A novel light luminaire is proposed and experimentally analyzed, which efficiently mixes and projects the tunable light from red, green and blue (RGB) light-emitting diodes (LEDs). Simultaneous light collimation and color mixing is a challenging task because most collimators separate colors, and most color mixers spread the light beam. Our method is simple and compact; it only uses a short light pipe, a thin diffuser, and a total internal reflection lens. We performed an experimental study to find a balance between optical efficiency and color uniformity by changing light recycling and color mixing.
© 2011 OSA
The unique controllability of LEDs is adding new dimensions of light utilization [1,2]. In contrast to conventional light sources, LED lamps can be easily controlled in: emission spectrum, polarization, temporal modulation, radiation pattern, and color. Multicolor LEDs offer real-time control of its color emission as never before in lighting history. This controllability of illumination is maximized by the ability of LED lamps to be easily integrated by small color LEDs, and to have their light output manipulated without inefficient color filters.
The color controllability of LEDs is very attractive in applications as down lighting, spot lighting, entertainment, architectural, floodlight, and show lighting, where a narrow color beam is projected at a distance. In contrast to electronic control, optical color mixing addresses color uniformity in three dimensions, i.e., not only uniformity across a plane, but also along a working distance. Traditionally, optical color mixing is done by wide-angle or long-distance projection. In a short distance with a limited spreading angle, well color mixing is difficult to achieve, even not considering the efficiency. In this paper, we propose a novel scheme to make color mixing and light projection of RGB LEDs. We also present an experimental study of its performance. We measure the color uniformity and optical efficiency for different conditions. And it is shown how by conveniently controlling light recycling the light of multiple color LEDs can be mixed and collimated efficiently.
2. New method of color mixing and projection
Light of multicolor LEDs can be directly combined without additional optics (Fig. 1(a) ), but strong color patterns occur if the illumination distance is not long enough [3,4]. Mixing rods, frosted glass, volume scattering, holographic and deterministic diffusers are used in illumination systems to provide a uniform output [5–8]. However, in opposite to light projection, all they significantly increase the width of the beam cone (Fig. 1(b)). In the other hand, non-imaging elements can reshape the radiation pattern of LEDs to collimate the light distribution [8–11]. But these projection devices assume that the source is a single monochromatic LED. When a cluster of RGB LEDs is used instead of a single color source (Fig. 1(c)), the result is a color fringing light projection [2,8,12]. This is a challenging problem because the solutions may be sensitive to the position of each LED, the optical design may be complex, or they may be difficult to fabricate [13,14].
We present a new design that overcomes most of these problems. It generates a uniform narrow beam whose spot size can be changed; and it requires only compact monotonic surfaces, which greatly simplifies the design and fabrication of the optics. The design concept is simple; first, a straight lightpipe is applied to make color mixing. However, the optimal length of a mixing rod could be too long when using multicolor LEDs [15,16]. In order to shorten it, we introduce a volume scattering diffuser in the light pipe to enhance the color mixing as shown in Fig. 2 . Note that the appropriate combination of the lightpipe and diffuser determines light recycling and color mixing . The exit face of the tube is attached at the entrance of a projecting lens. And then the optical design of the collimating lens becomes easy because now its input is a single-color homogenous light source. We designed a special TIR lens that efficiently projects the light of a Lambertian source [9,10,17]. The light pattern projected by the TIR lens has high color uniformity because the diffuser significantly improves the color mixing of the short light pipe. Figure 3 illustrates the difference between missing and using the diffuser. It is a photo of the projected bright spot over a white surface.
3. Optical efficiency
Optical efficiency is the percentage of light from the LED transmitted through the optical system. Since several optical elements interact with light, the optical efficiency becomes an important factor to evaluate the system performance. We define the optical efficiency as the ratio of the output luminous flux to the input luminous flux (both in lumens). In other words, it is the ratio of the luminous flux of the multicolor LED collimating lamp with respect to the light flux of the multicolor LED. And then to evaluate the efficiency, luminous flux should be known. Figure 4 shows the experiment setup we used to measure the optical efficiency. The multicolor LED was attached with a white base inside a large integrating sphere (Fig. 4(a)). It was a SphereOptics 40-inch diameter integrating sphere photometer. The lamp was attached to one external port of the integrating sphere (Fig. 4(b)). It must be noted that the light reflection and transmission through the lightpipe, diffuser and collimating lens are wavelength-dependent. And then the optical efficiency of the system should be a little different for different RGB color combinations. Next section discusses the other important parameter, the color uniformity.
