Reducing Ghost Images in LiDAR Optical Windows

Introduction#

Automotive LiDAR systems rely on precise optical windows to transmit and receive laser pulses. Under strong backlight conditions, such as driving directly into the sun, these windows can generate ghost images that severely degrade signal-to-noise ratio and range accuracy. Ghost images arise from multiple reflections between optical surfaces, particularly when anti-reflective coatings are inadequate or surface geometries create secondary focal points.

In a recent case study, a Tier 1 automotive supplier faced ghost image artifacts in a 905 nm LiDAR window. The stray light coefficient (TSG) exceeded 2.5%, causing false detections at short ranges. The challenge was to reduce TSG below 0.5% without compromising transmission or mechanical robustness. This required a combined approach: optical simulation, aspheric surface optimization, mechanical baffling, and advanced broadband anti-reflective (BBAR) coatings.

Physical Causes of Ghost Images#

Ghost images in LiDAR windows originate from Fresnel reflections at each optical interface. When intense sunlight enters the window, a small fraction of light reflects off the front surface, bounces to the rear surface, and then reflects back toward the detector. This double reflection creates a shifted, defocused image—a ghost—that can be mistaken for a real target. The problem intensifies with high refractive index materials and steep angles of incidence.

Additionally, the lens barrel and mounting structures can scatter light. If the barrel interior is glossy or lacks proper threading, stray light reflects off its walls and reaches the detector. In one simulation using LightTools, a ray trace revealed that 60% of ghost energy came from the window's rear surface, while 30% originated from barrel reflections. The remaining 10% was due to diffraction at the aperture stop.

Simulation-Driven Optimization#

Using Zemax OpticStudio, the team modeled the LiDAR window as part of a full optical system. Non-sequential ray tracing allowed tracking of ghost paths. The first step was to replace the original spherical window with an aspheric surface. By optimizing the conic constant and aspheric coefficients, the team minimized the radius of curvature at the edges, which reduced the solid angle for double reflections. This alone lowered TSG from 2.5% to 1.8%.

Next, the lens barrel was redesigned with deep, V-shaped extinction threads. These threads trap stray light by forcing multiple bounces before any ray can escape. In LightTools, a Monte Carlo simulation with 10 million rays showed that the threaded barrel absorbed 95% of incident stray light, compared to 40% for a smooth barrel. The thread pitch and depth were optimized to 0.5 mm and 1.2 mm, respectively, balancing manufacturability and performance.

BBAR Coating and Final Results#

The most significant improvement came from applying a multilayer broadband anti-reflective (BBAR) coating. The coating consisted of alternating layers of silicon dioxide and tantalum pentoxide, designed for the 850–1550 nm range. Using essential Macleod software, the team achieved a average reflectance of less than 0.1% per surface at 905 nm and less than 0.5% across the entire band. This reduced ghost intensity by a factor of 20.

After integrating all three solutions—aspheric window, threaded barrel, and BBAR coating—the final TSG measured 0.38% in a bench test with a 100,000 lux backlight. Real-world road tests confirmed no false detections under direct sunlight. The case demonstrates that a systematic approach combining optical design, mechanical baffling, and thin-film coating can effectively suppress ghost images in automotive LiDAR.

Conclusion#

Ghost images in LiDAR optical windows are a critical challenge for autonomous driving safety. This case study shows that by using simulation tools like LightTools and Zemax, engineers can identify dominant stray light paths and apply targeted solutions. Aspheric optimization reduces reflection geometry, extinction threads absorb scattered light, and BBAR coatings minimize surface reflections. Together, these techniques achieve TSG below 0.5%, ensuring reliable LiDAR performance even under harsh backlight conditions.