ConamArray: Compact Broadband Ultrasound Array

• ConamArray is a compact broadband airborne-ultrasound array that uses 32 MEMS loudspeakers to achieve delay-based beam steering.
• It employs a staggered two-row geometry and synchronized multi-channel drive electronics to overcome spatial aliasing and high-voltage limitations.
• Experimental validations in anechoic settings confirmed its broadband performance from 20 to 100 kHz with controllable grating lobes at extreme steering angles.

ConamArray is a compact broadband airborne-ultrasound array composed of 32 MEMS loudspeakers and introduced as a proof-of-concept platform for broadband beam steering in air (Laurijssen et al., 1 Sep 2025). It was developed to demonstrate that MEMS loudspeakers can be used not only as small ultrasound sources, but also as a steerable array with controllable radiation patterns over a wide frequency range. The system targets a practical limitation of conventional broadband airborne ultrasound transducers: many existing broadband emitters are either too large, require high driving voltages, or are difficult to scale into arrays without severe spatial aliasing and grating lobes. ConamArray addresses that limitation through a staggered two-row geometry and a synchronized multi-channel drive architecture, enabling beam steering across the ultrasonic band while keeping the hardware compact and low-voltage.

1. Research context and design rationale

The central motivation for ConamArray is the mismatch between broadband ultrasonic emission and compact beam-steerable array design. Conventional broadband transducers are described as poorly suited to compact arrays because large aperture elements lead to grating lobes at practical steering angles, while high-voltage requirements make the electronics bulky and less flexible (Laurijssen et al., 1 Sep 2025). The platform therefore explores a MEMS-based alternative: elements that are small enough to pack densely, broadband enough to support ultrasound applications beyond narrowband sensing, and compatible with dynamic steering.

At the device level, ConamArray uses 32 USound UA-C0603-3T MEMS loudspeakers. Each transducer is about 6 mm in diameter and 1.5 mm in height, and is specified to cover roughly 2 kHz to 80 kHz. The array itself is studied in the ultrasonic regime from 20 kHz up to 100 kHz. This operating range is significant because it extends beyond the nominal upper end of the individual transducer specification, and the measured data are reported to show substantial acoustic output from 20 kHz to 100 kHz. A plausible implication is that the array is intended less as a single-frequency emitter than as a broadband radiating structure whose utility depends on delay-domain steering and frequency-dependent radiation behavior.

The paper presents ConamArray primarily as a transmit beam-steering system. However, the hardware was also designed with pulse–echo use in mind, which places it within a broader class of airborne ultrasound platforms for imaging and localization rather than a single-purpose emitter.

2. Physical architecture and electronics

ConamArray is implemented as a stacked two-board system with a front-end PCB carrying the 32 MEMS loudspeakers and a back-end PCB carrying the control and drive electronics (Laurijssen et al., 1 Sep 2025). The two boards are connected through three board-to-board connectors, and the full assembly measures approximately 102 mm × 80 mm × 9.5 mm.

Subsystem	Components	Role
Front-end PCB	32 USound UA-C0603-3T MEMS loudspeakers	Ultrasonic transmission
Back-end PCB	STM32F429, FT231X, 32 DAC121S101, 16 ADA4099-2BRMZ, power regulation	Control and synchronized drive
Receive path	64 Knowles SPH0641LU4H-1 PDM MEMS microphones, RP2350B, FT232H	Pulse–echo and real-time receive beamforming

The 32 transmit elements are arranged in two staggered rows of 16 elements each. Within a row, the physical inter-element spacing is 6.1 mm. Because the rows are staggered, the projected spacing along the horizontal steering axis is effectively halved to 3.05 mm. That projected pitch is the key geometric parameter governing both the usable steering range and the onset of grating lobes.