4. Color uniformity
It is typical to display the color variation in the chromaticity coordinate system as a cluster of points surrounding the reference color point. However, it is practical and more objective to quantify the uniformity of a color pattern with one single and meaningful value. The traditional metrics quantify the “non-uniformity” of color distribution. But it has more sense to assess the uniformity instead of the non-uniformity. And then, based on a recent work , we calculate the color uniformity with3]
In order to measure the color coordinates, a colorimeter was positioned 40 cm from the exit end of the lamp (Fig. 5(a) ). We selected this measuring distance because the angular intensity (normalized) practically is the same for longer distances (see Fig. 5(b)). We used a chroma meter (Konica Minolta CL-200), whose performance was tested against an integrating sphere coupled with a spectrophotometer. To increase the spatial resolution, the aperture size was reduced to 0.7 cm diameter with an iris.
The light pattern is sampled across a circular grid of measurement points by both rotating the lamp, and linearly moving the colorimeter (see Fig. 5(a)). The colorimeter was moved with a translational stage along a line over the illuminated plane, and the luminaire was rotated about the optical axis. As seen in Fig. 6 , the result is a circular grid of 29 cm diameter , which should contain as much as possible points. But the measurement time increases with the number of measurement positions. Therefore we made several tests for choosing a suitable number. Figure 6 shows five configurations. Configuration (b) has many measurement positions (73), and then it is the reference scheme. The other configurations have only 37 and 25 points. We measured the color uniformity of the light pattern shown in Fig. 6(a). The values of color uniformity, using configurations (b)-(f), are: (b) 39.1%, (c) 37.6%, (d) 39.2%, (e) 44.5%, and (f) 42.6%. Note that configuration (d) has the uniformity most similar to that of the reference (b), and then we used the configuration (d) in all the following measurements. In other words, in Section 5 we calculate the color uniformity with Eq. (1) by using M = 37 with the configuration of Fig. 6(d). Although, it could be easier to measure a square array than a circular grid of points, we used the circular configuration because of two reasons. First the square grid gives equal importance to all points of the illumination pattern, but the points near the center of light pattern have more impact for the observer’s visual field. The circular approach solves this issue because the density of points is larger near the central region. The other reason is that a square grid is not compatible with a circular illumination pattern, and difficulties rise near the corner (points near the perimeter of the circular light pattern).
In addition, the color distribution was visually recorded with a camera. A translucent diffuse screen was positioned 40 cm from the light lamp, and then the light transmitted through the screen was imaged at a charge-coupled device (CCD) camera.
5. Experimental analysis
In this section the performance of the lamp is experimentally analyzed. For each measurement the light recycling and color mixing of the lightpipe/diffuser are varied. In particular we show the relationship between optical efficiency and color uniformity in function of: characteristics of diffuser, length of light guide, position of diffuser within light guide, and using two diffusers. We selected a 3-in-1 LED, where one red die, one green, and one blue die are put together in a single package (Fig. 7(a) ). Properly dimming each color die, this RGB LED produces a plenty of colors. In the experiment we generated white light with different correlated color temperature (CCT). In particular, we adjusted the individual drive currents to produce light with CCTs of 3000K, 4500K, and 6500K. However, the color can be freely changed with appropriate LED die control. And also the light pattern can be changed through different TIR lens.
We used a light guide with square cross section in all experiments. A circular cross section lightpipe was included in Section 5.1 for comparison purposes (Fig. 7(b)). We assembled and tested a wide variety of lightpipes. Silver scatter sheet was used for the reflective sidewalls (with an approximately 90% reflectivity). This type of sheet is usually employed in both lighting and display backlighting. In general, the lightpipe length L was 7 mm, and the cross-section D was 7 mm to facilitate the introduction of the RGB LED inside the light guide. We used the TIR lens shown in Fig. 7(c), which was designed to efficiently project the light of an extended Lambertian source [9,10,17]. This lens collimates the light into a narrow beam; its angular distribution is shown in Fig. 5(b).
