背景技术：
[0003]There are many situations where it is desired to create a light field that has a specified luminance profile. Light projection systems have a very wide range of applications from architectural lighting to the display of lifelike images. The projected light patterns can be dynamic (e.g. video), static (used for static images or static applications like the beams of typical car headlights projected through a lens onto the road, made by arbitrarily shaped optical surfaces, etc.). Light may be projected onto a wide range of screens and other surfaces which may be flat or curved. Such surfaces may be fully reflective (like a canvas used in a cinema, a wall or a building) or partially reflective (such as the windshield of a vehicle). Screens may be low-gain or high-gain, Lambertian or highly directional, high-contrast or lower in contrast. Light may be projected onto solid objects or onto a medium in a volume (such as fog).
[0004]Markets for and applications of light projectors include digital cinema, in-door and out-door advertising, medical imaging (both for display of images, as well as capture by a smart light source), large venue and live events or performances, automotive heads up displays, car head-lights and rear-lights, automotive entertainment and information displays, home-theatre, portable business projection, television and displays for consumer applications, military applications, aviation applications (like cockpit displays, smart landing-assistance, individual passenger entertainment displays), structured light sources for industrial applications, automotive headlights and other applications. Structured light may also be used for high precision applications, such as curing ink or other material for 2D or 3D printing, or steering light for laser micro-machining.
[0005]Various devices may be used to spatially modulate light. These may be called spatial light modulators (SLMs). Most SLMs provide a 2D array of independently and individually addressable pixels. Some examples of SLMs are reflective SLMs such as digital micro-mirror devices (DMDs), liquid crystal on silicon (LCoS) devices and transmissive SLMs such as LCD panels, transmissive LCD chips such as high-temperature polysilicon (HTPS) or low-temperature polysilicon (LTPS); and partially reflective/partially transmissive SLMs such as micro-electro-mechanical systems (MEMS) based systems in which some of incident light is transmitted and some of incident light is reflected. Most readily available spatial light modulation technologies are subtractive. These SLM technologies operate by absorbing or removing undesired light.
[0006]Other types of devices may controllably alter the nature and/or distribution of light using techniques that are not primarily subtractive. For example, the light redistributor may exploit interference of electro-magnetic waves (light), to modulate the distribution of light by controlling its phase characteristics and/or modulate the frequency of the light in order to change the apparent colour of light. Both of these examples show how light can be changed without converting energy from the light into wasted heat by absorbing the light.
[0007]Examples of dynamically-addressable focusing elements include: transmissive 2D arrays of controllable liquid crystal compartments with the property that the compartments can be controlled to selectively retard the phase of light, effectively causing a change in path-length. Devices that can controllably adjust the phase of light of different areas are called Phase Modulating Devices (PMD). PMDs may be transmissive or reflective. Some PMDs can individually control phase in a 2D array made up of a large number of pixels. A dynamically-addressable focusing element may also affect the polarization of light. Some devices may alter several light properties simultaneously.
[0008]Other types of dynamically-addressable focusing element comprise one or more scanning mirrors, such as a 2D or 3D microelectromechanical system (MEMS); and/or. one or more deformable lenses or mirrors or other optical elements. A dynamically-addressable focusing element may also or in the alternative comprise one or more optical switches.
[0009]Various sources can be used for illuminating SLMs, PMDs, imaging chips, or any other light re-distributing device, including arc lamps, light-emitting diodes (LEDs), LEDs plus phosphor, lasers, lasers plus phosphors. Each light source may emit light of different shapes, intensities and profiles. Traditional approaches to combining multiple light sources into a single higher-powered source include coupling light into optical fibres, and knife edge mirror beam combining, relaying into an integration rod, or some other optical averaging device.
