Light source module

Abstract

A light source module 50 includes multiple semiconductor sources 110, 120, 130, each emitting a beam at a different peak emission wavelength. The semiconductor sources have elliptical outputs beams with a more divergent, fast axis and a less divergent, slow axis. The fast and slow axes respectively correspond to vertical and horizontal directions. The module also includes beam circularising elements 511, 521, 531 which transform the beams from the semiconductor sources into beams that are circular defined by having an ellipticity greater than 87% or near circular defined by having an ellipticity greater than 70%. The circularised beams are then combined by a beam combiner 240 including one or more dichroic mirrors 122,132. The beam combiner 240 brings the individual beams into a common optical axis for output from the light source module. The beam circularising elements may have refractive power in the fast and slow axes. Alternatively, the beam circularising elements may have refractive power only in the fast axis and act to reduce divergence along the fast axis. The beam circularising elements may have refractive power only in the slow axis and act to increase divergence along the slow axis.

Description

BACKGROUND OF THE INVENTION The invention relates a light source module containing multiple light sources whose output beams are combined into a common output beam path. Light source modules which incorporate multiple light sources and combine their respective outputs into a common beam path are important components for a variety of applications, such as for RGB (Red, Green, Blue) projection systems. RGB light source modules may be used in virtual reality (VR) and augmented reality (AR) devices, such as glasses, goggles or visors, and head-up displays (HUDs). Light source modules are also used in the medical field for optical coherence tomography (OCT). The light sources used in light source modules are most usually semiconductor emitters such as laser diodes (LDs) and superluminescent diodes (SLDs). A given module may also combine LDs and SLDs, e.g. two LDs for green and blue and one SLD for red. The size of the light source module is an important consideration for many applications, e.g. for AR glasses. In turn, the size is in large part dictated by the components that are required to combine the beams, this being a 3-to-1 combination in the case of an RGB light source module. One known design approach for reducing the size of an RGB light source module is to perform the beam combination with a Photonic Integrated Circuit (PIC) chip. Despite the attractiveness of PIC-based designs for size reduction, free-space optics are in many respects easier to work with, so another strand of the prior art is to make free-space beam combiner designs as small as possible. Figure 1 show schematically a prior art light source module with a PIC-based beam combiner that follows the design approaches disclosed in each of: • Toshikazu Hashimoto and Junji Sakamoto "Visible-light Planar Lightwave Circuit Technology and Integrated Laser-light-source Module for Smart Glasses" NTT Technical Review Vol. 19 No. 3, pages 31-36, March 2021, and • Joe Kamei, "SEIREN KST: Compact Full Color Optical Engine for Smart Glasses" Proc. SPIE 11764, SPIEAVR21 Industry Talks II, 117641E (6 April 2021); https: / / doi.Org / 10.1117 / 12.2598230 The light source module 50 comprises red, green and blue emitters 110, 120 and 130 that are mounted on a common submount 140. The light output from each emitter is coupled into a PIC chip 200. The PIC chip 200 contains first to third input waveguides 210, 220 and 230. Light from the red emitter 110 is coupled into one end of the first input waveguide 210. Light from the green emitter 120 is coupled into one end of the second input waveguide 220. Light from the blue emitter 130 is coupled into one end of the third input waveguide 230. The other ends of the three input waveguides 110, 120 and 130 lead to a beam combiner 240 that is also integrated in the PIC chip 200. The beam combiner 240 is a wavelength multiplexer that combines the red, green and blue light from the three input waveguides 210, 220 and 230 into one end of a common output waveguide 250. The other end of the output waveguide terminates on the periphery of the PIC chip 200 and outputs RGB beams along a common optical axis, O. Two specific designs of 3-to-1 wavelength multiplexer combiners are disclosed in the above-referenced Hashimoto et al and Kamei publications. The RGB light beams emitted from the output waveguide 250 are transmitted through a collimating lens 260. The collimated RGB beams output from the collimating lens 260 are then transmitted out of the light source module along a common optical axis, O, through an optical window 150 that is integrated into an end wall of the light source module housing 100. One of the technical challenges with the waveguide device is the realization of the integrated beam combiner that would combine multiple input waveguides, each carrying a different colour or optical wavelength, into a common output waveguide with minimum excess losses. Another challenge is related to the chromatic dispersion of the refractive index of the optical waveguides, which results in different mode sizes for the different colours, especially in the common output waveguide 250, and thus in different emission angles, expressed as different far field angles or different numerical aperture (NA) values. Another challenge is related to the collimation lens at the output waveguide, more specifically to the chromatic dispersion of the lens material. In most cases, the chromatic dispersion of the lens is not zero, which means that the lens is not achromatic. Therefore, the collimation performance of the lens is different for the individual colours. For optimum collimation, the emitting spot of the output waveguide is positioned in the focal plane of the collimation lens. For a regular chromatic lens, the focal length changes with colour, which means that the single collimation lens may perform sufficient or full beam collimation for the centre colour, for example green, while it would underperform for the higher colour, for example red, resulting in divergent beam output, and overperform for the lower colour, for example blue, resulting in convergent beam output. Figure 2 shows a prior art light source module based on free-space beam combining optics as disclosed for three LD sources in: • N. Primerov, J. Ojeda, S. Gloor, N. Matuschek, M. Rossetti, A. Castiglia, M. Malinverni, M. Duelk, C. Velez "Ultracompact RGB laser diode module for near-to-eye displays" Proceedings Volume 11788, Digital Optical Technologies 2021; 117880Q (2021) https: / / doi.org / 10.1117 / 12.2594243 Essentially the same design is used for SLD sources in: • US 11,131,795 assigned to Exalos AG (see e.g. FIG. 6 therein). The light source module 50 comprises red, green and blue emitters 110, 120 and 130 that are mounted on a common submount 140. The divergent light beams output by the three emitters 110, 120 and 130 are collimated by respective collimation lenses 111, 121 and 131. The emitters are all arranged on a common submount 140 to emit their beams in parallel (upwards in the drawing). A plane mirror 112 reflects the collimated light beam from the first emitter 110 through 90 degrees so that it enters the beam combiner 240 at 90 degrees to the other collimated beams. The three collimated beams are then spectrally and spatially combined along a common optical axis, O, by a beam combiner 240 constituted by two dichroic filters 122 and 132 before exiting the module housing 100 through an optical window 150. In this free-space architecture, each collimation lens (CL) can be positioned at a different distance relative to the emitting spot of the light source, therefore allowing individual beam collimation despite each emitter emitting a different colour and despite each lens having chromatic dispersion and, therefore, having a different collimation performance and a different focal length for each emitted colour. If all the collimation lenses are of the same type, then the differences in the angular emission profiles of the light sources and the chromatic dispersion of the lens can be compensated to some extent by the individually adjusted working or focusing distance. A 4-emitter version of the same design as shown in Figure 2 is disclosed in: • US2019179151A assigned to Google Inc (see FIG. 12B therein) where there are not only R, G and B emitters but also a near-infrared (NIR) emitter. Generally, for each extra emitter the beam combiner requires one more dichroic filter, so this design of beam combiner can be used to combine 3, 4, 5 or more beams. The following patent publications • US2007252918A1 assigned to Panasonic Corporation, and • US2020174356A1 assigned to Sharp KK disclose similar optical arrangements as shown in Figure 2 for RGB display projectors, in which each of a plurality of emitters has its own collimating lens arranged prior to a beam combiner. The following patent publications • US2010226405A1 assigned to IPG Photonics Corporation, and • WO2020198406A1 assigned to IPG Photonics Corporation disclose variants of the optical arrangements shown in Figure 2, the difference being that instead of a single collimating lens being arranged after each emitter prior to the beam combiner, a lens pair is used, each lens pair consisting of fast axis collimation (FAC) lens and a slow axis collimation (SAC) lens. Figure 3 shows a prior art light source module based on free-space beam combining optics for 4-emitters also as disclosed in: • US2019179151A assigned to Google Inc (see FIG. 12C therein) The light source module 50 comprises red, green and blue emitters 110, 120 and 130 and a near-infrared (NIR) emitter 135 that are mounted on a common submount 140. The divergent uncollimated beams from the four emitters 110, 120, 130 and 135 are combined by a beam combiner 240 constituted by three dichroic filters 122, 132 and 142. The first emitter's beam is also deflected through 90 degrees by a plane mirror 112 prior to the beam combiner so it enters the beam combiner at 90 degrees to the other emitter beams. The beams output from the beam combiner 240 then have a common optical axis, O, and are collimated by a single collimation lens 260. We note that Figure 3 reproduces FIG. 12C of US2019179151A and the description of the module's operation is also as described in US2019179151A. Although the rays reflected from the plane mirror 112 and dichroic filters 122, 132 and 142 are depicted as being collimated, these rays must be divergent until they are collimated by the collimation lens 260. US2019179151A also proposes modifying the design of Figure 2 by adding a further collimation lens for the combined beams at the position of lens 260 in Figure 3, the further collimation lens 260 being said to have the function of fully collimating and / or correcting each of the combined beams. Figure 4 shows a prior art light source module based on tree-space beam combining optics as disclosed for three LD sources in: • Jorg Reitterer, Zhe Chen, Anna Balbekova, Gerhard Schmid, Gregor Schestak, Faraj Nassar, Manuel Dorfmeister, Matthias Ley, "Ultra-compact micro-electro-mechanical laser beam scanner for augmented reality applications," Proc. SPIE 11765, Optical Architectures for Displays and Sensing in Augmented, Virtual, and Mixed Reality (AR, VR, MR) II, 1176504 (27 March 2021); doi: 10.1117 / 12.2576704, and • EP2977808A1 assigned to Trilite Technologies GmbH The light source module 50 comprises red, green and blue emitters 110, 120 and 130 that are, as shown, mounted on a common submount 140. The three emitters 110, 120 and 130 are LDs that have their output beams collimated in their fast axes by a common cylindrical lens 270 to achieve fast axis collimation (FAC) and then the three beams that are collimated in their fast axes only are received by a monolithic block of three cylindrical lenses 280 that collimate the slow axes of the beams to achieve slow axis collimation (SAC) and thus output beams that are collimated in both their fast and slow axes. Performing beam collimation in this way with a crossed pair of cylindrical lenses is suitable for collimating an elliptical crosssection beam, as is emitted from an edge-emitting LD or SLD, where the elliptical crosssection arises from diffraction at the output facet of the semiconductor chip. (On the other hand, a vertical cavity surface emitting laser (VCSEL) will emit a circular cross-section beam so a single spherical lens is suitable for beam collimation.) With an elliptical cross-section