Color consistency should be bought as a measurable finished-product result, not as a loose promise about LED chips. A useful specification connects the intended appearance to CCT and Duv targets, a stated tolerance, a controlled measurement method, and separate acceptance rules for initial samples, production lots and reorders.
The practical objective is not to collect the smallest number on a datasheet. It is to prevent visible mismatch while preserving a test method that both buyer and supplier can repeat.
Start with the visual risk
Ask where differences will be noticed. Adjacent downlights on one ceiling, linear products joined end to end, wall washers aimed at the same surface and replacement lamps installed beside older stock create different levels of visual risk. Viewing distance, surface color, beam overlap, dimming level and ambient light can all affect perception.
Write the use case into the RFQ before selecting a tolerance. A demanding adjacency application may justify tighter control than products installed in separate rooms. This is an application decision, not evidence that one tolerance is universally correct.
If the decision is still at the application level, review CCT and CRI selection separately. CCT, color rendering and color consistency answer different questions.
Separate CCT, Duv and SDCM
Correlated color temperature, or CCT, describes the apparent warm-to-cool character of a near-white light source. It does not locate that source completely in chromaticity space. Duv indicates position relative to the Planckian locus and helps distinguish shifts that can appear greener or more magenta even when nominal CCT is similar.
SDCM, often discussed through MacAdam ellipse steps, expresses a tolerance region around a target chromaticity. A smaller step count usually describes a tighter region, but the number is incomplete without the target point, measurement conditions and whether it applies to initial measurements or maintained performance.
The U.S. Department of Energy’s LED color-characteristics guidance distinguishes CCT and Duv and explains why a single CCT label cannot fully describe chromaticity. Therefore, “same CCT” is not a finished acceptance criterion.
Do not treat binning as the finished result
LED manufacturers sort components into bins using defined measured characteristics. That information can support component control, but a luminaire or lamp adds optics, phosphor conversion, thermal conditions, drive current, mixing geometry and manufacturing variation.
Consequently, two procurement statements are not equivalent:
- “LEDs come from a named bin”; and
- “finished products measured under the agreed method remain inside the agreed chromaticity tolerance.”
The first describes an input-control route. The second describes the result the buyer sees and can accept. Ask for both when the risk warrants it, but make the finished-product requirement controlling. The distinction is especially important when comparing SMD and COB LED differences, because package construction and optical mixing routes differ.
Build the batch acceptance matrix
Use one matrix across qualification, first production and reorders.
| Control point | Target or limit | Measurement condition | Sample scope | Entscheidung |
|---|---|---|---|---|
| Nominal CCT | Buyer-defined target | Stabilized product at stated input | Defined units and positions | Record and compare |
| Duv or target chromaticity | Buyer-defined center or range | Same instrument and geometry | Same sample | Accept, investigate or reject |
| Initial SDCM tolerance | Application-specific limit | Agreed color space and reference | Unit-level results, not only average | Apply stated rule |
| Within-unit variation | Relevant for multi-emitter or linear products | Defined measurement positions | Each sampled unit | Check spatial mismatch |
| Between-unit variation | Relevant for adjacent installation | Same setup and sequence | Across sampled lot | Check clustering and outliers |
| Reorder match | Reference lot or approved range | Same method plus retained reference | New lot and retained sample | Approve or escalate |
| Maintained color | Only where required | Defined operating time and conditions | Qualified evidence plan | Review separately |
Do not write “3 SDCM” alone. State what is measured, around which target, at what time, with which decision rule.
Control the measurement method
A repeatable protocol should identify:
- the spectroradiometer or measurement-system class and current calibration status;
- input voltage, frequency, drive setting and dimming state;
- ambient and thermal conditions;
- stabilization criterion or warm-up procedure;
- measurement geometry, distance and product orientation;
- measurement positions for multi-emitter, linear or large-area products;
- reported values, including individual results rather than only averages;
- repeat-measurement rules when a result is close to a limit.
IEC 62717 addresses performance requirements for LED modules, including color-related quantities within its scope. It does not remove the buyer’s need to define the actual finished product, configuration and acceptance method.
