Light Analysis in lighting technology

Light Analysis in lighting technology by Norbert Unzner 2001 B&S Elektronische Geräte GmbH

Table of Contents: Part I 1 Introduction 3 2 Light sources 3 3 Methods of color measurements 4 3.1 Tristimulus Photometric Method and Spectrophotometric Method 4 3.2 Spectrophotometric Method used with Light Measuring Desk BS-LM-01 5 Part II 1 Spectrophotmetric Method according to DIN 5033 6 2 Correlated color temperature 9 3 Color rendering indices according to DIN 6169 10 Part III Solar simulation 12 A Appendix 14 A1 Bibliography 14 A2 Table of pictures 14 2

Part I 1 Introduction In modern lighting, according to the application, various light sources are used. Therefore knowledge about the characteristics of the emitted light is of great importance. The most interesting parameters are: Color Coordinates, the color impression a human observer gets when he looks at the light source. The color coordinates locate a single point in the, so called, color triangle, which contains all colors noticable for human eyes. Correlated Color Temperature, describes a light color which is closest to the light color of the, so called, "Planckian radiator", a light source with ideal light spectrum. This single value is given in Kelvin. Color Rendering Indices, indicate the rendering of colors of objects being irradiated by the investigated light source. Irradiance Illuminance The most accurate method to find out the necessary parameters, as for instance color temperature, is the spectrophotometric method. The fundamentals of this method shall be explained hereafter. 2 Light sources Two kinds of light sources are generally considered (Example see Fig. 1 and Fig. 2): 1. Thermal radiators 2. Discharge lamps Fig. 1: (left) Single ended 1000W halogen lamp with correlated color temperature of appr. 3200K, operating by the principle of heating up a filament caused by electric current. Fig. 2: (right) Double ended 1200W metal halide lamp similar to daylight spectrum. Between two electrodes inside the bulb an arc of plasma generates an intensive light source ignited by electrical high voltage pulses. 3

Incandescent lamps and halogen lamps belong to the first category, operating by the principle of heating up a filament. However, the luminous efficiency of these lamps is rather low, since most of the electric energy (input) is transformed into thermal radiation. Their spectra are of simple structure and therefore light analysis is not too complex. Since about 1936 in lighting also discharge lamps are used, having a much better luminous efficiency compared to incandescent lamps. The well known fluorescent lamps belong into this group, as well as Xenon- and metal halide lamps (HMI/HTI/HQI etc). Basically a discharge lamp consists of a closed glass bulb, containing mainly mercury, for operation as a discharge medium. Between two electrodes in the glass bulb an arc is ignited, which initiates an intense light emission. To achieve certain light qualities, a number of other components (f.i. halides of rare earth metals) are added. By varying combinations of different additives a wide range of discharge lamps was created. The example (Fig. 4) of an HMI-bulb (P= l8kw) shows the typical structure of the spectrum. The ingredients of the bulb were chosen in a way, that the continuous background of radiation and the superimposed spectral lines show a characteristic (quasi-continuous spectrum), very close to that of the sun. 3 Methods of color measurements 3.1 Tristimulus Photometric Method and Spectrophotometric Method Of particular interest are the colorimetric quantities of a light source, which allow the characterization of the light [1][2]. Color location, correlated color temperature and color rendering index are the values which are subject to DIN-Standardization. The most common methods to determine these values are the spectrophotometric method (see Spectrophotometric Method according to DIN 5033 part II, page 6) and the tristimulus photometric method. The tristimulus photometric method uses three different, special filters in order to reproduce the standardized spectral tristimulus values (see Fig.3). Therefore a measuring device based on this method produces the values X, Y and Z, which are used to calculate color coordinates according to DIN 5033. The great disadvantage of the latter is that there is no information about the spectral distribution. For instance it cannot be recognized if and in which wavelength spectral lines are present or if a light source radiates particularly intense in certain wavelength ranges. This kind of information cannot be found from the tristimulus photometric method, which stresses even more the importance of knowing the spectrum. Spectral tristimulus values according with DIN 5033 relating to 2 Standard Observer 2 1,5 1 0,5 0 380 480 580 680 780 Wavelength / nm x y z Fig. 3: Spectral tristimulus values according to DIN 5033 relating to 2 Standard Observer. 4