divergent beam, the different beam divergences in the two axes of the ellipse can therefore be independently corrected. Moreover, as done in the design of Figure 4, it is sensible first to collimate the more strongly divergent, fast axis and then the less strongly divergent, slow axis. The reverse order is less preferred, since the collimation lens for the strong axis would need to be larger. In the design of Figure 4, there is no beam combiner as such in the light source module - rather the three output beams (RGB) are directed onto a common point at the centre of a tiltable MEMS mirror 290 which can scan any one of the incident RGB beams in the desired direction to generate an image through rapid time slicing, i.e. writing each of the R, G and B beams onto the display surface sequentially with a repetition rate faster than the persistence time of the eye. In other words, the MEMS mirror enables each of the R, G and B beams to be output in the same output beam path but not at the same time; rather by positioning the MEMS mirror at different angles to take account of the different arrival angles of the R, G and B beams from the SAC lens(es) 280. Figure 5 shows a modified version of the design of Figure 4 as disclosed in: • EP3637171A1 assigned to Trilite Technologies GmbH (see e.g. Figs 9a & 9b therein) Compared to the design of Figure 4, each of the FAC and SAC lenses is provided as an individual lens. This is done so that the distance from light source to FAC lens and light source to SAC lens can be set individually for each beam. This individual positioning of the 5 FAC and SAC lenses takes account of the wavelength dependency of refractive index of the lens material, i.e. the effect that a given lens will have different focusing powers for each of the three emission wavelengths in the red, green and blue - or more precisely emission wavelength ranges in the red, green and blue. During manufacture of the light source module 50, the lenses are individually positioned in a calibration process to ensure that each beam is 10 precisely focused along both fast and slow axes on the MEMS mirror 290. Further beam combining light source modules are disclosed in: • US2022013975A1 assigned to Robert Bosch GmbH • WO21044409A1 assigned to Lumus Ltd • LIS2013215923 assigned to Corning Inc SUMMARY OF THE INVENTION According to an aspect of the invention there is provided a light source module comprising: a plurality of semiconductor sources, each configured to emit a beam at a different peak emission wavelength, wherein a property of the semiconductor sources is that their outputs beams are elliptical, each output beam being divergent with a more divergent, fast axis and a less divergent, slow axis, the fast and slow axes respectively corresponding to vertical and horizontal directions; beam circularizing elements arranged to transform respective ones of the beams from the semiconductor sources into beams that are circular as defined by having an ellipticity greater than 87% or near-circular as defined by having an ellipticity greater than 70%; and a beam combiner comprising one or more dichroic mirrors arranged to receive the beams from the beam-circularizing elements and to bring the beams into a common optical axis for output from the light source module. This design is particularly applicable to edge-emitting semiconductor sources whose output beams are inherently elliptical. Circularity of a beam can be defined in terms of ellipticity. According to the standard of the International Organization for Standardization (ISO): • ISO 11146-1:2021 "Lasers and laser-related equipment - Test methods for laser beam widths, divergence angles and beam propagation ratios - Part 1: Stigmatic and simple astigmatic beams" a circular power density distribution is a power density distribution having an ellipticity greater than 0.87 (see definition 3.7). Under this standard, a beam is defined as circular if it has an ellipticity of greater than 87%. We adopt that definition in this document. In addition, for this document, we define a near-circular beam as a beam with an ellipticity of greater than 70%. Beam ellipticity is defined in this document also according to ISO 11146-1:2021, namely beam ellipticity is the ratio of the smaller one of the divergence angles in the horizontal and vertical directions versus the larger one of the divergence angles in the horizontal and vertical directions. In some embodiments, each of the beam-circularizing elements has refractive power in the fast and slow axes. For example, such a dual-axis beam-circularizing element may be realized as lenses with at least one surface that is spheric or aspheric with a first curvature in one axis and that is spheric or aspheric with a second curvature in the other axis. Another example is to form such a dual-axis beam-circularizing element as a free-form lens, which can be manufactured with an arbitrary amount of diffractive power in each of the fast and slow axes. In other embodiments, each of the beam-circularizing elements has refractive power only in the fast axis, each beam-circularizing element acting to reduce divergence along the fast axis. For example, the beam circularizing elements may be fast-axis reduction (FAR) lenses. In still further embodiments, each of the beam circularizing elements has refractive power only in the slow axis, each beam-circularizing element acting to increase divergence along the slow axis. For example, the beam-circularizing elements may be slow-axis expansion (SAE) lenses. For example, a FAR or SAE lens may be realized with a lens having at least one surface that is flat in one axis and that is curved to be spheric or aspheric in another axis (e.g. a cylindrical lens). In some embodiments, the beam circularizing elements are ellipsoid-shaped ball lenses, with the ellipsoid having a smaller diameter in the vertical, fast-axis direction and a larger diameter in the horizontal, slow-axis direction. An ellipsoid-shaped ball lens is can modify the beam divergence independently in the horizontal and vertical axes as well as outputting a collimated beam, i.e. a beam with equal and near-zero divergence in both axes. In some embodiments, after passage through the beam-circularizing elements, all the beams have similar divergence angles. More specifically, the beams, as formed by the beamcircularizing elements, have horizontal-direction divergence angles that lie within a range of 1, 2, 3, 4 or 5 degrees of each other and vertical-direction divergence angles that lie within a range of 1, 2, 3, 4 or 5 degrees of each other. As well as emitting elliptical beams, a further property of edge-emitting semiconductor sources is that their output beams have astigmatism in respect of their fast and slow axes, most commonly in the range 5 to 25 micrometres. This astigmatism can be defined by the size of the offset along the optical axis between the virtual emission point of the output beam in the horizontal direction and the virtual emission point in the vertical direction. In certain embodiments, the beam-circularizing elements are configured to provide for that the divergent beam formed by each of the beam-circularizing elements has an offset between their horizontal and vertical virtual emission points of less than at least one of 1,2, 3, 4, 5, 6, 7, 8, 9 and 10 micrometres. Reducing or eliminating astigmatism also facilitates the generation of beams that are as near as possible circular in cross-section and with equal divergence angle in each axis. While astigmatism is a property of edge-emitting semiconductor sources, other semiconductor sources may also have astigmatic output beams, so this design approach is also applicable to cancel out astigmatism generally. In some embodiments, the module further comprises beam expanding or contracting lenses arranged in respective ones of the divergent beams output from the beam-circularizing elements and before the beam combiner, the beam-expanding or beam-contracting lenses being configured to increase or decrease the beam cross-sections. In some embodiments, the module further comprises a common focusing element arranged in the common optical axis from the beam combiner to form an image from all the beams in a common image plane. In other embodiments, the module further comprises individual focusing elements arranged in respective ones of the divergent beams output from the beam-circularizing elements and before the beam combiner to form an image from all the beams in a common image plane. Various options for realising the common focusing element or the individual focusing elements exist as lenses or mirrors will be known to the person skilled in the art. For example, such a focusing element may have at least one surface that is flat, spheric or aspheric with a first curvature in one axis and that is flat, spheric or aspheric with a second curvature in the other axis (e.g. a cylindrical lens with its axis aligned orthogonally to the first single-axis focusing elements thereby to form a crossed pair). In some embodiments, the common image plane is located at a finite reference distance from the light source module. Some example values for the reference distance at which the image is formed on the common plane are 10 mm, 20 mm, 50 mm or 100 mm. Each of the beams is preferably brought to a focus, within a certain tolerance, in the common image plane. This can be defined in terms of the beams at the common image plane each having a beam diameter that is as close as possible to the beam waist diameter, e.g. no more than 1.3, 1.4,   (=1.414) or 1.5 times the beam diameter at the beam's beam waist. In other embodiments, the output beams from the light source module are collimated to minimize the divergences along both axes. In still further embodiments, the common image plane is virtual, so that the beams are divergent when output from the light source module to form a virtual image. The semiconductor sources are any combination of: superluminescent light-emitting diodes (SLDs); and laser diodes (LDs). The SLDs or LDs may be of the edge-emitting type. The SLDs or LDs may include semiconductor optical amplifiers (SOAs), see for example • US2020251610A1 for an SLD-SOA source. The light source module may be for emitting three primary colours as needed for display or projector applications. Namely, the module has semiconductor sources with peak emission wavelengths with red, green and blue colours to provide an RGB light source module. In addition to modules with emitters that have peak wavelengths in the visible wavelength region (approximately 380 to 750 nm), the modules may incorporate one or more semiconductor sources with peak emission wavelengths in the near-infrared (e.g. from 750 nm up to 2 micrometres). In some embodiments, the module further comprises one or more beam-shaping optical elements arranged in the common optical axis and after the beam combiner to reshape the beams before they are output from the light source module. In some embodiments, the module further comprises a beam-steering element arranged in the common optical axis after the beam combiner, the beam-steering element being configured and operable to controllably vary the direction which the beams are output from the light source module. In some embodiments, each of the beam-circularizing elements is a lens or lens combination. In other embodiments, all but one of the beam-circularizing elements are lenses or lens combinations, the other being a concave mirror which is arranged to direct its output onto a back surface of a first one of the dichroic mirrors. In a still further embodiment, each of the beam circularizing elements is a mirror. In yet further embodiments, it would be possible to use a mixture of mirrors and lenses for the beam circularizing elements. Various options for realising and arranging the dichroic filters of the beam combiner will be known to the person skilled in the art. For example: • The dichroic filters may have a single edge at one specified wavelength such that longer wavelengths are reflected while shorter wavelengths are transmitted. • The dichroic filters may have a single edge at one specified wavelength such that shorter wavelengths are reflected while longer wavelengths are transmitted. • The dichroic filters may have a double edge at two specified wavelengths such that wavelengths shorter than the first wavelength or longer than the second wavelength are reflected while wavelengths longer than the first wavelength and shorter than the second wavelength are transmitted. • The dichroic filters may have a double edge at two specified wavelengths such that wavelengths shorter than the first wavelength