LED Lighting – Error Analysis in Illuminance Measurement
One of the most important characteristics of a luxmeter is matching to the sensitivity of the human eye V(λ). V(λ) is the spectral luminous efficiency of the human eye in daylight, and is defined in DIN 5031. The required matching is accomplished by means of an optical filter whose quality dictates the accuracy class. Permissible deviation is specified in DIN 5032 and DIN EN 13032 for the individual classes.
Permissible deviation for V(λ) matching makes reference to light type A (incandescent light at 2856 K). In the case of lamps with an extremely discontinuous spectrum or a very narrow-band spectrum, error may be significantly greater. Typical lamps with these spectra include LEDs.
In the case of LEDs, white light is generated either from a blue LED with a conversion coating (usually phosphorus) or in accordance with the RGB principle by means of one red, one green and one blue LED. With both processes, large spectral components occur on the rising or falling edge of the spectral sensitivity of the human eye in daylight V(λ). The quality of V(λ) matching is especially important within these ranges, for which reason the use of more precise, class B devices in accordance with DIN 5032-7 and DIN EN 13032-1 is recommended. When examining class B and class C luxmeters by performing measurements at white and single-color LEDs, we see how important f´1 error is for the accuracy of the measurement results.
Various white and colored LEDs were examined in the following with GOSSEN’s top quality filtering for luxmeters. For the purposes of this investigation, two typical filterings with an f´1 of less than 3% (corresponds to class A) were used, as well as a third filtering with an f´1 error of 6.5% (corresponds to class C).
LED Lighting – Error Analysis in Illuminance Measurement
One of the most important characteristics of a luxmeter is matching to the sensitivity of the human eye V(λ). V(λ) is the spectral luminous efficiency of the human eye in daylight, and is defined in DIN 5031. The required matching is accomplished by means of an optical filter whose quality dictates the accuracy class. Permissible deviation is specified in DIN 5032 and DIN EN 13032 for the individual classes.
Permissible deviation for V(λ) matching makes reference to light type A (incandescent light at 2856 K). In the case of lamps with an extremely discontinuous spectrum or a very narrow-band spectrum, error may be significantly greater. Typical lamps with these spectra include LEDs.
In the case of LEDs, white light is generated either from a blue LED with a conversion coating (usually phosphorus) or in accordance with the RGB principle by means of one red, one green and one blue LED. With both processes, large spectral components occur on the rising or falling edge of the spectral sensitivity of the human eye in daylight V(λ). The quality of V(λ) matching is especially important within these ranges, for which reason the use of more precise, class B devices in accordance with DIN 5032-7 and DIN EN 13032-1 is recommended. When examining class B and class C luxmeters by performing measurements at white and single-color LEDs, we see how important f´1 error is for the accuracy of the measurement results.
Various white and colored LEDs were examined in the following with GOSSEN’s top quality filtering for luxmeters. For the purposes of this investigation, two typical filterings with an f´1 of less than 3% (corresponds to class A) were used, as well as a third filtering with an f´1 error of 6.5% (corresponds to class C).
f´1 <3% class A/B
Using the filters identified as no. 9 and no. 36, f´1 error is less than 3% in both cases. f´1 error is the sum of the surfaces which deviate from the V(λ) curve relative to total surface area underneath the V(λ) curve. In the case of filter no. 9, error occurs almost exclusively within the blue range from 400 to roughly 500 nm, and with filter no. 36, error is uniformly distributed over the entire wavelength range. All possible variants between these two extremes can occur with GOSSEN’s class B(A) filters.
f´1 = 6,5% class C
Filter no. 10 is a typical filter for a class C meter. As we can see, deviation from the V(λ) curve is distributed relatively uniformly over the entire wavelength range in this case as well, although on the whole it’s larger.
White LEDs for Lighting
The OSRAM warm white 6 W is a typical white LED.
It’s based on the principle of a blue
LED with a florescent coating.
OSRAM warm white 6 W
As is demonstrated by this LED, error in matching to the sensitivity of the human eye doesn’t change. For all three filterings, V(λ) matching error is practically identical to error for light type A.
The COB (chip on board) is another white LED. With this design the blue LED is attached directly to the aluminum base and covered with a yellow-orange florescent coating.
Chip on Bord cool white
With this LED as well, f1 error remains unchanged. It’s practically identical to error for light type A with all three filterings.
Next we’ll examine a white LED based on a combination of three LEDs in the colors red, green and blue. Warm or cold white light can be generated by controlling the three LEDs differently.
RGB warm white
RGB cool white
For these two white LEDs, by means of which white light is generated by variably controlling one red, one green and one blue LED, filtering error is greater relative to light type A, although it’s not yet significant.
Color LED’s for Effects Lighting
In the case of red, orange, green and blue-green LEDs, no deviation of f1 error can be detected as compared with light type A. Blue LEDs are the exception.
Luxeon blue
In the case of the blue LED with filter no. 9, a considerable difference relative to error with light type A becomes apparent.
Addendum: Error Analysis, Philips Lumiblade OLED
The Lumiblade OLED from Philips is used for lighting purposes. It produces warm, white light which is generated in an organic layer between two glass panels.
Phillips Lumiblade OLED
As is the case with all LEDs which are used for lighting purposes, error relative to error with light type A is negligible.
Conclusions
Class A/B and C luxmeters with matching to the sensitivity of the human eye are very good for measuring LEDs used for lighting purposes In the case of LEDs used for background illumination in LCDs (e.g. in automobile dashboards), error may be considerably greater than is the case with light type A. These observations hold equally true for GOSSEN luminance meters, because they’re equipped with the same filtering as the illuminance meters.
Further Information
GOSSEN Foto- und Lichtmesstechnik GmbH, Lina-Ammon-Str. 22, D-90471 Nuremberg, Germany