[0010]However, in some cases, the useful characteristics of individual low-powered light sources are not preserved when combined using the traditional approaches, and higher-powered single-emitters are either not available, or have a prohibitively high cost per watt of light. For example, when light from multiple laser diodes is combined some of the characteristics affected are:[0011]Coherence: When coupling light from multiple discrete laser diodes or laser diode bars into a multi-mode fibre, or combining multiple laser beams into a single beam using a knife edge mirror array plus lens, coherence is lost.[0012]Polarization: The light at the output of a multi-mode fibre is no longer polarized, so some polarization recovery techniques must be used for applications that require polarized light.
There is a need for light sources and projectors that effectively combine light from multiple light sources. There is a particular need for cost-effective light sources and projectors in which light from multiple light sources can be manipulated to yield desired light patterns having desired optical characteristics.
具体实施方式：
[0053]Throughout the following description, specific details are set forth in order to provide a more thorough understanding of the invention. However, the invention may be practiced without these particulars. In other instances, well known elements have not been shown or described in detail to avoid unnecessarily obscuring the invention. Accordingly, the specification and drawings are to be regarded in an illustrative, rather than a restrictive sense.
[0054]Several novel approaches have been devised for tiling light from multiple sources in a parallel, collimated fashion. It is advantageous in some applications that the tiled light patches have minimal overlap with one another. In any of these approaches the light patches may be arrayed on a surface of a dynamically-addressable focusing element such as a phase modulator (PMD).
Adjustable Diode Array, Fixed Mirrors
[0055]A two-stage array of knife-edge mirrors can be used to tile a two dimensional array of lower-powered light sources, such as LEDs or laser diodes, to cover the active area of an imaging chip with discrete, non-overlapping patches of light. Each discrete laser diode (or other light emitter) is mounted in a holder with built-in X, Y and angular adjustment, and a lens for capturing and collimating the light. The holders each comprise a two-axis stage, and a holder with a tip/tilt adjustment in some embodiments.
[0056]This limits the compactness of the light source arrangement, so two arrays of knife edge mirrors are used, oriented 90 degrees to each other. FIGS. 1A and 1B show an example light source 100 which applies this type of approach. Light source 100 includes light emitters 102, lenses 103, knife-edge mirrors 104A and 104B. Mirrors 104A and 104B are arranged at 90 degrees to one another.
[0057]If space or design constraints force more complex geometry, two knife edge arrays can be combined in other orientations to achieve closely spaced parallel beams.
[0058]The knife edge mirrors serve to reduce the spacing of the individual beams, and clip the edges to minimize overlap and maximize coverage. Each beam may diverge slightly, and to different degrees along the long and short axis, so the mirror assembly is kept compact. The distance from the output of the mirror assembly to the imaging chip is kept as small as is practical.
Fixed Diode Array with Adjustable Mirrors
[0059]In another example embodiment, a 2-dimensional grid of light emitters (e.g. laser diodes) is fit into a fixed mount machined to tight manufacturing tolerances, with inset collimating lenses. Light beams emitted by the array of light emitters are directed at a two-stage array of knife-edge mirrors. In this embodiment the light emitters remain fixed and alignment is achieved by moving the mirrors. Once alignment has been achieved the mirrors may be permanently set, such that the output of the opto-mechanical system is a two-dimensional array of discrete, non-overlapping patches of light. This embodiment may otherwise be very similar or the same as the embodiment illustrated in FIGS. 1A and 1B.
[0060]This adjustment can be achieved by mounting each mirror on a pivot joint, which can be set with an adhesive once adjusted, or a flexible structure that can be shimmed into place.
[0061]In some embodiments knife edge mirrors are split into segments for each source, with individual tip-tilt control. This facilitates de-coupling individual sources from a row or column. Adjustable segments may be adjusted to apply any corrections to the direction or divergence of the individual beams.
[0062]Some embodiments provide adjustments to the positions and/or orientations of light emitters 102 as well as to the angles of mirrors 104.