or longer than the second wavelength are transmitted while wavelengths longer than the first wavelength and shorter than the second wavelength are reflected. Light source modules according to the invention may be used as components of a display projector. There may thus be provided a display projector comprising: a light source module according to the invention; and a beam scanner for raster scanning the beams output by the light source module to form an image. The beam scanner may be external to the light source module or be partly or wholly integrated in the light source module, e.g. to scan beams within the light source module in one axis or two axes before output. Moreover, such display projectors may be used as components of a head-mounted vision system. BRIEF DESCRIPTION OF THE DRAWINGS This invention will now be further described, by way of example only, with reference to the accompanying drawings. Figure 1 is a schematic drawing of a prior art light source module based on a PIC chip beam combiner. Figures 2 to 5 are schematic drawings of different prior art light source modules based on free-space beam combining optics. Figure 6 is a schematic perspective drawing of a divergent beam of circular cross-section emitted from a point source. Figure 7 is a schematic perspective drawing of a divergent beam of elliptical cross-section emitted from a point source. Figures 8a & 8b show the basic chromatic dispersion dependence of the focusing power of a lens. Figures 9a & 9b compare the basic chromatic dispersion dependence of the refractive power of a standard lens (Figure 9a) and how this dependence is removed with an achromat lens (Figure 9b). Figure 10 is a schematic plan view of a light source module according to an embodiment of the invention. Figure 11a shows the orthogonal top and side views of an edge-emitting semiconductor light source that emits an elliptical, astigmatic beam output with different beam sizes and different divergence angles in the horizontal and vertical direction. Figure 11b shows the edge-emitting semiconductor light source of Figure 11a in combination with a one-axis circularizing lens, which transforms the elliptical, astigmatic beam output from the source into a circular, astigmatism-free beam with equal beam size and equal divergence angles in the horizontal and vertical direction by modifying the beam propagation in the vertical direction and reducing the fast-axis divergence. Figure 11c shows the edge-emitting semiconductor light source of Figure 11a in combination with a one-axis circularizing lens, which transforms the elliptical, astigmatic beam output from the source into a circular, astigmatism-free beam with equal beam size and equal divergence angles in the horizontal and vertical direction by modifying the beam propagation in the horizontal direction and increasing the slow-axis divergence. Figure 11 d shows the edge-emitting semiconductor light source of Figure 11a in combination with an elliptical ball lens, represented by a smaller ball lens for the fast axis, which collimates the divergent beam output in the vertical direction, and a larger ball lens for the slow axis, which collimates the divergent beam output in the horizontal direction, generating a collimated, circular, astigmatism-free beam with equal beam size and equal divergence angles in the horizontal and vertical direction. Figure 12, Figure 13 and Figure 14 are schematic plan views of light source modules according to further embodiments of the invention. Figure 15a shows a schematic graph of a comparative example to illustrate how the horizontal beam diameter H and the vertical beam diameter V may change as a function of propagation distance, z, for a beam output from a semiconductor source, such as a LD or SLD, when a single, standard collimation lens is used. Figure 15b shows a schematic graph of how the horizontal beam diameter H and the vertical beam diameter V may change as a function of propagation distance, z, for a beam output from a semiconductor source, such as a LD or SLD, that was circularized with a first circularizing lens and collimated with a second lens according to certain embodiments of the invention. Figures 16 to 21 are schematic drawings of light source modules according to further embodiments of the invention. Figures 22a and 22b are schematic plan views in two different planes of a light source module according to an embodiment of the invention based on circularizing mirrors. Figures 23a and 23b are schematic plan views in two different planes of a light source module according to another embodiment of the invention based on circularizing mirrors. Figures 24 and 25 show two examples of near-to eye projection system incorporated in a vision system having a spectacles format. DETAILED DESCRIPTION In the following detailed description, for purposes of explanation and not limitation, specific details are set forth in order to provide a better understanding of the present disclosure. It will be apparent to one skilled in the art that the present disclosure may be practiced in other embodiments that depart from these specific details. Numerical values for limits, ratios, ranges etc. are to be interpreted exactly in this document and in particular when construing the claims and not rounded up or down using significant figures, decimal places or otherwise. Figure 6 is a schematic perspective drawing of a divergent beam 4 of circular cross-section emitted from a source point, S. This would be the situation for an output by: a VCSEL; an optical fibre laser; an optical fibre at the end of a cascade of fibre couplers used for combining beams; for an edge-emitting LD or SLD that has an integrated lens at its output face to correct for the different divergences of the chip facet output; or for an edge-emitting LD or SLD that is otherwise carefully designed to have identical mode sizes in the horizontal and vertical axis at the output facet, for example by deploying horizontal or vertical tapers or other beam-shaping structures. Figure 7 is a schematic diagram of a divergent beam 4 with an elliptical cross-section as output by an edge-emitting LD or SLD 5 emitted from a source point, S. The beam has a vertical, fast axis and a horizontal, slow axis. For an elliptical cross-section beam there will be both a major diameter (along the fast axis in the x-direction labelled V for vertical) and a minor diameter (along the slow axis in the y-direction labelled H for horizontal). Figures 8a & 8b show the basic chromatic dispersion dependence of the refractive power of a lens. Figure 8a is a schematic diagram of a lens 6 arranged to collimate three different beams 1,2, 3 of three different wavelengths 11, A2, X3, e.g. colours, the beams all having a common source point, S. Figure 8b is a schematic graph showing the dependence of wavelength, X, on refractive index, n, for an arbitrary lens material, e.g. as is typical for common transparent glasses and crystals. Any lens material and hence any given lens will have a different refractive power n1, n2, n3 that is dependent on the wavelength of the light X1, A2, X3. The difference in the refractive power stems from the chromatic dispersion of the lens material. Using a ray model and considering Snell’s law of refraction, it is evident that a change in the refractive index of the lens material will result in different refraction angles when the light passes any of the lens surfaces. If the two lens surfaces are not parallel, a lens is formed that is capable of changing the beam propagation direction dependent on the colour of the light. This may result in beam parameters (waist position, waist size, divergence angles, etc.) being different for different wavelengths A1, A2, A3. As shown in Figure 8a, a standard lens may thus have different focal lengths for different colours, therefore resulting in the output beams being either convergent, collimated or divergent, dependent on their optical wavelength, when being emitted from the same source point S. Figures 9a & 9b show the basic chromatic dispersion dependence of the refractive power of a standard lens and how this dependence is removed with an achromat lens. Figure 9a is a schematic diagram of a standard lens 6 arranged to focus a collimated polychromatic input beam, for example white light, consisting of three different beams 1, 2, 3 of three different wavelengths A1, A2, A3, e.g., colours. In this example, lens 6 is a standard collimation lens that would focus the beams to different focal points S1, S2, S3, based on their wavelengths A1, A2, A3. Figure 9a also demonstrates that, in the reverse direction, generating a collimated white output beam with a standard collimation lens 6 would require emitters of different colours A1, A2, A3 to be placed at different source points S1, S2, S3 and the different emitters would be required to emit the light with different divergence angles in order for the collimation lens 6 to generate a collimated light beam with the same output beam diameter and the same output beam divergence for all the different colours. Figure 9b is a schematic diagram of an achromatic lens 7 arranged to focus a collimated polychromatic input beam, for example white light, consisting of three different beams 1, 2, 3 of three different wavelengths A1, A2, A3, e.g., colours. With an achromat lens, the emitters of different colours can be positioned at the same source point or distance S with similar divergence in order to generate a collimated white output beam of equal beam diameter and beam divergence. Figure 10 is a schematic plan view a light source module 50 according to an embodiment of the invention. The light source module 50 comprises red, green and blue emitters 110, 120 and 130 that are mounted on individual submounts 140, although the emitters may be also mounted on a common submount with a suitable geometry. Each emitter emits a divergent beam which has an elliptical cross-section, although the light source module design may also be used for emitters that emit a circular cross-section divergent beam. Each divergent beam is focused in at least one of the two axes by individual lenses 511, 521 and 531, for example free-form or fast-axis reduction (FAR) lenses, that convert the divergent, elliptical beam output from each emitter into a divergent, circular beam. The lenses 511, 521 and 531 are referred to in the following as circularizing lenses CIRL1, CIRL2 and CIRL3, respectively. The beam from emitter 110 has an elliptical beam shape before the circularizing lens, meaning that it has different beam diameters in the fast axis and slow axis and different divergence angles in both axes. The beam from emitter 110, after having passed through the circularizing lens 511, has a circular beam shape, meaning that is has nearly equal beam diameters in both fast axis and slow axis and nearly equal divergence angles in both axes. The circularizing lenses CIRL2 and CIRL3 transform the elliptical beams emitted from emitters 120 and 130 respectively in a similar manner. The beam as output from the circularizing lens 511 is deflected by 90 degrees by a plane mirror 112 prior to entering a beam combiner 240 formed by dichroic filters 122, 132. Ina variant, the deflection angle of the plane mirror 112 may deviate somewhat from 90 degrees, e.g., by up to ±20 degrees. The three beams that are input into the beam combiner 240 are thus still diverging in both axes and this is also the case for the three beams as output by the beam combiner 240. The beam combination thus takes place by the dichroic filters 122, 132. The dichroic filters could be edge filters, i.e. single-edge filters, in which case the emitters would need to be arranged in wavelength order (either descending as illustrated or ascending). On the other hand, if the dichroic filters are bandpass or band-reflect filters, i.e. two-edge filters, there would be no requirement for any particular wavelength sequence; rather any arbitrary sequence would be possible. The combined beams output from the beam combiner 240, i.e., the three individual beams that now share a common optical axis, O, are then incident on a common collimation lens (CCL 750) to complete the collimation. (In an alternative embodiment, the common collimation lens may be exchanged for an equivalent common collimation mirror.) The collimated beams are then output from the module 50 along the common optical axis, O, via a window 150 arranged in an end wall of a housing or enclosure 100 that contains the various components. Optionally, the window 150 may be omitted to provide an open output port for the module 50. In our description of the light source module, we use the