Dichroic or Polarizing Beamsplitter? The Combiner Choice That Decides Waveguide Efficiency

AR/VR and Smart Glasses · 2026-10-06 · 3 min read

Near-eye displays have a brutal optical budget. Light from a micro-display a few millimetres wide has to be collimated, in-coupled, expanded through a waveguide, out-coupled toward the eye, and still arrive bright enough to be readable outdoors. Every optical interface takes a cut, and the combiner is where the largest single cut happens.

Dichroic or Polarizing Beamsplitter? The Combiner Choice That Decides Waveguide Efficiency

Two component families do that job, and they fail in different ways.

Dichroic combiners

A dichroic stack separates wavelengths by interference: reflect the display band, transmit the rest. It does not care about polarisation, which is its main advantage in a waveguide where stress birefringence and curved surfaces scramble the polarisation state. It can also be designed for RGB in a single stack, with three reflection bands and transparent windows between them.

The cost is angular sensitivity. A dichroic edge shifts toward shorter wavelengths as the angle of incidence increases, and in a waveguide the beam hits the out-coupler across a range of angles by design. A stack that is centred at 532 nm at 45° will not be centred at 532 nm at 55°. The result is a colour shift across the field of view — the classic complaint that the image looks green at the centre and magenta at the edge.

Polarizing beamsplitters

A PBS separates by polarisation rather than wavelength, so it is spectrally broadband and comparatively insensitive to wavelength drift. In a system that already manages polarisation — typically with a quarter-wave retarder and a reflective combiner — a PBS can reach high extinction and good uniformity.

Its weakness is exactly that dependency. Any depolarisation in the waveguide, at the in-coupler, or from the display itself leaks straight into the wrong channel, and the leakage shows up as veiling glare rather than as a sharp artefact. Polarisation management also adds components, alignment steps and cost.

How the choice is actually made

Three questions decide it in practice. Does the system already control polarisation? If not, a PBS is a liability and a dichroic is the safer architecture. How wide is the angular range at the combiner? Wide ranges punish dichroics and favour polarisation-based designs. Is the display narrowband — a laser or an LED with a defined band — or broadband? Narrowband sources let a dichroic be designed with tight bands and high transmission between them; broadband sources force wider stacks and more layers.

The manufacturing detail that gets missed

Whichever family is chosen, uniformity across the aperture becomes the governing specification once the part is larger than a few millimetres. A combiner is not evaluated at one point. It is evaluated over the whole field, which means the coating thickness distribution across the substrate has to be controlled to a fraction of a percent. That is a chamber-geometry problem as much as a design problem, and it is the reason two suppliers can quote the same stack and deliver visibly different images.

Ask for transmission and reflection maps across the aperture at the working angle, with the measurement geometry stated. A single-point spectrophotometer curve is not evidence of a combiner.

← All 400 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