Relation between the illuminance responsivity of a photometer and the spectral power distribution of a source

46 9, 093607 September 2007 Relation between the illuminance responsivity of a photometer and the spectral power distribution of a source Ferhat Sametoglu, MEMBER SPIE National Metrology Institute of Turkey TÜBİTAK-UME 41470 Gebze Kocaeli, Turkey E-mail: ferhat.sametoglu@ume.tubitak.gov.tr Abstract. The illuminance responsivity dependencies on spectral power distributions SPDs of different types of light sources are studied. In this work we used three types of calibrated photometers, one of which was home-made and two were commercial photometers. A monochromatorbased facility was used to scan SPDs of light sources. The dependencies of spectral mismatch correction factors F S t,s s and the illuminance responsivities of photometers versus the SPDs of tungsten-filament incandescent, fluorescent, high-pressure sodium, and metal-halide light sources and white/colored light-emitting diodes are presented throughout this work. 2007 Society of Photo-Optical Instrumentation Engineers. DOI: 10.1117/1.2786888 Subject terms: photometer head; illuminance responsivity; correlated color temperature; spectral power distribution; incandescent lamp; fluorescent illuminants; high-pressure illuminants; light-emitting diodes. Paper 060778R received Oct. 6, 2006; revised manuscript received Mar. 6, 2007; accepted for publication Apr. 13, 2007; published online Sep. 26, 2007. 1 Introduction Different measurement methods and standards are used to realize photometric units. All of these methods are achieved via source-based or detector-based standards. 1 7 By comparing two methods, detector-based standards are preferred by the national metrology institutes NMIs in photometric realizations because they have the best reproducibility, repeatability, and transportation flexibility. Photometric detectors photometers are divided into three main groups: standard photometers, radiometers with appropriate filters, and spectrophotometers. A V -corrected or a V -corrected plus flat diffuser type of photometer is used as a standard photometer. A photometer with a V -corrected filter is normally employed with a standard incandescent lamp placed on the optical axis of the photometer at a sufficient distance to provide normal incident light with a small divergence angle. V -corrected types of photometers without a diffuser are suitable for the luminous intensity, retroreflection, and goniophotometer-based luminous flux measurements. 3,5,8,9 V -corrected types of photometers with a diffuser are more subject to stray light due to a large acceptance angle, but less subject to errors for a large-size lamp at shorter distances. Generally, diffuser-type photometers are compatible for use in the illuminance, luminance, and integrating sphere-based luminous flux realizations. 10 12 Spectrophotometers with a scanning grating and a detector combination or a constant grating and an array diode combination are alternative devices for photometric measurements. Light sources are scanned between the visible 0091-3286/2007/$25.00 2007 SPIE range, and photometric quantities X v are defined in relation to the corresponding radiometric quantity X e by the equation X v = K m X e V d, 1 where K m is the maximum spectral luminous efficiency for photopic vision 683 lm/w that relates radiometric quantities to photometric quantities, and V is the spectral luminous efficiency function, which is adopted by the Commission Internationale de l Eclarge CIE as an average action spectrum for the visual response of the human eye. In the spectrophotometric method, errors coming from the CIE-V matching are eliminated because the standard CIE-V function is used to separate the photometric range from the full measured radiometric spectrum. To measure photometric quantities with a photometer or a radiometer, the device s illuminance responsivity should be precisely calibrated and the selected device must have a relative spectral responsivity matched to CIE-V. Matching the spectral response of photometers to the V function is the most important criterion of photometers. Thus, photometers are characterized for spectral mismatch to the CIE-V function by the calculation error factor f l. 13 The luminous intensity is a quantity that describes the photometric output of a light source while the illuminance responsivity is used to describe a photometer. To precisely calculate the photometric quantity, the spectral responsivity of the photometer and the distribution temperature of the light source should be known. Photometers are generally calibrated against CIE Illuminant A 2856 K Plankian radiation. An error occurs when a photometer measures a light source that has a spectral power distribution SPD 093607-1

