Design of a light-guide used for the real-time monitoring of LCD-displays

Transcription

1 Design of a light-guide used for the real-time monitoring of LCD-displays W. Meulebroeck *a, Y. Meuret a, C. Ruwisch a, T. Kimpe b, P. Vandenberghe b, H. Thienpont a a Vrije Universiteit Brussel, Dept. of Applied Physics and Photonics (FirW-TONA), Pleinlaan 2, B-15 Brussels, Belgium b Barco View, Kennedy Park 35, B-85 Kortrijk, Belgium ABSTRACT We describe the design of a light-guide that is part of a sensor which continuously measures the quality of Cold Cathode Fluorescent Lamp (CCFL) backlights. This sensor gives information to the electronic system that compensates the screens degradations. The light-guide should fulfill several conditions. First of all the acceptance angle of the lightguide should be in accordance with a predefined value, this for reasons of calibration. Secondly the energy-loss inside the light-guide should be minimal. This is indispensable especially when using the light-guide for color screens. In this case the flux density on the photodiodes surface is much lower in comparison with monochromatic screens due to the color filters which are present. Finally we want the light-guide to be low-cost and we have to make sure that the design could easily be built inside a screen. For the optical design of this light-guide we used both sequential and non-sequential raytracing software. We started our simulations with an L-shaped light-guide profile. The dimensions of this guide were chosen in such a way that the element easily fits into a standard screen. To make sure that the sensor only captures light within the desired acceptance angle we proposed a system using a lens on top of this light-guide. In this paper we describe the design and simulation results of this L-shaped light-guide with extra lens. After making some final changes to the design to reduce the influence of environmental stray light, we ended up with a light-guide which fulfilled most of the predefined specifications. Keywords: optical sensor, LCD-screens, optical design, light-guide, CCFL-backlights, optical simulations, ray-tracing 1. INTRODUCTION More and more Cathode Ray Tube (CRT) displays are replaced by Liquid Crystal Displays (LCD s). Not only for consumer purposes like televisions or personal computer screens, but also in some high-end applications. LCD-screens are ultra flat, show a high brightness and offer a perfect geometry as opposed to their CRT based counterparts, which suffered from geometric imperfections due to the technology or magnetic influences of the environment. In addition they consume less energy. On the other hand this new technology still has some disadvantages and shortcomings. LCDscreens are for example very sensitive to backlight aging and screen degradation. Screen degradations not only occur after a period of time, they also depend on the temperature of the environment and will have a negative impact on the image quality. For some applications this could be a problem; especially those applications where high quality images are required and this during the entire life cycle of the screen. Awaiting the development of LCD-screens with better specifications concerning aging and degradation, one could use a sensor that continuously measures the screen quality and gives the necessary information to an electronic compensation circuit. The sensor consists of a light-guide which captures some light coming from the screen and a photodiode which is placed at the end of the guide. This measurement method enables to handle not only the backlight aging but also the aging of the polarizer, filters and the LCD itself. We illustrate the basic set-up of an LCD-screen with sensor in Figure 1. * wmeulebr@vub.ac.be; phone ; fax ;

