MEMS mirror

# MEMS Scanning Mirror Beam Steering

A mechanical LIDAR system needs to sweep its laser beam across an entire scene to build up a full point cloud, and a MEMS scanning mirror does that by tilting a microscopic silicon mirror back and forth at its natural mechanical resonant frequency, redirecting the beam across a wide field of view without any bulky external optics or motor assembly. Steering a laser beam across a scene requires some mechanism to continuously change the beam's direction — older systems used full mechanically rotating assemblies, spinning an entire laser-and-detector module. A MEMS mirror instead etches a tiny, torsion-hinged mirror directly into silicon using the same fabrication techniques as other microelectromechanical systems, and drives that mirror to oscillate back and forth, reflecting the laser beam across a swept angular range as the mirror tilts — all within a package orders of magnitude smaller and lighter than a full rotating assembly.

Driving the mirror at its mechanical resonant frequency rather than at an arbitrary oscillation rate is what makes large-angle, high-speed scanning achievable with a small, low-power actuator, because a resonant structure amplifies a small driving force into a much larger oscillation amplitude than the same drive force could produce off-resonance. A MEMS mirror's torsion hinge and mirror mass together have a natural frequency at which they want to oscillate, the same way a swing has a natural frequency at which gentle periodic pushes produce large swinging arcs. Driving the mirror right at that resonant frequency lets a comparatively weak actuator — built into the same tiny MEMS structure — build up a large angular sweep over many oscillation cycles, achieving a scan range that direct, non-resonant actuation at that power level could never reach.

Operating at resonance comes with a real constraint: the mirror's sweep speed and angular velocity profile are tied directly to its resonant frequency rather than freely programmable, so the scanning pattern a resonant MEMS mirror produces — typically a sinusoidal back-and-forth sweep — is dictated by the mirror's own mechanical properties, limiting how the system can arbitrarily point the beam compared to a fully steerable optical system without that mechanical constraint. This is why resonant MEMS scanning tends to produce a fixed, repeating scan pattern rather than an arbitrarily addressable one — the mirror naturally wants to oscillate through its own mechanical sweep, and deviating meaningfully from that pattern requires fighting the resonance rather than exploiting it, which is a real tradeoff against alternative beam-steering approaches that trade mechanical simplicity for more flexible pointing.

The SPAD discussed in the prior round handles the timing side of a LIDAR measurement; the MEMS scanning mirror handles the spatial side, determining exactly where in the scene that timed laser pulse is actually pointed at any given instant — together, timing precision and beam direction are the two halves of what turns a single laser-and-detector pair into a complete, swept 3D point cloud of a scene.