Fibre-Couple Each Diode
[0063]FIGS. 2A and 2B illustrate another embodiment wherein the outputs of laser diodes 102 or other light emitters are guided by optical fibers 203 to create a desired array of light patches. For example, the light patches may be arrayed on a surface of a dynamically-addressable focusing element such as a phase modulator (PMD). In this design an array of light emitters 102 each has an associated lens 103 that captures the emitted light, and couples it into a single-mode optical fiber 203. Fibers 203 are bundled, and the output of the bundle is relayed onto the imaging chip (e.g. a PMD). This approach can be used to transform a source array into any shape, spacing, or configuration. A single-mode fibre will maintain polarization and coherence of laser light, but this same approach can be implemented with multi-mode optical fibers, with optimal diameter and geometry for increasing coupling efficiency without significant losses in coherence or excessive divergence.
“Christmas Tree” Mirror Mount
[0064]Instead of tiling patches in a rectangular pattern as shown, for example, in FIGS. 1A and 2A, a radial pattern of light patches can be achieved by using a mirror having a generally conical “Christmas tree” design. One potential configuration 300 is illustrated in FIGS. 3A to 3C. In this embodiment, light emitters 102 are mounted radially facing inward toward a Christmas tree mirror 304. In the illustrated embodiment, mirror 304 comprises a plurality of generally conical axially spaced-apart mirror surfaces 304A and 304B. The approach exemplified by FIGS. 3A to 3C can also be expanded, by machining lens curvature into the mirror surfaces 304A and 304B to collimate the beams, reduce the spacing, and aperture the beams in a single step.
[0065]A large source spacing may be used to improve alignment at the expense of the overall size of the opto-mechanical system.
[0066]If consistent polarization is required, angles may be taken into account to maintain consistent polarity in each beam as it hits the imaging chip or other destination.
Parabolic Mirror for Combining Radial Beams
[0067]FIGS. 3D, 3E and 3F show an arrangement 300A which is similar to arrangement 300 but uses a parabolic lens 304A to deflect beams from a radial arrangement of light emitters 102A, 102B into a parallel, closely spaced arrangement. In the illustrated embodiment, the emitters include light emitters 102A which direct beams at mirror 304A at a first angle to the axis of symmetry of mirror 304A and light emitters 102B which direct light beams at mirror 304A at a second angle to the axis of symmetry of mirror 304. This concept is similar to the “Christmas Tree” approach illustrated in FIGS. 3A to 3C, but without the aperturing effect of mirror edges.
De-Magnification of Over-Sized Beam Grid
[0068]The target illumination area, such as an imaging chip, SLM, or PMD, may be small compared to an array of sources. Any of the above approaches, including the knife-edge mirror approach described with reference to FIGS. 1A and 1B can be used to create a tiled pattern of non-overlapping patches of light (i.e. an array of parallel light beams) on a scale that is larger than the target area. In some embodiments, the scale of the patches of light is a factor of 2 or more times larger than the target area (e.g. the patches of light cover an area 4 or more times that of the target area in some embodiments). An optical system can be used to reduce the area of the resulting cluster of beams by demagnifying the light beams such that the array of patches of light is adjusted to the size required.
[0069]An example arrangement 400 is shown in FIGS. 4A and 4B. An optical system 401 (for example any of the systems described above) creates an array of beams 402. An optical system 404, in this example comprising lenses 404A and 404B de magnifies the array of beams.
[0070]Gaps may be maintained between the parallel beams, so divergence or distortion in the de-magnified light profile may be avoided.
Cascading Knife-Edge Mirror Stages
[0071]In some embodiments Individual modules each comprising a plurality of light emitters as described above can be produced in configurations which each yield an array of light patches (for example a 3×3 or 3×2 configuration). Beams output by two or more of such modules may be de-magnified as illustrated in FIGS. 4A and 4B, and tiled using an adjustable mirror technique, as described above (e.g. using arrays of tiltable mirrors arranged as shown in FIG. 1A).
[0072]FIG. 5 illustrates an example system 500 in which light output by three modules 401 is deflected by mirrors 502 to form an array 503 of patches of light. Light from any suitable number of modules 401 may be combined in this manner. Array 503 may combine arrays from modules 401 in a linear manner, as shown, or may combine arrays from modules 401 to yield an output array 503 that is larger in each of two dimensions than the arrays of patches of light from individual modules 401. For example, the arrays of patches of light from modules 401 may be arranged to make a composite array having plural rows and columns of arrays from individual modules 401.