2 Signal to compensation circuit Photodiode Electronics Backlight LCD panel Light Light-guide SENSOR Figure 1: The basic set-up of an LCD-screen with sensor 2. DESIGN, SIMULATION AND EVALUATION OF AN L-SHAPED LIGHT-GUIDE WITH EXTRA LENS We have to keep in mind that the light from the screen is captured at the front of the screen, while the electronic compensation circuit is found at the back of the screen. Because of this the light-guide of the sensor should have an L- shaped profile. In Figure 2 we illustrate the basic shape of the light-guide. Light coming from the screen is captured on the entrance window of the light-guide and reaches a first slope window on which the rays are totally internal reflected. As the light-guide lowers the effective size of the screen, the dimensions of the entrance window should be as small as possible. They are chosen in such a way that enough light is captured at the photodiode. We worked with an entrance window of three by eight millimeter. The part of the light-guide between the entrance window and the first sloped window is called the nose of the lightguide. For practical reasons, the length of the nose should also be kept as small as possible. We chose a value lower than three millimeter for this length. The light travels further on into the central part of the guide and is at the end of this part again totally internal reflected by a second slope window where after the light reaches the end leg of the system. The photodiode is placed at the end surface of this leg; the exit window. Because of the smaller dimensions of the photodiode relative to the entrance window, a tapered end leg is used. We assume an exit window of three by three millimeter. This corresponds with the dimensions of the active surface of commercial available photodiodes. The material of the light-guide is Poly-Methyl-MethAcrylate (PMMA). Slope Central part window 1 Slope window 2 Nose Entrance window End leg Exit Light coming window from the screen Photodiode Figure 2: Basic geometry of the L-shaped light-guide. Together with the photodiode this guide forms the sensor. The scope of our work was to design a light-guide for which the acceptance angle is in accordance with a predefined value, in our case an acceptance angle of +/ During the design we had to take care of the condition that the lightguide should be used both for monochromatic and color screens. Therefore we not only simulated the acceptance angle of our system under test but we also made simulations concerning the energy yield of the system which we define as the ratio between the light at the exit and the light at the entrance window. It is obvious that we only obtain a maximum yield when the energy-loss inside the light-guide is minimal. Especially for color screens it is very important to obey this condition. In the latter case the flux density on the photodiodes surface is much lower in comparison with monochromatic screens due to the color filters which are present.

3 We started with the study of the L-shaped light-guide from figure Light-guide without lens We started with the simulation of the acceptance angle of the L-shaped light-guide without lens. For the simulations we used the non-sequential ray-tracing software Speos (Optis). As the index of refraction of PMMA has a relative constant value in the visual part of the electromagnetic spectrum (1.51 for 4 nm and 1.49 for 78 nm) we made the simulations only for one wavelength, more in particular for 58 nanometer. The optimum number of rays we selected by successively tracing a higher amount of rays through the system and looking to the energy yield. At a certain point the energy yield will converge towards a constant value. From the latter analysis we concluded that tracing one million rays through the system is most optimal. After we defined our system we simulated the acceptance angle of the system. The working principle consisted in irradiating the system with a collimated beam with a power of one Watt and to investigate the power at the position of the detector. This simulation was repeated for different angles of incidence. The ratio between the flux density at the exit and entrance window gives us the yield of the system as a function of the angle of incidence. There are no coatings or scattering properties defined for the PMMA-air transitions. We only consider Fresnel reflection and transmission. The simulation result is given in Figure 3. Light-guide without lens Figure 3: Energy yield as a function of the incidence angle for the light-guide without lens This result showed us that for the L-shaped light-guide without lens there is still some light captured for almost all angles between -5 and Light-guide with lens To decrease the acceptance angle we suggested a system where we put a plano-convex lens on top of the light guide. The specifications of the lens are such that the light is focused on one of the sloping windows of the guide, to which we will further refer as the slope window. For a collimated incident beam parallel with the optical axis, the light will be focused at the centre of this window. If the collimated incident beam makes a small angle with the optical axis (= angle of incidence), only a small part of the window will be illuminated. We call this illuminated window the focal window. The size of this focal window increases when the angle of incidence increases We considered two different situations (see Figure 4); a first scenario where the light is focused at the first sloping window and a second option where the light is focused at the second sloping window.

