Currently the major issues in applying the laser as an illumination source for projectors are beam shaping and laser speckle. We present a compact total solution for both issues by using a diffractive beam shaper associated with a cylindrical lens for the illumination optics and a vibrating motor attached to the beam shaper to eliminate speckle on the projection screen. The diffractive beam shaper features a double-sided microlens array with a lateral shift to each other. The illumination pattern is free of zero diffraction order mainly due to the continuous and spherical surface relief of the lenslet, which can be accurately fabricated with diamond turning and injection molding without quantizing surface relief, so that the illumination pattern on the microdisplay can match the design very well with high diffraction efficiency. In addition, the vibration of the diffractive beam shaper in the longitudinal mode has been found effective for eliminating the dot pattern in the illumination and reducing laser speckle on the projection screen. The proposed laser illuminator has been implemented on a three-panel LCoS projector engine to replace the traditional UHP lamp. The uniformity and speckle contrast are measured to be 78% and 5.5% respectively, which demonstrates the feasibility and potential of the proposed scheme.
© 2012 OSA
With the ever increasing of luminous efficiency and reliability, lasers have been regarded as the potential illumination light source for next generation projectors, especially for miniature projectors, due to their high collimation and hence high collection efficiency. For most conventional light valve projection displays illuminated by UHP or LED lamps, the illumination modules consist of multiple optical components, which makes it difficult to reduce the form factor of the whole projection engine. In addition, the light collection efficiency on the microdisplay panel is constrained by the large source etendue [1–3], and this limitation becomes more aggravated with the reduction of panel active area. Laser light sources provide potential solutions for those issues. High collimation of lasers is helpful not only for the collection efficiency but also for the compactness of the illumination optics. With the additional feature of coherency, diffractive optical elements (DOE) become available in the illumination light path. The advantage of laser source for projectors is even more prominent if the microdisplay requests polarized light, such as liquid crystal on silicon (LCoS) panels.
In order to achieve uniform illumination on the microdisplay by converting the Gaussian laser beam into a flat-top profile, several schemes have been proposed, such as double-sided microlens array , bi-convex aspheric lens , chirped microlens arrays , binary pixelated beam shapers , and random microlens arrays . Among which diffractive elements exhibit better compactness. However, the existence of zero-order diffraction in the illumination pattern becomes the major issue. Zero order diffraction mainly comes from the quantization or fabrication error of the DOE surface relief. Staircase surface relief can be more easily fabricated with high accuracy but has inherent limit of diffraction efficiency, whereas piecewise continuous surface relief exhibits high diffraction efficiency but cannot be easily realized with high accuracy, and hence the diffraction efficiency losses still. Solutions for eliminating the zero-order diffraction were developed by using a tapered substrate for a chirped microlens arrays component  or by randomizing parameters of one-side microlens arrays component .
Another major issue in laser-based display system is the speckle pattern on the screen due to the coherency of laser sources. The formation and statistical analysis of speckle has been completely discussed in literatures [9, 10]. Several schemes of speckle reduction have been developed and those could be categorized into light source modulation [11, 12], spatial averaging , digital image processing , and temporal averaging [15–19]. Among which the concept of temporal averaging could be the most widely used, and the speckle contrast can be reduced from about 80% to 4-13% by rotating diffractive optical elements , ultrasonic devices , fast 2-D scanning micro-mirror , moving screen , motionless diffractive optical element  or shift-averaging .
The abovementioned two issues for laser projectors are mostly resolved individually with corresponding dedicated devices. In this paper, a vibrating diffractive beam shaper associated with a cylindrical lens is proposed as a compact total solution for illumination optics of laser illuminated projectors to resolve both issues. The feature size of the lenslet on the beam shaper is around hundred micrometers but it reveals strong diffraction effects. The design, fabrication and characterization of the diffractive beam shaper are described first, followed by the evaluation on the overall performance of illumination quality and speckle contrast tested on a three-panel LCoS projector engine. The schematic system architecture is shown in Fig. 1 .