Light Emitters
[0073]Any of a wide variety of light emitters may be used in the embodiments described above. for example, the light emitters may comprise lasers. Solid state lasers such as laser diodes are practical for a range of applications. Other examples of light emitters include solid-state light emitters such as light emitting diodes (LEDs); plasma light emitters; cold cathode light emitters; lamps, etc. In some embodiments the light emitters emit coherent light. In some embodiments the light emitters emit polarized light.
[0074]Light emitters may be provided in the form of discrete devices or may be packaged together in packages combining plural light emitters. For example, light emitters in embodiments as described above may be provided using systems comprising multiple light emitters, such as diode bars, with appropriate emitter count and spacing. Such embodiments may be advantageous for reducing the number of separately-mounted components in a light source.
[0075]Some light emitters may emit light in a form which is advantageously corrected to yield a beam with desired properties (e.g. a beam that is well collimated and directed in a desired direction). Custom optics may be provided for beam conditioning and correction for beam path in some embodiments.
[0076]FIG. 6 shows an example apparatus 600 comprising an edge-emitting diode array 602 which provides a plurality of individual light emitters 102C. Each light emitter 102C has a fast and slow axis. Apparatus 600 includes collimating optics 605 which includes a lens 605A for collimating in one axis and a plurality of lenses 605B for collimating in a second axis to yield a line of collimated output light beams. Two or more sets of apparatus 600 may be stacked to provide a two dimensional array of emitters.
[0077]Especially where an off-the-shelf diode bar is used for array 602, some conditioning and “smile correction” may be provided in the optical system in the case that the line of emitters has some curvature. An aperture, or reverse knife edging may be provided to increase the separation between beams. This can facilitate substantial elimination of overlap between adjacent beams.
Dealing with Path Length and Divergence
[0078]Most light emitters do not emit perfectly collimated beams of light. A beam of light from a light emitter will generally exhibit some divergence. It is desirable to reduce the effect of such divergence. In cases where divergence of the beams from different emitters can be substantially eliminated, an output array of patches of light may have the patches spaced very close to one another without any significant overlap between the patches. Some light emitters emit light that diverges differently in different directions. A direction in which the divergence is large may be called the fast axis. A direction in which divergence is smaller may be called a slow axis. Where a light emitter has a fast and a slow axis, a single symmetrical lens can approximately collimate a beam from the light emitter along the fast or slow axis, but the beam will continue to diverge in the other axis.
Pairs of Mirrors to Fix Path Length
[0079]Some embodiments equalize path length from the light emitters to the corresponding patches at the target area. Making path lengths equal for all beams is advantageous at least in part because beam divergence can differ for different path lengths. Where the path lengths are equal, divergence of all of the beams may be approximately equal.
[0080]Mirrors can be used to fold the light path to equalize the path length across all beams, as shown in apparatus 700 of FIG. 7. Using multiple mirror stages, the path length for each beam can be identical without complex geometry. Apparatus 7 includes light emitters 102 that emit light beams 103-1, 103-2 and 103-3. Each beam interacts with a pair of mirrors that fold its path. the mirrors are spaced so that the path lengths from each light emitter 102 to output beams 703 are equal. Apparatus 700 includes mirror pair 701A-1 and 701 B-1 which acts on beam 103-1; mirror pair 701A-2 and 701 B-2 which acts on beam 103-2; and mirror pair 701A-3 and 701B-3 which acts on beam 103-3.
Asymmetrical Lenses
[0081]Light profiles of source light beams may not be radially symmetrical, either in terms of shape, or rate of divergence. For example, the fast and slow axis of a laser diode have different rates of divergence. This can be corrected by introducing a lens that is not circularly symmetrical in the beam direction (e.g. a cylindrical lens) in the light path.