terms collimation lens and collimated output beam in a somewhat loose manner (which is not uncommon in the art). It is namely a design choice whether the light source module forms an image at infinity, which would imply parallel rays for the output beams in a simple ray optical consideration of the situation (i.e. collimated beams in a strict ray optics sense), or to form an image at a finite distance away from the light source module, e.g. at a distance in the range 10 mm to 100 mm, in which case the output beams would not be collimated (in a simple ray optics sense). The term collimation lens therefore does not imply that the collimation lens necessarily outputs a beam that forms an image at infinity from a monochromatic point source object but also includes forming an image at a finite distance. In some embodiments, in the common plane, each output beam has a beam diameter that is as close as possible to the beam waist diameter, e.g. no more than 1.3, 1.4,   (—1.414) or 1.5 times the beam diameter at the beam's beam waist. In other words, each of the output beams is at or close to its focus, i.e. within a certain tolerance, in the common plane. The circularizing lenses 511, 521, 531 may all be the same while nevertheless being able to somewhat compensate for the three beams having different vertical far field (VFF) values and / or different horizontal far field (HFF) values and / or different colours by positioning each circularizing lens at a defined distance to its emitter, i.e., the distance from each emitter to its circularizing lens may all be different and set individually. However, such a circularizing lens might be designed for a specific aspect ratio HFF:VFF, for example 1:3, 1:4, 1:5 or higher values or even lower values, and for a specific wavelength. Therefore, deviating strongly from the designed aspect ratio of the circularizing lens or of the design wavelength may result not in a circular output beam with similar divergence angles in X- and Y-direction but, for example, in a slightly elliptical beam and / or in slightly different divergence angles in X-and Y-direction. Therefore, it might be preferable to use individual circularizing lenses that are specifically designed for the beam profile and for the wavelength of the emitter. In order to achieve the same or similar beam diameters and beam divergence angles after the common collimation lens among the three beams, the propagation distance between the common collimation lens CCL 750 and each of the emitters 110, 120, 130 is carefully defined in the layout of the optical module so that the wavelength-dependent differences in the refractive power and / or the wavelength-dependent focal distance of the common collimation lens for each of the three beams is compensated for. The common collimation lens could be a spherical or aspherical lens, a free-form lens or could be also realized as a lens that is composed of individual lenses. For example, it might be preferable for the optical layout and for the optical performance of the optical module to use an achromatic lens as the common collimation lens such that the refractive power and thus the focal and working distance is nearly independent of the emitter’s wavelength. A common type of an achromatic lens is a doublet, which is composed of two individual lenses made from glasses with different amounts of dispersion. For example, one element is a negative or concave element made from a material with high dispersion and the other element is a positive or convex element made from a material with low dispersion. The two lens elements are mounted together to form a single optical lens where chromatic and spherical aberrations are counterbalanced and minimized. Similar spot size for each of the beams can be defined in terms of having similar beam diameters at a common plane located at a reference distance from the light source module, such as 10 mm, 20 mm, 50 mm or 100 mm, or infinity, for example. Beam diameter similarity may be defined by the ratio of the beam diameters of the smallest one of the beam diameters versus the largest one of the beam diameters. For example, the spot sizes may be deemed to be sufficiently similar if this ratio is greater than 50%, 60%, 70%, 80% or 90%. The beam diameters and the divergence angles of the output beams of the light source module is defined by the beam divergence angles after the circularizing lenses and by the propagation distance between the circularizing lenses and the common collimation lens. The divergence angles after the circularizing lenses depend on the initial divergence angles and on the wavelength of the emitters and on the optical performance of the circularizing lenses. For example, an emitter may have a VFF of 36° and a HFF of 12°, and the circularizing lens may be designed to reduce the VFF with an aspect ratio of 1:3, meaning that the divergence angles after the circularizing lens would be 12° in both axes. In this example here, the VFF and HFF values are referenced as the half angles at a relative intensity of 1 / e2 (13.5%). However, it is equally common to specify VFF and HFF values as full angles at 50% relative intensity or full angles at 1 / e2 relative intensity. The full beam diameter w for a collimated beam (at a relative intensity of 1 / e2) after the common collimation lens with a focal length f can be estimated as w = 2 ■ f ■ tan(0) with 0 being the half divergence angle of the circular beam after the circularizing lens (at a relative intensity of 1 / e2). For example, with 0=12° and a common collimation lens with a focal length / =3.0 mm, the diameter of the collimated output beam would be w=1.28 mm. Realizing smaller beam diameters at the output of the light source module would require, for example, a common collimation lens with a smaller focal length, for example / =2.5 mm or / =2.0 mm, and thus a smaller working distance. Depending on the size of the optical components, in particular the beam combiners, and on the overall optical layout, this might be challenging to achieve. Therefore, it might be preferable to have smaller divergence angles 0 after the circularizing lenses, for example half divergence angles 0=10° or 0=8°. For collimated output beams with significantly smaller beam diameters, it might be therefore preferable to place individual collimation lenses after the circularizing lenses. Realizing larger beam diameters can be realized by selecting a common collimation lens with a larger focal length, for example / =4.0 mm or / =5.0 mm or even larger values. Naturally, the size of the common collimation lens would become larger but also the working distance would increase, which means that the distances between the common collimation lens and the emitters have to be increased as well. Alternatively, the divergence angles 0 after the circularizing lenses could be increased to a certain extent while still ensuring that the beam size of the divergent beam at the entrance of the common collimation lens does not experience beam clipping and unwanted diffraction artifacts. While this architecture provides certain flexibility in tailoring the beam diameters at the output of the light source module, there are certain limitations and trade-offs that need to be considered. As already explained, there might be a certain minimum focal length required for the common collimation lens in order to accommodate the free-space beam combiner optics. Alternatively, space constraints for the light source module may limit the maximum focal length of the common collimation lens. It is further noted that one degree of design freedom can be provided by the ability to vary the horizontal divergence (HFF) of the emitters through the chip design. Foran edge-emitting device with an optical waveguide, for example an LD or an SLD, the horizontal mode size can be tailored over a relatively wide range through the waveguide width or the waveguide height or its index profile. For instance, a horizontal taper could be realized near the output facet of the chip such that the HFF for a particular emitter is matched to the refractive power of the circularizing lens and to a desired reference propagation distance inside the module where the combined beams are to have particular specified properties in terms of beam divergence / collimation / convergence and beam cross-sectional shape. Similarly, for an edgeemitting device, such as an LD or an SLD, the vertical mode size and thus the vertical divergence (VFF) can be tailored to some extent by modifying the optical waveguide structure in the vertical axis, for example by adjusting the refractive index profile of the epitaxial layer structure or the thickness values of certain layers of this structure. Another degree of design freedom is to use circularizing lenses that have a refractive power not only in one axis but in both axes. Using the same example as above, an emitter may have a VFF angle of 36° (fast axis) and a HFF angle of 12° (slow axis). Using a circularizing lens that is designed with a refractive power only in the fast axis with an aspect ratio of 1:3, the divergence angles after the circularizing lens would be 12° in both axes. However, a different design of a circularizing lens or mirror may have refractive power in both the fast axis and the slow axis and could, therefore, generate a circular beam with smaller divergence angles, for example of 10°, 8° or even smaller values in both axes. Alternatively, a circularizing lens or mirror may have refractive power in both axes and may generate a circular beam with larger divergence angles, for example of 14°, 16° or even larger values in both axes. A circularizing lens or mirror may be also designed for other HFF:VFF aspect ratios. For example, an emitter may have a HFF angle of 10° and a VFF angle of 40° and may thus require a circularizing lens or mirror with an aspect ratio of 1:4 to generate a circular beam with equal divergence angles at the output. Alternatively, an emitter may have a HFF angle of 9° and a VFF angle of 45° and may thus require a circularizing lens or mirror with an aspect ratio of 1:5 to generate a circular beam with equal divergence angles at the output. The aspect ratio can be designed without difficulty to provide any value from 1:1 to 1:10. It could also have larger aspect ratio values greater than 1:10, although this is unlikely to be required for realistic light sources at the input. It could also have smaller values than 1:1 even though this may not be required for realistic light sources at the input. Figures 11a shows orthogonal views of an edge-emitting semiconductor light source as shown in Figure 7, such as a LD, SLD or SOA, with X being the horizontal direction, Y the vertical direction and Z the light propagation direction. The light-generating active region is indicated as a stripe in both the side view and top view. A first optical characteristic of the output beam from an edge-emitting semiconductor light source is, as discussed previously with reference to Figure 7, is that the output beam is elliptical. The optical mode shape and size is quite different in the vertical and horizontal directions. Typical for semiconductor-based light sources, the vertical size of the lightgeneration waveguide and thus the vertical size of the optical mode is much smaller than the horizontal size of the waveguide or of the optical mode. For example, the light-generating active region, typically a quantum well or multi-quantum well, can have a vertical thickness ranging from a few nanometres to a few tens of nanometres while the horizontal width of the light-guiding waveguide is typically on the order of a few microns for lateral single-mode devices. The optical mode shape and size is consequently quite different in the vertical and horizontal directions, as discussed previously with reference to Figure 7. Typical mode sizes for single-mode emitters, for example in the visible-wavelength range, are around 0.5 pm in vertical direction and 1.5-2.5 pm in horizontal direction. The different optical mode sizes in horizontal and vertical direction inside the waveguide of the LD, SLD or SOA devices also results in an elliptical beam profile in the far field or after standard collimation optics but also in different divergence angles in horizontal and vertical direction, i.e., the smaller vertical mode size inside the waveguide, the so-called near field, translates to a larger divergence angle cpy in the far field, while the larger horizontal mode size inside the waveguide translates to a smaller divergence angle cpx in the far field. A second optical characteristic of the output beam from an edge-emitting semiconductor light source is