4 Focal window Focal window Lens Light from focal window Lens Acceptance angle Acceptance angle Light from focal window Figure 4: L-shaped light-guide with extra lens to focus the light at the first sloping window (left) or at the second sloping window (right) 2.1. Selection of the sloping window at which the light is focused System 1: focusing on the first sloping window In a first step we calculated the lens parameters such that the light is focused at the first sloping window. We started the design by calculating the parameters for a spherical lens. We assumed our lens to be constructed of the same material as the light-guide. As we place the lens directly in front of the entrance plane of the light-guide, the size of the base-plate of our lens should equalize the size of the entrance plane which is three by eight millimeter. Due to the relative big dimensions of the base-plate, the curvature of the lens can not be very small and as a consequence of this also the focal length of the lens should equal a certain minimum value. In Figure 5 we show the lens parameters and the systems geometry. For the calculation we used the following formulas: R = f(n-1) R² - r² = (R-h)² (1) With R the lens-curvature; f the focal length; n the refractive index of our material; r the semi-diameter and h the sag of the lens Using these formulas we were able to calculate the lens dimensions. For this calculation we worked with an index of refraction of 1.49 (PMMA). After this calculation we optimized the system parameters with the sequential ray-tracing software Solstis (Optis) such that the lens aberrations were minimal for all visible wavelengths. At the end we obtained a focal length f of 9.72 mm, a lens-curvature R of 4.41 mm and a sag h of 3.31 mm. The distance D between the lens baseplate and the first focal window equals 8.22 mm. 3 mm Focal window D r Y X 8 mm R Y Z h r Base-plate lens = entrance plane light-guide Lens Figure 5: Definition of the systems geometry and the lens parameters Although we optimized the system such that the aberrations are as small as possible, we had to conclude that the rays are not well focused in the centre of the first focal window; this because of the many aberrations in the system due to the strong curvature of the lens surface. We illustrate this in Figure 6 where we show the tracing of the rays through the system after the illumination with a collimated beam. We simulated both directions (X and Y) separately by placing a diaphragm in front of the lens. For the X and Y direction the diaphragm diameter equaled respectively three and eight millimeter; the dimensions of our base-plate. Although we knew from the start that using a spherical lens will lead to large spherical aberrations due to the small angles of incidence, we nevertheless wanted to perform the simulation. In case the aberrations would have been acceptable, we would have had immediately a system which fulfilled the predefined conditions.

5 X direction First sloping window Y direction 3 mm 8 mm D spherical lens D Figure 6: Ray-tracing towards the first sloping window using a spherical lens The strong spherical aberrations in the system inspired us to make the lens surface aspherical. As the rays are focused inside material and not into air, we know from theory that we have to apply an elliptical profile [1]. We found the acquired degree of ellipticity for our set-up by optical simulations in the sequential ray-tracing software Solstis. This degree is given by the conical constant E. For an explanation of this constant, we refer to Figure 7. Equation conical surface: ρ Z = 1. R ρ² (1+E). ρ² R² Figure 7: Equation of a conical surface [2] After optimization we ended up with a focal length f of 1.8 mm, a lens-curvature R of 4.41 mm; a lens sag h of 3.31 mm and a conical constant E of The distance D between the lens base-plate and the first focal window equals 8.58 mm. The optimization parameter was the spot size of the focused beam at the first sloping window. As we show in Figure 8, for this system the rays are well focused at the first focal window. Z X direction First sloping window Y direction 3 mm 8 mm D aspherical lens D Figure 8: Ray-tracing towards the first sloping window using an aspherical lens Finally we calculated the dimensions of the focal window such that we have an acceptance angle between +/ After calculation we checked the results by performing some additional simulations on our system (see Figure 9). At the slope window we placed a diaphragm with a diameter equal to the calculated dimensions of the focal window. Next we illuminated the system with a collimated beam under a certain angle. For different angles starting from -2.5 and going until 2.5, we checked if the rays were traced through the diaphragm. We concluded that the sides of the focal plane should be 1.11 mm. If we use for example a focal window with sides of only one millimeter this will lead to an obstruction of the rays as we show in Figure 9. Taking into account that the focal plane of the real system is placed under an angle of 45, the value of 1.11 mm should be increased until 1.57 mm.