[0082]Some embodiments provide an array of cylindrical lenses to correct the divergence along the slow axis of beams from a plurality of light emitters such as laser diodes. This approach is well-suited for the case where diode bars or a diode bar stack provide the light emitters. This approach may also be applied to an array of discrete diodes or other light emitters. An example is illustrated in FIG. 6.
[0083]Various possible optical arrangements to correct divergence of beams having a fast and a slow axis include using pairs of cylindrical lenses for each axis, using a single shared spherical lens and a cylindrical lens for each axis and so on.
Long Geometry
[0084]As described above, the effects of divergence can be mitigated by designing the source array such that distance from the source array to the mirrors which guide the light into parallel beams (e.g. knife edge mirrors as shown in FIG. 1A) is large, relative to the distance between the knife-edge mirrors and the target area (e.g. a dynamically addressable focusing element or other imaging chip). In some embodiments the distance from the light emitters to the mirrors is at least 3, 5, 10 or 18 times larger than the distance from the mirrors to the imaging chip.
[0085]With this approach, increasing the distance from the light emitters to the mirrors makes any relative differences in path lengths for different beams smaller. The mirrors may be close enough to the imaging chip that the amount of divergence between the mirrors and the imaging chip is small, such that no undesirable overlap occurs at the imaging chip.
Desired Light Source Characteristics for Freespace Laser Projector Using Phase and Amplitude Modulation
[0086]In a traditional digital projector, it is important that the amplitude SLM (DLP, LCD, LCoS) which is imaged onto the projection screen via the projection lens is illuminated uniformly.
[0087]In some embodiments a light source illuminates a dynamically-addressable focusing element non-uniformly. In such embodiments dynamically-addressable focusing element (e.g. a phase modulator) may be controlled to provide structured illumination (which varies from location to location in a known way) on an amplitude SLM. While it remains beneficial to illuminate the phase modulator uniformly (even heat dissipation, uniform light profile on SLM when a flat phase is addressed or in case the phase SLM fails), intensity variation across the phase SLM can be accounted for and the lensing pattern can be adjusted to ‘correct’ for it (for example to provide uniform illumination on the amplitude SLM when desired).
Alignment Example
[0088]A light source for a particular application desirably achieves specifications required by the application for beam quality and stability in a package that is simple and compact. Ideally the light source may be installed as a single module which can be aligned at time of manufacture using adjustments that facilitate rapid accurate alignment (e.g. orthogonal adjustments such that adjustment of one beam property does not change other beam properties).
[0089]In some embodiments alignment is performed in a ‘bottom-up’ approach in which light beams from individual light emitters are centered and collimated, the light emitters are assembled into banks, the alignment of the beams is adjusted, and then beam shrinking optics are adjusted to deliver the output light to a desired target area (in some embodiments the target area is a few mm per side, for example 12×7 mm). After each alignment step the adjustment may be fixed using a settable material such as a suitable epoxy, glue, solder or the like.
[0090]Table I provides three sets of example design specifications. Some of these specifications are achieved in some embodiments. One or more of these sets of specifications are achieved in some embodiments.
TABLE IExample Design SpecificationsGoodBetterBestPriorityFeaturespecsspecsspecsUnitComments/Metrics1Z-Parallelism5002100.5arc secFast axis, measureon optical axisat .1 m, .5 m and 5 mdistance 6.4 umresolution, 300 mmaway2XY1520.5DegreesParallelism3XY-Shifting50100.1px(tiling)(LETO)4Divergencemrad(fast axis)Divergencemrad(slow axis)5Fill-factor on409599% LETOFill factor is aLETOareafunction of beamsize and diodecount. Need tomodify one or theother6Intensity407595%uniformity onrelativeLETOto peak7MeasurementabsentPresent,absentportremovable8Beamunknownquantifiedconstrained%Meet 1, 2, 3 after 1 h,stability overvariation6 h, 24 htime9BeamunknownquantifiedconstrainedMeet 1, 2, 3 afterstability over20 C., 40 C., 60 C.temperature10Beamunknownquantifiedconstrained(Profile, Centrestability overLocation, Intensity)powerConsistency &predictability overvarious modulationapproaches, PWM,current control, etc.11Throughput607590%
Example Light Source Block
[0091]In this example, eight laser diodes, each with a corresponding collimating lens, are positioned in an array, with 10 mm separation distance. Each diode is pressed into a copper block with integrated cooling fins, and attachment features. Eight lenses are mounted in a fixed block with 10 mm spacing. A jig holds the fixed lens array stationary relative to an alignment pattern some arbitrarily large distance away, in the far field. This alignment pattern includes reference lines indicating desired beam positions, with 10 mm spacing.
