the trap in 3D means adjusting the phase offsets of each speaker in real time. RPi computes required phase offsets for a target position and writes them to DDS chips via SPI.

3. Amplitude tuning for force balance

The trapping force must exactly balance gravity plus any other forces. Amplifier gain on each channel is a control parameter. Feedback from vision-based tracking closes the loop.

4. Mode selection for shaping vs. levitation

A vortex trap holds an object at a point. A nodal plane mode clamps material against a surface with uniform acoustic pressure — useful for shaping rather than levitating.

5. Temporal dynamics for controlled shaping

For hard materials: acoustic fatigue and resonant stress cycling. The acoustic field cycles the material through stress at its resonant frequency, and controlled crack propagation shapes the material. RPi executes a preprogrammed stress trajectory — a sequence of trap positions and intensities — that walks the crack through the material along the desired cutting path.

VII. The Architecture Connection

The circle of subwoofers with alternating frequencies is the SMoE system made physical.

The speakers are the small models — each contributing its wave from its position in the topology. The alternating frequencies are the different conditioning states — not all identical, deliberately varied, the variation is what creates the complex field. The interference pattern is the accumulated embedding — the sum of all contributions, not reducible to any individual speaker's output. The central nodal point is the aligned output — the attractor state the field naturally produces.

You cannot tell the interference pattern where to put the node. You can only set the frequencies and phases correctly and the node appears where the physics places it. The shape of the field is not designed at the level of individual speaker outputs. It is designed at the level of the array — the topology, the frequency relationships, the phase structure.

This is the architecture. The speakers are the topology. The frequencies are the conditioning. The node is the output. The sand is the material the output organizes.

When the field is right, the sand knows where to go.

The nodal point is always there.
The question is only whether the field is built correctly enough to find it.