6 diaphragm First sloping window size diaphragm = 1.11 mm size diaphragm = 1 mm Figure 9: Simulations to determine the dimensions of the focal window Although this system shows theoretically satisfying results, we rejected this set-up because the practical implementation requires an increase in length of the nose of the light-guide. The distance D between the lens base-plate and the first focal window equals 8.58 mm. This can only be obtained by increasing the length of the nose. Because this is not acceptable, we switched to the study of a second system where we focus the light at the second sloping window System 2: focusing on the second sloping window Analogue like for system one, we calculated the lens specifications and optimized and simulated the system again with the sequential ray-tracing software Solstis. Due to the larger distance between the lens and the slope window, we can work with a spherical lens. This lens has a focal length f of 26.2 mm, a lens-curvature R of 9.31 mm and a lens sag h of 1.38 mm. The focal window should be 1.7 mm by 1.7 mm to obtain an acceptance angle of +/- 2.5 degrees. These results were found using the same method as explained in the previous paragraph. Here the distance between the lens and the focal plane is large enough, so we can keep the original nose length. We decided that a focusing of the light at the second sloping window is preferable. For this system we can use a spherical lens instead of an aspherical one and the original length of the nose can be retained Acceptance angle and energy yield of the light-guide with lens (non-symmetrical end-leg) After this design we checked the performance of the light-guide with lens. Both the acceptance angle and the energy yield were simulated and evaluated. We started with the simulation of the acceptance angle. For this we followed the same working principle as explained in paragraph 1. We only consider Fresnel reflection and transmission. In Figure 1 (left) we give the simulated energy yield as a function of the incidence angle. We concluded that the acceptance angle of the light-guide with lens is indeed much lower in comparison with the acceptance angle of the light-guide without lens. Only for incident angles between -3 and +5 there is light captured on the detector. Although the extra lens lowers the acceptance angle, we have to remark that the system is still not perfect. The profile is not symmetrical and the energy yield is very low. If we look for example to the energy yield of the system for an angle of incidence of zero degrees, we learn that only 23% of the incident light reaches the exit plane of the system. To explain these observations we repeated the same simulation with this difference that all the side walls except the first sloping and the focal window were made perfectly absorptive. As can be derived from Figure 1 (right) the profile is now smaller. This means that the earlier observed broadening is a result of scattering of light next to the focal window of the second sloping window. A part of this scattered light will reach the exit window. The energy yield is still as low as before (17%) and we also remark that the profile is not centered round zero degree but that there is a shift of about -.5. The next step of our research was to find an explanation for these two effects.

7 Light-guide with lens (Fresnel reflection and transmission on side walls) Light-guide with lens (absorptive side walls) Figure 1: Energy yield as a function of the incidence angle for the light-guide with lens; respectively with Fresnel reflection and transmission at the walls (left) and with absorptive walls (right) In Figure 11 we give the energy yield for two different positions of the detector plane. In situation A the detector plane is placed just after the second sloping window; situation B reproduces the real situation where the detector plane is placed at the position of the photodiode. We simulated the energy yield for an angle of incidence of. From Figure 11 and Figure 12 we conclude that the small energy yield is due to the big energy losses in the end leg of the light-guide. In the latter figure we show the result of a non-sequential ray tracing through the light-guide. It is clear that most of the light is lost in the tapered non-symmetrical end leg of the system (energy yield of 48.27% at position A and 8.15% at position B). The non-symmetrical end leg is also responsible for the earlier observed shift. Light falling in under an angle of incidence of zero degrees will be stopped by the end leg. Energy yield for two different positions of the detector plane for an angle of incidence of. A , ,15 5 B A B Figure 11: Comparison of the energy yield of the light-guide with lens for two different positions of the detector plane Figure 12: The low energy yield is due to the high energy loss in the non-symmetrical tapered end-leg

8 To increase the energy yield of our system and make it symmetrical, we suggested to make the end leg symmetrical. To obtain the predefined acceptance angle we have to make sure that only the light coming from the focal window will reach the end of the light-guide. All the other rays are stray light and will only cause a change in acceptance angle which is unacceptable. In the next paragraph we describe the simulation results of the light-guide with lens and symmetrical endleg. We compare the results of different options we proposed to eliminate the stray light Acceptance angle and energy yield of the light-guide with lens (symmetrical end-leg) All the simulations which we discuss in this paragraph are performed with the non-sequential ray-tracing software ASAP (BreaultResearch). This was necessary because our Speos license did not allow us to apply surface coatings to the geometry. As we will further on explain, we tried to apply surface coatings to eliminate stray light. The simulation strategy remained the same. After the definition of the correct geometry we performed four successive simulations where we calculated the energy yield for four different windows: at the position of the photodiode (window 1), at the end-leg (window 2), at the region of the second sloping window which does not form part of the focal window (window 3) and at the first sloping window (window 4). For the four simulations we used a different source definition. We started from a point source and changed the divergence angle of the source. For the first simulation the source emitted light under angles between zero and 2.5, for the second simulation between zero and 5, for the third simulation between zero and 7.5 and for the last simulation between zero and 1. From this data we could calculate the energy yield for the four windows under test for angles between -2.5, , and In this way we get a clear insight into where the light is lost in the system. The energy yield of the light-guide corresponds with the energy yield at the photodiode (window 1). Ideally this value should be 1%. Another target is to have an acceptance angle between -2.5 and 2.5. To get an idea about the acceptance angle we introduced the term angular contrast which we define as the ratio between the energy yield at the photodiode for angles between -2.5 and The higher the angular contrast, the more the system converges to the predefined acceptance angle. We started with the simulation of the light-guide with lens and with symmetrical tapered end-leg. In Figure 13 we give a side view of the former non-symmetrical and the new symmetrical tapered end-leg. We applied no coatings or scattering properties to the element, so the light within the guide is reflected and transmitted following the Fresnel laws. Light from focal window is mostly lost Light from focal window reaches the detector Figure 13: By using a symmetrical tapered end-leg (right) instead of the former asymmetrical tapered end-leg (left) we minimize the energy loss inside the system As can be learned from Figure 14a the energy-loss in the end-leg is now minimal. There is only a small amount of light which escapes at the walls of the end-leg. The energy yield of the system is high (76.8%). The angular contrast on the other hand is very low (= 1). This is caused by the many rays reflected at window 3 which reaches the photodiode. This is illustrated in Figure 14b.