[0092]A single diode block is held in a three-, four-, five-, or six-axis positioning stage, and positioned such that the emitter is centered with respect to the corresponding lens, the output beam is collimated (neither diverging or converging in the far-field), oriented so that the polarization is consistent with the light source specification, and directed such that the beam is coincident with the corresponding position indicated on the alignment pattern.
[0093]The z-axis position of the diode emitter (parallel to the beam direction) controls the divergence of the beam. The x- and y-axis position of the diode emitter control the x- and y-location of the laser spot on the alignment pattern. Rotation about the z-axis controls the polarization orientation. The x- and y-axis position of the diode emitter can be adjusted to correct for distortions in the beam shape.
[0094]When the diode is suitably positioned, it is fixed to the lens block. This mechanical connection can be achieved in a number of ways such as:[0095]1. The diode block has tabs that are soldered to pads on the lens block[0096]2. The diode block remains in the jig, and an adhesive such as epoxy or a suitable UV-curing adhesive or a suitable thermal curing adhesive is applied to fix the diode position,[0097]3. The diode block is spot welded to lens block,[0098]4. The diode block is initially aligned coarsely and is precisely deformed in the jig for fine alignment.
Techniques for Improving Alignment Precision
[0099]With a printed alignment mask on the example system described, collimation and parallelism accuracy is limited to on the order of +/−1 mm at a distance of 3 m, or 0.015 degrees.
[0100]For improved accuracy, more advanced techniques can be implemented. Some examples are:[0101]1. A diffraction grating can be positioned in the light path, to produce larger alignment patterns with diffractive imaging, and magnified for increased measurement accuracy.[0102]The collimation can be adjusted to achieve the optimal point spread function to a much higher precision.[0103]The parallel beam alignment precision can be improved by registering two alignment patterns rather than attempting to centre an amorphous beam dot on an alignment grid visually.[0104]2. A dynamic diffractive optical element, such as a phase-only Spatial Light Modulator, can also be used to improve alignment precision, by dynamically changing the diffractive alignment pattern for a multi-step alignment approach.[0105]Sets of alignment patterns can be generated, starting with coarse patterns, and moving to progressively finer alignment.[0106]Different patterns may be better suited to achieving different aspects of alignment. For example, horizontal lines for XZ-plane alignment, vertical lines for YZ-plane alignment, a suitable horizontally and vertically symmetrical pattern for collimation adjustment, or for optimizing the beam angle about the optical axis etc.