that it is astigmatic in respect of the horizontal and vertical directions, which is a consequence of the active region dimensions being different in the horizontal (X) and vertical (Y) directions. There is a non-uniformity of the gain profile in the thicker X-direction compared to the thinner Y-direction, especially at higher output power levels. The non-uniform gain profile results in a non-uniform index profile, which causes an amplitude and phase distribution of the optical mode inside the waveguide such that it appears as if the horizontal mode is emitted from further inside the waveguide than the vertical mode. In other words, in the fast axis, cpY, the emission appears to be from closer to the output facet of the semiconductor chip than in the slow axis, cpX. The difference in z-direction between the virtual emission point in the horizontal and vertical direction is referred to as an astigmatism AZa and has typical values of 5-25 pm but in some cases could be larger or smaller than that. For many applications and optical systems, it is also beneficial to have the minimum amount of astigmatism. For example, when coupling light from free space to single-mode or polarization-maintaining fibres or waveguides using a focusing lens, it will be beneficial for higher coupling efficiency to have not only a circular beam profile but also to have zero astigmatism because any astigmatism will cause the focused light to have two different focal points, which will reduce the coupling efficiency. These astigmatism effects may cause drawbacks in other optical systems where lenses, diffractive elements or other optical components are used, for example scanner systems, imaging systems or display architectures where ideally a high-quality image of the light source should be generated for best performance, so it will also be important to reduce the astigmatism of the light source to a minimum in such other optical systems. Figures 11b and 11c show two example configurations where a circularizing lens placed in the output beam of an astigmatic, edge-emitting semiconductor source can perform both circularization and astigmatism correction. In both example configurations, the circularizing lens has refractive power only in one axis, i.e. either the slow axis or the fast axis. Figure 11b shows the case of refractive power only in the fast, vertical axis, a single-axis circularizing lens may be realized as a fast-axis reduction (FAR) lens, where the larger vertical divergence angle from the source is reduced and matched to the smaller horizontal divergence angle. Figure 11c shows the case of refractive power only in the slow, horizontal axis, a single-axis circularizing lens may be realized as a slow-axis expansion (SAE) lens where the smaller horizontal divergence angle is increased and matched to the larger vertical divergence angle. Figure 11b shows the same two orthogonal views of an edge-emitting semiconductor light source as shown in Figure 11a in combination with a one-axis circularizing lens (GIRL). The one-axis circularizing lens transforms the elliptical, astigmatic beam output from the source into a circular, astigmatism-free beam with equal beam size and equal divergence angles in the horizontal and vertical direction by modifying the beam propagation in the vertical direction and reducing the fast-axis divergence. In this particular example, the circularizing lens has at the input side that receives the light beam from the edge-emitting light source, a convex surface in the vertical (Y) direction and a straight, uncurved surface in the horizontal (X) direction. At the opposite output side, the circularizing lens has a concave surface in the vertical (Y) direction and a straight surface in the horizontal (X) direction. Consequently, this particular circularizing lens will not alter the divergence angle of the emitted beams in the horizontal plane (xz-plane) whereas in the vertical plane (yz-plane) the convex surface will have a refractive effect and may fully collimate the light beam such that the light rays inside the circularizing lens propagate parallel to the Z-axis (solid line ray paths inside the lens in Figure 11b)), or it may partially collimate the light beam such that the light rays propagate inside the circularizing lens in a divergent manner (dashed line ray paths inside the lens in Figure 11 b) or a convergent manner (not shown in Figure 11b). It may be preferred to design the shape of the convex input surface of the circularizing lens such that it is matched as closely as possible to the vertical beam profile and the vertical divergence angles of the light source, resulting in a fully collimated beam inside the circularizing lens with the rays being parallel to the Z-axis (solid lines inside the lens in Figure 11b). At the output side of the circularizing lens, the concave surface in the vertical direction will diffract the light beams in the fast axis so that a divergent optical beam is generated. Here, the concave output surface is designed such that the generated beam profile and divergence angles in the Y-direction are matched as closely as possible to the beam profile and divergence angles in the X-direction. The thickness of the circularizing lens is designed so that, at the second concave output surface, the beams in the horizontal and vertical direction have the same size. In that case, ideally a circular beam profile is generated with ideally equal divergence angles in both axes. In this example, the circularizing lens will not modify the beam divergence angle in the horizontal direction but its optical thickness and propagation length inside the circularizing lens will modify the beam size in the horizontal direction. Namely, the thicker the circularizing lens is the larger the horizontal beam size will be at the output. Therefore, adjusting the thickness of the circularizing lens can be used as a design parameter to eliminate the astigmatism and to generate equal beam diameters in both the X- and Y-directions, while the shape of the concave output surface can be used as a design parameter to generate equal divergence angles in both axes. Therefore, it is possible to design a circularizing lens to match the output of any particular semiconductor light source in order to transform the beam output by the semiconductor light source into an exactly circular beam and a beam with zero astigmatism. Figure 11c shows the edge-emitting semiconductor light source of Figure 11a in combination with a one-axis circularizing lens, which transforms the elliptical, astigmatic beam output from the source into a circular, astigmatism-free beam with equal beam size and equal divergence angles in the horizontal and vertical direction by modifying the beam propagation in the horizontal direction and increasing the slow-axis divergence. The circularizing lens of Figure 11c may be realized as a slow-axis expansion (SAE) lens where the smaller horizontal divergence angles are increased and matched to the larger vertical divergence angles. The example configuration of Figure 11c will be understood with reference to the discussion of Figure 11b above. An SAE lens of this kind may generate a circular, astigmatism-free output beam with similar divergence angles in both axes. Compared with an FAR lens, an SAE lens will cause the beam size to increase much more rapidly over a given propagation distance, which is a useful option when it is desired to enlarge the beam over as short a distance as possible within a module to save space. Comparing Figure 11b and 11c with Figure 11a, the circularizing lenses FAR and SAE not only harmonize the divergence angles and beam sizes in the slow and fast axis but also eliminate the astigmatism AZa of the emitter’s output beam. While in Figure 11a the virtual emission spot in the horizontal axis is further inside the waveguide compared to the vertical axis, which is close to the output facet, the FAR lens in Figure 11b shifts the virtual emission spot in the vertical axis inside the emitter’s waveguide to a position that is identical to the position of the horizontal axis’ emission spot, thereby reducing the astigmatism to zero. Similarly, the SAE lens in Figure 11c shifts the virtual emission spot in the horizontal axis closer to the output facet or even outside the device in front of the output facet. Because of diffraction at the lens interfaces, also the virtual emission spot in the vertical axis is modified and is in the same plane as the horizontal axis’ emission spot, thereby again reducing the astigmatism to zero. Furthermore, when realizing an optical light source module, it may be desired to have circularizing lenses that have a refractive power in both axes. For example, such a circularizing lens could have a convex or concave surface in the horizontal direction to either decrease or increase the divergence angle in the X-axis. The output surface in the vertical direction would have to be adjusted accordingly such that the resulting divergence angle in the Y-axis is matched to the divergence angle in the X-axis. Such a dual-axis circularizing lens would provide greater design flexibility for various optical architectures, since smaller or larger beam diameters could be generated at the output of the light source module, as explained earlier. Figure 11 d shows the edge-emitting semiconductor light source of Figure 11a in combination with ball lenses to modify the beam divergence in either the horizontal or vertical direction and to, ultimately, collimate the output beam. Ball lenses are another type of lens that can be used for embodiments of the invention. A standard ball lens has a constant diameter / ) in all directions and is made of a material with a refractive index n at a given wavelength. The effective focal length (EFL) of a standard ball lens is then defined as EFL = nD / 4(n - 1). A ball lens with a larger diameter D will therefore have a larger EFL and a ball lens with a smaller diameter D will have a smaller EFL. To collimate the divergent and elliptical beam output from semiconductor light sources, such as edge-emitting LDs or SLDs, an ellipsoidshaped ball lens is suitable, with the ellipsoid having a smaller diameter in the vertical, fastaxis direction and a larger diameter in the horizontal, slow-axis direction. This is diagramatically represented by showing a smaller diameter ball lens for the fast axis, which collimates the divergent beam output in the vertical direction, and a larger diameter ball lens for the slow axis, which collimates the divergent beam output in the horizontal direction. Such an ellipsoid-shaped ball lens is also able to compensate for any astigmatism of the light source, since the EFL values are different in the horizontal and vertical directions. In addition, the collimated output beam would have a circular cross-section with equal beam diameters and thus equal beam divergence angles in the horizontal and vertical directions. Therefore, an ellipsoid-shaped ball lens is an example of a circularizing lens, with the special characteristic of generating a collimated output beam. Similar to having half-ball lenses for certain optical architectures, a circularizing lens may also be realized with a half-ellipsoid ball lens. Figure 12 shows an embodiment of a light source module 50 containing three emitters 110, 120, 130 which each emit a divergent beam. After each emitter 110, 120, 130 there is arranged a circularizing lens GIRL 511, 521, 531. The beam divergence angles are further modified in both axes after transmission of the beams through the respective circularizing lenses 511, 521, 531 by further individual lenses (L) 611, 621, 631. Similar to as described above with reference to Figure 10, each emitter 110, 120, 130 is mounted on a suitable submount 140. The elliptical output beams of each emitter, which also have astigmatism, are first propagated through individual circularizing lenses GIRL 511, 521, 531, which are also mounted on the same individual submounts 140 as the emitter devices. Alternatively, the circularizing lenses can be mounted separately and not on the same submounts 140 as their associated emitters. After propagating through the circularizing lenses, the individual output beams have a circular beam profile with similar divergence angles in both horizontal and vertical axis and with zero astigmatism. Each output beam then propagates through a second set of individual lenses 611, 621, 631 that either increase or decrease the divergence angles for both axes. The function of the second lenses 611, 621, 631 is to enlarge or reduce the respective beam diameters output by the subsequent common collimation lens CCL 750 at the