Dark field illumination design aims to make surface defects visible while keeping the surrounding surface dark to the camera. For a ring light, the useful design question is how illumination angle, surface scatter, and camera acceptance interact across the inspection area. TracePro can help compare these variables before the LED board and housing are fixed. [1]
Establish the dark field geometry
A low-angle ring light illuminates the part obliquely while the camera views it from a different direction. On a suitably oriented smooth surface, the dominant specular reflection misses the camera. Scratches, edges, and other features can redirect light toward it. Ring height, diameter, part curvature, and the camera’s field of view all affect whether this geometry holds across the image. [2]
Start with the actual working distance and inspection area. Include the lens entrance pupil or an appropriate acceptance model. Check the field edges as well as the center, and include expected part tilt. A layout that works at one point can admit background reflections elsewhere.
Give the surface a credible scatter model
The clean surface determines the background against which a defect must be found. Use measured or otherwise justified bidirectional scatter data when available, with the wavelength and surface orientation recorded. A roughness value alone does not uniquely specify the angular scatter distribution.
Represent the defect at a level appropriate to the optical question. A geometric groove can support an initial ray analysis when its dimensions justify geometrical optics. Very small features may require a different physical treatment or measured scatter data. Do not present an arbitrary groove as a validated model of every scratch.
For brushed or directionally finished surfaces, check whether the chosen scatter model and data preserve the relevant orientation dependence. Rotate the part through likely production orientations. A model that averages away directional scatter may miss the most difficult background condition.
Model the complete ring light
Use an LED source description that captures the emission relevant to the lens and part, ideally supported by supplier measurements or ray data. A nominal viewing angle is not a complete spatial and angular source model. Include individual LED positions, any diffuser or secondary optics, apertures, and mechanical obstructions.
Compare ring diameter, height, LED orientation, and source spacing as controlled variants. More LEDs may change azimuthal coverage, flux, thermal loading, and packaging; it does not guarantee improved defect visibility. Keep electrical or optical normalization explicit when comparing designs.
TracePro supports illumination modeling and importance sampling for suitable analyses. If sampling is directed toward a small target, use the resulting weighted flux and verify convergence. The number of displayed rays reaching the camera is not itself a brightness measurement. [1, 3]
Separate optical collection from image detection
Collected flux at a modeled pupil is useful for screening designs, but it is not automatically pixel-level defect contrast. Compare equal or appropriately normalized surface regions and state how the defect and background contributions were isolated.
To predict an image, account for imaging optics, magnification, blur, pixel sampling, and the detector response. The inspection system’s exposure, noise, and processing also affect whether the defect is detected. A ray-traced improvement in collected defect-to-background signal should therefore be presented as an optical result, not an asserted production detection rate.
Keep the metric and thresholds tied to the inspection requirement. Save both signal components alongside any ratio so a large ratio caused by an almost vanishing denominator does not hide an impractically weak signal.
Test production variation
Sweep plausible part positions, tilt, defect orientations, and surface finishes. Include the expected range of LED output and the mechanical tolerances that change illumination angle. Track worst conditions rather than reporting only the most attractive nominal view.
If the background remains too bright, examine alternative angular distributions, wavelengths, and illumination arrangements supported by the application. Any use of polarization should be modeled and tested with the relevant optical properties. There is rarely one universal adjustment that resolves all finishes and defects.
Validate with representative parts
Build a comparison using good parts and known defects spanning the required range. Match the camera settings, geometry, and surface condition to the model, then compare the patterns and relative signals. Investigate discrepancies before using the model to select final hardware.
Download a free TracePro trial to evaluate ring-light geometry with your source and surface data. Use the simulation to narrow the physical experiments and document why the chosen illumination is robust across the parts that matter.
