The color rendering index (CRI) determination experiment measures how accurately a light source reproduces the colors of illuminated objects compared with a reference illuminant, making it one of the most important photometric tests in modern lighting quality assessment. Performed in accredited third-party laboratories with calibrated spectroradiometers and standardized procedures defined by CIE 13.3, CIE 15, IES TM-30, and GB/T 5702, the experiment delivers the general index Ra, special indices R1 through R14, and complementary fidelity and gamut metrics. This technical guide explains the principles, sample requirements, test items, measurement methods, and practical applications of CRI determination.
Technical Overview
The color rendering index (CRI) determination experiment is a standardized photometric and colorimetric test that quantifies the ability of a light source to reveal the true colors of illuminated objects in comparison with a reference illuminant of the same correlated color temperature (CCT). Results are expressed as dimensionless indices on a scale of 0 to 100, where 100 represents rendering indistinguishable from the reference source. The governing methodology is CIE 13.3-1995, supported by the colorimetric computation rules of CIE 15:2018 and ISO/CIE 11664, extended by CIE 224:2017 for high-fidelity sources above R90, and complemented by the IES TM-30-20 two-measure evaluation system. The corresponding Chinese national standard, GB/T 5702, is technically aligned with the CIE framework.
During the experiment, the spectral power distribution (SPD) of the test source is measured with a calibrated spectroradiometer. Fourteen defined test color samples are then mathematically rendered first under the test source and then under the reference illuminant. The color differences observed for each sample are converted into special color rendering indices R1 through R14, and the average of the first eight indices (R1 to R8) yields the general color rendering index Ra. A larger chromatic shift means a lower index; incandescent sources achieve Ra values near 100 at low correlated color temperatures, while typical LED products range from Ra 70 to Ra 98.
Because the eight-sample average can conceal weaknesses in saturated colors, contemporary specifications frequently require separate reporting of R9 (saturated red), the complete set of special indices, and TM-30 fidelity (Rf) and gamut (Rg) metrics, providing a more complete picture of color rendering performance.
Testing Samples
LED Lamps and Luminaires
Finished LED products, including omnidirectional bulbs, tubes, downlights, panel lights, high-bay luminaires, and street lighting modules, constitute the majority of submitted samples. Each unit should be a representative production item, complete with its intended diffusers, lenses, and optical covers, since any spectral modification by the optics influences the measured result. Samples are operated at their rated electrical input, typically alternating current at the declared voltage and frequency, in their designated mounting orientation.
Conventional Discharge and Filament Sources
Fluorescent lamps, metal halide lamps, high-pressure sodium lamps, and halogen or incandescent lamps remain important test categories, particularly for legacy installations, replacement programs, and reference-source studies. Discharge and filament lamps are generally seasoned before measurement, commonly 100 hours of aging in accordance with established IEC test conventions, so that their spectral output reflects stabilized service behavior rather than initial run-in characteristics.
LED Packages, Modules, and Light Engines
Component-level samples, including mid-power packages, chip-on-board arrays, flexible strips, and remote-phosphor elements, are tested on standardized heat sinks at the specified junction conditions, such as a defined case temperature and drive current, as declared by the manufacturer. Testing components under controlled thermal conditions is essential because phosphor conversion efficiency, and therefore both CCT and color rendering, shifts with temperature.
Sample Quantity and Conditioning
To account for production spread, laboratories normally test several units per model, commonly three to five, and report the arithmetic mean together with the observed range. Samples must be clean, undamaged, electrically intact, and accompanied by their rated electrical and photometric data. Before measurement, every sample is thermally and electrically stabilized inside the test enclosure until its light output varies by less than approximately 0.5 percent over a 15-minute interval.
Testing Items
The CRI determination experiment yields a defined package of colorimetric results, typically including the following items:
- General color rendering index Ra, the arithmetic mean of special indices R1 through R8.
- Special color rendering indices R1 to R14, notably R9 (strong red), R10 (strong yellow), R11 (strong green), R12 (strong blue), R13 (skin tone), and R14 (leaf green).
- Correlated color temperature (CCT) in kelvin and the distance from the Planckian locus (Duv), which together locate the source chromaticity and determine the correct reference illuminant.
- Spectral power distribution from approximately 380 nm to 780 nm, the primary measured quantity from which all derived indices are computed.
- Supplementary TM-30-20 results, including fidelity index Rf, gamut index Rg, and the color vector graphic, when a broader rendering characterization is required.
- Extended long-CRI fidelity scoring above R90 in accordance with CIE 224:2017 for premium high-fidelity products.
Testing Methods
Instrumentation and Calibration
The core instrument is a bench or array spectroradiometer covering at least the 380 nm to 780 nm band, with a recommended spectral bandwidth of 5 nm or less and fine wavelength sampling for accurate colorimetric integration. Irradiance responsivity is calibrated with a standard lamp traceable to a national metrology institute, and the wavelength scale is verified with emission-line reference lamps. Depending on source geometry, flux is collected either inside an integrating sphere in a 2π or 4π configuration or with a cosine-corrected receptor placed at a fixed distance inside a light-tight enclosure. The laboratory environment is held at 25 °C ± 1 °C, and the supply voltage and frequency are regulated to the rated values with low total harmonic distortion.
Measurement Procedure
The experiment follows a fixed sequence designed to guarantee reproducibility across laboratories:
- Registration and inspection: record model data, rated electrical parameters, and nominal photometric claims; verify mechanical and electrical integrity.
