Light Measurement for LED, Displays and Optical Light Sources

Light measurement for LED and displays quantifies how much optical radiation a source emits and how that radiation is perceived by the human eye or by an image sensor. The central technical goal is the traceable separation of radiometric quantities (physically present radiant power) and photometric quantities (brightness weighted by eye sensitivity). The central challenge is that the two evaluation systems diverge markedly for narrowband sources - coloured LEDs, OLED subpixels, quantum-dot converted displays - and a simple broadband sensor does not reliably resolve that divergence. The decisive physical quantities are therefore radiant flux, luminous flux, luminance, colour coordinates, peak wavelength and, for switched sources, the temporal modulation of brightness. Only a spectrally resolved measurement yields all quantities derived from it consistently.

The market for LED displays (directly emitting modules using SMD, mini-LED, micro-LED or COB technology, excluding LCD backlights) is put at around USD 19.7 billion for 2025, with an expected annual growth rate of around 5.4 % through 2031 (Mordor Intelligence, 2025). Within the overall display market, micro-LED is regarded as the fastest-growing technology segment, with a rate expected for 2031 of around 9.7 % against around 7.2 % for the overall market (Mordor Intelligence, Display Market, 2025). For the LED lighting market alone, estimates range between roughly USD 100 and 110 billion (2025) at growth rates of 8 to 13 % depending on market definition (Grand View Research, 2025; Fortune Business Insights, 2026). These figures come from commercial market studies with differing market definitions and should be read as an order of magnitude, not as official statistics.

The technological picture from the International Energy Agency is on firmer ground: lighting accounts for roughly 8 % of global electricity demand in buildings and outdoor applications (around 2,200 TWh, 2024); the average luminous efficacy of LEDs sold today has nearly doubled since 2015 to close to 100 lm/W, with leading products reaching up to 230 lm/W (IEA, 2026). Around 110 countries now have binding minimum energy performance standards (MEPS) for lighting. For measurement technology this development means a shift from internal quality control to regulatory-grade, evidentiary testing: luminous efficacy, colour point and ageing behaviour increasingly have to be documented in a standard-compliant, traceable way, not merely plausibility-checked in-house.

A further driver is EU Ecodesign Regulation (EU) 2019/2021 for electronic displays, which since 2023 has prescribed, among other things, a dedicated efficiency category for micro-LED displays and fixed requirements for peak white luminance under defined ambient illuminance (100 lx) (EUR-Lex, Regulation (EU) 2019/2021). A revision of this framework has been under way since November 2024; the European Commission plans drafts for early 2026 (European Commission, 2024). Market growth in mini-/micro-LED and OLED is thus meeting a tightening regulatory testing obligation - together, both raise the requirements on measurement accuracy, spectral resolution and automation in manufacturing and incoming inspection.

Radiometric quantities - irradiance, radiance, radiant flux - describe the physically present radiant power independently of the human eye. Photometric quantities - illuminance, luminance, luminous flux - describe the same process weighted with the spectral luminous efficiency function V(λ) of the human eye. V(λ) is zero outside roughly 380 to 780 nm: radiation in that range physically exists but does not contribute photometrically, yet it can still cause material ageing, photobiological effects or sensor response. The conversion runs through the constant K_m = 683 lm/W, defined at the maximum of V(λ) at 555 nm (CIE 018:2019). Because V(λ) is a fixed, tabulated curve, a spectrally resolved measurement lets you derive both radiometric and photometric quantities as well as colour coordinates, colour temperature and colour rendering index - the reverse path, from a single photometric number back to the spectrum, is not possible.

Displays and LEDs differ fundamentally in how light is generated and modulated. For measurement technology, what matters is whether a source emits broadband or narrowband light, whether it is self-emissive or backlit, and how it is driven over time.

Luminous flux or radiant flux describe the total power a component emits in all directions and are the reference quantity for efficacy and binning; they can only be captured direction-independently with an integrating sphere, because point measurements do not capture the radiation pattern. Luminance is direction-dependent, since it describes what a surface emits toward a specific viewing direction; it determines how bright a display appears to a viewer at a defined angle, and must therefore be captured at defined image positions and viewing angles. Colour coordinates, colour temperature and colour rendering index depend directly on the spectrum and determine colour fidelity; they are particularly sensitive to the spectral resolution of the measurement system for narrowband primaries. Forward current and junction temperature affect luminous flux, peak wavelength and colour point of an LED simultaneously, because higher temperature shifts the semiconductor band gap and thus the emission wavelength - comparing two measurements is only valid at the same electrical and thermal operating point. How optical output shifts over service life is described, for the UV types, under UV LEDs for UVA, UVB and UVC. Flicker metrics (P_st LM, SVM, flicker index) quantify the temporal modulation of brightness in PWM-dimmed sources and are decisive, because the same average luminance can have a completely different physiological effect depending on modulation frequency and depth.