Fig. 1 a Normalized relative spectral responsivities of photometer heads. b Differences between the CIE-V functions. different from the calibrated source. If the photometer s relative spectral responsivity and the spectral distributions of the test and standard light sources are known, the spectral error can be corrected by the spectral mismatch correction factor F S t,s s as given by 14 S t V d S s s rel d F S t,s s =, 2 S t s rel d S s V d where S t and S s are the SPDs of the test and standard light sources, and s rel is the relative spectral responsivity of a photometer. The photometer signal is multiplied by this factor to eliminate spectral mismatch errors. From the photometric point of view, there is an alternative equation to calculate the spectral mismatch correction factor of a photometer for an incandescent light source 15 : F T = T T A m, 3 T A is the color temperature of the standard Illuminant A type light source Plankian radiator operated at 2856 Kelvin, where m is the mismatch index, and T is the correlated color temperature CCT of the source. The tungsten filament light source is operated at two different CCTs T 1 and T 2 and the mismatch index is calculated according to the following equation: m = log I 2 T 2 I 1 T 1 y1 T 1 y 2 T 2 log T 2 4 T 1, where I 1 T 1 and I 2 T 2 are luminous intensities, and y 1 T 1 and y 2 T 2 are photocurrents generated at the output of the photometer obtained at temperatures T 1 and T 2. This study presents illuminance responsivity variations of a home-made photometer a trap-detector based radiometer and two commercial photometers versus the SPDs of different light sources. Variations of illuminance responsivities that depend on tungsten-based incandescent lamps, fluorescent and high-pressure illuminants, and different light-emitting diodes LEDs are introduced in Sec. 3. 2 Photometers Used in Experiments The first photometer head is a temperature-controlled, V -filtered photometer made by the National Metrology Institute of Turkey UME model no. FR 25.0-1. The basic components of the photometer head, which is described in Ref. 2, are a black, anodized, thin bimetal aperture with a 0.1-cm 2 area; a PRC Krochmann-manufactured V -correction filter; and a trap detector three Hamamatsu S1337-11 series photodiodes in a reflectiontype trap configuration. A circular thermoelectric Peltier element and a Pt-100 temperature sensor are located in the photometer housing to adjust and monitor the housing temperature. The photocurrent generated at the output of the photometer is measured via a transimpedance amplifier Lab Kinetics, Ltd., SP-042 and a digital multimeter combination HP 3458 A. The temperature of the photometer housing was adjusted to 25.0 C±0.1 C. The second and third photometers are temperature-stabilized photometer heads manufactured by PRC Krochmann model no. TH15BA and LMT Lichtmesstechnik GmbH model no. P30SCT. The PRC Krochmann-manufactured photometer head has an aperture 10 mm in diameter and a V -correction filter assembled in a cylindrical housing. The temperature of the housing is thermostatically stabilized at 35.0 C ± 0.1 C. The LMT Lichtmesstechnik GmbH-manufactured photometer head has an aperture 30 mm in diameter, colored glass filters, and a diffuser at the entrance. The temperature of the housing is thermostatically stabilized at 25.0 C±0.1 C. The spectral mismatch error factor of the photometer heads are f 1 =1.5%, f 1 =1.4%, and f 1 =0.5%, respectively. The relative spectral responsivity curves of the photometer heads and the differences between their CIE-V functions are shown in Figs. 1 a and 1 b. As shown in 093607-2

Table 1 Dependencies of the F S s S t factors and the illuminance responsivities of photometer heads versus the CCTs of incandescent light sources. F S t S s factor Influence on the illuminance responsivity, R V % CCT K FR 25.0-1 TH15BA P30SCT FR 25.0-1 TH15BA P30SCT 2000 1.0009 1.0005 1.0009 0.09 0.05 0.09 2200 1.0007 1.0004 1.0006 0.07 0.04 0.06 2400 1.0004 1.0003 1.0004 0.04 0.03 0.04 2600 1.0002 1.0002 1.0002 0.02 0.02 0.02 2800 1.0000 1.0000 1.0000 0.00 0.00 0.00 2856 1.0000 1.0000 1.0000 0.00 0.00 0.00 2900 0.9999 0.9999 0.9999 0.01 0.01 0.01 3200 0.9997 0.9997 0.9998 0.03 0.03 0.02 Fig. 1 b, differences between the CIE-V functions and the relative spectral responsivities of the photometer heads manufactured by UME and PRC Krochmann are similar. Major variations are observed over wavelength ranges from 495 to 535 nm. The maximum variation from the CIE-V function in the wavelength range is an order of 2.3% at 500 nm for both photometer heads. Differences from 380 to 495 nm and from 535 to 780 nm fluctuate within ±1.0%. The variation of the relative spectral