What a colorimeter is
Colour is a perception produced by a light source, a surface and an observer together, and a colorimeter replaces all three with fixed standards so that a colour can be written down as numbers. That is what makes colour a specification: a target is agreed as coordinates, a tolerance is agreed as a difference, and from then on a sample either falls inside it or does not, regardless of who is looking at it or under which lamp.
The instrument works by illuminating the sample in a defined way and measuring the light that returns from it through a set of filters or a dispersive element, from which coordinates in a standard colour space are calculated. Two families are distinguished by their optics. A filter instrument, usually built around the forty-five degree geometry, illuminates at an angle and views at another, and reads a surface the way an observer standing beside it would. A spectral instrument measures across the wavelength range and computes the same coordinates from the spectrum, which allows it to do more with the data.
The distinction that decides most specifications is not the family but the geometry, and specifically whether the gloss of the surface is included in the measurement. A sphere, or integrating, geometry collects both the light reflected in the specular direction and the light scattered by the surface, so a sample is measured as though the gloss were part of its colour. A forty-five degree geometry excludes the specular component by design. The same paint measured under the two arrangements gives different numbers, and neither is wrong.
How a colorimeter measures
The measurement begins with the illuminant and the observer. A colour is only defined relative to a light source, so the instrument contains or assumes one, and a standard observer function stands in for the human response. Both are conventions agreed by international committee rather than properties of the instrument, and a reading quoted without them is not reproducible.
From the measured reflectance the instrument computes coordinates in a colour space, which describe the sample as a lightness and two chromatic components. A difference between a sample and a standard is then a distance in that space, expressed as a single number. The formula used to compute that distance matters, because different formulae weight the components differently and a pair of samples can sit on either side of a tolerance depending on which one is applied.
Repeatability and agreement are two different figures and are often quoted as one. Repeatability is the spread of readings taken on the same instrument of the same sample, and it sets the smallest change the instrument can detect. Agreement between two instruments is worse, because it depends on their individual calibration, their geometry and their spectral response. Two instruments can each be repeatable to a small fraction of a unit and disagree with each other by more than the tolerance being applied.
Configuration choices
Most of the specification is fixed by these decisions rather than by the accuracy class. Each one constrains the next, which is why they are settled in order.
| Configuration | Options to fix | What it actually decides |
| Measurement geometry | Forty-five degree illumination with a normal view, or a sphere that includes the specular component | Whether gloss is part of the colour. A textured, grained or gloss-varied surface behaves differently in each, and the choice has to follow the sample and the tolerance. |
| Aperture size | A small or large measurement area, and whether more than one is available | Whether a small feature can be measured without picking up its surroundings, and whether a textured sample is averaged over enough area to be representative. |
| Colour space and difference formula | The coordinates reported, and the formula used to compute a difference between sample and standard | Whether a quoted tolerance is being evaluated on the basis it was written for. The same pair of samples can pass under one formula and fail under another. |
| Illuminant and observer | The light source assumed in the calculation, and the standard observer function applied | Whether a reading taken at one site can be compared with a reading taken at another. Without a common convention the numbers are not comparable at all. |
| Repeatability | The spread of repeated readings on the same sample, and the conditions under which it is specified | The smallest difference the instrument can be relied on to detect, which has to be comfortably smaller than the tolerance it is being used to police. |
| Captured data and library | Standards stored on the instrument, a colour library of measured samples, or a connection to a phone or computer | Whether a judgement can be made at the machine and whether a record of it exists afterwards. A tolerance that is not recorded cannot be defended. |
| Calibration arrangement | The white reference used to zero the instrument, and how often it is applied | Whether the readings stay comparable over a shift and between instruments. This is the practice that keeps a colour programme honest, more than the resolution of the display. |
The geometry is the first decision, because it decides what the number means. Everything else refines how precisely that number is repeated.
How to choose a colorimeter
| Decision | What settles it | What to ask |
|---|
| The surface being measured | Flat and uniform, or textured, grained and glossy | Texture and gloss decide the geometry and the aperture together. A sample with a directional surface needs the arrangement that averages it rather than one that catches it at the wrong angle. |
| The size of the area to be judged | A large panel, or a small printed detail | Aperture size decides whether the instrument measures the feature or the average of the feature and everything beside it. |
| Who sets the tolerance | A customer specification, or a figure the producer chooses | A tolerance quoted in a difference unit is only meaningful with the formula and the geometry attached; a tolerance without them is not a specification. |
| The margin the process can hold | A wide tolerance, or one close to the repeatability of the instrument | Where the tolerance is narrow, the repeatability of the instrument and the calibration practice become the limiting factors rather than the accuracy. |
| Whether the judgement is made on the line | A decision at the machine, or a sample sent to a laboratory | A portable instrument puts the decision where the work is and needs a display, a battery life and a tolerance check built into it. |
| Whether colours have to be stored | A standard or two, or a library of approved colours | A library makes the standard reproducible when a physical sample is lost or has faded, which is the usual reason a colour programme breaks down. |
| Whether the reading has to be reported | A pass or fail marked on a traveller, or data exported to a system | The export path is specified with the instrument, because a measurement that has to be retyped will eventually be retyped wrongly. |
If any of these are still open, send what is known. A quotation can be issued against a duty description and refined once the gaps are filled.
The nearest alternatives
These are the types an enquiry for a colorimeter usually turns out to be, and the point at which the choice is genuinely decided.
| Nearest alternative | What separates it from a colorimeter |
| Spectrophotometers | Measure how much light a sample absorbs or transmits across a range of wavelengths, in order to determine what the sample contains. A colorimeter measures the colour of a surface against a standard. The instruments share an optical bench and answer different questions, and where a concentration is wanted the spectrophotometer is the one that is calibrated against a chemical standard. |
| TOC analyzers | Report the total organic carbon in a water sample by oxidising it and measuring the result, which is a destructive determination of content rather than an assessment of appearance. |
| Water quality analyzers | Report the condition of a water by several methods at once. Where colour is one of them, it is usually measured as absorbance in a cell rather than as the appearance of a surface, and the two meanings of colour should not be carried across. |
| Particle size analyzers | Report a size distribution rather than a colour. A powdered sample is measured by both instruments and the two results describe different properties of the same powder. |
| Coating thickness gauges | Measure how thick a coating is by a magnetic or eddy-current method. Colour and film thickness are the two figures a painting or coating specification usually sets together, and each needs its own instrument. |
The boundary worth settling first is what the number is for: the appearance of a surface, or the concentration of a substance in a liquid. Both are called colour measurement and they are not interchangeable.
What to include in the enquiry
The first reply can be a quotation rather than a set of questions if the enquiry carries these. Where a figure is not known, send the duty description and we will confirm what it implies.
- The surface to be measured, including whether it is textured, glossy or grained
- The size of the area that has to be judged, and therefore the aperture required
- The tolerance to be applied, and the colour space and difference formula it is written in
- The illuminant and observer conventions the specification assumes
- Whether a portable instrument is wanted, or a bench instrument in a fixed location
- Whether a standard or a library of standards has to be stored and recalled
- Whether the readings have to be exported, and to what system
- The repeatability the process requires relative to the tolerance being policed
- Quantity, destination port and required delivery date