2. Diffractive laser beam shaper for illumination optics
Microlens array (MLA) or fly-eye lens has been widely used in light valve projection display illuminated with UHP lamps or LEDs. The concept is to divide a non-uniform illumination pattern from the source into several segments, and each segment is optically mapped onto and illuminates the whole active area of the microdisplay. The finer the segmentation, the better the uniformity. The feature size of the lenslet is normally at the scale of a few millimeters and refraction dominates the light path behavior. However, when the feature size of the lenslet is reduced to the scale of hundred micrometers and the microlens array is illuminated by a highly coherent source, such as lasers, diffraction becomes dominative. The intensity distribution starts with circular Gaussian profile at the exit of the microlens array and gradually develops into a dot array pattern at the far field, which manifests the domination of diffraction. The distribution pattern of the dot array is similar to the shape of the lenslet and the pitch of the dot array increases with the propagation distance . As long as there is no strong peak diffraction order in the pattern and the gap between the dot peaks is sufficiently small down to a few microns, the uniformity can be improved with temporal averaging technique which will also be used for reducing laser speckle on the screen. Considering the tolerance and manufacturability of the microlens array, the surface profile of the lenslet is chosen to be identical and spherical. With this constraint, a single side microlens array was found difficult to achieve uniform dot array with the desired distribution size. Therefore, a double-sided microlens array with more design parameters was proposed, where the microlens array on two sides are totally identical but with a lateral shift to each other , as shown in Fig. 2 . This double-sided microlens array still behaves as a diffractive element to shape the laser beam into a square illumination pattern composed of a dot array with very fine dot pitch.
Figure 3 shows the parameters in optimizing the double-sided microlens array with a lateral shift SL. The phase distribution function ψ(ζ,η) provided by the microlens array needs to turn the circular Gaussian beam Ui into a rectangular top-hat distribution Uo(x,y;z) on the microdisplay panel at the specified distance z0. According to scalar diffraction theory, Uo (x,y;z) is the Fresnel diffraction pattern of Um(ζ,η), which is the complex field distribution at the exit of microlens array. Equation (1) expresses the mathematical representation of the relationship.
The design parameters which can modulate ψ(ζ,η) are the pitch and the curvature of the lenslet, and the lateral shift SL. Distance z0 can also be used as a parameter with limited range, which is the adjustable range of the distance between the beam shaper and microdisplay. On the other hand, the criteria for optimizing Uo(x,y;z) on the microdisplay side are the area of diffraction pattern which should match the active area of microdisplay and a collection efficiency above 80% in the active area. Those design parameters couple together and iteration process employing fast Fourier transform (FFT) algorithm is required before all the criteria can be met. Once the phase distribution function for the desired Uo(x,y;z) is found, the thickness distribution of the beam shaper can be obtained from Eq. (2).
Furthermore, in order to determine the texture of the beam shaper, T(ζ,η) is related to the parameters of the lenslet geometry as expressed in Eq. (3).
The design target is based on a three-panel spatial convergence color LCoS projector, where the beam shaper will be attached for evaluation, as shown in Fig. 1. The projector uses a 0.7-inch panel with an active area of 14.8mm × 8.3mm and pixel size of 7μm. The distance z0 is calculated to be 164mm considering the refractive index of all the glass material in the color separation unit. The design process shows difficulty to obtain a uniform diffractive pattern with rectangular shape lenslets, and the result came out with square lenslets having the radius of curvature of 325.8μm and the pitch of 140μm respectively. The lateral shift between two side microlens arrays became 90.5μm. Figure 4 shows the simulated field distribution at the distance z0 = 164mm (target) and 5m respectively. It shows that the dot pitch increases with the propagation distance, and the uniformity degrades in the mean time. The optimization process has ensured that the dot pitch is less than 10μm, roughly the size of the pixel, and there is no dramatic change of local uniformity in the whole pattern. The aspect ratio of the illumination pattern will be reshaped to match the microdisplay with an attached cylindrical lens.
The diffractive beam shaper was fabricated with the ultra-precision injection molding process and PMMA is used for the material, as shown in Fig. 5 . The molds shown in Fig. 6 were made with its textures grooved on Ni-coated layer by a 5-axis free-form ultra-precision diamond turning machine. The confocal microscopy inspection of the master mold is shown in Fig. 7 , which measures the exact feature size in two dimensions of the lenslet.
Table 1 lists the measured structure parameters of the lenslet on the master mold, and it shows that the deviation of pitch along x-axis and y-axis are smaller than 1μm. The accuracy of lateral shift is also crucial for the performance of the beam shaper and the tolerance was estimated to be 10μm. A dedicated fine-tuned mechanism was used in the injection chamber. Measurement shows that the deviation of lateral shift in x and y directions are 6μm and 6.2μm respectively, which lead to a misalignment of 8.6μm along the diagonal shift. Figure 8 shows the diffractive beam shaper, which has a dimension of 6mm × 6mm and weighs 18mg.