Electrical power draw is not an optical quantity: the efficiency of an LED in converting electrical into optical power is typically well below 100 % and is itself temperature- and current-dependent; watts alone give no basis for deriving luminous flux or irradiance. A time value alone is not sufficient for pulsed or PWM-dimmed sources, because the measurement result depends on the instrument's integration time - the same source yields different apparent brightness for short and long integration times if the dimming frequency and integration window are not matched. The operating point is often underestimated: a measurement in the cold state immediately after switch-on and a measurement after burn-in are two different measurements, because luminous flux, peak position and colour coordinate shift with junction temperature. A broadband sensor only approximates V(λ) approximately; for metameric but spectrally different sources - for example a warm-white LED and a halogen lamp with the same colour impression - the same sensor delivers two different readings, of which at most one is correct. A measurement outside the actual point of interest is insufficient for displays, because luminance and colour point vary across the surface and across viewing angle; a single-point measurement at image centre captures no edge shading, no blooming and no angular dependence. These assumptions each hold only under specific conditions: a broadband measurement remains adequate as long as the sources being compared are sufficiently similar spectrally - for narrowband or unknown spectra that condition no longer holds.

Two spectra that look identical to the human eye are called metameric. The narrower the primary colours of a source - as with coloured LEDs or QD-OLED displays with spectral FWHM values of only 20 to 40 nm (Samsung Display, 2023) - the more strongly the results of different sensor types diverge for metameric comparison sources. Three-filter colorimeters approximate the human eye's colour-matching functions using three broad filter curves; for narrowband spectra this approximation becomes inaccurate, because the source's actual spectral distribution no longer lies within the filter curves' tolerance. The characterisation of such spectral mismatch is specified normatively in CIE 015:2018 (colorimetry, including new standard illuminants for various LED types) and in CIE 233 (calibration and characterisation of array spectroradiometers) (CIE 015:2018; CIE 233).

Geometrically, displays and micro-LEDs add the complication that the radiation pattern cannot be treated as point-like: for very small emitters (micro-LED chip size in the single-digit micrometre range), near- and far-field effects overlap, and for mini-LED backlights with local dimming, stray light between neighbouring zones has to be captured direction-resolved. In a published test case, a bright image area of 340 nits was measured against an adjacent, notionally dark frame of only 0.0021 nits; the spectroradiometer used had to resolve down to 0.0005 cd/m² and was operated at five defined distances to shield bright image areas (Allion Labs, 2023). What matters here is not peak luminance but the dynamic range and stray-light suppression of the measurement system.

The relationship between radiometric radiant flux and photometric luminous flux is defined by the integral over the spectral distribution:

Φ_v = K_m · ∫ Φ_e,λ(λ) · V(λ) dλ

Here Φ_v is the luminous flux in lumens, Φ_e,λ(λ) the spectral distribution of radiant flux in W/nm, V(λ) the photopic spectral luminous efficiency function (dimensionless, maximum 1 at 555 nm), and K_m = 683 lm/W the photometric radiation equivalence constant (CIE 018:2019 - The Basis of Physical Photometry). The model assumes photopic vision (light-adapted eye, 2° field of view); for very low luminance levels (mesopic or scotopic vision) a different weighting function applies, and V(λ) describes a statistical standard observer, not individual perception. The practical consequence: two radiation sources with identical radiometric power but different spectral position produce completely different luminous flux - the conversion is wavelength-dependent and not linear with electrical or radiometric power.

1. Assumptions. Three monochromatic LED sources each emit exactly 1 mW of optical radiant power - a blue LED at 470 nm, a green LED at 555 nm (the maximum of V(λ)) and a red LED at 625 nm. Tabulated V(λ) values per CIE 018:2019: V(470 nm) = 0.0910; V(555 nm) = 1.0000; V(625 nm) = 0.3210 (BIPM, Principles Governing Photometry, 2019).

2. Formula.
Φ_v = K_m · V(λ) · Φ_e, with K_m = 683 lm/W.

3. Calculation

4. Result. At an identical radiometric power of 1 mW, the resulting luminous flux ranges from around 0.062 lm (blue) to 0.683 lm (green) - a factor of about 11.

5. Technical interpretation. Datasheet figures in milliwatts of optical power are not directly comparable for judging visual brightness as long as wavelength is disregarded. Anyone comparing LEDs of different peak wavelength by their radiant power is not comparing perceptually equivalent quantities. Only a spectrally resolved measurement allows the correct, wavelength-specific conversion into luminous flux.