This is particularly obvious with the phenomenon of the "metameric colors". These are different colors causing exactly the same "color impression", when irradiated with a certain kind of light. However, when changing the kind of light they appear definitely different, even though both kinds of light have the same color location. The different color impression at the same color location of the light sources immediately can be explained by the difference of their spectra, using the spectrophotometric method, where the tristimulus photometric method could not disclose this phenomenon. A further disadvantage of the tristimulus photometric method compared to the much more accurate spectrophotometric method is its restriction to the visual range. Spectral measurement will also supply information about the neighboring UV- and IR-ranges. Also practical experience has shown that, particularly with discharge lamps, devices using the tristimulus method sometimes supply differing results, which restricts their use. Fig. 4: The typical spectrum of an 18kW metal halide lamp shows the particularly high light emission in the visual range (380nm.. 780nm). 3.2 Spectrophotometric Method used with Light Measuring Desk BS-LM-01 The spectrophotometric method is much more accurate. It uses the whole spectral range of a light source to characterize it. Therefore radiance intensity in certain wavelength ranges can be stated. The Light Measuring Desk BS-LM-01 is based on the spectrophotometric method and therefore allows extensive analysis of light sources. The light source must be placed in front of the measuring device before the controlling computer starts the measuring sequence. The automatically running spectral analysis provides the spectrum of a light source, the composition of the light according to its containing light colors. Without any further measurement the computer of the measuring desk calculates the characteristics for light quality, the color coordinates, correlated color temperature and color rendering indices. 5

Part II 1 Spectrophotometric Method according to DIN 5033 For the aforementioned reasons a measuring desk using the spectrophotometric method was developed. Main component of this unit is a PC-controlled spectrometer, that is able to analyse, highly accurate and with high spectral resolving power, the complete spectral range of any light source. The angle of beam is adapted to the so called 2 Standard-Observer, which was defined to avoid errors through color-shifting, occurring subjectively when observation takes place under larger angles of beam. After the spectrum of the light source has been analysed, the standardized spectral tristimulus values "X", "Y" and "Z" can be computed from the function of radiation Φ λ by integration from k=380nm to 1=780nm in the visual spectral range (Equ.1). Equ. 1 X a Φ x( λ) dλ l = λ l k Y a Φ y( λ) dλ = λ k l Z a Φ z( λ) dλ = λ k Functions x (λ), y (λ), z (λ), depending on wave length, represent the standard functions of spectral tristimulus values of the 2 -Standard Observer in accordance with DIN 5033, part 2 (see Fig.1). They are given in steps of 5nm from λ = 380nm to λ = 780nm. The constant value a is not calculated because the next step (Equ. 2) eliminates this value. Spectral tristimulus values according with DIN 5033 relating to 2 Standard Observer 2 1,5 1 0,5 0 380 480 580 680 780 Wavelength / nm x y z Fig. 1: Spectral tristimulus values according to DIN 5033 relating to 2 Standard Observer. These functions describe the characteristics of the human eye (sensitivity, color sensation), which have been evaluated empirically with test persons and which are given as standard references according to DIN. Scaling of the spectral tristimulus values according to Equ. 2, 6

