Tag: MEMS Microphones

Beyond traditional handheld and stand-mounted studio transducers lies a rapidly advancing frontier of specialized acoustic capture technology. From microscopic lithographically etched silicon MEMS (Micro-Electro-Mechanical Systems) arrays embedded in spatial computing headsets and smartphones, to hemispherical Pressure Zone Boundary (PZM) transducers and tetrahedral Spherical Harmonics Ambisonic arrays for 360-degree virtual reality audio, modern acoustic engineering merges solid-state physics with real-time digital beamforming.

Over three billion MEMS (Micro-Electro-Mechanical Systems) Microphones are manufactured annually, powering active noise-canceling (ANC) headphones, smart speakers, automotive voice recognition, and miniature broadcast lavaliers.

Unlike conventional electret condenser capsules assembled mechanically from stretched Mylar films and stamped metal rings, a MEMS transducer is etched directly onto a crystalline silicon wafer using photolithography and deep reactive-ion etching (DRIE):

• Polysilicon Diaphragm and Perforated Backplate: A sub-micron silicon nitride or polysilicon membrane (measuring less than 1 millimeter in diameter) is suspended a few micrometers above a rigid perforated silicon backplate.
• Co-Packaged ASIC Die: Mounted inside the same tiny 3mm x 2mm surface-mount metal lid is a dedicated Application-Specific Integrated Circuit (ASIC) containing a charge pump (to bias the silicon capacitor), a low-noise preamplifier, and—in digital MEMS variants—a fourth-order Delta-Sigma Modulator outputting direct 1-bit Pulse Density Modulation (PDM) or I2S digital audio!

1. Solder Reflow Thermal Immunity: Traditional Mylar electret capsules lose their electrostatic charge and warp when exposed to 260°C automated PCB solder reflow ovens. Silicon MEMS capsules withstand extreme thermal cycling with zero degradation.
2. Ultra-Tight Phase and Sensitivity Matching (+/- 0.25 dB): Because thousands of MEMS dies are etched simultaneously on the same silicon wafer with nanometer precision, their phase and amplitude curves match within fractions of a decibel and degree—a mandatory requirement for multi-microphone acoustic beamforming arrays.
3. Vibration and G-Force Insensitivity: Because a polysilicon diaphragm has orders of magnitude less mass per unit area than Mylar, MEMS capsules exhibit vastly lower mechanical sensitivity to chassis handling noise and physical vibration.

How does a smart speaker or conferencing bar located 4 meters across a reverberant room isolate a human voice while music is playing? It deploys a multi-element MEMS Phased Beamforming Array.

Delay-and-Sum Linear Beamformer Equation (N Elements spaced by distance d):
Arrival Delay at Element n for Angle theta:
tau_n(theta) = (n * d * sin(theta)) / c

Beamformer Output Signal y(t) Steered to Target Angle theta_0:
y(t) = (1 / N) * Sum_{n=0}^{N-1} [ x_n( t + tau_n(theta_0) ) ]

Where:
- N       = Total number of MEMS microphone elements in the array
- d       = Inter-microphone physical spacing (typically 15mm to 40mm)
- c       = Speed of sound in air (343 m/s)
- theta_0 = Desired look-direction steering angle

By applying programmable sub-millisecond digital delays (tau_n) to each microphone channel before summing them in a DSP processor, acoustic wavefronts arriving from the target angle theta_0 align in exact constructive phase, while off-axis interference and room reverberation sum destructively. Advanced MVDR (Minimum Variance Distortionless Response) adaptive beamformers dynamically steer deep -30 dB cancellation nulls directly toward moving noise sources in real time!

When a conventional stand-mounted microphone records an actor speaking or a piano playing near a hard reflective surface (such as a stage floor, conference table, or grand piano lid), sound reaches the capsule via two distinct paths: the direct line-of-sight path (D_1) and the reflected bounce path off the surface (D_2).

Because path D_2 is longer than D_1, the delayed reflection creates severe hollow comb-filtering notches across the vocal midrange.

Developed by Ed Long and Ron Wickersham in 1978 and commercialized by Crown International, the Pressure Zone Microphone (PZM) eliminates boundary comb filtering by placing a miniature condenser capsule directly inside the boundary layer—mounted face-down just 0.5 to 1.0 millimeter above a rigid metal boundary plate sitting flush on the floor or wall:

• Zero Path Length Differential: Within this 1mm pressure zone, the direct sound wave and the boundary-reflected sound wave arrive at the diaphragm at the exact same instant (Delta_d approx 0), pushing all phase cancellation notches above 20,000 Hz (completely outside human hearing!).
• +6 dB Coherent Sensitivity Boost: Because the direct and reflected waves sum coherently in phase at the boundary surface, acoustic pressure doubles, yielding a natural +6 dB increase in sensitivity and a +3 dB improvement in direct-to-reverberant sound ratio.
• Uniform Hemispherical Pickup: A boundary microphone exhibits a flat frequency response across an entire 180-degree hemisphere above the mounting surface, making it ideal for stage footlights, boardroom tables, kick drum interiors, and grand piano lids.

