Introduction: Spiral sensor placement changes how a microphone array samples sound in space, and that geometry decides how cleanly a UAV direction estimate comes out.
When people first look at a drone acoustic localization system, the microphone count gets most of the attention. But 64 sensors do not automatically beat 32 if they sit in the wrong places. The positions themselves — the array geometry — decide how the array samples an arriving sound wave, which directions it can separate, and which directions it can confuse. Two arrays with identical microphones, identical electronics, and identical software can produce very different direction maps purely because one uses a square grid and the other uses a spiral. The sections below explain why spiral spacing suppresses side lobes and spatial aliasing, and why channel matching matters just as much as the layout.
Why Array Geometry Controls Direction Finding
A microphone array works out direction by comparing what each sensor receives: small differences in arrival time and small differences in level across the array. Those comparisons only make sense because the sensors sit at known positions. In effect, the array samples the sound field in space the way an ADC samples a signal in time, and the spacing between sensors plays the same role the sampling interval plays in time-domain processing. Get that spacing wrong and the array starts reporting directions that were never there. Get it right and a single board can resolve a drone in a noisy sky. Geometry also sets resolution through aperture — the physical spread of the outermost sensors. A wider array produces a narrower main beam, so two drones a few degrees apart stay distinguishable instead of merging into one blur. This is why a physically large board is not wasted material: a 460 × 460 × 20 mm array with 64 MEMS microphones in a spiral arrangement spends its area on aperture and then fills that aperture with useful sample points. Channel count alone does not buy resolution; the positions those channels occupy do.
1. Regular Grids Can Create Strong Side Lobes and Aliasing
A square or rectangular grid repeats the same spacing along two axes. That regularity is convenient to draw and easy to simulate, but it leaves a fingerprint on the beam pattern. Because the spacing repeats, the array responds almost equally well to a set of other directions that share the same delay pattern — the grating lobes. Once the distance between neighbouring sensors grows past roughly half the wavelength of the incoming sound, those false peaks can approach the strength of the real one, and the array can no longer say which direction the drone is actually in. Axis-aligned rows also concentrate sidelobe energy into a few predictable directions, which is exactly the situation where a strong reflection off a building starts to look like a second aircraft.
2. Spiral Layout Spreads Sensors to Reduce Repeated Spatial Intervals
A spiral — typically several curved arms wrapping around a common centre — breaks that repetition. Sensors sit at many different radii and angles, so the distances between pairs of sensors form a broad, spread-out set instead of a handful of repeated values. When no single spacing dominates, the beam pattern cannot pile its secondary peaks into a few strong grating lobes; the leftover energy spreads across many angles at lower levels. That spread also holds up across frequency, which suits MEMS microphones with a 20 Hz to 80 kHz band. And because a spiral grows outward, it fills a large circular aperture efficiently, giving a designer wide spacing near the rim for resolution and tight spacing near the centre for high-frequency work at the same time.
How Spiral Spacing Affects Side Lobes and Spatial Aliasing
Two numbers govern most of the behaviour. The smallest gap between any two sensors sets how high in frequency the array can work without ambiguity, while the largest gap — usually measured across the full diameter — sets how narrow the main beam becomes. A spiral lets a designer tune both at once. The closest pairs should stay comfortably below half the wavelength of the highest frequency of interest; for a 5 kHz signature that means roughly 34 mm, so inner turns are usually kept tighter than that. The outer arms then reach to the rim and stretch the aperture, which is what sharpens the beam. Side lobes and spatial aliasing are separate problems, and it helps to keep them apart. Side lobes are energy that leaks into angles you did not ask about while the main beam still points correctly; they raise false alarms in busy airspace. Aliasing is worse, because it creates ambiguity: the array reports a completely wrong direction with full confidence. Spiral spacing attacks both, lowering the average side lobe level and breaking up the periodic structure that produces hard aliases. It does not remove the need for adequate sampling density. That is why adding channels to the same board, up to the 192-channel expansion the LS8118F hardware supports, mainly means more sample points inside one aperture: the main beam narrows slightly, the alias-free band widens, and the pattern becomes smoother. The number of arms in the spiral is a design choice with a visible effect. More arms give more even angular coverage and a response that stays stable as the source moves around the array; too few arms leave directions where the pattern is rough. Designers usually compare candidate layouts by plotting the beam pattern across the whole working band rather than judging one frequency, because a layout that looks clean at 1 kHz can look uneven at 6 kHz. That habit matters more than any single rule of thumb.