X Equ. 2 x = X + Y + Z y = X Y + Y + Z then directly leads to the values x and y known as chromaticity coordinates, defining the color location in the chromaticity diagram according to DIN 5033. Fig.2 shows the DIN 5033 color triangle for example. Other well known color triangles are the CIE UCS 1960- (u, v ) and CIE UCS 1976 color triangles (u, v ). Equ. 3 4x u = 2x + 12y + 3 6y v = 2x + 12y + 3 Simple transformation of (x, y) with Equ.3 and Equ.4 leads to the color coordinates of the mentioned systems.both should interpret the color difference of two color locations better then DIN 5033. Equ. 4 4x u = 2x + 12y + 3 9y v = 2x + 12y + 3 Fig. 2: Chromaticity diagram according to DIN 5033. Point 1 marks color coordinates of 4kW metal halide lamp spectrum. The black line near by point 1 is the Planckian radiator curve, connecting the color locations of a Planckian radiator at different color temperatures. 7

Knowledge of coordinates x and y allows to check for instance the chromaticity of a light source or to find out whether its color location is in the required limits of a certain ellipse of tolerance in the chromaticity diagram (Fig. 4). These criteria are important for the development and production of bulbs and lights, but also for lighting in Film and Television. Fig. 3: The typical spectrum of an 18kW metal halide lamp shows the particularly high light emission in the visual range (380nm.. 780nm). Color coordinates of different measurements are shown in Fig.4. Fig. 4: A cut-out from DIN 5033 chromaticity diagram of an 18kW metal halide lamp installed in a fixture with Fresnel lens, with dimming curves at different power (P=9kW..18kW) included. Operation with Fresnel lens (point 1..7) and with open fixture (w/o Fresnel lens, point 8..14). Also shown the Juddian lines, ellipse of tolerance for metal halide lamps, and the Planckian curve, connecting the color locations of a Planckian radiator at different temperatures. 8

As an example for their scope of application, in Fig.4 the dimming curves of an metal halide lamp in a fixture are shown. With the light fixture open (w/o Fresnel lens) the dimming curves of the lamp are entirely located in the ellipse of tolerance. By closing the light fixture (w/fresnel lens), the dimming curve changes its location and partly moves out of the area of tolerance bordered by the ellipse. The slightly inclined vertical lines represent the Juddian straight lines (called after the physicist D.B. Judd). From Fig. 4 it can be taken that by dimming electrical power between 100% and about 80% even a slight correction of color temperature might be possible. The line left and above of the ellipse shows the color locations of the Planckian radiator at different color temperatures. As one might expect, changes in electrical operating conditions of the bulb will influence the light emission, at least luminous flux is reduced when electrical power is decreased. The ideal lamp would have constant color location and color temperature if f.i. nominal power is decreased from 100% to 50%. With thermal radiators, for example halogen lamps, this cannot be achieved, since they operate almost like a Planckian radiator. The dimming curve of modern metal halide lamps however show relatively small changes in color location, which means a high quality of the generated light. Further to the mentioned colorimetric values, with the spectrophotometric method other important terms can be calculated. Fig. 5: Spectrum and colorimetric values of the halogen lamp shown in Part I, Fig. 2. The spectrum looks like the radiation emitted by Planckian radiator at T=3133K. 2 Correlated color temperature The correlated color temperature T n ( K for Kelvin) characterizes the color of specific radiation. Only near by the Planckian radiator curve this temperature is well defined (see Fig.2). Physically speaking, the correlated color temperature T n of a radiator is the temperature in correlation to the Planckian radiator at which the chromaticity of the Planckian radiator comes closest to the one of the investigated radiator. 9