responsivity of the photometer head manufactured by LMT Lichtmesstechnik GmbH from the CIE-V function is observed within ±1.1% the maximum variation is at 525 nm. 3 Experiments and Results Four different types of light sources are used in the experiments to determine dependences of illuminance responsivities versus SPDs. The groups of sources are based on tungsten-filament incandescent lamps, fluorescent illuminants, high-pressure sodium and metal-halide illuminants, and various LEDs. 3.1 Relations Between Illuminance Responsivities of Photometer Heads and SPDs of Tungsten- Filament Incandescent Sources Two types of lamps the Osram Wi41/G typical 6 A/180 W and the Sylvania 1000-W quartz-tungsten halogen lamp ANSI designation, free electron laser lamps typical 8.1 A/1000 W were used as the tungstenfilament lamps in the measurements. The Osram Wi41/G lamp was a gas-filled incandescent lamp having a reverseconical shaped bulb. The Sylvania 1000-W lamp had a coiled-coil filament, mechanically clamped at both ends with no middle support. The SPDs of these lamps were nearly the same as the blackbody distribution. Therefore, it was not necessary to correct the illuminance responsivity of a photometer at 2856 K by the spectral mismatch correction factor because F S t,s s was equal to unity. To change the CCT of the lamps, the applied current was varied in the experiments. The CCT of the Osram Wi41/G extended to 2900 K, whereas the CCT of the Sylvania lamp extended up to 3200 K. The dependences of the spectral mismatch correction factors of the photometer heads on the SPDs of the lamps, and the influence of these variations on illuminance responsivities, were characterized using a double-monochromator-based measurement facility. Since the measurement system has been described in detail in earlier publications, 1,16 only a brief overview of the setup is provided here. The measurement facility was based on the Bentham Instruments, Ltd.-manufactured doublemonochromator DTMc300, which had a wavelength repeatability of 0.01 nm. The incandescent light sources were sequentially focused onto the entrance slit of the monochromator. A PTN 150-20 type of Heinzinger dc power supply was used to operate the lamps at a constant-current mode. The photometer heads were held on a carriage placed at the output of the monochromator and translated via the computer control. The carriage also carried the reference trap detector 17 used to measure the SPDs of both the Osram Wi41/G and the Sylvania light sources. Compensations for changes in both light sources during the experiment were made by using the signal from the monitor detector, described in Ref. 16. The SPD results were used to calculate the spectral mismatch correction factors F S t S s over the range 2000 3200 K with steps of 200 K according to Eq. 2, and to determine the influences of the F S t S s factors on the illuminance responsivities of the photometer heads Table 1. The normalized SPD curves of sources having CCTs of 2000 K, 2400 K, 2856 K, and 3200 K on the visible region, and the variations in the F S t S s factors of the photometer heads versus the CCTs of tungsten-based incandescent sources, are shown in Figs. 2 a and 2 b. It was observed that the F S t S s factors for all the photometer heads had approximately the same behavior Fig. 2 b. Table 1 demonstrates that the F S t S s factor of the UME- 093607-3

Fig. 2 Characterization results for incandescent lamps. a SPD curves for four CCTs. b Variations of the F S t S s factors versus CCTs. made photometer head FR 25.0-1 varied from 1.0009 at 2000 K to 0.9997 at 3200 K. The variation of the F S t S s factor of the FR 25.0-1 resulted in the illuminance responsivity of the photometer head within an order of 0.09% below 2856 K at 2000 K and 0.03% above 2856 K at 3200 K. The contributions of the F S t S s factors to the illuminance responsivities of the PRC-made TH15BA and the LMT-made P30SCT photometer heads at the same CCTs varied from 0.05 to 0.03% at 2000 K and 3200 K for the TH15BA, respectively, and from 0.09 to 0.02% for the P30SCT, respectively Table 1. The ratio of the second term on the right side of Eq. 2, S s s rel d / S s V d, is constant for each type of light source and equal to 1.0030 the FR 25.0-1, 1.0000 the TH15BA, and 1.0007 the P30SCT. The multiplier S t V d, the first term on the right side of Eq. 2, exhibits the same behavior for all photometer heads. The sum of the multiplier increases due to the CCT of the lamp 3.33 at 2000 K and 11.56 at 3200 K. The multiplier S t s rel d plays an active role in the calculation of the F S t S s factor. The variations in S s s rel d / S t s rel d of each photometer head for the CCT range from 2000 to 3200 K are depicted in Fig. 3 a. Figure 3 a shows that