Ask how bright an LED is and the answer is a lumen value. That number does not come from a camera or a light meter pointed at the die. It comes, in almost every production and laboratory setting, from an integrating sphere. Understanding how the sphere works explains why two datasheets for similar LEDs can state different flux for what looks like the same part.
Why Flux Cannot Be Measured Directly
Luminous flux counts light leaving in every direction. A detector placed anywhere sees only a narrow cone, weighted by its angular position. To total the flux you would need to measure from every angle and integrate, which is slow and error-prone for sources with unusual beams. An LED with a narrow lens and a bare die of equal output would need entirely different sampling schemes.
The Sphere Principle
An integrating sphere is a hollow cavity coated with a diffuse white reflectance material. Light entering the cavity bounces many times, and after each bounce the illumination becomes more uniform. The result is a radiance on the sphere wall that is proportional to the total flux entering it, independent of the original beam shape. A small port with a detector reads that uniform signal, and the sphere geometry converts it back to flux.
The key property is spatial integration: the reading no longer depends on where the source points, only on how much light it emits.
Setup Variants: 2-Pi and 4-Pi
Two mounting schemes dominate. In a 4-pi setup the source sits at the sphere center and radiates into the full solid angle; this is standard for bare LED dies. In a 2-pi setup the source is mounted at a port and radiates into half the sphere; this suits luminaires and boards that emit forward only. Each scheme has its own self-absorption and port-loss corrections, and the wrong choice biases the flux reading by several percent.
Detector Options: Photometer or Spectroradiometer
| Detector | Output | Strengths | Limits |
|---|---|---|---|
| Photometer head | Lux-equivalent signal, converted to lm | Fast, low cost, production friendly | Spectral mismatch error on narrow-band LEDs |
| Spectroradiometer | Full SPD, then integrated to lm | Accurate color and flux together | Slower, higher cost |
For colored and phosphor-converted LEDs, spectral data matters as much as flux, so a spectroradiometer at the sphere port is the laboratory default.
Calibration and Uncertainty
A sphere does not know lumens by itself. It is calibrated against a standard lamp with known luminous flux, traceable to a national metrology institute. Errors accumulate from coating aging, dust, source self-absorption, temperature drift and stray light through ports. Well-run labs state an expanded uncertainty of a few percent for LED flux; a datasheet number given without any stated uncertainty should be read as a typical value, not a guarantee.
Reading a Flux Specification
A complete flux claim has three parts: the value, the measurement conditions, and the tolerance. When a supplier states 120 lm typical with no conditions, the range behind that average can be wide, because binning usually splits one package type into several flux steps. Production parts are sorted after manufacture, so the typical value describes the center of the distribution, not the floor. For engineering purposes, design the optical budget against the minimum flux step you will actually receive, not the datasheet headline.
Key Takeaways
The integrating sphere converts any beam shape into one uniform wall signal proportional to total flux, which is why it anchors every LED flux specification. The measurement chain is only as good as its calibration, its setup geometry and its detector. When comparing flux numbers, ask what sphere setup, calibration source and uncertainty stand behind them.
Related reading
Color quantities also come from sphere-coupled spectra: Spectral Power Distribution: Reading the SPD Curve.
More optical fundamentals in the LED Academy archive.
