Stray Light Is a Specification: Blackening, Edge Treatment and Unbudgeted Reflections

Optical Design and Stray Light Control · 2026-10-20 · 4 min read

When an imaging system underperforms, the lens design gets blamed first. In practice a large share of contrast loss traces to light that took a path nobody modelled: a reflection off a ground edge, a bounce between two filter surfaces, a scatter halo from a polished chamfer, or a glint from the inside of a mount.

Stray Light Is a Specification: Blackening, Edge Treatment and Unbudgeted Reflections

None of those are in the optical design file. All of them are in the hardware, and all of them can be specified.

The three paths stray light takes

It helps to separate the mechanisms, because each has a different fix.

Surface scatter comes from roughness and from defects: scratches, pits, sleeks and coating pinholes. It produces a broad halo around bright sources and reduces contrast everywhere at once. Its cure is better polishing and better cleanliness, and its specification is scratch-dig plus, increasingly, a roughness or total scatter figure.

Specular reflections from unintended surfaces come from edges, chamfers, mounting rims and the uncoated back faces of windows. These produce discrete ghosts, and they are the most common cause of a bright artefact that appears only when a source is in a particular position. Their cure is geometry and blackening.

Multiple internal reflections occur inside transmissive components: light bouncing between the two surfaces of a window or a filter, or between a filter and the sensor cover glass. These create faint, offset copies of the image and a general veiling glare. Their cure is AR coating plus, critically, wedging or tilting so that the reflected light misses the aperture.

Blackening: what it can and cannot do

Edge blackening is the standard treatment for ground and chamfered surfaces, and it is often specified loosely — "edges blackened" on a drawing, with no definition of what that means.

The useful specification is optical: reflectance of the blackened surface, over the operating band, at grazing incidence. A blackening process that reads 2 percent at normal incidence can read far higher at the shallow angles that actually matter, because edge surfaces in a barrel are usually seen at grazing incidence. Ink-based blackening, anodised metal and chemically etched glass all behave differently at those angles, and the difference is often larger than the difference between two coating designs.

Two practical cautions. First, many blackening processes are not compatible with high-power laser fluence; a blackened edge that overheats is a damage site. Second, blackening adds a process step that must be compatible with cleaning — a blackened edge that sheds particles into a clean assembly is a defect source.

Aperture stops and baffling

The cheapest stray-light fix is usually geometric: stop the light before it reaches the problem surface. An aperture stop placed at a conjugate plane, a baffle inside a barrel, or a simple light trap can remove a ghost that no coating change would eliminate.

The design discipline is to ask, for each surface, "what can see what?" A filter at 45 degrees in a beamsplitter assembly, for example, can usually see the sensor package, the barrel wall, and the source — and the reflection path between the sensor and the filter is the one most often missed.

Wedging, tilting and the filter-specific case

Thin filters are particularly prone to internal reflection artefacts because their surfaces are parallel and their coatings can be highly reflective at some wavelengths, even when they transmit well in band.

Three mitigations are common. Tilting the filter a few degrees moves the ghost off the detector. Wedging the substrate — polishing the two faces at a small angle — spreads the ghost into a smear rather than a point. And coating the second surface with a broadband AR layer reduces the amplitude of the reflection before it can bounce.

Each has a cost. Tilting introduces spectral shift, which matters for narrowband filters. Wedging adds polishing cost and complicates metrology. The choice depends on whether the ghost is worse than the shift, which is a system question, not a component question.

What to put on the drawing

  • Reflectance of edge and mounting surfaces, stated at grazing incidence and over the operating band.
  • Scratch-dig for optical surfaces, and a separate criterion for surfaces that are not in the optical path but can scatter into it.
  • Chamfer geometry and its treatment — an untreated chamfer is a bright line waiting for a bright source.
  • For transmissive parts, whether wedge or tilt is permitted, and by how much.

Why this is worth attention now

Sensors keep getting more sensitive and systems keep getting more compact, and both trends make stray light more visible. A ghost that was lost in a large, dark housing becomes obvious when the same optical function is packed into a 20 mm module next to an illuminated surface.

The parts that cause the problem are rarely the expensive ones. They are the edges, the chamfers, the spacers and the uncoated faces — the components that appear on a mechanical drawing rather than an optical one. Specifying them is not glamorous work, but it is where contrast is won.

← All 410 articles

Need a Quote or Engineering Review?

Send your drawings, specifications or coating requirements — we reply within 24 hours.

Request a Quote

Related Articles

Request a Quote