the variations of the photometer heads ranging from 2600 to 3200 K were less than 0.005% and varied more below 2600 K 0.1% at 2000 K. The experimental light sources having CCTs below 2600 K had poor SPDs compared with those above 2600 K Fig. 2 a ; therefore, the influence of the changes on the relative responsivity of the photometer heads Fig. 1 b was more effective for the SPDs below 2600 K. 3.2 Relations Between Illuminance Responsivities of Photometer Heads and SPDs of Fluorescent and High-Pressure Sources Fluorescent and high-pressure illuminants have more energy outputs in the 300- to 400-nm wavelength range, over which tungsten-based incandescent lamps do not have a Fig. 3 a Variations of S s s rel d / S t s rel d ratios. b Variations of S t V d / S t s rel d ratios. 093607-4

Table 2 Dependencies of the F S s S t factors and the illuminance responsivities of photometer heads versus the CCTs of fluorescent and high-pressure illuminants. F S t S s factor Influence on the illuminance responsivity, R V % CIE illuminants CCT K FR 25.0-1 TH15BA P30SCT FR 25.0-1 TH15BA P30SCT FL4 2940 1.0009 1.0016 1.0003 0.09 0.16 0.03 FL3 3450 1.0005 1.0011 0.9999 0.05 0.11 0.01 FL6 4150 1.0000 1.0005 0.9996 0.00 0.05 0.04 FL2 4230 1.0000 1.0003 0.9996 0.00 0.03 0.04 FL5 6350 0.9992 0.9991 0.9989 0.08 0.09 0.11 FL1 6430 0.9992 0.9990 0.9989 0.08 0.10 0.11 FL9 4150 0.9998 1.0000 0.9995 0.02 0.00 0.05 FL8 5000 0.9994 0.9993 0.9992 0.06 0.07 0.08 FL7 6500 0.9992 0.9988 0.9989 0.08 0.12 0.11 FL12 3000 1.0033 1.0034 1.0018 0.33 0.34 0.18 FL11 4000 1.0031 1.0030 1.0014 0.31 0.30 0.14 FL10 5000 1.0028 1.0026 1.0010 0.28 0.26 0.10 FL3.1 2932 1.0008 1.0016 1.0002 0.08 0.16 0.02 FL3.2 3965 1.0000 1.0004 0.9996 0.00 0.04 0.04 FL3.3 6280 0.9991 0.9991 0.9989 0.09 0.09 0.11 FL3.4 2904 1.0002 1.0010 0.9999 0.02 0.10 0.01 FL3.5 4086 0.9998 0.9999 0.9995 0.02 0.01 0.05 FL3.6 4894 0.9994 0.9993 0.9992 0.06 0.07 0.08 FL3.7 2979 1.0030 1.0031 1.0015 0.30 0.31 0.15 FL3.8 4006 1.0027 1.0028 1.0011 0.27 0.28 0.11 FL3.9 4853 1.0028 1.0027 1.0010 0.28 0.27 0.10 FL3.10 5000 1.0025 1.0021 1.0009 0.25 0.21 0.09 FL3.11 5854 1.0027 1.0024 1.0008 0.27 0.24 0.08 FL3.12 2984 1.0005 1.0009 1.0001 0.05 0.09 0.01 FL3.13 3896 1.0001 1.0000 0.9997 0.01 0.00 0.03 FL3.14 5045 0.9995 0.9991 0.9993 0.05 0.09 0.07 HP1 1959 0.9989 0.9993 0.9998 0.11 0.07 0.02 HP2 2506 1.0046 1.0049 1.0027 0.46 0.49 0.27 HP3 3144 0.9985 0.9989 0.9990 0.15 0.11 0.10 HP4 4002 0.9972 0.9975 0.9981 0.28 0.25 0.19 HP5 4039 0.9982 0.9982 0.9988 0.18 0.18 0.12 093607-5

Fig. 4 Normalized SPD curves of fluorescent illuminants versus wavelength. significant emission in the UV. 18 Therefore, fluorescent or high-pressure illuminants are preferable sources, especially in colorimetric measurements. In comparison with incandescent lamps, fluorescent illuminants have some advantages. Fluorescent illuminants are more efficient than incandescent lightbulbs in terms of an equivalent luminance. A large portion of the consumed energy is converted to usable light and less is converted to heat, whereas an incandescent lamp may convert only 10% of its power input to visible light. However, the main disadvantage of fluorescent or high-pressure illuminants is that they require a ballast to stabilize the lamp and provide the initial striking voltage required to start the arc discharge. 19 The CIE-standardized six groups of illuminants, representing typical fluorescent lamps, were used to determine illuminance responsivity dependence. 20 These include standard fluorescent illuminants FL1 to FL6, broadband fluorescent illuminants FL7 to FL9, narrowband fluorescent illuminants FL10 to FL12, standard fluorescent halophosphate illuminants FL3.1 to FL3.3, DeLuxe-type fluorescent illuminants FL3.4 to FL3.6, three-band fluorescent illuminants FL3.7 to FL3.11, and multiband fluorescent illuminants FL3.12 to FL3.14. A standard high-pressure sodium illuminant HP1, a color-enhanced high-pressure sodium illuminant HP2, and three types of high-pressure metal-halide illuminants HP3-5 standardized by CIE were also used. These illuminants cover the CCT range from 1950 to 6500 K. The SPDs of these illuminants differ from the distribution of the Plankian radiator. The SPD values for each illuminant tabulated in Ref. 20 were normalized to unity at their maximum values, and the F S t S s factors and the variations of the illuminance responsivities of the photometer heads were calculated Table 2. The normalized SPD curves of a standard fluorescent FL5, a broadband fluorescent FL9, a narrowband fluorescent FL12, a standard fluorescent halophosphate FL3.1, a DeLuxe-type fluorescent FL3.4, and a multiband fluorescent FL3.12 illuminant are presented as examples from their groups in Fig. 4. Figure 5 demonstrates Fig. 5 Normalized SPD curves of high-pressure sodium and metal-halide illuminants versus wavelength. 093607-6