Originally pioneered by mathematician and audio engineer Michael Gerzon at Oxford in the 1970s, Ambisonics is a full-sphere surround sound technique that captures and reproduces the complete three-dimensional soundfield (including horizontal azimuth theta and vertical elevation phi) around a single point in space.

A First-Order Ambisonic (FOA) microphone (such as the Sennheiser AMBEO VR Mic, Rode NT-SF1, or SoundField MKV) houses four matched subcardioid or cardioid capsules mounted on the faces of a regular geometric tetrahedron:

1. FLU: Front-Left-Up
2. FRD: Front-Right-Down
3. BLD: Back-Left-Down
4. BRU: Back-Right-Up

The raw 4-channel audio coming directly from these four tetrahedral capsules is called A-Format. To make the recording usable for spatial mixing and head-tracked VR rendering, a DSP matrix converts the A-Format signals into First-Order B-Format, which decomposes the soundfield into four orthogonal Spherical Harmonics (W, X, Y, Z):

First-Order Ambisonic (FOA) A-Format to B-Format Matrix Conversion:
1. W (Omnidirectional Pressure - 0th Order):
   W = FLU + FRD + BLD + BRU

2. X (Front-to-Back Figure-8 Gradient - Cosine Azimuth):
   X = FLU + FRD - BLD - BRU

3. Y (Left-to-Right Figure-8 Gradient - Sine Azimuth):
   Y = FLU - FRD + BLD - BRU

4. Z (Up-to-Down Vertical Figure-8 Gradient - Sine Elevation):
   Z = FLU - FRD - BLD + BRU

Once encoded into 4-channel B-Format (using the modern AmbiX standard: WYZX channel order with SN3D normalization), the recording is completely speaker-agnostic! In post-production, an engineer can decode the single B-Format recording into Mono, Stereo, 5.1 Surround, 7.1.4 Dolby Atmos beds, or Binaural 360° headphones with real-time 3-axis head-tracking rotation (Yaw, Pitch, Roll).

The following matrix compares specialized acoustic capture technologies across geometry, output format, and application domains:

While First-Order Ambisonics (4 channels) captures full 360-degree directionality, its spatial resolution is limited to first-order virtual polar patterns (Cardioid, Hypercardioid, Figure-8). To achieve razor-sharp spatial localization across large multi-speaker domes, acoustic researchers utilize Higher-Order Ambisonics (HOA).

The number of required spherical harmonic channels K for an Ambisonic system of order N follows the quadratic law:

Total Channels K = (N + 1)^2
– 1st Order (FOA): (1 + 1)^2 = 4 Channels
– 3rd Order (HOA): (3 + 1)^2 = 16 Channels
– 5th Order (HOA): (5 + 1)^2 = 36 Channels
– 7th Order (HOA): (7 + 1)^2 = 64 Channels (e.g., mh acoustics Eigenmike em64)

When calibrating studio monitors with room correction software (such as Sonarworks, Trinnov Optimizer, or Room EQ Wizard / REW), a standard studio cardioid vocal microphone cannot be used. Measurement microphones (such as the Earthworks M30, Brüel & Kjær 4007, or DPA 4006) adhere to strict metrological criteria:

• True Pressure Omnidirectional Capsule: Eliminates proximity effect and maintains uniform sensitivity regardless of speaker reflection angles.
• Ultra-Small Diameter (1/4-inch / 6mm Tip): Minimizes acoustic diffraction and pressure buildup at the capsule tip, keeping high-frequency response flat within +/- 0.5 dB from 9 Hz to 30 kHz.
• Individual Calibration File (.cal): Every measurement microphone is tested in an anechoic chamber against a primary laboratory reference standard, providing an ASCII frequency-compensation file loaded into the analysis DSP.

Modern specialized microphone architectures extend acoustic capture far beyond single-axis studio recording:

• Deploy Silicon MEMS Phased Arrays for embedded voice interfaces, conferencing hardware, and ANC wearables requiring thermal stability, sub-decibel phase matching, and real-time beamforming.
• Utilize Boundary / PZM Transducers on reflective surfaces (grand piano lids, kick drum shells, theatrical stages, and boardroom tables) to eliminate boundary comb filtering and gain +6 dB of coherent acoustic sensitivity.
• Standardize on Tetrahedral First-Order or Spherical Higher-Order Ambisonic Arrays when capturing 360° VR spatial audio, game soundscapes, and three-dimensional room impulse responses.
• Always employ calibrated 1/4-inch Omnidirectional Pressure Measurement Microphones when executing studio room acoustic analysis and monitor DSP equalization.