Automation of Light Source Alignment
[0107]The alignment process can be automated, diode-by-diode, using computer-controlled 4-, 5-, or 6-axis alignment stages, and either a machine vision camera directed at a screen, or by relaying the output beams onto an optical sensor, such as a CCD, or CMOS. The following is an example algorithm that may be applied for automated or semi-automated alignment. The algorithm begins with a fixed block of lenses, mounted to an alignment jig. For the duration of the alignment procedure, the lens array may remain fixed relative to the all other elements, excluding the diodes. The alignment jig holds the lens array block, pointed at a dynamic diffractive optical element (e.g. a phase modulator). Light output from the phase modulator is resized using standard optics, and relayed onto an optical sensor, or projected onto a screen and captured by a machine vision camera. The diffractive optical element and screen or optical sensor are placed at a distance that is very large compared to the focal length of the lenses in the fixed array.[0108]1. Begin[0109]2. Position the light emitter (e.g. a laser diode) in approximate alignment with the corresponding lens and clamp the laser emitter in a jig, providing 4-, 5-, or 6-axis micro-positioning. The jig may comprise a stage, and a holding device.[0110]3. Proceed with the automated alignment procedure:[0111]a. Apply a flat phase pattern to the dynamic diffractive optical element, and adjust the focus by moving the light emitter so that the beam neither converges nor diverges. This can be achieved by sampling the beam profile at several distances along the optical axis and adjusting the distance between the laser diode and the lens until all samples are the same width. This can also be achieved using a beam splitter and a phase sensor, and adjusting the position of the laser emitter until the beam profile is maximally flat.[0112]b. Insert a polarizer into the light path (if the dynamic diffractive optical element is not polarized, or if it is not polarized in the desired direction). Adjust the angle of the light emitter about the optical axis, holding all other adjustments constant, until the beam reaches maximum brightness.[0113]c. Apply alignment patterns to the dynamic diffractive element and adjust the position of the light emitter position until the patterns are registered optimally. This process can be repeated for various aspects of alignment, including XZ-plane parallelism, YZ-plane parallelism, rotations about the X or Y axes.[0114]d. This alignment procedure can be repeated in multiple steps, from coarse to fine alignment[0115]4. When sufficiently precise alignment is achieved for the light emitter-lens pair, fix the light emitter to the lens block using one of the methods described above.[0116]5. Repeat the above steps for each additional light emitter adjusting for beam characteristics as above and also ensuring that the beam is parallel to the beams of previously-aligned light emitters added to the block.
Combined System
[0117]Once a block of light-emitter-lens pairs has been aligned to yield collimated, parallel beams, with identical polarization orientation the block may be combined with other elements which function to create a more compact beam array, and to shape and resize the combined light profile to cover an imaging chip. Such a system might comprise:[0118]a mount to hold the array of diode-lens pairs;[0119]an array of knife-edged mirrors arranged to decrease the spacing between adjacent beams;[0120]one or more lenses or mirrors for expanding or contracting the beam to suit the desired application;[0121]mirrors for folding the light path to achieve a compact footprint and/or to equalize path lengths or different beams;[0122]cooling for the heat-generating elements (e.g. suitable heat sinks and/or active coolers such as Peltier elements); and[0123]control electronics for the light emitters.
[0124]Some embodiments also include monitoring sensors such as temperature sensors attached to measure operating temperatures of the light emitters and/or other elements and/or measurement ports at which beam profiles may be evaluated.
[0125]In an example embodiment the light emitters comprise 500 mW lase diodes such as the model ML501P73 laser diodes available from Mitsubishi electric. these laser diodes output light at 638 nm. An example display includes 6 to 20 such laser diodes.
Control Electronics
[0126]It is not always desired to have light emitters running at full brightness at all times due to the fact that some images do not contain much light. It is possible to steer unneeded light into a dump area (e.g. through suitable control of a dynamically-addressable focusing element), but it would be more ideal to reduce the light output of the light emitters and reduce energy consumption and heat output. Reducing the output of the light emitters for darker images may also improve black level by reducing scattered light.
[0127]Laser diodes can be dimmed by reducing the amount of current passing through them and or turning them off and on at a sufficiently rapid speed to not be noticeable by a human observer—known as pulse width modulation (PWM). It is more difficult to achieve precise intensity control by controlling the current than by PWM.
[0128]When using PWM control a duty cycle (% of the time that a light emitter is ON) can be thought of as controlling the output light intensity. For example one way to implement 8 bit control over light intensity is to clock a counter at 256xx the PWM frequency such that an output is held in a state corresponding to the light emitter being ON until the counter value reaches the 8 bit intensity value. The output would be in a state corresponding to the light emitter being OFF at other times during each PWM cycle.