In consumer electronics, the combination of luminance, colour gamut and viewing-angle dependence determines display quality; what matters most here is colour measurement on narrowband primaries (QD-OLED, micro-LED) and contrast measurement under local dimming. In the automotive industry, LED tail lights, matrix headlamps and increasingly large in-vehicle displays are tested photometrically and colorimetrically; what counts there is the luminous intensity distribution at defined angles and colour stability across temperature range and service life. In medical technology, surgical displays and diagnostic monitors must maintain a defined, calibrated luminance and greyscale response, because image interpretation depends directly on photometric consistency. In semiconductor and LED manufacturing, the process quantity is the reproducible assignment of luminous flux, peak wavelength and colour point to batches (binning); integrating spheres and spectroradiometers are the standard tools here, because only they capture the operating-point-dependent total luminous flux direction-independently. In research and development - for example when characterising micro-LED chips in the single-digit micrometre range - the critical quantity is size-dependent efficiency, since smaller chips show an altered ratio of radiative to non-radiative recombination paths and therefore a different wavelength and efficiency characteristic (Fan et al., Optics Express, 2023). In digital signage, peak white luminance under ambient light is particularly relevant metrologically, because it feeds directly into the energy efficiency class under regulation (see market context above). In occupational safety and photobiological safety assessment, spectral irradiance in the 380-780 nm range is the decisive quantity for assessing blue-light hazard under IEC 62471 - a topic area covered in depth by the market page Occupational safety and photobiological safety. For the neighbouring ultraviolet range, comparing sources by their spectrum is likewise shown by the spectral database of UV lamps - the same representation, the same question, only in a different wavelength range.

For micro-LED displays with millions of individual emitters, sample testing at panel level is not sufficient: even at a transfer yield of 99.99999 %, roughly two defective subpixels per 4K panel remain by calculation (MicroLED Association, 2023); a 100 % inline inspection of every individual emitter with high spatial resolution is required. For OLED displays with very low black luminance (true black) combined with high peak luminance, an extremely high dynamic range of the measuring instrument is required, since bright and dark image areas have to be captured in the same measurement setup without stray light from bright areas corrupting the dark reading. Testing under EU Ecodesign requirements prescribes measurement under a defined ambient illuminance of 100 lx, which calls for a controlled, reproducible measurement room rather than a dark-room measurement. For OLED long-term and burn-in testing, IEC 62341-5-3 standardises a measurement methodology spanning weeks with a stable reference and colour-difference evaluation per CIEDE2000, because point single measurements do not capture pixel-individual, content-dependent ageing (IEC 62341-5-3:2019).

The application determines which measurement quantity is needed and where the measurement has to be taken. For the total luminous or radiant flux emitted by a component, an integrating measurement arrangement is required, because only an integrating sphere integrates over all directions of emission and thus measures independently of the individual radiation pattern. For luminance and colour point on displays, measurement has to be taken at defined image positions and viewing angles, since both quantities are direction-dependent; a single point at image centre gives no information about edge homogeneity or viewing-angle dependence.

A broadband measurement is adequate as long as the sources being assessed are sufficiently similar and broadband, and no tight colorimetric specification is required. A spectral measurement becomes necessary as soon as narrowband primaries, unknown or changing spectra, tight colour-point tolerances or a photobiological assessment are involved - because only the spectrum provides the basis for a sensor-type-independent conversion. Spatial measurements (scanning, imaging luminance measurement systems) become necessary when homogeneity, blooming or local defects have to be assessed across a surface; time-resolved measurements with sufficient bandwidth are needed when PWM dimming, temporal dithering or flicker metrics (P_st LM, SVM) are to be evaluated, because otherwise the instrument's integration time distorts the result.

Measurement uncertainties for luminance and illuminance meters cannot be stated as a blanket figure, but depend on the instrument class: CIE 231:2019 defines, for the first time, an internationally aligned classification system for illuminance and luminance meters with quality indices, where the worst individual index always determines the instrument class (CIE 231:2019). Metrological traceability of such instruments - via calibrated luminance and radiance standards and via accredited calibration laboratories - is a prerequisite for measurement values to remain comparable across sites, manufacturers and test houses. As a reference quantity for this, the luminance and radiance standard is available; metrological traceability is handled by the accredited calibration laboratory. For spectral analysis and batch comparison, the SR900 and the rest of the spectrometer line-up are available; as an integrating measurement arrangement with a spectrophotometer, the Cary 60 integrating sphere is suitable.