Fig. 6 a Variations of F S t S s factors. b Values of S t V d / S t s rel d for standard fluorescent illuminants. the normalized SPD curves of the standard and colorenhanced high-pressure sodium illuminants H1 and H2 and the metal-halide illuminants HP5. Wien s displacement law is not generated for fluorescent illuminants because they are not thermal sources like incandescent lamps. The SPDs of fluorescent sources are more complicated compared to those of incandescent lamps, because poweragainst-wavelength graphs for their light outputs show many sharp peaks, not just one smooth curve. Fluorescent sources have critical sharp peaks at 405-nm, 436-nm, and 546-nm wavelengths, which originate from the excited mercury vapor in tubes. The difference between the SPDs of each type of fluorescent illuminant and a standard source a tungsten filament light source operated at 2856 K is the largest in the shorter wavelengths because the standard source has insufficient spectrum at 2856 K. Table 2 shows that the maximum variation of the F S t S s factor was generally observed for those illuminants that had the lowest CCT in their group. It was also observed that the SPD values of the fluorescent illuminants wavelength ranges extended from 380- to 550-nm increases, depending on the incremental portion of the illuminant s CCT. The critical wavelength for the SPDs was 550 nm, because the illuminant with the lowest SPD under 550 nm pushed the higher SPD beyond 550 nm. The inverse profile was obtained just above 550 nm for SPDs. The F S t S s factors of the photometer heads were influenced more by use of both the narrowband FL10 to FL12 and the three-band fluorescent illuminants FL3.7 to FL3.11. Illuminance responsivities varied within an order of 0.3% for the FR 25.0-1 and the TH15BA photometer heads for both illuminants stated above, and less than 0.2% for the P30SCT photometer head. Figures 6 and 7 demonstrate the variations of the F S t S s factors of the photometer heads and the S t V d / S t s rel d ratios for the standard fluorescent illuminants. Figure 6 b shows that by increasing Fig. 7 a Variations of F S t S s factors. b Values of S t V d / S t s rel d for three-band fluorescent illuminants. 093607-7

Fig. 8 a Variations of F S t S s factors. b Values of S t V d / S t s rel d for high-pressure illuminants. the CCT from 2940 K FL4 to 6430 K FL1, the S t V d / S t s rel d ratios varied from 0.9979 to 0.9962 for the FR 25.0-1, from 1.0016 to 0.9990 for the TH15BA, and from 0.9995 to 0.9982 for the P30SCT. The same variation characteristics were observed for the F S t S s factors of photometer heads Fig. 6 a. The sum of the multiplier S t V d changed from 9.67 to 6.71. Differences between multipliers S t s rel d and S t V d varied from 0.02 to 0.03, from 0.02 to 0.01, and from 0.005 to 0.01 for the FR 25.0-1, the TH15BA, and the P30SCT, respectively. Figure 7 b shows that by increasing the CCT from 2979 K FL3.7 to 5854 K FL3.11, the S t V d / S t s rel d ratios varied from 1.0000 to 0.9997 for the FR 25.0-1, from 1.0031 to 1.0024 for the TH15BA, and from 1.0008 to 1.0000 for the P30SCT. The sum of the multiplier S t V d was less sensitive to the CCTs of illuminants and fluctuated around 3.5± 0.2. Differences between multipliers S t s rel d and S t V d at all CCTs was less than 0.01. Figure 8 b shows that by increasing the CCT from 1959 K H1 to 4039 K H5, the S t V d / S t s rel d ratios varied from 0.9959 H1 to 1.0016 H2, 0.9955 H3, 0.9942 H4, and 0.9952 H5 for the FR 25.0-1. The variations of the above ratios for the TH15BA and the P30SCT photometer heads, respectively, were 0.9993 and 0.9990, 1.0049 and 1.0020, 0.9989 and 0.9983, 0.9975 and 0.9974, and 0.9982 and 0.9980. The sum of the multiplier S t V d changed from 4.37 to 11.11. The differences between multipliers S t s rel d and S t V d varied from 0.02 to 0.05, from 0.003 to 0.02, and from 0.004 to 0.02 for the FR 25.0-1, the TH15BA, and the P30SCT, respectively. Negative differences for both of the multipliers were observed only for the color-enhanced illuminant 0.01, 0.04, and 0.02. 3.3 Relations Between Illuminance Responsivities of Photometer Heads and SPDs of White and Colored LEDs An LED is a solid-state device that has an extremely long life span typically 10 years, twice as long as the best fluorescent bulbs, and 20 times longer than the best incandescent bulbs. These solid-state devices have no moving parts, no fragile glass environments, no mercury, no toxic gasses, and no filament. They also are much more efficient and are more mechanically robust than incandescent lightbulbs and fluorescent tubes. 21 Table 3 Dependencies of the F S s S t factors and the illuminance responsivities of photometer heads versus the CCTs of white LEDs. F S t S s factor Influence on the illuminance responsivity, R V % Model CCT K FR 25.0-1 TH15BA P30SCT FR 25.0-1 TH15BA P30SCT MWGC 4165 1.0007 1.0017 1.0003 0.07 0.17 0.03 MW1D 5759 0.9997 1.0008 0.9994 0.03 0.08 0.06 LW6C 7411 0.9992 1.0004 0.9992 0.08 0.04 0.08 093607-8