A defining feature of the system is the dual-microcontroller back-end. On transmit, an STMicroelectronics STM32F429 generates synchronized DAC updates. A timer peripheral provides the common sample clock, and DMA-driven GPIO writes update all DAC inputs in lockstep on each rising edge. This makes the DAC outputs phase-coherent without CPU intervention, which the paper identifies as crucial for beam steering. Runtime steering is performed by streaming new waveform samples or delays over USB.

The transmit chain is explicitly described as waveform-agnostic. Rather than restricting the hardware to single-frequency phase steering, it can support chirps, multisine signals, and sinusoidal bursts. This is an important architectural distinction because the paper’s stated aim is broadband ultrasound rather than narrowband operation.

The receive path consists of a separate array of 64 Knowles SPH0641LU4H-1 PDM MEMS microphones arranged in a 16-by-4 layout. These microphones are controlled by a Raspberry Pi RP2350B dual-core ARM-M33 using PIO state machines, and data are streamed out through an FTDI FT232H in synchronous FIFO mode. The reception path is described as enabling real-time receive beamforming, although the main experimental results concern the transmit array.

3. Steering formulation and time-domain implementation

The steering model is geometric and delay-based rather than narrowband and phase-only (Laurijssen et al., 1 Sep 2025). The array is treated in a right-handed coordinate system with $X$ as the depth or array-normal direction, $Y$ as the horizontal axis, and $Z$ as the vertical axis. For an element $m$ at position

$\mathbf{p}_m = [x_m, y_m, z_m]^{\mathsf T},$

and steering direction defined by elevation $\theta$ and azimuth $\phi$, the unit steering vector is

$$\mathbf{u}(\theta,\phi)= \begin{bmatrix} \cos\theta\cos\phi \ \cos\theta\sin\phi \ \sin\theta \end{bmatrix}.$$

The transmit delay for each element is

$$\tau_m = -\frac{1}{c}\mathbf{p}_m^{\mathsf T}\mathbf{u}(\theta,\phi) = -\frac{1}{c}\bigl(x_m\cos\theta\cos\phi + y_m\cos\theta\sin\phi + z_m\sin\theta\bigr),$$

where $c$ is the speed of sound.

For horizontal-plane steering, with $Y$0, the paper gives

$Y$1

with $Y$2 the inter-element spacing along the $Y$3-axis.

Implementation is performed in the time domain through integer sample delays,

$Y$4

where $Y$5 is the sample rate. The paper emphasizes that this delay-based formulation is waveform-agnostic. That matters because broadband beam steering cannot, in general, be reduced to a single set of narrowband phase shifts without frequency-dependent distortion. In this sense, ConamArray is better understood as a delay-steered broadband emitter than as a conventional phased array optimized for a single tone.

4. Spatial sampling, projected pitch, and grating-lobe onset

A central technical issue in ConamArray is the relation between element spacing, wavelength, and steering angle (Laurijssen et al., 1 Sep 2025). The staggered two-row geometry reduces the projected spacing to

$Y$6

This reduced effective pitch is what makes beam steering possible over part of the ultrasonic band, but it also determines the grating-lobe onset frequency.

Using the projected spacing, the paper reports a grating-lobe onset frequency of 58.2 kHz. The derivation in the paper is described as somewhat garbled in the typeset formulas, but the intended condition is the standard spatial aliasing limit: grating lobes appear when spacing becomes larger than about half a wavelength in the steering direction. For broadside extreme steering conditions, the paper states the equivalent relationship as

$Y$7

The practical interpretation is direct. Lower ultrasonic frequencies can be steered more cleanly, whereas larger steering angles and higher frequencies quickly introduce visible grating lobes. This establishes an intrinsic tradeoff between compactness and steering envelope. The staggered geometry improves the spatial sampling relative to a single row with 6.1 mm pitch, but it does not remove the fundamental bandwidth limitation imposed by the finite projected spacing.