Use stabilization, not an arbitrary stopwatch
Color can change while a product reaches thermal equilibrium. A fixed warm-up time is useful only if it reliably produces stable conditions for the product under test. A better protocol defines either a validated time for that model or a stability criterion that the lab can observe.
Record when measurements were taken. If sample qualification uses a stabilized laboratory method but incoming inspection uses an immediate handheld reading, the two datasets should not be treated as interchangeable. The methods can serve different screening roles, but their limits need separate validation.
Inspect distributions, not only averages
A batch average can sit near the target while individual units form two visible clusters. Request individual chromaticity coordinates, CCT and Duv where the visual risk justifies them. Plotting the points can reveal drift, clustering or isolated outliers that an average hides.
Define whether acceptance depends on every sampled unit, a statistical rule, or both a population window and an outlier limit. If a supplier reports only a pass label, the buyer cannot determine how close the batch was to the boundary or whether the process is moving.
Separate initial color from color shift
Initial consistency and color maintenance are different controls. The U.S. Department of Energy’s LED Systems Reliability Consortium describes color shift as a system-level issue influenced by multiple components and operating stresses. A tight initial batch does not prove that all units will follow the same chromaticity path over time.
Where maintained appearance is critical, define the operating duration, environment, measurement intervals and allowable shift separately. Avoid turning an initial SDCM result into a lifetime claim. Also avoid assuming that lumen maintenance evidence proves color maintenance; those outputs require distinct review.
Qualify the sample, then lock the production method
Verwenden Sie die Checkliste zur Musterbewertung to record the approved configuration. The color section should capture the LED source identity or control route, optics, diffuser, drive setting, relevant firmware, measurement file, instrument information and visual reference.
After approval, identify which changes require notification or requalification. Examples can include LED source or bin route, phosphor system, optic or diffuser material, drive current, thermal construction and a measurement-method change. The list should match the actual product architecture rather than being copied blindly.
Create a reorder reference rule
Reorders fail when “same color” refers to memory instead of a controlled reference. Choose one or more references:
- numerical target and tolerance;
- approved sample retained under defined storage conditions;
- signed measurement report and raw data;
- a limit sample or paired visual standard;
- an earlier production lot identified by code and date.
A retained physical sample can age, become contaminated or differ from the installed population. For that reason, it should complement numerical data rather than replace it. State which reference controls if the sources disagree.
Use a three-stage release protocol
Qualification stage: confirm the target, measurement method and visual suitability using representative samples. Record individual results and approve the configuration.
First-production stage: sample the manufactured lot, repeat the agreed method and compare both the numerical distribution and visual result with the approved reference. Investigate systematic offsets even if the lot narrowly passes.
Reorder stage: compare the new lot with the controlled numerical range and the retained reference. Review any notified material or process change. Do not assume a previous order’s approval transfers automatically.
This protocol turns color from a descriptive adjective into a release decision.
RFQ fields to send the supplier
Include the intended application and adjacency risk; nominal CCT; target chromaticity or Duv range if used; SDCM tolerance and reference point; initial versus maintained requirement; product state and stabilization rule; measurement geometry; sample size or lot-plan reference; within-unit and between-unit rules; reorder reference; raw-data format; change-notification list; and disposition for borderline or failed results.
If the product family is still being selected, review relevant LED-Lampenfamilien without assuming that one family-wide page proves batch performance. The RFQ, approved sample and production evidence must all refer to the same finished configuration.
Redaktionelle Quellen
- IEC 62717, LED modules for general lighting — Performance requirements: https://webstore.iec.ch/en/publication/7393
- U.S. Department of Energy, LED-Farbmerkmale: https://www.energy.gov/sites/prod/files/2016/08/f33/led-color-characteristics-factsheet.pdf
- U.S. Department of Energy, Zuverlässigkeit von LED-Leuchten: Einfluss der Farbverschiebung: https://www.energy.gov/sites/default/files/2019/10/f67/lsrc_colorshift_apr2017.pdf