In industrial LED and display manufacturing, photometric and colorimetric testing is increasingly built into the production line rather than performed as a downstream sample check in the lab. In technical terms this means: every component or panel is measured at the end of the line under a defined electrical and thermal operating point, the result is compared in real time against tolerance windows for luminous flux, colour point and - for micro-LED - individual subpixels, and the classification (binning) is fed straight back into material-flow control. For panels with millions of emitters, automated, high-resolution image sensing with downstream spectral referencing replaces a complete individual measurement of every subpixel, because a spectroradiometric single scan would not be feasible within the required cycle time. The technical significance lies in the fact that process deviations - a drift in junction temperature or an inhomogeneous phosphor coating, for instance - feed straight back into the photometric measurement quantity and are therefore detected early in the process, rather than only showing up as a finished-product failure.

The shift from discrete LED modules to micro-LED direct emitters moves the testing task from panel level to individual-emitter level: with millions of subpixels per display, 100 % inline inspection of individual emitters becomes a precondition for economical manufacturing, since repair and redundancy concepts only work if defects are located without gaps (Yu et al., Journal of Physics D: Applied Physics, 2024). Mini-LED backlights with local dimming demand measurement systems with very high dynamic range, because assessing stray light (blooming) requires measuring bright and extremely dark image areas in the same setup. For OLED displays, the normative anchoring of burn-in and image-sticking tests (IEC 62341-5-3, 2nd edition) is driving standardisation of long-term measurement with colour-difference evaluation per CIEDE2000. Quantum-dot-based displays (QD-OLED, QLED) are shifting colorimetric testing from filter colorimeters to spectrally resolving methods, because their narrowband primaries (FWHM 20-40 nm) amplify the approximation errors of classical three-filter sensors. On the regulatory side, EU Ecodesign Regulation (EU) 2019/2020 for light sources is tightening flicker and stroboscopic-effect limits in stages - SVM ≤ 0.9 since September 2021, SVM ≤ 0.4 since September 2024 - turning time-resolved light measurement into a mandatory regulatory test rather than an optional add-on (EUR-Lex, consolidated version). Ecodesign and energy-label regulation for electronic displays itself is currently being revised, with drafts expected in early 2026 and a stated focus on durability and repairability (European Commission, 2024).

Six peer-reviewed papers provide the methodological basis for the central claims made in this text on spectral mismatch, integrating-sphere metrology, flicker assessment, micro-LED efficiency and OLED ageing:

1. Martinsons and Hlayhel describe a method for measuring the spectral responsivity of a photometer using optical excitations of arbitrary spectral distribution - the methodological basis for quantitatively determining spectral mismatch errors of the kind seen with narrowband LED and display primaries. Martinsons, Christophe; Hlayhel, Raed: “A method to measure the spectral responsivity of a photometer using optical excitations with arbitrary spectral distributions”, Lighting Research & Technology, 2022, DOI 10.1177/14771535211026322.
2. Poikonen et al. present a multifunctional, absolute integrating-sphere setup for luminous flux measurements on LEDs in 2π and 4π geometry, including calibration procedure, uncertainty budget and comparison with a goniophotometer - relevant to the metrological traceability of LED luminous-flux measurements in manufacturing. Poikonen, Tuomas; Manninen, Pasi; Kärhä, Petri; Ikonen, Erkki: “Multifunctional integrating sphere setup for luminous flux measurements of light emitting diodes”, Review of Scientific Instruments 81(2), 023102, 2010, DOI 10.1063/1.3285263.
3. Miller et al. provide a systematic review of temporal light modulation (flicker): stimulus quantities, physiological responses and a comparison of competing metrics (percent flicker, flicker index, P_st LM, SVM) - the basis for placing regulatory flicker limits in context. Miller, N. J.; Leon, F. A.; Tan, J.; Irvin, L.: “Flicker: A review of temporal light modulation stimulus, responses, and measures”, Lighting Research & Technology, 2023, DOI 10.1177/14771535211069482.
4. Fan et al. analyse size-dependent optoelectronic properties of red micro-LEDs using an electro-opto-thermal coupling model and show how current crowding and etch damage at the chip edge affect the efficiency of small emitters - central to interpreting measurements on individual micro-LED chips. Fan, Kaili, et al.: “Analysis of size-dependent optoelectronic properties of red AlGaInP micro-LEDs”, Optics Express 31(22), 36293-36303, 2023, DOI 10.1364/OE.503024.
5. Jiang and Xu develop a unified OLED ageing model that couples operating point, current amplitude and temperature to predict pixel-individual efficiency decline - the basis for burn-in compensation algorithms and for standard-compliant long-term measurement protocols. Jiang, Xingtong; Xu, Chihao: “A unified OLED aging model combining three modeling approaches for extending AMOLED lifetime”, Journal of the Society for Information Display 29, 768-784, 2021, DOI 10.1002/jsid.1064.
6. Yu et al. summarise, in a review, the key yield factors of micro-LED display manufacturing across chip fabrication, mass transfer and detection/repair - the technical background for the growing importance of 100 % inline testing. Yu, Binhai; Li, Yong; Li, Jiasheng; Ding, Xinrui; Li, Zongtao: “Challenges of high-yield manufacture in micro-light-emitting diodes displays: chip fabrication, mass transfer, and detection”, Journal of Physics D: Applied Physics 57, 463001, 2024, DOI 10.1088/1361-6463/ad6ce3.