In order to calculate the correlated color temperature following working cycle must be processed. First perform relative spectral analysis of light source. Calculate color coordinates (x, y) according to DIN 5033 Transformation of (x, y) into CIE UCS 1960 coordinates (u, v) with Equ.3 Search shortest straight line between (u, v) and Planckian radiator curve and locate point of intersection. The temperature of Planckian radiator at this point is called correlated color temperature T n. This simple instruction goes more complicated caused by physical circumstances. First there is the radiance of the Planckian radiator (Equ. 5), inserted in Equ.1 and Equ.2 results a system of equations which could not be solved for temperature T n analytically. Therefore in practice using the rational second order Chebyshev approximation of the Planckian radiator and an iterative procedure to find the correlated color temperature in the CIE UCS 1960 color chart is a helpful and precise method. Equ. 5 2hc Lλ ( λ, T ) = 5 λ 2 hc exp λk BT T = Temperature λ = Wavelength k B = Boltzmann s constant h = Planck constant c = Speed of light 1 1 All by itself the correlated color temperature does not supply too much information about the investigated light because, particularly with discharge lamps, the specific characteristics of a light source cannot be described definitely by it. However, as an additional characteristic and eliminating criterion it will be very valuable as a comparison quantity. Common values for incandescent lights range from 2000 K up to 3200K (halogen lamps). Metal halide lamps with their light similar to that of daylight have a closest color temperature of typically 5600K. An essential means to locate the correlated color temperature without using any computer algorithm are the Juddian straight lines drawn in the DIN-diagram, representing the color locations at equal color temperatures. Therefore T n can be found by looking up the Juddian line going through the color location of the tested radiator (see Fig.4). 3 Color rendering indices according to DIN 6169 Other important colorimetric values are the color rendering indices according to DIN 6169 [4], which are also calculated without further means by a computer program of the spectral measuring desk. These values are indicating the rendering of colors of objects being irradiated with the light of the investigated radiator. It can be estimated for instance, whether a surface appearing red by daylight, will appear the same or will produce a different "color impression" for the observer with artificial light. This is of particular importance in taking films, in order get all color shades reproduced as close to the original ones as possible. The mentioned color rendering indices are related to the color rendering characteristics of light sources in lighting technology and are determined with the test-color-procedure according to DIN 6169, part 2. This method uses the standardized spectral radiance factors of the test colors l...14, being available as a data file (DIN 6169, part 2). 10

The "Reference lights according to DIN 6169" which represent, depending on temperature, either the Planckian radiator or the different daylight phases, are used for light reference. The calculation of the color rendering indices considers according to standard the color transfer in CIE-UCS-1960 color triangle. To calculate the difference Ei between the color impression of a test color No. i when irradiated with the investigated light and the color impression of the same test color i when irradiated with the reference light, the color distance formula CIE 1964 is used. The special color rendering index R is then calculated with Equ.6: i Equ. 6 R i = 100 4, 6 E i=1..14 i The general color rendering index R a is the arithmetic average value from the special color rendering indexes R i for test colors No. 1 through No. 8. If the difference Ei becomes Zero, color rendering is at its optimum and the color rendering index reaches 100. Below this R i is not limited and may even become a negative value, when color rendering is bad. If possible, for the definition of a light source all special color rendering indices should be given, because Ra by itself as an average value is less meaningful. Typical values of R i for halogen lamps are close to 100, because the black radiator serves as reference light. Metal halide lamps also reach Ra -values of well above 90. Fluorescent lamps have worse color rendering quality, with R a typically around 50, having some Ri -values even much lower than average value R a. The reason is the poor light emission in the long wave red spectral range. The spectrophotometric method presented in this article offers the fundamental means for a comprehensive and accurate analysis of light, enabling the manufacturer of lamps and lights to get valuable information about the kind and quality of his light sources. 11