9 Light-guide with lens (uncoated; no scattering properties) Window 1 (photodiode) Window 2 (tapered end leg) Window 3 Window 4 (first sloping window) Other window a b Figure 14: Simulation results of the uncoated light-guide with lens (no scattering properties) To lower the number of rays that reach the photodiode coming from window 3, we investigated two main options. In Figure 15 we give a schematic view of both options. Roughness Focal window Different slopes on focal and slope window Focal window Light from focal window Light from focal window Figure 15: To get rid of the stray light we investigated two different options. A first system where we make some surfaces rougher than others (left) and a second option where we work with different slopes on the focal and second slope window (right) In a first trial we started from the light-guide with lens and tried to get rid of the stray light by playing on the roughness of certain walls. We hoped that due to the rougher surfaces the rays coming from window 3 experienced more reflections at the end-leg and as a consequence the chance of absorption of these rays becomes higher. In a second approach we changed the slope of window 3. The slope of the focal window remains the same. The light rays falling in onto the slope window are totally internal reflected. Thanks to the choice of the two different slopes, only the light which is focused by the lens onto the focal window will reach the end surface of the light-guide. The other rays are leaving the system at the slope window.

10 Light-guide with surface roughness Because it was very difficult to have a link with the reality concerning scattering properties, we started with a system with rather arbitrarily chosen scattering properties. On the first sloping window and on the focal window of the second sloping window we have Fresnel reflections and transmissions. The surfaces to which we applied scatter properties we defined to specularly reflect the incident light for 2%, to transmit 4% of the incident light and to scatter the rest of the light (4%) according to a Lambertian model. On all the other windows we also have Fresnel reflections and transmissions. We simulated two different systems. First we only applied scatter properties to window 3; in a second simulation session we applied scatter properties both to window 3 and all the walls of the end-leg ( window 2 ). We show the simulation results in Figure 16. For the first system we have a high energy yield (68.6%). Moreover we have a higher angular contrast (= 5) compared to the light-guide without scattering properties. For the second system the energy yield is much lower, only 25.8% of the incident light between zero and 2.5 degree is captured on the photodiode. The angular contrast on the other hand is better (= 13). The higher angular contrast for the second system can be explained by the higher number of rays coming from window 3 which do not reach the detector. Unfortunately also a huge amount of rays coming from the focal window are lost in the end leg. This explains the low energy yield (25.8 %) for the second system. Light-guide with lens (uncoated; roughness on window 3) Light-guide with lens (uncoated; roughness on window 2 and 3) Window 1 (photodiode) Window 3 Other Window 2 (tapered end leg) Window 4 (first sloping window) Window 1 (photodiode) Window 3 Other Window 2 (tapered end leg) Window 4 (first sloping window) a b Figure 16: Simulation results of the uncoated light-guide with lens with a roughness on window 3 (a) and a roughness on window 2 and 3 (b) A possible solution exists in shortening the end-leg. With this geometry, we expect that most of the light lost in the endleg is coming from window 3 while the light reflected at the focal window will reach the photodiode (see Figure 17a). We shortened the end-leg from 3 millimeter to 2 millimeter. With this new system we achieve an energy yield of 4.3% and an angular contrast of 12 (see Figure 17b). As a first general conclusion we can say that we theoretically came to a geometry which meets our predefined values concerning the acceptance angle and energy yield. The acceptance angle is between +/- 2.5 and we have an energy yield of 4.3%. Future research should learn us if it is easy to implement this geometry in practice. If we fabricate the geometry with a milling-machine, we will already have a certain surface roughness due to the fabrication process. With an optical surface profilometer we can measure the surface profile. With this information we are able to define the precise structure of the light-guide in the ray-tracing software ASAP and as such work with realistic scatter properties. If this simulation gives satisfactory results we can think on a mass-fabrication of this component. Taking into account the shape and dimensions of the component, we expect that milling 15 pieces of the light-guide per day would be possible (6 per hour). Moreover we can also think on replicating this structure; this under the condition that the replication process will keep the surface roughness of the original component.