A seventh dissertation, listed in our own publication index, examines the same question from the perspective of display development: how blue-light content in display backlighting can be reduced using LED clusters with a controllable colour spectrum. Habich, Nils; Homeyer, Ing Kai: “Untersuchung zur Vermeidung der Blaulichtanteile bei der Hintergrundbeleuchtung von Displays unter Einsatz von LED-Clustern mit steuerbarem Farbspektrum”, dissertation, University of Hildesheim, 2018. Further work on light, vision and photobiological effects is compiled on the customer publications topic page.

• CIE 018:2019, The Basis of Physical Photometry, 3rd Edition - basic conventions of physical photometry, definition of K_m and V(λ).
• CIE 015:2018, Colorimetry, 4th Edition - standard observers, calculation of tristimulus values and colour differences, standard illuminants for LED types.
• CIE 231:2019, Classification System of Illuminance and Luminance Meters - classification system for photometric instruments.
• IEC 62471-7:2023, Photobiological safety of light sources and luminaires primarily emitting visible radiation.
• ICDM/SID, Information Display Measurements Standard (IDMS), Version 1.3, 2025 - standardised display measurement procedures.
• Commission Regulation (EU) 2019/2021 of 1 October 2019 laying down ecodesign requirements for electronic displays.

How does a photometric measurement differ from a radiometric one?
A radiometric measurement captures the physically present radiant power without weighting. A photometric measurement weights the same radiation with the human eye's luminous efficiency function V(λ) and is zero outside roughly 380-780 nm. Both systems can be derived from a spectral measurement, but not the other way round.

Why do two instruments give different readings on the same LED?
If the LED emits narrowband light and the instruments use broadband sensors with different spectral responsivity, both approximate V(λ) with different accuracy. For metameric but spectrally different comparison sources, this leads to diverging, sometimes contradictory measurement results.

Which quantity matters for displays: luminance or illuminance?
For the brightness perceived by a viewer of a display surface, luminance (cd/m²) is decisive, since it describes, direction-dependently, what the surface emits toward the viewing direction. Illuminance (lux), by contrast, describes how much light falls onto a surface and is relevant for ambient-light measurements.

How is flicker measured on LED and OLED sources?
A time-resolved light measurement with sufficient bandwidth and a defined sampling rate is required, from which metrics such as P_st LM or SVM are computed per the underlying standard formulas. A simple average-value measurement does not capture the temporal modulation.

Why is an LED's electrical power rating not enough to judge brightness?
The efficiency of converting electrical into optical power depends on the component, the current and the temperature, and is well below 100 %. In addition, the photometric effect depends strongly on emission wavelength. Watt figures alone therefore give no basis for inferring luminous flux or luminance.

Which standard governs OLED burn-in testing?
IEC 62341-5-3 sets out measurement methods for determining image sticking and service life of OLED modules, including an HDR measurement method and colour-difference evaluation per CIEDE2000.

How is colour measured on narrowband displays such as QD-OLED?
Because of the narrow spectral FWHM of the primaries (20-40 nm), spectrally resolving spectroradiometers are preferable to three-filter colorimeters, since filter colorimeters only approximate the colour-matching functions inaccurately for narrowband spectra.

From what point is a spectral measurement required instead of a broadband one?
As soon as narrowband primaries, unknown or changing spectra, tight colour-point tolerances or a photobiological assessment are involved. For broadband, spectrally similar sources, a broadband measurement is generally sufficient.

Adjacent application fields: Photovoltaics and solar simulation, Occupational safety and photobiological safety.

Not sure whether a broadband measurement is enough, or whether your LED or display application requires a spectral measurement?

Whether a simple sensor is sufficient or a spectroradiometer is required depends on the spectral width of your source, the required colour-point tolerance and the operating point at which the measurement is to be taken. Jointly clarifying the actual testing task - individual component, module or finished display - shows which measurement quantity, which measurement geometry and which traceability are actually required. Get in touch.