Part III Solar simulation In science and technology irradiation of objects in a reproductive way, in order to examine the influence of natural sun radiation for instance, became more and more importance. Natural sun radiation by itself includes unstable intensity and therefore is less suitable as radiation source. On the other hand artificial radiation sources are able to meet the requirements depending on the application: Simulation of average global radiation according to AM 1,5 spectrum (Fig.1) Irradiation of about 1000W/m² adjustable Homogeneous irradiation on testing plane Time-continuous irradiation on testing plane. Low divergence angle of irradiation In the case of greater areas to be irradiated (e.g. testing solar cells) metal halide lamps are a good choice because they have a spectrum similar to the daylight without additional filtering and a greater efficiency than tungsten lamps. For testing solar cells the basic norm IEC 82 (CO) 15 is used. This norm assumes the average spectral irradiation of natural sun radiation in summer at noon and cloudless sky called AM 1,5 spectrum. Fig.1 shows this spectrum including simulator classes A, B and C used in the case of different accuracy. Class A allows smallest differences and represents highest accuracy. A radiation source meets the norm if average irradiation of each wavelength interval is part of corresponding tolerance areas. Table 1and Table 2 show standardized values. Interval of wavelength in µm Percentage of radiant intensity of entire interval 0,4µm... 1,1µm 0,4.. 0,5 18,5 0,5.. 0,6 20,1 0,6.. 0,7 18,3 0,7.. 0,8 14,8 0,8.. 0,9 12,2 0,9.. 1,1 16,1 Table 1: IEC 82 (CO) 15 given reference spectrum. Simulator-class A B C Percentage difference ± 25 ± 40 60 + 100 Table 2: IEC 82 (CO) 15 allowed difference of reference spectrum (Table 1). 12

3000 Intensity [ W / (m² µm) ] 2500 2000 1500 1000 Class C Class B Class A AM 1,5 (IEC 82(CO)15) 500 AM 1,5 Spectrum 0 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 Wave Length [µm] Fig. 1: AM1,5 spectrum and bandwidth of simulator class A, B and C according to IEC 82 (CO) 15 13

A Appendix A1 Bibliography [1] Richter, M., Einführung in die Farbmetrik. de Gruyter, Berlin 1981 [2] Wyszecki, G. u. Stiles, Color Science. Wiley & Sons, New York 1982 [3] Deutsche Normen, DIN 5033, Farbmessung 1979 [4] Deutsche Normen, DIN 6169, Farbwiedergabe [5] Osram, Technisch-wissenschaftliche Abhandlungen der Osram-Gesellschaft, Band 12, Springer, Berlin 1986 A2 Table of Pictures Part I Fig. 1: (left) Single ended 1000W halogen lamp with correlated color temperature of ca. 3200K, operating by the principle of heating up a filament caused by electric current... 3 Fig. 2: (right) Double ended 1200W metal halide lamp similar to daylight spectrum. Between two electrodes inside the bulb an arc of plasma represents intensive light source ignited by electrical high voltage pulses... 3 Fig. 3: Spectral tristimulus values according to DIN 5033 relating to 2 Standard Observer.... 6 Fig. 4: The typical spectrum of an 18kW metal halide lamp shows the particularly high light emission in the visual range (380nm.. 780nm)..... 8 Part II Fig. 1: Chromaticity diagram according to DIN 5033. Point 1 marks color coordinates of 4kW metal halide lamp spectrum. The black line near by point 1is the Planckian radiator curve, connecting the color locations of a Planckian radiator at different temperatures... 7 Fig. 2: The typical spectrum of an 18kW metal halide lamp shows the particularly high light emission in the visual range (380nm.. 780nm). Color coordinates of different measurements are shown in Fig. 6.... 8 Fig. 3: A cut-out from DIN 5033 chromaticity diagram of an 18kW metal halide lamp installed in a fixture with Fresnel lens, with dimming curves at different power (P=9kW..18kW) included. Operation with Fresnel lens (point 1..7) and with open fixture (w/o Fresnel lens, point 8..14). Also shown are the Juddian lines, ellipse of tolerance for metal halide lamps, and the Planckian curve, connecting the color locations of a Planckian radiator at different temperatures... 8 Fig. 4: Spectrum and colorimetric values of the halogen lamp shown in Fig. 1. The spectrum looks like the radiation emitted by Planckian radiator at T=3133K... 9 Part III Fig. 1: AM1,5 spectrum and bandwidth of simulator class A, B and C according to IEC 82 (CO) 15... 13 14