[0129]In some projectors Digital Light Processing (DLP) devices are used to create the final image. It is desirable to provide a light source as described herein which is compatible with downstream DLP devices. In a DLP device a binary modulator flips a micro mirror back and forth between an “on” state where it sends light to the screen and an “off” state where it sends light to a “dump” area. Each pixel has a corresponding micro mirror. The DLP creates greyscale by flipping the micro-mirror back and forth rapidly. The micro mirror is controlled to spend more time in the “on” state to make a brighter pixel or more time in the “off” state to make the pixel dimmer.
[0130]In an example DLP driving scheme, each pixel has an 8 bit (or more) greyscale drive value per frame of video (usually 60 fps), these are translated into 8 mirror flip periods, with one period for each bit. The period corresponding to the least significant bit is short. The period doubles for each bit and is longest for the most significant bit.
[0131]Whether a bit is set to 0 or a 1 determines whether the mirror is flipped to the “on” or the “off” position for the corresponding period. FIG. 8 shows that the shortest period that the mirror may be in the “on” or “off” state is for the lowest-order bit (b0). This shortest period may be called the “flip period”.
Asynchronous Light Pulses
[0132]If a pulsed light source is used (for example to produce light at 50% of the maximum level), flickering will occur if the “off” and “on” pulses of the light emitter are asynchronous to the mirror flipping and the periods of “off” and “on” significantly differ from frame to frame on a static image due to a low pulse frequency for the light state.
Slow Asynchronous Light Pulses
[0133]This is illustrated in FIGS. 9A and 9B. In frame 1 of FIG. 9A the viewer perceives two light pulses during the time the DMD transmits light. In frame 2 of FIG. 9B the viewer perceives three light pulses during the same DMD open period. This 50% change in light intensity is due to the light pulses being asynchronous to the mirror flips.
Fast Asynchronous Light Pulses
[0134]If the “off” and “on” light source periods are short relative to the “mirror flip” period, the difference between “off” and “on” periods between static frames should be drastically reduced and be imperceptible to the human eye. For example, FIGS. 10A and 10B show an example in which the light emitter is modulated significantly faster than the DLP flip period.
[0135]In FIGS. 10A and 10B only a single minimum width mirror flip is shown depicting a drive value of 1. The viewer perceives a light intensity corresponding to 27/54 in FIG. 10A and to 28/54 in FIG. 10B. A disadvantage of this solution is that a large amount of electromagnetic interference (EMI) can be produced by switching powerful lasers off and on very quickly. Also, tighter timing tolerances are required by the circuitry to minimize duty cycle distortion between lasers.
Synchronous Light Pulses
[0136]If the light emitter “off” and “on” periods are synchronous to mirror flips, there should be practically no difference between static frames and the light source pulse generator need only run at the period of the mirror flips, drastically reducing generated EMI and allowing for slacker timing considerations. FIG. 11 illustrates an example embodiment in which light output from a light emitter is synchronized to DLP flip cycles.
[0137]When a new frame arrives, the mirror flip logic for all pixels can be updated simultaneously via a double buffering scheme (or in blocks from top to bottom if desired).
Synchronizing the Light Source to the DLP
[0138]Some DLP driver chips provide a “trigger out” pin that indicates the start of a mirror flip cycle. In absence of this an independent “mirror timing recovery” circuit can be constructed. When the mirror is in the “off” state it sends light to a “dump” area. Placing a photoreceptor in the dump area will send a voltage back to the circuit when a mirror flips to the “off” state. During a “training mode” the lasers are constantly on and the DLP sends only the least significant bit to the dump area (i.e. drive level 254 for a DLP with 8 bit control). Using a high speed reference clock and counters the period of the shortest mirror flip can be determined and the timing of subsequent mirror flips can be predicted. Similar methods are employed in telecom applications for clock and data recovery from a single wire serial data stream. A jitter attenuator may be provided depending on the amount of error in the recovery system.
[0139]With the recovered mirror flip period, the light source can synchronize PWM for the light emitters to the mirror flip periods such that a deterministic light intensity can be produced for the shortest mirror flip period (and all longer periods). During longer mirror flip periods the PWM cycle could simply repeat (twice for bit 1, four times for bit 2, 8 times for bit 3 etc.), or to further reduce EMI the PWM