output of the light source module. These second lenses 611, 621, 631 can be realized with spherical, aspherical or free-form lenses. As explained earlier, the functionality of these second lenses 611, 621, 631 could, in principle, be integrated into the design of the circularizing lenses 511, 521, 531. In the illustrated design implementation, the output beams of the second lenses 611,621,631 are not collimated but are either convergent or divergent as they pass through the beam combiner unit 240 with its combiner elements 122 and 132. Similar to as described with reference to Figure 10, only once the output beams have propagated through the common collimation lens 750 need they have the same divergence angles in both the vertical and horizontal directions, e.g. to be collimated or to have a common real or virtual focal point (or correspondingly located focal planes for forming an image). The optical layout and specifically the propagation distance of the light from the emitters 110, 120, 130, or from the circularizing lenses 511, 521, 531 or from the second lenses 611,621,631, to the common collimation lens 750 is intentionally chosen by the design such that the three output beams have the same divergence angles in both axes, which may be non-divergent (i.e. collimated), convergent (i.e. real focus) or divergent (i.e. virtual focus). Figure 13 shows another embodiment of a light source module 50 where the beam divergence angles after the circularizing lenses are again modified in both axes using separate lenses. Similar to the example of Figure 12, each emitter 110, 120, 130 is mounted on a suitable submount 140, together with individual circularizing lenses GIRL 511, 521, 531, which serve to convert the elliptical and astigmatic output beams of each emitter into circular, astigmatism-free beams with equal divergence in both axes. After the circularizing lenses, the individual output beams propagate through a second set of individual lenses 111, 121, 131 that act to modify the divergence angles of the light beams in both axes such that the output beams of lens 111, 121, 131 are made the same in both axes before entering the beam combiner unit 240 with its combiner elements 122 and 132. In this embodiment, the combined beams are output from the module 50 without further manipulation. In other words, compared with the embodiment of Figure 12, there is no common collimation lens. The propagation distance of the light from the individual emitters 110, 120, 130 to the individual lenses 111, 121, 131 may be intentionally chosen by the design so that three output beams are generated with the same divergence angles in both axes, which may be non-divergent (i.e. collimated) as illustrated, convergent (i.e. real focus) or divergent (i.e. virtual focus). The design of Figure 13 may be preferred over the embodiments shown in Figure 10 or Figure 12 when smaller beam diameters at the output are required. Figure 14 shows another embodiment of a light source module 50 where the elliptical, astigmatic and divergent beam output of each emitter 110, 120, 130 passes through ellipsoid-shaped, circularizing ball lenses GIRL 561, 571, 581, which serve to convert the output beams of each emitter into circular, astigmatism-free beams that are collimated, i.e. having equal and near-zero divergence in both axes. In this embodiment, the circularization and the collimation of each beam is performed with a single optical element before entering the beam combiner unit 240. In other words, compared with the embodiment of Figure 13, there are fewer optical components needed as the functionality of each pair of circularizing lens 511, 521, 531 and collimation lens 111, 121, 131 is replaced by an ellipsoid-shaped, circularizing ball lens GIRL 561, 571, 581. Therefore, such an arrangement may be preferred for more compact designs or designs of lower cost. One or more of the above described designs can be implemented to provide one or more of the following advantages: • Ina three-source module with a common collimation lens, two optical components are eliminated compared with having three individual collimation lenses, since the common collimation lens performs the dual-axis collimation for all three beams. With a higher number of sources, e.g. four or more, the benefit is further increased. Reducing the component count is not also a cost saving but also saves manufacturing time by reducing the number of components that need to be aligned optically before fixation. Having a lower component count may also improve manufacturing yield. • Performing beam circularization followed by beam focusing or beam collimation allows for generation of circular optical beams that have similar beam diameters and similar beam divergence angles in both horizontal and vertical direction. Optical modules with such architectures can be designed to emit circular beams that all diverge at the same rate while maintaining their size, shape and aspect ratio. Further, optical modules with such architectures can be designed to emit collimated, circular beams that propagate with the same size, shape and aspect ratio over long propagation distances. Further, optical modules with such architectures can be designed to emit astigmatism-free, circular beams that converge in the same manner and that generate a similar beam waist for the horizontal and vertical direction at the same image position. Figure 15a describes, by way of a comparative example, the typical performance that is achievable when using a single collimating lens, for example an aspheric lens, to form a beam from the output of a semiconductor, edge-emitting light source, such as an LD or an SLD. A single collimation lens has the role of simultaneously collimating the horizontal, slow axis and the vertical, fast axis. In more detail, Figure 15a shows how the beam diameters w of a collimated beam in the fast-axis, vertical direction V and in the slow-axis, horizontal direction H may evolve over propagation distance z. In this example, the vertical collimated beam emitted by the light source module may first converge until the beam diameter reaches a minimum at the distance z_wO-V. The minimum beam diameter wO-V is also referred to the beam waist. Over a certain propagation distance Az_R-V around this minimum in the beam diameter, the vertical beam diameter remains approximately constant within a certain tolerance, for example not increasing by more than one of 30%, 40%, ^2 (41.4%) or 50% compared with the beam diameter at the beam waist. For Gaussian-shaped beams, this propagation distance may refer to twice the Rayleigh length over which the beam diameter increases by ^2. Elliptical beams and beams with astigmatism may show two different beam waists at different propagation distances, i.e., the beam waist for the vertical axis, wO-V, might be different to the beam waist for the horizontal axis, wO-H, as illustrated. Furthermore, the propagation distance of the beam waist for the vertical axis, z_w0-V, might be different to the distance of the beam waist for the horizontal axis, z_w0-H, which is referred to as astigmatism. For semiconductor-based, edge-emitting light sources, it is also common that the horizontal beam waist is larger compared to the vertical beam waist, thereby resulting typically in a larger propagation distance Az_R-H around the waist position. It is possible that the crossing point of horizontal and vertical beam diameter, z_rb, at which the collimated optical beam has perfect circularity, may occur between the vertical waist position, z_wO-V, and the horizontal waist positions, z_wO-H, or at least close to both. Around this crossing point of perfect circularity, z_rb, a certain propagation distance Az_rb can be defined over which the collimated beam has sufficient circularity, meaning that the ellipticity is above a certain minimum value, for example 70% as illustrated by the rectangular box around z_rb. Here, the ellipticity is defined as the ratio of the smaller beam diameter of the two axes versus the larger beam diameter of the two axes. Consequently, a perfectly circular beam would have an ellipticity of 100% (i.e. unity) and a beam is defined as being circular, according to ISO 11146-1:2021, if its ellipticity is greater than 87%. As illustrated by way of example in Figure 15a, an elliptical beam output from a light source that is collimated with a single collimation lens may result in a collimated beam with a strong astigmatism where the horizontal and vertical beam waist positions have a larger distance z relative to each other such that the beam divergence in either horizontal or vertical direction at the crossing point has a larger slope, thereby resulting in only a smaller range Az_rb with acceptable circularity and with acceptable beam shape. Figure 15b is to be compared with Figure 15a and shows the improved performance that is achievable as in embodiments of the invention when a lens pair of a circularizing lens and a collimation lens is used to form a collimated beam from the output of an edge-emitting, semiconductor light source, such as an LD or an SLD. A lens pair of a circularizing lens and a collimation lens generates a circular, astigmatism-free optical beam, which is divergent with similar divergence angles in both axes, and then collimates the circular beam. The circulating lens and the collimation lens can be optimized, and their position can be adjusted, such that the vertical beam waist wO-V and the horizontal beam waist wO-H not only have similar values to each other but also occur at similar propagation distances z_w0-V and z_w0-H, respectively, from the light source module, which means that the astigmatism is greatly reduced. Consequently, the crossing point of horizontal and vertical beam diameter, z_rb, at which the collimated optical beam has perfect circularity, is also close to these distances. In this scenario, the resulting beam divergence is similar for the vertical, fast axis and the horizontal, slow axis, meaning that also the propagation distance of Az_rb with sufficient circularity can be significantly larger - indeed much larger than Az_R-V and Az_R-H, as illustrated. By way of comparison, using a single collimation lens, the propagation distance Az_rb of an acceptably circular (i.e. round) beam could be tens of millimetres while it may be tens of centimetres when using a separate circularizing lens and collimation lens, which is clearly beneficial for many system architectures. Furthermore, the generated output beams have a circular shape, as desired for many applications, and have a minimum amount of astigmatism. Figures 16 to 21 are schematic plan views of light source modules according to further embodiments of the invention. The same or corresponding features are labelled with the same reference numerals as used previously and are not all explicitly mentioned again in the following description. Figure 16 shows a variant of the design of Figure 10 according to another embodiment in which the common collimation lens CCL 750 is arranged in an aperture in the end wall of the module housing or enclosure 100 thereby dispensing with the need for the separate window 150. The length, L, of the module enclosure can thus be reduced. Moreover, a further optical component is eliminated. The common collimation lens CCL 750 thus also functions as an output window. Figure 17 shows a variant of the design of Figure 10 according to another embodiment in which the plane mirror 112 is dispensed with and instead the first emitter 110 is arranged to emit in the lengthwise direction of the module. Although the plane mirror 112 is dispensed with, this design will generally result in an increased length, L, of the module and also dictate against being able to mount all the emitters on the same submount 140 (or greatly increase the size of the submount). Figure 18 shows a variant of the design of Figure 10 according to another embodiment in which the functions of the plane mirror 112 and circularizing lens 511 are combined by a circularizing mirror CIRM 112 / 511 with a concave mirror surface in, at least, the fast-axis orientation. One fewer component is therefore needed. The concept of using a concave mirror, for example a dual-axis parabolic metal mirror, for beam circularization can be extended to all emitters 110, 120, 130 when the light of those concave mirrors is reflected vertically (out of plane of the paper). In this case, each emitter would emit an elliptical, astigmatic beam that is circularized by an individual concave mirror 511 or lens 521, 531, with the output beams then being combined by the beam combiner 240 before being collimated by a common collimation