Table 4 Dependencies of the F S s S t factors and the illuminance responsivities of photometer heads versus LED colors. F S t S s factor Influence on the illuminance responsivity, R V % Model LXHL Color FR 25.0-1 TH15BA P30SCT FR 25.0-1 TH15BA P30SCT LR5C Royal blue 1.0461 1.0175 1.0201 4.61 1.75 2.01 MB1D Blue 1.0209 0.9961 1.0075 2.09 0.39 0.75 ME1D Cyan 0.9897 0.9936 0.9937 1.03 0.64 0.63 MM1D Green 0.9925 0.9964 0.9952 0.75 0.36 0.48 ML1D Amber 0.9969 0.9969 0.9986 0.31 0.31 0.14 MH1D Red-orange 1.0166 1.0163 1.0109 1.66 1.63 1.09 MD1D Red 1.0148 1.0129 1.0094 1.48 1.29 0.94 In recent years, white and colored LEDs have been used more as informative indicators in various types of embedded systems, in transmitting digital information, and in illumination applications. It is also known that due to the stability and modern features of LEDs, they are preferred for use in photometric and spectrophotometric applications. 22,23 To determine dependencies of the illuminance responsivities of photometer heads versus the CCTs of white LEDs and color features of colored LEDs, the temperaturestabilized high-power Luxeon Star-type LEDs manufactured by Lumileds Lighting, LLC were used. Each LED consists of a Luxeon emitter mounted to a hexagonal aluminum submount. The white LEDs have batwing wideangle radiation pattern and Lambertian flat radiation pattern radiation patterns, whereas the colored LEDs have Lambertian patterns. Each LED was operated using a PTN 150-20 series Heinzinger dc power supply. The SPD of each LED was measured using a double-monochromator-based facility. The LEDs were separately attached to the input opening of an integrating sphere, and the output opening of the sphere with a diffuser was attached to the input of the monochromator. The bandpass of the monochromator and the measurement steps were aligned to 5 nm, and the F S t S s factors were calculated by scanning each LED between 380 and 780 nm. The measurement results for the white and colored LEDs are shown in Tables 3 and 4. White LEDs are composed of an InGaN chip coated with a phosphor composition. In Fig. 9 a, the same phosphor additive was used for the LXHLLW6C 7411 K and LXHLMW1D 5759 K types of LEDs, while a different additive was used for the LXHLMWEC 4165 K. The measurements show that the influences of the F S t S s factors on the illuminance responsivities of the photometer heads were within ±0.1% for the FR 25.0-1, from 0.2% at 4165 K to 0.04% at 7411 K for the TH15BA, and from Fig. 9 a Normalized SPD curves of white LEDs. b Values of S t V d / S t s rel d versus CCTs. 093607-9

Fig. 10 Normalized SPD curves of colored LEDs versus wavelength. 0.03% to 0.08% for the P30SCT see Table 3. Figure 9 b shows that by increasing the CCT from 4165 to 7411 K, the S t V d / S t s rel d ratios varied from 0.9977 to 0.9967 and 0.9962 for the FR 25.0-1. The variations of the S t V d / S t s rel d ratios for the TH15BA and P30SCT photometer heads, respectively, were 1.0017 and 0.9996, 1.0008 and 0.9987, and 1.0004 and 0.9984. The sums of the multiplier S t V d were 13.94, 17.13, and 17.05 at 4165 K, 5759 K, and 7411 K, respectively. The differences between multipliers S t s rel d and S t V d varied from 0.03 to 0.06, from 0.02 to 0.01, and from 0.01 to 0.03 for the FR 25.0-1, the TH15BA, and the P30SCT, respectively. The illuminance responsivities of the photometer heads were more influenced by the colored LEDs. The normalized SPD curves of the colored LEDs are demonstrated in Fig. 10. The basic chip used for the royal-blue LXHL-LR5C, the blue LXHL-MB1D, the cyan LXHL-ME1D, and the green LXHL-MM1D LEDs is InGaN, whereas an AlInGaP chip is used for the amber LXHL-ML1D, the red-orange LXHL-MH1D, and the red LXHL-MD1D LEDs. All the LEDs consist of Luxeon emitters mounted to hexagonal aluminum submounts. The LXHL-LR5C royal blue is part of the world s brightest LEDs, with power 700 mw at 700 ma and superb lumen maintenance that far exceeds other standard and high-flux LEDs. The maximum continuous current for other LEDs is 350 ma. Unlike white LEDs, colored LEDs emit incoherent narrowspectrum light and have narrowband SPDs. As shown in Fig. 10, the dominant wavelengths of the LEDs were 460 nm royal blue, 470 nm blue, 515 nm cyan, 530 nm green, 595 nm amber, 625 nm red-orange, and 640 nm red. Full widths at half-maximums FWHMs of the LEDs were 20 nm amber, 25nm royalblue, redorange, and red, 30 nm blue and cyan and 35 nm green. Figure 11 b shows that the S t V d / S t s rel d ratios for the colored LEDs varied the most for the royal-blue LED 1.043 for the FR 25.0-1, 1.018 for the TH15BA, and 1.019 for the P30SCT. The reason for this high variation is that the royal-blue Fig. 11 a Variations of F S t S s factors. b Values of S t V d / S t s rel d for colored LEDs. 093607-10