The illumination pattern from the diffractive beam shaper has been evaluated by shining red, green and blue lasers. Blue (473nm) and green (532nm) sources are diode-pumped solid- state (DPSS) lasers and the red (635nm) one is a diode array. The beam shape of the blue and the green lasers are both circular with a beam diameter of 2.5mm, whereas the red one features rectangular shape of 5 × 8mm2. The pattern is examined at the distance z0 = 25mm and z0 = 164mm. Table 2 shows the result and it indicates that the uniformity maintains quite well but the pitch of the dot array increases with a distance. Being a diffractive element performing strong diffraction effects, the most notable feature is no zero-order diffraction exists and the dispersion effect is negligible as shown in Fig. 9 . It indicates that the difference of FWHM among the red, green and blue illumination patterns is lower than 5%. The weak dispersion phenomenon can be considered as a minor factor affecting illumination efficiency in the projection system. The test result also implies that the diffractive beam shaper designed for z0 = 164mm can still be useful for smaller panel size with shorter illumination light path.
The optical performance of the double-sided MLA beam shaper is listed in Table 3 , including transmittance, collection efficiency, polarization ratio and uniformity.
3. Dot pattern elimination and laser speckle reduction with vibrating beam shaper
Being an intrinsic property of the diffractive beam shaper, the high contrast dot pattern in the illumination pattern has to be properly eliminated to acquire acceptable illumination quality. In addition, speckle pattern inherent to the coherent light source can also be seen on the screen superimposed with the dot pattern. The proposed scheme to resolve both issues is vibrating the beam shaper so as to temporally mix slightly different illumination patterns on the microdisplay. Both the dot pattern and laser speckle should be averaged out on the screen if the illumination pattern changes sufficiently within the effective averaging time period. The evaluation results from simulation and experiment on both the microdisplay plane and the projection screen are discussed in the following sections.
3.1 Illumination quality analysis on the microdisplay with the vibrating beam shaper
In order to generate multiple illumination patterns on the microdisplay plane to perform the temporal-averaging operation, a DC-current-controlled vibration motor has been used for vibrating the beam shaper. The current-dependent response of the motor was measured on amplitude and frequency (f), and the result is shown in Fig. 10 . Above input current of 30mA, the amplitude starts to become steady at 0.3mm c.a.
By changing the orientation of the vibrator, the attached beam shaper can be oscillated in longitudinal mode (along z-axis) or in transverse mode (perpendicular to z-axis). Effectiveness of eliminating dot pattern and speckle pattern in these two modes are evaluated and compared with each other. For evaluating the effectiveness of temporal integration based on simulation, the illumination pattern at various vibrating position of beam shaper is calculated with Eq. (1) and then superimposed together as the overall illumination pattern on the microdisplay. Figure 11(a) shows 30 simulated illumination patterns at the central area of microdisplay with transverse shifts of the beam shaper from 0 to ± 300μm at an interval of 20μm, and the illumination pattern from superimposing all 30 patterns is shown in Fig. 11(b), where the measured illumination pattern is also shown for comparison. The result from simulation and experiment match quite well, which indicates that the dot pattern is not sufficiently eliminated. The simulated illumination pattern with the transverse shifts of the beam shaper at 0, 160 and 300μm are highlighted in Fig. 12 , which shows that the illumination pattern does not change sufficiently even at the extreme position of the transverse shift. It in turns implies that the diffraction pattern is not sensitive to the transverse shift of the beam shaper.
Similarly, Fig. 13(a) shows 30 simulated illumination patterns at the central area of microdisplay with longitudinal shifts of the beam shaper from 0 to ± 300μm at an interval of 20μm, and the illumination pattern from superimposing all 30 patterns is shown in Fig. 13(b), where the measured intensity distribution is also shown for comparison. It indicates that the dot pattern has been effectively eliminated, and the effectiveness can be further illustrated with highlighting the illumination pattern with the longitudinal shifts of the beam shaper at 0, 160 and 300μm, shown in Fig. 14 . The pattern change has been sufficiently large to fill the gap between dot peaks within the vibration stroke.