lens 750. Figure 19 shows a variant of the design of Figure 10 according to another embodiment in which a common collimation mirror 750 is used instead of a common collimation lens. The common collimation mirror 750 receives the combined output beams from the beam combiner 240 and simultaneously collimates the beams along their fast and slow axes and deflects them by 90 degrees, so that the beams are output in the opposite direction to the propagation direction of the beams as output from the emitters 110, 120, 130. Another option would be if the common collimation mirror 750 were rotated clockwise by 180 degrees compared with the illustration in which case the beam would be output in the same direction as the propagation direction of the beams as output from the emitters 110, 120, 130. In further variants, the common collimation mirror 750 could also be rotated in other ways compared to how it is illustrated, for example to direct the output beam orthogonal to the plane of the drawing either upwards or downwards. Figure 20 shows a variant of the design of Figure 10 according to another embodiment in which additional beam-shaping optics are positioned after the common collimation mirror 750, for example a pair of prisms, namely a first prism 550 and a second prism 560. The beam-shaping optics receives the combined output beams from the common collimation mirror 750 and changes the beam diameters of all beams simultaneously in one or both axes by a fixed magnification factor. For example, an amorphic prism pair placed in the horizontal plane, as shown in Figure 19, may either increase or decrease the horizontal beam diameters of all beams such that a residual beam ellipticity is removed and circular beams are generated. For such an amorphic prism pair, the propagation direction through the prism pair will decide whether the beam diameters will be enlarged or reduced. Such beamshaping optics may also be employed to intentionally generate a certain beam ellipticity at a defined reference distance, z_ref, away from the light source module (which may be unity for a circular beam). Alternatively, such beam-shaping optics could be placed such that the beam diameters in the vertical direction are either increased or decreased by a certain amount. Another option would be to position such beam-shaping optical elements after a common collimation lens (e.g., to modify the embodiment of Figure 17) instead of after the common collimation mirror as provided in the present embodiment. Figure 21 shows a variant of the design of Figure 10 according to another embodiment in which an additional beam-steering element 290 is positioned after the common collimation lens 750. The beam-steering element could be, for example, a micro-electro-mechanical mirror (MEMS) that is mounted to oscillate or rotate about one axis (e.g. with a hinge mount), two perpendicular axes (e.g. four tilt actuators arranged in a rectangle or square which are independently drivable to form arbitrary pairings) or an arbitrary axis (e.g. ball mount). The beam-steering element allows steering of the received collimated beams from the emitters 110, 120 and 130 to provide a time-dependent propagation direction through the common output window 150. The beam-steering element can be used to steer all beams together when the emitters 110, 120, 130 are on simultaneously. The beam-steering element can also be used in display applications where the emitters are time-division multiplexed such that only one primary colour is emitted at any one time in order to steer each primary colour beam separately. It shall be understood that the addition of a beam-steering element inside the common package could be realized also for any of the other embodiments mentioned. Figure 22a and 22b shows a light source module according to another embodiment, which uses mirrors instead of lenses as in the embodiment of Figure 10. Compared with the embodiment of Figure 10, individual circularizing lenses are replaced with circularizing mirrors. The elliptical, divergent output beams of emitters 110, 120, 130 are reflected by circularizing mirrors 511, 521, 531 that are positioned in a common, first plane with the emitters. The circularized, reflected beams are reflected out of the plane of the drawing, for example vertically upwards or downwards. Figure 22b shows a second plane of the same light source module 50. The divergent beam from source 110 after reflection by the circularizing mirrors 511 is received by the plane mirror 112 and deflected into the beam combiner 240. The divergent beam from sources 120, 130 after reflection by the circularizing mirrors 521, 531 respectively are received by the dichroic filters 122, 132 respectively of the beam combiner 240. In this architecture, the orientation of the deflector 112 and of the dichroic filters 122, 132 is such that the beams are received perpendicular to the drawing plane but deflected within the drawing and transmitted towards the common collimation lens 750 before exiting the optical module through the window 150. The functionality of such reflective, concave mirrors could be enhanced by also adding the functionality of the lenses 611, 621, 631 of Figure 12, in which case these reflective mirrors could not only circularize the input beams but also increase or decrease the beam divergence angles in both axes. Figure 23a and 23b are schematic plan views of a light source module according to another embodiment of the invention based on circularizing mirrors instead of lenses and may be compared with the lens-based embodiment of Figure 13. The circularizing mirrors 511, 521, 531 are positioned in the same plane as the emitters 110, 120, 130 and reflect circularized, collimated optical beams onto a different plane where the plane mirror 112 and beam combiner 240 are positioned. The plane mirror 112 and dichroic filters 122, 132 of the beam combiner 240 bring the beams into a common optical axis O for output from the module via a window 150. In this embodiment, the circularizing mirrors 511, 521, 531 generate beams that are both circularized and collimated, i.e. non-divergent, and therefore combine the functionality of the circularizing lenses 511, 521, 531 and the collimation lenses 111, 121, 131 of Figure 13. A variant of this approach would be to provide collimation lenses 111, 121, 131 as shown in Figure 13 but positioned between the emitters 110, 120, 130 and the circularizing mirrors 511, 521, 531 to provide the collimation, in which case the circularizing mirrors 511, 521, 531 would provide the circularization but not the collimation. Furthermore, in the above description, it has been mentioned that, for each beam, a first lens performs beam circularization followed by an optional second, lens that performs dual-axis manipulation of the combined output beams for collimation, real focusing or virtual focusing. Depending on the embodiment, the second lens may be common to all beams and arranged after the beam combiner or individual to each beam and arranged before the beam combiner. Moreover, in one embodiment each pair of first and second lenses is implemented as a single ellipsoid-shaped ball lens performing both circularization and collimation. The lenses of any of these embodiments can be realized with a so-called “free-form lens” that have a defined refractive power with an arbitrary and freely selective surface in either horizontal or vertical axis, or even a radially symmetric or asymmetric surface profile. Moreover, the paired combination of circularizing lens and collimation lens acting on each beam (either with individual emitter-specific collimation lenses as in Figure 13 ora common collimation lens arranged after the beam combiner as, for example, in Figure 10 or 12) could be realized by not having well-defined optical surface profiles in one or the other optical axis at the input or output side of the lens but by realizing a well-defined profile of the refractive index in one or the other optical axis at either the input side or output side of the lens. Lenses of this type are commonly referred to as graded-index (GRIN) lenses. Furthermore, the concept of a first circularizing lens and a second collimation lens (either with individual emitter-specific collimation lenses as in Figure 13 or a common collimation lens arranged after the beam combiner as, for example, in Figure 10 or 12) can be extended to so-called “metalenses” that have a flat optical surface instead of a concave or convex optical surface to generate a refractive optical power, for example to focus or collimate light beams. Metalenses are a new type of lenses with a flat metasurface made of dielectric or plasmonic materials and nanostructures that typically induce a phase change on a subwavelength range to modify the amplitude distribution or wavefront of an optical beam. The expectation is that metalenses will induce lower amounts of beam distortions and aberrations compared to traditional spheric or aspheric lenses, which might be important for certain applications, including display applications, where beam quality is important. Suitable light sources for light source modules according to embodiments of the invention include semiconductor emitters based on pn-junction emission such as LDs, SLDs and semiconductor optical amplifiers (SOAs). Commercially, the dominant materials system for current blue or green LDs and SLDs is based around gallium nitride and related materials, principally those in which gallium is partially or wholly substituted with aluminium and / or indium in the quaternary system GaAIInN. The structure of a blue or a green LD or SLD chip may be made from the GaAIInN materials system. In red LD and SLD chips, the semiconductor heterostructure may be made of one or a multiple number of light-emitting layers that are sandwiched between doped layers of different type. The active layers may contain In, Al, Ga, As or In, Al, Ga, P elements. P-type layers are above, towards the surface of the device. N-type layers are below, in between the light emitting layers and a substrate (GaAs). Both n- and p-layers may contain In, Al, Ga, As or In, Al, Ga, P elements. While all the illustrated embodiments show three emitters, all these designs are readily applicable to greater numbers of emitters with each extra emitter requiring one more dichroic filter to be added to the beam combiner. The designs may also be used with two emitters in which case only one dichroic mirror would be needed. Embodiments of the invention may provide light source modules with various combinations of sources. Particular combinations of interest are: • 3xSLDs(RGB) • 3 x LDs (RGB) • at least 1 LD and at least 1 SLD of another colour -> example 1: blue LD, green LD, red SLD -> example 2: blue LD, green SLD, red SLD • Any of the above RGB sources + 1 NIR SLD -> example 1: NIR SLD used for OCT imaging -> example 2: NIR SLD used for eye tracking (e.g., SLO) • Any of the above RGB sources + 1 NIR LD -> example: NIR LD used for MEMS scanner control or for eye tracking Figure 24 shows a wearable headset 600 in spectacles format incorporating a near-to-eye display projector. A housing 680 is integrated midway along each temple (i.e. arm) 640 and houses a light source module 50 embodying the invention. Combined RGB output beams emitted from the light source module 50 are directed to a beam scanning unit 690 which projects the output beams onto the glass surface 650. The beam scanning unit 690 may be based on a MEMS light engine. The RGB light beams are then reflected from the glasses to form of a virtual image in front of a wearer's eye, i.e. the left eye El or the right eye Er, through an outcoupling element. Projection in front of both eyes allows for additional imaging possibilities, such as stereoscopic imaging for 3D. The display projector can thus be incorporated into glasses, spectacles or another head-mountable unit. Figure 25 shows another kind of wearable headset 600 in a spectacles format incorporating a near-to-eye display projector. Here, the combined RGB output beams emitted from the light source module 50 are directed to a beam scanning unit 690 that steers the output beams into the glasses 660 through a suitable entry window. The glasses have embedded waveguide 5 structures to guide the RGB light beams and also have embedded optical elements, for example diffractive grating elements, to extract the RGB light beams from the glasses and redirect towards the wearer's eyes. It will be clear to one skilled in the art that many improvements and modifications can be made to the foregoing exemplary embodiments without departing from the scope of the 10 present disclosure.

