Wiki / Mirrors / Retroreflector

Retroreflector

A right-angle pair of mirrors that reflects any incoming ray back antiparallel to its incidence direction, independent of angle. Its delay-line motion starts at the placed position and periodically slides the whole element away along its own apex axis, only ever lengthening the round-trip optical path over a user-set range — a physical model of a mechanical retroreflecting delay stage.

Open in the canvas →

Click the retroreflector to see its live specs and try its parameters — this mini canvas can't be moved, deleted, or added to.

In the real world

A single flat mirror sends a ray back at whatever angle the law of reflection dictates — tilt the mirror even slightly and the returned beam walks off target. A corner retroreflector solves that by pairing two flat mirrors at exactly a right angle. Each bounce still obeys the ordinary law of reflection, but the composition of two perpendicular reflections has a special property: the outgoing ray is always exactly antiparallel to the incoming one, independent of the angle of incidence, for any ray that enters within the device's aperture.

The three-dimensional version of this idea — three mutually perpendicular mirror facets meeting at a corner, called a corner cube — is why bicycle reflectors and road signs throw a car's headlights straight back at the driver regardless of the exact angle the light arrives from, and why the retroreflector arrays left on the Moon by the Apollo missions still return laser pulses fired from Earth decades later with sub-arcsecond alignment tolerance. The 2D version modeled here — two mirrors at 90°, sometimes called a "roof" or "porro" reflector — is the working element inside a Michelson interferometer arm that needs alignment-insensitive retroreflection, and inside mechanical delay lines: mounting one on a translation stage and sliding it changes the round-trip path length by twice the stage's travel, without ever needing to re-align the returned beam.

d^=d^\hat{d}' = -\hat{d}
The defining property of a corner retroreflector: the outgoing direction is exactly the negative of the incoming one, for any incidence angle within the aperture — unlike a single flat mirror, whose return direction depends on incidence angle.
ΔL=2Δx\Delta L = 2\,\Delta x
Translating a retroreflector by Δx along its own axis changes the round-trip optical path by twice that distance — the basis of every retroreflecting mechanical delay line, from tabletop pulse stretchers to gravitational-wave interferometer arms.

In OpticalSetup

The Retroreflector is built from the same two flat mirror surfaces, each obeying the exact vector law of reflection used by the plain mirror, joined at a shared apex at exactly 90°. Ray tracing finds the first mirror hit, reflects it, then finds the second mirror hit and reflects again — two ordinary reflections, composed — which is enough for the antiparallel-return property to fall directly out of the vector reflection law rather than being special-cased.

Its delay-line movement section adds an optional periodic motion: set to Periodic linear, the whole element slides back and forth along its own apex axis, rotation-aware, so it works at any angle you place it on the table. The motion always starts at the position you placed it — the shortest path — and moves only in the direction that adds path length, sweeping up to the configured travel range (50 mm by default, up to 200 mm) at the configured frequency, then back. Because it's a true retroreflector rather than an abstract path-length tag, this doubles as a physical model of a mechanical retroreflecting delay stage: moving it by Δx really does add 2Δx of round-trip path, computed from the actual traced geometry.

Simplified vs. reality

Reflectivity is a single flat percentage applied identically to both mirror surfaces, with the same caveats as the plain mirror: no angle- or polarization-dependence, and no wavelength-dependent coating behavior. The delay-line motion is an idealized triangle wave — no modeled stage inertia, servo settling time, or velocity ripple — and, like the piezo stage's scanning, it drives the traced geometry directly rather than a separate abstract path-length parameter.

Related components

Further reading