Table 5 Uncertainty budget of the F S t,s s factor determination. 100 relative uncertainty Incandescent source LED source Source of uncertainty FR 25.0-1 TH15BA P30SCT FR 25.0-1 TH15BA P30SCT Optical power measurement 0.02 0.02 0.02 0.04 0.04 0.04 Wavelength scale 0.01 0.01 0.01 0.01 0.01 0.01 Wavelength repeatability 0.01 0.01 0.01 0.01 0.01 0.01 Stray light 0.01 0.01 0.01 0.01 0.01 0.01 Current stability 0.01 0.01 0.01 0.03 0.03 0.03 Relative spectral responsivity 0.06 0.08 0.10 0.06 0.08 0.10 Combined standard uncertainty 0.07 0.08 0.10 0.08 0.10 0.11 Expanded uncertainty k=2 0.13 0.17 0.21 0.16 0.19 0.23 LED s discrimination of the SPD is too high compared with that of the standard illuminant. The same variation characteristics were observed for the F S t S s factors of the photometer heads Fig. 11 a. The influences of the F S t S s factors on the illuminance responsivities of the photometer heads for the royal-blue LED were 4.61% for the FR 25.0-1, 1.75% for the TH15BA, and 2.01% for the P30SCT. The variations in the F S t S s factors between the photometer heads for each of the LED colors are explained by the differences in the relative spectral responsivities of the photometer heads from the CIE-V function. The F S t S s factor of a photometer head is influenced more by the LED with the narrowest FWHM. The uncertainty budget to determine the F S t,s s factor for the incandescent and LED sources is given in Table 5. There are six main uncertainty components that directly influence the calculation of the F S t,s s factors. These components are the wavelength accuracy, the wavelength repeatability and stray light of the monochromator, the uncertainty in the SPD measurements, the current stability of the light source, and the relative spectral responsivity of the photometer head. The combined uncertainty was taken as the root of the sum of squares of the uncertainty components. The uncertainty of the photometer head in the relative responsivity determination was the most influential component on the total uncertainties: 0.06% for the FR-25.0-1, 0.08% for the TH15BA, and 0.10% for the P30SCT. The uncertainty in the relative responsivity was determined using uncertainties from the responsivity of the reference trap detector, the power instability of the monochromatic beam, fluctuations in the background radiation, the temperature setting, noise levels in the transimpedance amplifier, etc. 16 The other main uncertainty components used to determine the F S t,s s factors were the SPD measurements of light sources using the reference trap detector, and the current setting stabilities of the light sources. 4 Conclusion Dependencies of the illuminance responsivities of three types of photometer heads with the CCTs of different types of light sources were studied. The characterized photometer heads were a UME-made trap detector-based photometer head the FR 25.0-1, a PRC Krochmann-manufactured photometer head the TH15BA, and a LMT Lichtmesstechnik GmbH-manufactured photometer head the P30SCT. A double-monochromator-based facility was used to measure the SPDs of tungsten filament incandescent light sources and LEDs, and to calculate the F S t S s factor of each photometer head with an expanded uncertainty of 0.2% k=2. The influences of the F S t S s factors on the illuminance responsivities of photometer heads for the fluorescent and high-pressure illuminants were determined using spectral data for each type of illuminant tabulated by the CIE. The measurements and calculations prove that the illuminance responsivity values of the photometer heads calibrated at 2856 K are affected more by the use of LED-type sources. The principal reason for this conclusion is that the LEDs have high SPDs in the restricted and narrow spectral bandwidth of the visible region. Moreover, some differences between the relative spectral responsivities of the photometer heads and the CIE-V function are observed, depending on wavelength, because the photometers are equipped with V -corrected filters. Therefore, some variations on the F S t S s are obtained for each photometer head. In particular, when using the royal-blue LED, the illuminance responsivity for the FR 25.0-1, the TH15BA, and the P30SCT changed from 4.6% to 1.8% and 2.0%, respectively. For the red-orange and red LEDs, the variations on the illuminance responsivity are 1.5%, 1.3%, and 0.9%. When using the white LEDs, the variation on the illuminance responsivity remains restricted within ±0.1% 4000 to 7500 K because white LEDs have SPDs on the 093607-11