Claims

1. A light source module comprising:a plurality of semiconductor sources, each configured to emit a beam at a different peak emission wavelength, wherein a property of the semiconductor sources is that their outputs beams are elliptical, each output beam being divergent with a more divergent, fast axis and a less divergent, slow axis, the fast and slow axes respectively corresponding to vertical and horizontal directions;beam circularizing elements arranged to transform respective ones of the beams from the semiconductor sources into beams that are circular as defined by having an ellipticity greater than 87% or near-circular as defined by having an ellipticity greater than 70%; anda beam combiner comprising one or more dichroic mirrors arranged to receive the beams from the beam circularizing elements and to bring the beams into a common optical axis for output from the light source module.

2. The module of claim 1, wherein each of the beam circularizing elements has refractive power in the fast and slow axes.

3. The module of claim 1, wherein each of the beam circularizing elements has refractive power only in the fast axis, each beam circularizing element acting to reduce divergence along the fast axis.

4. The module of claim 1, wherein each of the beam circularizing elements has refractive power only in the slow axis, each beam circularizing element acting to increase divergence along the slow axis.

5. The module of any preceding claim, wherein the beam circularizing elements are ellipsoid-shaped ball lenses, with the ellipsoid having a smaller diameter in the vertical, fastaxis direction and a larger diameter in the horizontal, slow-axis direction.

6. The module of any preceding claim, wherein after passage through the beam circularizing elements the beams have horizontal direction divergence angles that lie within a range of 1, 2, 3, 4 or 5 degrees of each other and vertical direction divergence angles that lie within a range of 1, 2, 3, 4 or 5 degrees of each other.

7. The module of any preceding claim, wherein a further property of the semiconductor sources is that their output beams have astigmatism in respect of their fast and slow axes, as defined by an offset between respective virtual emission points of the output beam in the horizontal direction and in the vertical direction, and wherein the beam circularizing elements are configured to provide for that the divergent beam formed by each of the beamcircularizing elements has an offset between their horizontal and vertical virtual emission points of less than at least one of 1,2, 3, 4, 5, 6, 7, 8, 9 and 10 micrometres.

8. The module of any preceding claim, further comprising:beam expanding or contracting lenses arranged in respective ones of the divergent beams output from the beam circularizing elements and before the beam combiner, the beam expanding or contracting lenses being configured to increase or decrease the beam crosssections.

9. The module of any preceding claim, further comprising:a common focusing element arranged in the common optical axis from the beam combiner to form an image from all the beams in a common image plane; orindividual focusing elements arranged in respective ones of the divergent beams output from the beam circularizing elements and before the beam combiner to form an image from all the beams in a common image plane.

10. The module of claim 9, wherein the common image plane:is located at a reference distance from the light source module;is located at a distance of infinity from the light source module to output collimated beams from the light source module; oris virtual so that the beams are divergent when output from the light source module.

11. The module of claim 9, wherein, the common image plane is located at a reference distance from the light source module and, in the common image plane, each output beam has a beam diameter that is no more than one of 1.3, 1.4, or 1.5 times the beam diameter at its beam waist.

12. The module of any preceding claim, wherein the semiconductor sources are any combination of:superluminescent light-emitting diodes; andlaser diodes.

13. The module of any preceding claim, wherein respective ones of the semiconductor sources have peak emission wavelengths with red, green and blue colours to provide an RGB light source module and optionally there is also a semiconductor source with a peak emission wavelength in the near-infrared.

14. The module of any preceding claim, further comprising one or more beam-shaping optical elements arranged in the common optical axis and after the beam combiner to reshape the beams before they are output from the light source module.

15. The module of any preceding claim, further comprising a beam-steering element5 arranged in the common optical axis after the beam combiner, the beam-steering element being configured and operable to controllably vary the direction which the beams are output from the light source module.

16. The module of any preceding claim, wherein each of the beam circularizing elements is a lens or lens combination.10   17. A display projector module comprising:a light source module according to any of claims 1 to 16; and a beam scanner for raster scanning the beams to form an image.

18. A vision system configured to be placed on a human head incorporating a display projector module according to claim 17.15

Images