Why Channel Matching Matters as Much as the Layout
A clever layout only pays off if every channel behaves like every other channel. Beamforming combines channels with delays and weights derived from the geometry; if one microphone is a decibel hotter or a few degrees out of phase, the sum is no longer the clean sum the layout implied. Amplitude mismatch lifts the sidelobe floor, filling the empty parts of the pattern with energy that should not be there. Phase mismatch is more direct: it corrupts the arrival-time differences the whole method depends on, shifting the estimated angle even when the geometry itself is perfect. Mismatch creeps in from several directions at once. MEMS microphones ship with part-to-part sensitivity tolerance, and the acoustic port, gasket, and board mounting each add their own variation. Trace routing differences change the electrical delay between sensor and converter, and timing skew in the acquisition chain shows up as an apparent direction error. Over a 460 mm board that must operate from -40 °C to 70 °C, sensitivity drift varies with position, so two channels that matched on the bench may diverge on a cold morning deployment. This is the part of array design that is easiest to underestimate. Two boards built to the same drawing can behave differently once assembled, because placement accuracy, solder quality around the MEMS port, and mechanical flatness all shift channel phase and level. Teams working with a custom acoustic array PCBA service tend to ask about fine-pitch placement experience and channel-consistency measurement, not just the schematic. It is also why a spiral layout is a starting point rather than a finished answer: a stated single-array azimuth accuracy such as ±3° is a nominal product figure tied to specific test conditions, and real performance still depends on calibration, background noise, and how the array is integrated.
Conclusion
Spiral geometry earns its place in UAV acoustic localization for a simple reason: it spreads sensor spacing instead of repeating it. That spread keeps the beam pattern from growing strong grating lobes, widens the band over which directions stay unambiguous, and lets a large circular aperture hold many channels efficiently. Aperture sets resolution, and matching keeps it. The LS8118F hardware shows the shape of the idea in practice, with a 64-channel Infineon IM72D128V01 MEMS array in a spiral layout on a 460 × 460 × 20 mm board, expansion up to 192 channels, and a stated nominal single-array azimuth accuracy of ±3°. Anyone who wants the full specification can find it on the listing.
FAQ
Q：Why do microphone arrays use spiral layouts instead of square grids?
A：Square grids repeat the same spacing in two directions, and that repetition produces grating lobes — false peaks that can compete with the real direction once spacing exceeds about half a wavelength. A spiral distributes sensors across many radii and angles so no single spacing dominates. Secondary peaks spread out and drop in level, the pattern stays usable over a wider frequency band, and the outward-growing shape fills a large circular aperture without wasting area on empty rows.
Q：What is spatial aliasing in a microphone array?
A：It is the spatial version of the aliasing that affects sampled signals. When the distance between sensors is too large relative to the wavelength of the sound, two different arrival directions produce the same set of arrival-time differences. The array then reports a false direction with full confidence, and nothing downstream can tell the two apart. Keeping the closest sensor spacing comfortably below half the shortest wavelength of interest is the standard way to avoid it.
Q：Does a larger microphone array always improve drone sound localization?
A：No. A larger aperture narrows the main beam, which helps separate nearby sources, but a big board also tends to spread sensors further apart, and wide spacing invites aliasing at higher frequencies. Large boards are harder to build consistently as well, because mechanical flatness, mounting, and channel-to-channel matching all get tougher. Size helps resolution when sampling density and channel matching scale with it.
Sources / References
Discrete-Time Signal Processing | MIT OpenCourseWare
Microphone Array Signal Processing | Springer Nature Link
Derivational Complexity Is an Invariant Cost Model | Springer Nature Link
No comments:
Post a Comment