5. Simulation methodology and experimental validation

The simulation program had three stated purposes: predicting main-lobe steering, estimating side-lobe and grating-lobe behavior, and computing per-element delays for arbitrary broadband waveforms (Laurijssen et al., 1 Sep 2025). The simulations used the measured or known geometry of the 32-element array and swept frequency from 20 kHz to 100 kHz while varying steering angle.

Three representative azimuth angles were reported: $Y$8, $Y$9, and $Z$0. At $Z$1, the simulations showed a single strong main lobe with low side-lobe energy across most of the band. At $Z$2, the main lobe broadened and side-lobes rose, with grating lobes becoming visible around 70 kHz. At $Z$3, beam quality degraded strongly and the grating-lobe threshold around 58.2 kHz became clearly evident.

Experimental validation was performed in an anechoic chamber. The array was mounted on a FLIR pan–tilt unit, and a calibrated microphone was placed 1.5 m away on the PTU centerline, aligned with the midpoint between the two MEMS rows. A logarithmic chirp from 100 kHz down to 20 kHz was used to characterize broadband response and steering performance. The chirp allowed examination of beam behavior across the full ultrasonic band in a single measurement.

The measured data matched the simulations closely. At broadside, the array produced a well-defined main lobe with minimal side-lobe energy, confirming that the staggered geometry and synchronized drive electronics were functioning as intended. At $Z$4 steering, the measured beam broadened and grating lobes became visible around 70 kHz, consistent with the model. At $Z$5 steering, the beam pattern was strongly degraded, and grating lobes dominated the high-frequency ultrasonic region.

The paper also reports the measured broadband frequency response at broadside. It shows substantial acoustic output from 20 kHz to 100 kHz, confirming the broadband nature of the MEMS devices in the assembled system. A noticeable dip near 60 kHz is attributed to destructive interference inside the MEMS speaker cavity. The authors note that this could be mitigated somewhat by removing the protective film on the transducer, though at the cost of reduced ingress protection.

6. Contributions, constraints, and prospective developments

The main contribution of ConamArray is the demonstration that a compact MEMS-based array can perform broadband beam steering in air over the ultrasonic band, a function that is described as difficult to realize with conventional large or high-voltage transducers (Laurijssen et al., 1 Sep 2025). The combination of synchronized multi-DAC waveform generation and a staggered two-row geometry shows that beam steering can be achieved with small-form-factor MEMS devices while preserving compactness and low-voltage operation.

The main limitation is the expected one for any compact array: steering range is constrained by grating lobes, especially at higher frequencies. The paper explicitly notes that steering to large angles such as $Z$6 causes severe degradation, and that practical use in the ultrasonic regime is constrained to more modest steering angles if grating lobes are to be suppressed. This suggests that ConamArray is best interpreted as a broadband steerable platform with a bounded angular operating region rather than as a uniformly steerable wide-angle emitter.

Future work proposed in the paper includes baffled designs or acoustic waveguides to effectively reduce spacing, potentially via 3D-printed channels. It also proposes evolving the current dual-microcontroller back-end into a single FPGA-based controller to unify transmit and receive processing and enable adaptive, real-time beamforming. These directions indicate that the present system is intended as a platform architecture rather than a finalized application-specific instrument.

The application domains named for ConamArray are ultrasonic imaging, localization, bio-inspired or biomimetic robotics, and broader airborne ultrasound sensing and manipulation tasks. Because the paper includes a reception path designed for pulse–echo operation, a plausible implication is that future versions may move from transmit-only beam steering demonstrations toward integrated ultrasonic sensing systems.

A common source of confusion is nomenclature. ConamArray is distinct from the Concentric Rectangular Array, a sparse active planar array for co-located transmit/receive sensing (1803.02219), and from cylindrical conformal metasurface array design methods in electromagnetics (Yoo et al., 2021). In the literature represented here, ConamArray refers specifically to the 32-element broadband MEMS ultrasound transducer array described in (Laurijssen et al., 1 Sep 2025).