11 Light-guide with lens (uncoated; roughness on window 2 and 3; shorter end leg) Window 1 (photodiode) Window 3 Other Window 2 (tapered end leg) Window 4 (first sloping window) a b Figure 17: Simulation results of the uncoated light-guide with lens with a roughness on window 2 and 3 and with a shorter end-leg Light-guide with different sloping angles on window 3 The idea of this approach consists of changing the slope angle of window 3. The slope angle of the focal window remains the same. The light rays falling in onto the slope window are totally internal reflected. Thanks to the two different slope angles, only the light which is focused by the lens onto the focal window will reach the end surface of the light-guide. The other rays are leaving the system at the slope window. In Figure 18 we give the simulation results of this geometry. These are the results of an uncoated, polished light-guide. We do not take scatter properties into account. The rays are following the Fresnel laws for reflection and transmission. This system works like we want. The light rays falling in onto window 3 are leaving the system. The rays coming from the focal window are guided towards the photodiode. In Figure 18a we illustrate how the rays are traced through the system in the environment of the focal window. This system gives rise to an energy yield of almost 7% and an angular contrast of 16 (see Figure 18b). Light-guide with lens and different sloping angles (uncoated; no scattering properties) Window 1 (photodiode) Window 3 Other Window 2 (tapered end leg) Window 4 (first sloping window) a b Figure 18: Simulation results of the uncoated light-guide with lens with different sloping angles We conclude that the last described system gives the best results so far. It has a high energy yield and angular contrast. Finally we have to remark that all the environmental light should be eliminated. We have to make sure that environmental light can not enter the light-guide. A possible solution exists in coating the light-guide. We simulated a system where we put a metallic coating (95% reflection) on all the walls except on the entrance and exit window and on window 3. With this system we finally came to a system which fulfilled all our predefined specifications: a high energy yield (66.4%), a high angular contrast (17) and almost no environmental light which could enter the system. The simulation results we show in Figure 19.

12 Light-guide with lens and different sloping angles (metallic coating; no scattering properties) Window 1 (photodiode) Window 3 Other Window 2 (tapered end leg) Window 4 (first sloping window) Figure 19: Simulation results of the coated light-guide with lens with different sloping angles 3. CONCLUSIONS We succeeded in designing a low-cost, high-performance light-guide which is part of a sensor that could be used for the continuous compensation of the degradation of LCD-screens. The light-guide is made of PMMA and has an L-shaped profile with a spherical lens at the entrance plane. This lens focuses the incident light at the second sloping window of the light-guide. By applying different sloping angles to the second sloping window we can make sure that only the light within a predefined acceptance angle reaches the photodiode at the end of the light-guide. We designed a system with an acceptance angle between +/ The total energy yield at the detector is 66.4%. Environmental light can be eliminated by coating certain parts of the guide. The set-up should be quite flexible as we could choose the dimensions of the focal window in accordance with the acceptance angle we want. REFERENCES 1. Pedrotti & Pedrotti, Introduction to optics, Prentice Hall International editions, 1996, second edition, chapter 3 paragraph User-manual optical ray-tracing software Solstis (v. 4.6), p. 5-3, July 1998 ACKNOWLEDGMENTS The authors want to express their gratitude to het Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen (IWT) for the funding of this research.