5.1 Dependence upon the type of diffuser
Inherent to the idea of light homogenization is the use of a diffuser plate. As shown in Fig. 3, the diffuser is a key element to enhance the color mixing. This is why using the correct diffuser is important. Figure 8 shows the lamp performance for different diffuser types. We used four commercially available diffusers for this experiment: (D1) DP1309, (D2) TP228, (D3) PKK0030-300, and (D4) PKK0030-500. Their optical characteristics are shown in the table attached to Fig. 8(a). By transmittance we mean the single shot measurement of transmittance . Figure 8(a) includes the measurement of lamps assembled with light guides of both circular and square cross section. The plot shows that a luminaire using a square lightpipe has better color uniformity than one using a circular light guide, but has a little less optical efficiency. This is because a square cross section tube is a better homogenizer than a circular one , but the optical efficiency is a little lower. The graph also shows how a thin diffuser with wide full width half maximum (FWHM) angle is better for color mixing, and how a thin diffuser with narrow FWHM angle helps with better optical efficiency.
The optical efficiency of the lamp with circular cross section lightpipe is around 5.5% better than one using a square cross section tube. However, the color uniformity of a lamp with square lightpipe is about 30.0% higher than one using a circular tube. Therefore we used in all following experiments square cross section lightpipes. Figure 8(b) shows the image of the light pattern projected by the lamp with square cross section tube. It can be noted that color fringes are nearly impossible to observe.
5.2 Dependence upon tube length
Total color mixing is achieved if the light pipe is infinitely long. In the real world, a luminaire is of finite size, and most of the times it needs to be compact. Hence we analyzed the influence of length within a range from 2.1 to 10.5 mm. Figure 9(a) shows the relationship between optical efficiency and color uniformity for several lightpipe lengths. The graph shows how color uniformity increases with length, and how a short pipe helps with better optical efficiency. Figure 9(b) shows efficiency and uniformity in function of lightpipe length. From this figure we can see that when L/D is larger than 0.8 the uniformity is larger than 70%. Similarly, when L/D is smaller than 1.2, the optical efficiency is larger than 52%. Therefore, a length-width ratio 0.8<L/D<1.2 assures good color mixing and efficiency.
5.3 Dependence upon diffuser position
The diffuser not only increases color mixing, but also controls light recycling. Light that is reflected back is recycled by all the reflective walls of the tube. Therefore, the height of the diffuser influences the overall optical efficiency. Hence we analyzed the effect of the diffuser position inside the lightpipe of length L = 7 mm. We varied the position within a range from 2.1 to 7 mm. Figure 10(a) shows the relationship between optical efficiency and color uniformity for different diffuser locations. The graph shows how the diffuser at top is better for color mixing, and how the diffuser near the RGB LED helps with better optical efficiency. The light patterns are shown in Fig. 10(b). We observe a somehow good color mixing for uniformities larger than 50%, and it is nearly impossible to observe color fringes for uniformities larger than 70%.
5.4 Lightpipe with two diffusers
Using two diffusers instead of one improves color mixing, reduces the length of tube, and makes the lamp very compact. This increases the versatility in the design to improve the lamp performance. Many possible combinations of different types of diffusers can be used. We tested several combinations, and we found that using D3 on the top, and D5 inside the tube was the best option. Diffuser D5 is the commercially available diffuser TP292, which has a 75μm thickness, 89.7% one-shot transmittance, and FWHM = 22°. Figure 11(a) shows the relationship between optical efficiency and color uniformity for different positions of the inner diffuser. The graph shows how color uniformity significantly increases when the distance between diffusers increases. And the color uniformity is almost perfect when the separation between the two diffusers is maximal. This could be because the cavity enclosed by the diffusers is a better color mixer if it is large. However, the optical efficiency is quite larger if the diffusers are putted together. Note that using L1/L = 1 in a shorter tube improves compactness and increases the optical efficiency keeping high color uniformity. It is interesting to note that the optical efficiency is almost the same for all other diffuser positions, and even more, the efficiency increases with uniformity. The light patterns are shown in Fig. 11(b). All the patterns show excellent color uniformities, and it is impossible for us to observe color fringes.