3.2 Evaluation of speckle reduction on the projection screen
Vibration of beam shaper has shown its effectiveness for eliminating dot pattern, especially in the longitudinal mode, it is further evaluated for its performance of speckle reduction on the projection screen. The experiment setup is shown in Fig. 15 . The whole projection unit is a three-panel LCoS projector engine with LCoS panels replaced with mirrors attached with quarter wave plates for emulating full bright state. The original UHP lamp is then replaced with red, green and blue lasers as described in section 2. Those three lasers are all single mode type. The spectral widths of the blue and green lasers are both 2nm and that of the red one is 10nm. f-number of the projection lens and CCD camera used in the measurement conditions are 2.9 and 32 respectively. Throw distance is 2.5m and the projection screen is Lambertian type.
Figure 16 shows the variation of speckle contrast versus driving current of vibrating beam shaper in the longitudinal vibration mode, where the corresponding vibration amplitude to each driving current is also given. This speckle contrast is denoted as CL. It indicates that CL drops significantly with the increasing of driving current below 30mA, where vibration amplitude increases largely with the increasing driving current. There becomes no significant change of CL above 30mA where vibration amplitude becomes steady with the increase of driving current, although the vibrating frequency still keeps increasing.
For comparison, Fig. 17 shows the variation of speckle contrast versus driving current of vibrating beam shaper in the transverse vibration mode, and this speckle contrast is denoted as CT. It indicates that CT shows similar trend as CL in corresponding to the change of driving current. However, the effectiveness of the speckle reduction is much less than the case of the longitudinal vibration mode. The achievable lowest contrast is 22%.
The effectiveness of speckle reduction in the longitudinal vibration mode is further examined, and Fig. 18 shows the cross-sectional distribution at the central area of the projection pattern without vibration and with driving current of 60 mA, depicted with dash line and solid line respectively. It indicates that the peak distribution is considerably eliminated with the vibration.
Figure 19 shows the cross-sectional intensity distribution at the central area of the projection pattern with the longitudinal displacement of the beam shaper at 0, 160 and 300μm. It indicates that sufficient variation of the intensity distribution helps the reduction of speckle contrast.
Both simulation and experiment results manifest that the longitudinal displacement of the beam shaper helps more than the transverse displacement for generating sufficient amount of different illumination patterns, hence speckle patterns, in the temporal averaging process. This can be explained as follows. The diffractive beam shaper has a periodic structure profile and the transverse displacement does not change the diffraction pattern much at the target illumination plane. Whereas the longitudinal displacement of the beam shaper is equivalent to generating diffraction pattern at different propagation distance on the target illumination plane, which could have significant change especially for the case of near field diffraction. The more change of the diffraction pattern, the more number of the speckle patterns play into the temporal averaging process, and the lower the speckle contrast. Similarly, the larger the vibration amplitude, the more different diffraction patterns generated, and hence the more speckle patterns play into the temporal averaging process.
Based on the result of speckle contrast evaluation, the longitudinal vibration mode of the beam shaper with driving current of 60mA gives the lowest speckle contrast of 5.5%, and therefore is used as the modulation condition for prototyping and evaluating the three-panel LCoS projector as shown in Fig. 1. where the size of the LCoS panel is 0.7”. Due to an aspect ratio of 16:9 for the panel, a cylindrical lens with focal length of 19.5mm is put against the beam shaper so as to make the illumination pattern match the active area of the LCoS, as shown in Fig. 20(a) . The cylindrical lens will keep stationary when the beam shaper is vibrated. The schematic light path for the three-panel LCoS projector with laser sources and the vibrating beam shaper is shown in Fig. 20(b).
Table 4 shows the full red, green and blue pictures of the projector without and with vibration of the diffractive beam shaper. With vibration of the beam shaper, the speckle contrast can be reduced to 2.5% for blue, to 5.5% for green and to 2% for red.
Video has also been demonstrated on the projector engine, and Figs. 21(a) and 21(b) show a snapshot of an animation film without and with vibrating the beam shaper respectively. The improvement on uniformity of the projection image and speckle contrast with the vibration is obvious.
Based on the evaluation result of uniformity up to 78% ANSI uniformity and speckle contrast down to 5.5% on a three-panel spatial convergence color LCoS projector, the proposed vibrating diffractive beam shaper associated with a cylindrical lens has demonstrated its effectiveness as a compact solution for illumination optics of light valve projection displays with laser light sources. As the speckle reduction mechanism is embedded in the illumination path of the projector engine, its performance is independent to the projection screen and can be available for mobile devices. Due to the feature of light weight and compact size, the vibration of diffractive beam shaper can be easily modulated with different frequency, amplitude and mode for accommodating to different application conditions to achieve desired illumination quality and speckle contrast. In addition, the proposed scheme also provides the potential for integrating all the components, including lasers, optics and vibrating mechanism, on a micro bench and becoming a micro opto-electro-mechanical device for miniature projectors.