- Seasoning: operate discharge and filament lamps for the prescribed aging period; allow LED products to reach thermal equilibrium in the test setup.
- Mounting: install the sample in the geometrically defined position with its declared operating orientation and any permanent external control gear.
- Stabilization: energize the sample until successive photometric readings satisfy the stated stability criterion.
- Spectral acquisition: record the SPD over repeated scans, apply dark-current and stray-light corrections, and average the accepted scans.
- Computation: derive CCT, Duv, chromaticity coordinates, and all color rendering indices using the standardized algorithms.
- Reporting: document results for each unit, statistical summaries across units, operating conditions, instrument traceability, and measurement uncertainty.
Calculation Principle
From the measured SPD, tristimulus values are computed with the CIE 1964 10-degree standard observer, and the chromaticity is projected into the 1960 UCS diagram to obtain CCT and Duv. A reference illuminant is then selected: a Planckian radiator for sources below 5000 K and a daylight D-series illuminant for sources at or above 5000 K. The fourteen test color samples are rendered under both the test source and the reference illuminant, and the test-source data are adjusted with the chromatic adaptation transform defined in CIE 13.3. Color differences for each sample are evaluated in the CIE 1964 U*, V*, W* uniform color space and converted through the relation Ri = 100 − 4.6 × ΔEi, from which Ra is the arithmetic mean of R1 to R8. By convention the indices are capped at 100, while poorly rendering narrowband sources can produce negative special indices.
Uncertainty and Quality Control
Principal uncertainty contributions include wavelength-scale error, spectral bandwidth mismatch, detector nonlinearity, stray light, calibration-lamp uncertainty, source drift, temperature sensitivity, and unit-to-unit variability of the samples. Reputable laboratories control these effects through scheduled recalibration, working-standard verification before each session, control charts on reference lamps, participation in interlaboratory comparisons, and full documentation of correction algorithms. Typical expanded uncertainty (coverage factor k = 2) for Ra lies in the region of one to two index points, with somewhat larger uncertainty for R9 because of its strong dependence on deep-red spectral content.
Applications
Color rendering data generated by the determination experiment are used throughout the lighting industry and beyond:
- Regulatory compliance and market access: many national regulations and energy-labeling schemes set minimum thresholds, commonly Ra of at least 80 for general indoor lighting and Ra of at least 90 for color-critical environments, supported by harmonized reporting under frameworks such as EN 13032-4 and the IEC 62722 performance series.
- Manufacturing quality control: end-of-line verification of phosphor blending, LED binning consistency, and color stability across production batches and thermal conditions.
- Color-critical facilities: specification and acceptance testing for museums, art galleries, textile and printing inspection, medical examination and dental shade matching, food retail display, and film or television studios, where fidelity of skin tones and saturated hues is decisive.
- Product development: optimization of phosphor mixtures, multi-channel spectral engines, and human-centric or circadian lighting designs that must balance color rendering with spectral power in specific wavelength bands.
- Procurement verification: independent confirmation by specifiers, contractors, and facility owners that delivered luminaires meet the color rendering values declared in technical datasheets.
FAQ
What CRI value is considered good for general lighting?
For most residential, office, and commercial interiors, Ra of at least 80 is regarded as the baseline. Values of Ra 90 and above are recommended wherever color appearance matters, such as retail, hospitality, healthcare, and education, while Ra 95 to 98 is typically specified for museums, studios, and medical color assessment.
Why is R9 reported separately from Ra?
R9 describes the rendering of strong red, and it is not included in the eight-sample average that produces Ra. Many phosphor-converted LED sources with Ra around 80 exhibit weak or even negative R9 because of limited deep-red emission, which visibly degrades the appearance of skin tones, meat, flowers, and red surfaces. Consequently, professional specifications commonly require R9 greater than zero, and often 50 or higher for demanding applications.
How does CRI differ from the TM-30 method?
The traditional CRI system averages eight moderately saturated samples in a 1960s-era color space and rewards only fidelity to the reference. TM-30-20 evaluates 99 color evaluation samples in the modern CAM02-UCS color appearance model and reports both a fidelity index (Rf) and a gamut index (Rg) together with a color vector graphic, distinguishing sources that increase colorfulness from those that dull colors. TM-30 therefore resolves well-known blind spots of Ra while remaining fully complementary to it.
Can a light source achieve a CRI above 100?
No. The index scale is anchored so that the reference illuminant scores exactly 100, and calculated values are capped at that maximum. Conversely, narrowband sources that severely distort object colors can return negative special indices, which are reported as less than zero or truncated to zero depending on the reporting convention.
What equipment is essential for the experiment?
A calibrated spectroradiometer with an appropriate flux-collection geometry, an integrating sphere or a cosine-corrected receptor, standard lamps for wavelength and spectral responsivity calibration, a stabilized power supply, a temperature-controlled dark enclosure, and compliant analysis software implementing the CIE 13.3 and TM-30 algorithms constitute the minimum laboratory configuration.
Does dimming affect the measured color rendering index?
Yes. The chromaticity, and occasionally the spectral shape, of a source can change with drive level, especially for dimmed LED systems in which phosphor behavior shifts with junction temperature. The experiment is therefore performed at the declared rated operating point, and any dimmed-condition values are separately measured and clearly identified in the test report.