FMCW lidar

Frequency-modulated continuous-wave lidar ranges by chirping a laser's frequency, splitting off a local oscillator, and beating the delayed return against it on a photodiode pair. A target at distance d produces a beat at f = (2d/c)·(df/dt), so range is read as a frequency, and any radial motion adds a Doppler term that an up-down chirp separates cleanly. Every point in the cloud therefore arrives with its own velocity, which time-of-flight systems must infer across frames.

The detection physics is the draw. Coherent mixing amplifies the return by the local oscillator, giving shot-noise-limited sensitivity at milliwatt-class eye-safe powers near 1550 nm, and only light coherent with the transmitted chirp beats at all, so sunlight and other vehicles' lidars fall out as broadband background rather than false returns. The price is paid in laser quality: the coherence length must cover the round trip, which for 200 m class automotive range means linewidths in the hundred-kilohertz regime, a spec that connects directly to the linewidth measurement methods covered on this site, and the chirp must be linear to a fine tolerance, enforced with predistortion or an optical phase-locked loop, or corrected in post-processing by resampling against a reference interferometer.

The trend line is integration. The interferometer, splitters, balanced detectors, and increasingly the beam distribution, whether an optical phased array or a focal-plane switch array of grating emitters, are moving onto silicon photonics, with the narrow-linewidth source the last off-chip holdout. On the bench, FMCW work is dominated by three measurements: chirp nonlinearity against a fiber delay reference, laser frequency noise, and balanced-receiver common-mode rejection.

References: B. Behroozpour et al., IEEE Commun. Mag. 55(10), 135 (2017); C. Rogers et al., Nature 590, 256 (2021).