We have proposed a new multicolor LED projection luminaire, which has been constructed, and experimentally analyzed. This color tunable lamp efficiently mixes and projects the light from RGB LEDs, which is very useful in many applications for color or CCT changing. The method is simple, compact and effective, it is composed of only three parts: one short, straight, and high reflective lightpipe; a thin volume scattering diffuser; and a compact TIR collimating lens. We performed an experimental study to find a balance between optical efficiency and color uniformity. In order to objectively assess the optical efficiency and color uniformity they were defined, and the setup for their measurement was explained. For the experimental analysis, we varied light recycling and color mixing in the cavity integrated by the lightpipe and the diffuser. Four conditions were analyzed: (1) the dependence upon the diffuser type, (2) dependence upon lightpipe length, (3) dependence upon diffuser position, and (4) the effect of using two diffusers. In the first condition, we found that the diffuser thickness and FWHM are the key diffuser properties. In the second condition, we found that there is an optimal range of length-width ratios for the lightpipe. In the third, the diffuser at top of tube increases color mixing, and at bottom increases the optical efficiency. The fourth condition shows how the color uniformity is almost perfect when the separation between the two diffusers is maximal. Depending on the color uniformity tolerances, it is important to consider how the efficiency-uniformity relation affects the system performance. In general, we observed a somehow good color mixing for uniformities larger than 50%. And, it was hard to observe color features for uniformities larger than 70%.
This study was sponsored by the National Science Council of the Republic of China with the contracts of no. 97-2221-E-008-025-MY3, 99-2623-E-008-002-ET and NSC100-3113-E-008-001. The authors would like to thank Breault Research Organization and Howard Huang for the support of simulation with ASAP.
References and links
2. W. J. Cassarly, “High-brightness LEDs,” Opt. Photon. News 19(1), 18–23 (2008). [CrossRef]
4. J. T. Dong, R. S. Lu, Y. Q. Shi, R. X. Xia, Q. Li, and Y. Xu, “Optical design of color light-emitting diode ring light for machine vision inspection,” Opt. Eng. 50(4), 043001 (2011). [CrossRef]
6. C. C. Sun, W. T. Chien, I. Moreno, C. T. Hsieh, M. C. Lin, S. L. Hsiao, and X. H. Lee, “Calculating model of light transmission efficiency of diffusers attached to a lighting cavity,” Opt. Express 18(6), 6137–6148 (2010). [CrossRef] [PubMed]
7. J. Muschaweck, “Randomized Micro Lens Arrays for Color Mixing,” Proc. SPIE 7954, 79540A (2011). [CrossRef]
8. W. J. Cassarly, “Recent Advances in Mixing Rods,” Proc. SPIE 7103, 710307 (2008). [CrossRef]
12. D. Esparza and I. Moreno, “Color patterns in a tapered lightpipe with RGB LEDs,” Proc. SPIE 7786, 77860I (2010). [CrossRef]
13. E. Bailey, “Narrow beam RGB array optic,” Proc. SPIE 6669, 666917 (2007). [CrossRef]
14. R. P. van Gorkom, M. A. van As, G. M. Verbeek, C. G. A. Hoelen, R. G. Alferink, C. A. Mutsaers, and H. Cooijmans, “Etendue conserved color mixing,” Proc. SPIE 6670, 66700E (2007). [CrossRef]
15. C. M. Cheng and J. L. Chern, “Illuminance formation and color difference of mixed-color light emitting diodes in a rectangular light pipe: an analytical approach,” Appl. Opt. 47(3), 431–441 (2008). [CrossRef] [PubMed]
17. C.-C. Sun, T.-X. Lee, S.-H. Ma, Y.-L. Lee, and S.-M. Huang, “Precise optical modeling for LED lighting verified by cross correlation in the midfield region,” Opt. Lett. 31(14), 2193–2195 (2006).
19. We chose a 29 cm diameter because more than 95% of the luminous energy is enclosed within this circle. At this radial position the view angle of the circle is 20 degrees. From Fig. 5b it can be seen that the intensity is quite low at this emission angle.