The research work is funded by National Science Council Taiwan under project number 982A14. The authors would like to thank to Cinetron Ltd. Inc for the support on prototyping the LCoS projector engine, and Instrument Technology Research Center (ITRC), Taiwan for fabricating the diffractive beam shaper. The help from Buena Vista Film Co. for acquiring the authorization to borrow the animation film picture from Walt Disney Studios Motion Pictures, US is also much appreciated.
References and links
3. J. W. Pan, C. M. Wang, H. C. Lan, W. S. Sun, and J. Y. Chang, “Homogenized LED-illumination using microlens arrays for a pocket-sized projector,” Opt. Express 15(17), 10483–10491 (2007). [CrossRef] [PubMed]
4. P. C. Chen, C. C. Chen, P. H. Yao, and C. H. Chen, “Double side lenslet array for illumination optics of laser projector,” Proc. SPIE 7232, 72320X, 72320X-9 (2009). [CrossRef]
5. S. Zhang, “A simple bi-convex refractive laser beam shaper,” J. Opt. A, Pure Appl. Opt. 9(10), 945–950 (2007). [CrossRef]
7. C. Dorrer and J. D. Zuegel, “Design and analysis of binary beam shapers using error diffusion,” J. Opt. Soc. Am. B 24(6), 1268–1275 (2007). [CrossRef]
8. R. M. Tasso, Sales, Geoffrey Gretton, G. Michael Morris, and Daniel H. Raguin, “Beam shaping and homogenization with random microlens arrays,” in Diffractive Optics and Micro-Optics, R. Magnusson, ed., Vol. 75 of OSA Trends in Optics and Photonics Series (Optical Society of America, 2002), paper DMA3.
9. J. W. Goodman, Speckle Phenomena in Optics: Theory and Applications (Roberts & Company, 2006).
10. J. W. Goodman, “Some fundamental properties of speckle,” J. Opt. Soc. Am. 66(11), 1145–1150 (1976). [CrossRef]
11. A. Furukawa, N. Ohse, Y. Sato, D. Imanishi, K. Wakabayashi, S. Ito, K. Tamamura, and S. Hirata, “Effective speckle reduction in laser projection displays,” Proc. SPIE 6911, 69110T, 69110T-7 (2008). [CrossRef]
12. G. M. J. Craggs, F. Riechert, Y. Meuret, H. Thienpont, U. Lemmer, and G. Verschaffelt, “Low-speckle laser projection using farfield nonmodal emission of a broad-area vertical-cavity surface-emitting laser,” Proc. SPIE 7720, 772020, 772020-8 (2010). [CrossRef]
14. T. Iwai and T. Asakura, “Speckle reduction in coherent information processing,” Proc. IEEE 84(5), 765–781 (1996). [CrossRef]
15. L. L. Wang, T. Tschudi, T. Halldórsson, and P. R. Pétursson, “Speckle reduction in laser projection systems by diffractive optical elements,” Appl. Opt. 37(10), 1770–1775 (1998). [CrossRef] [PubMed]
16. L. L. Wang, T. Tschudi, M. Boeddinghaus, A. Elbert, T. Halldorsson, and P. Petursson, “Speckle reduction in laser projections with ultrasonic waves,” Opt. Eng. 39(6), 1659–1664 (2000). [CrossRef]
17. M. N. Akram, Z. M. Tong, G. M. Ouyang, X. Y. Chen, and V. Kartashov, “Laser speckle reduction due to spatial and angular diversity introduced by fast scanning micromirror,” Appl. Opt. 49(17), 3297–3304 (2010). [CrossRef] [PubMed]
18. E. G. Rawson, A. B. Nafarrate, R. E. Norton, and J. W. Goodman, “Speckle-free rear-projection screen using two close screens in slow relative motion,” J. Opt. Soc. Am. 66(11), 1290–1294 (1976). [CrossRef]
19. G. Ouyang, Z. M. Tong, M. N. Akram, K. V. Wang, V. Kartashov, X. Yan, and X. Y. Chen, “Speckle reduction using a motionless diffractive optical element,” Opt. Lett. 35(17), 2852–2854 (2010). [CrossRef] [PubMed]