How Filter Curves Are Actually Measured: Spectrophotometer Geometry and the Numbers Behind the Datasheet

Filter Metrology and Reliability · 2026-10-04 · 3 min read

Every filter buyer compares transmission curves. Very few ask how those curves were produced — and yet the measurement geometry can change the curve as much as the coating design does. Two suppliers can ship the same physical filter and quote transmission values that differ by several percent, with both being honest.

How Filter Curves Are Actually Measured: Spectrophotometer Geometry and the Numbers Behind the Datasheet

The spectrophotometer basics that matter

A transmission measurement compares the light through the sample against a reference beam. The instrument reports the ratio. What is inside that ratio depends on choices the operator makes: spot size, angular cone, polarisation state, sample orientation, and whether the measurement is single-beam corrected or double-beam referenced.

Spot size and inhomogeneity

A spectrophotometer with a large spot averages over a region of the filter. If coating thickness varies across the part — and it always does, because chamber geometry produces a thickness distribution — a large spot averages the good and the bad together and reports a curve the actual aperture does not show. A small spot reports the local curve, which may be excellent at one point and unrepresentative of the working field.

Neither choice is wrong. But a supplier quoting with a large spot is reporting a different part than a buyer measuring with a small one. The rule: the measurement spot should approximate the actual aperture used in the application, or a map should be requested.

The cone of light

Filters are specified at normal incidence, but spectrophotometers illuminate with a finite angular cone — typically a few degrees. An interference filter shifts blue with angle, so a converging cone smears the passband: the measured curve is the average across the cone, with the edge softened and the peak flattened. The wider the cone, the worse the apparent steepness — for a filter that may be perfectly steep when used in a collimated beam.

This is the single most common cause of "the received filter looks worse than the datasheet". It is not the filter. It is the cone.

Polarisation

At non-normal incidence, s and p polarisation see different effective layer thicknesses and shift differently with angle. An unpolarised measurement averages the two. For a dichroic at 45°, the s curve and the p curve can differ by several nanometres in edge position; the unpolarised average sits between them and matches neither. Buyers of angle-critical parts should request s and p curves separately — a measurement capability not every supplier has.

Blocking and the measurement floor

Optical density is where measurement limitations bite hardest. A spectrophotometer's stray light sets a floor around OD 4–5; beyond that, the instrument reports its own stray light, not the filter's blocking. Deep blocking is verified differently: with a laser-based setup at specific wavelengths, or by calculation from the design combined with sampled laser measurements. A datasheet claiming OD 6 across a wide range should state how that range was measured. If it does not, ask.

Practical requests to make

  • State the measurement geometry you will accept: spot size, angular cone, polarisation, sample temperature.
  • For angle-critical parts, request curves at the working angle, with s and p separated.
  • For steep edges, request the edge slope figure with the measurement cone stated.
  • For deep blocking, request laser-based spot checks at your critical wavelengths.
  • When comparing suppliers, compare measurement conditions first. A 2% transmission difference between two datasheets may vanish — or double — once geometry is normalised.

A transmission curve is an engineering document, not a marketing image. The conditions under which it was produced are part of the specification, and the buyers who put those conditions in writing are the ones who stop losing arguments with their own incoming inspection.

Product page: optical filters and coated components — specifications, coating options and lead times.

← All 430 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