whole visible region. The same variation was also observed for the incandescent light sources. A light source with a CCT of 2000 K influences the illuminance responsivity of the photometers an order of 0.1%. Among fluorescent and high-pressure illuminants, the light sources affecting illuminance responsivity are narrowband and three-band fluorescent illuminants 0.3%. The reason for the difference is that such illuminants have abrupt spectral slopes in the narrow spectral bandwidth such as LEDs. High-pressure and color-enhanced illuminants, the SPD of which changes the wavelength range sharply from 570 to 615 nm, have the highest SPD distribution of all the illuminants in the red region of the spectrum and affect the illuminance responsivity the most 0.5%. References 1. F. Samedov and O. Bazkir, Realization of photometric base unit of Candela traceable to cryogenic radiometer at UME, J. Eur. Appl. Phys. 30 3, 205 213 2005. 2. F. Samedov, M. Durak, and O. Bazkır, Filter-radiometer based realization of Candela and establishment of photometric scale at UME, Opt. Lasers Eng. 43 11, 1252 1256 2005. 3. Y. Ohno, Improvement photometric standards and calibration procedures at NIST, J. Res. Natl. Inst. Stand. Technol. 102, 323 331 1997. 4. C. L. Cromer, G. Eppeldauer, J. E. Hardis, T. C. Larason, and A. C. Parr, National Institute of Standards and Technology detector-based photometric scale, Appl. Opt. 32, 2936 2948 1993. 5. J. Metzdorf, Network and traceability of the radiometric and photometric standards at the PTB, Metrologia 30, 403 408 1993. 6. L. P. Boivin, A. A. Gaertner, and D. S. Gignac, Realization of the new candela 1979 at NRC, Metrologia 24, 139 152 1987. 7. P. Toivanen, P. Karha, F. Manoochehri, and E. Ikonen, Realization of the unit of luminous intensity at the HUT, Metrologia 37, 131 140 2000. 8. F. Samedov, O. Celikel, and O. Bazkır, Establishment of a computer-controlled retroreflection measurement system at the National Metrology Institute of Turkey UME, Rev. Sci. Instrum. 76 9, 096102 2005. 9. G. Sauter, Goniophotometry: new calibration method and instrument design, Metrologia 32, 685 688 1995/96. 10. F. Sametoglu, New traceability chains in photometry and radiometry at the National Metrology Institute of Turkey, Opt. Lasers Eng. 45 1, 36 42 2007. 11. F. Samedov and M. Durak, Realization of luminous flux unit of lumen at UME, Opt. Appl. 34, 265 274 2004. 12. P. Toivanen, J. Hovila, P. Karha, and E. Ikonen, Realizations of the units of luminance and spectral radiance at the HUT, Metrologia 37, 527 530 2000. 13. CIE, The Basis of Physical Photometry, Publication No. 18.2 1983. 14. Y. Ohno and A. E. Thompson, Photometry the CIE V function and what can be learned from photometry, in Measurements of Optical Radiation Hazards A Reference Book Based on Presentations Given by Health and Safety Experts on Optical Radiation Hazards, Sep. 1 3, 1998, Gaithersburg, MD, ICNIRP 6/98, CIEx016-1998, pp. 445 453 1998. 15. W. Erb and G. Sauter, PTB network for realization and maintenance of the candela, Metrologia 34, 115 124 1997. 16. F. Sametoglu, Establishment of illuminance scale at UME with an absolutely calibrated radiometer, Opt. Rev. 13 5, 326 337 2006. 17. O. Bazkir and F. Samedov, Characterization of silicon photodiode based trap detectors and establishment of spectral responsivity scale, Opt. Lasers Eng. 43, 131 141 2005. 18. G. Wyszecki, Development of new CIE standard sources for colorimetry, Die Farbe 19, 43 76 1970. 19. CIE, The Measurement of Luminous Flux, Publication No. 84 1989. 20. CIE, Colorimetry, Publication No. 15.3 2004. 21. E. Mills, The specter of fuel-based lighting, Science 308, 1263 1264 2005. 22. C. F. Jones and Y. Ohno, Colorimetric accuracies and concerns in spectroradiometry of LEDs, in Proc. CIE Symposium 99 75 Years of CIE Photometry, Budapest, Hungary, pp. 173 177 1999. 23. F. Samedov, M. Durak, and A. K. Turkoglu, Photometric characterizations of light emitting diodes, in Proc. 2nd Balkan Conference on Lighting, Istanbul, Turkey, pp. 150 155 2002. Ferhat Sametoglu graduated from the physics department, Baku State University, Azerbaijan, in 1993. He worked at the Photoelectronic Institute of the Academy of Science of Azerbaijan from 1993 to 1998 and received his PhD degree there in 1998. He has been working in the Optics Laboratory at the National Metrology Institute of Turkey UME since 1999 and headed the laboratory from 2001 to 2006. He has been a member of the Optical Society of America and SPIE since 2005. His main interest is optical metrology. 093607-12