Light engine structure, communication device and data center system

Abstract

The embodiment of the present application relates to a kind of light engine structure, communication equipment and data center system, the light engine structure, comprising: drive chip, the drive chip includes driving circuit and welding electrode, the drive chip is used in the way of gold wire bonding with substrate or PCB board connection;Flip VCSEL array, the flip VCSEL array is flip soldered on the welding electrode of the drive chip, and is electrically connected with the driving circuit;Lens array is set on the light path of the flip VCSEL array, for the light emitted by the flip VCSEL array is collimated or focused.

Description

Technical Field

[0001] This application relates to the field of optoelectronic packaging technology, and in particular to optical engine structures, communication equipment, and data center systems. Background Technology

[0002] With the rapid development of data centers, high-performance computing, and artificial intelligence applications, the demand for high-speed optical interconnect technology is increasing. Traditional optical engine structures typically use upright VCSEL chips and connect them to the driver circuit via wire bonding.

[0003] However, due to factors such as signal integrity, heat dissipation efficiency, package size, and coupling efficiency, this connection method makes it difficult to simultaneously ensure normal electrical connection and compact chip arrangement density in multi-channel array design, making it unsuitable for large-scale high-speed optical transmission arrays. Summary of the Invention

[0004] The main objective of this invention is to provide a light engine structure, communication device, and data center system, aiming to solve the aforementioned technical problems in the prior art.

[0005] In a first aspect, this application provides a light engine structure, including: A driver chip, comprising a driver circuit and a welding electrode, is used to connect to a substrate or PCB board via gold wire bonding. A flip-chip VCSEL array is flip-chip soldered onto the soldering electrode of the driver chip and electrically connected to the driver circuit. A lens array is disposed on the light output path of the flip-chip VCSEL array to collimate or focus the light emitted by the flip-chip VCSEL array.

[0006] In one embodiment, the optical engine structure further includes an optical fiber array, the light input end of which is coupled to the lens array for receiving optical signals processed by the lens array; Each VCSEL unit in the flip-chip VCSEL array corresponds one-to-one with each lens unit in the lens array, and each lens unit in the lens array corresponds one-to-one with each fiber unit in the fiber array.

[0007] In one embodiment, the VCSEL cells in the flip-chip VCSEL array adopt a flip-chip structure, the substrate of the VCSEL cell is located on the light-emitting side, and the substrate is provided with a light-emitting window.

[0008] In one embodiment, the driver chip further includes a control circuit for independently controlling the drive current of each VCSEL cell in the flip-chip VCSEL array.

[0009] In one embodiment, the lens array is fixed above the driver chip by an optical bracket, and a cavity is formed between the optical bracket and the driver chip to accommodate the flip-chip VCSEL array.

[0010] In one embodiment, the light-incident end face of the fiber array and the light-exiting end face of the lens array are precisely aligned using an optical coupling adhesive or a mechanical alignment structure.

[0011] In one embodiment, the fiber units in the fiber array are arranged in a linear array or a two-dimensional array.

[0012] In one embodiment, each lens unit in the lens array is a convex lens, a microlens, or an integrated optical lens.

[0013] In one embodiment, a heat dissipation substrate is further included, on which the driver chip is mounted, and the heat dissipation substrate is provided with pins or pads that are electrically connected to the driver chip.

[0014] In one embodiment, the heat dissipation substrate is a ceramic substrate, a metal substrate, or a silicon substrate.

[0015] Secondly, this application also provides a communication device, a host computer, and a motherboard; and The light engine structure as described in any of the above embodiments is connected to the motherboard.

[0016] In one possible implementation, the light engine structure is detachably connected to the motherboard.

[0017] Thirdly, this application also provides a data center system, including: Server racks; and, The communication device as described in any of the above, wherein the communication device is located in the cabinet.

[0018] The present invention has at least the following beneficial effects: The optical engine structure provided by this invention achieves direct electrical connection between the VCSEL array and the driver chip through flip-chip bonding, eliminating the parasitic inductance introduced by traditional wire bonding. This allows high-frequency modulation signals to be transmitted to the VCSEL units more accurately, supporting higher data transmission rates. In the flip-chip structure, the light-emitting area of ​​the VCSEL chip is close to the heat dissipation substrate of the driver chip, and heat can be directly conducted to the driver chip and the heat dissipation substrate through the solder bumps, significantly reducing thermal resistance and improving the reliability and operational stability of the device. Furthermore, flip-chip bonding eliminates the wiring space required by wire bonding, enabling high-density integration of multi-channel VCSEL arrays in a smaller area, which is beneficial for the miniaturization design of optical modules. On this basis, an external lens array collimates or focuses the emitted light from the VCSEL and precisely couples it with the fiber array, significantly improving optical coupling efficiency and reducing optical signal transmission loss. In summary, the optical engine structure provided by this invention achieves high-frequency signal transmission performance and structural stability by flip-chip bonding a flip-chip VCSEL array to the bonding electrodes of the driver chip, making it electrically connected to the driver circuit, and placing a lens array on the light output path of the flip-chip VCSEL array. Based on this, the driver chip is connected to the substrate or PCB board by gold wire bonding. This mutually complementary optical engine architecture can achieve modular high integration, high bandwidth, low parasitic capacitance and efficient thermal management, and is suitable for large-scale high-speed optical transmission arrays. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the optical engine structure in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of a communication device in one embodiment of this application; Figure 3 This is an exploded view of a communication device according to an embodiment of this application.

[0020] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish one element from another. For example, without departing from the scope of this application, a first client may be referred to as a second client, and similarly, a second client may be referred to as a first client.

[0023] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. "Multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. "Several" means at least one, such as one, two, etc., unless otherwise explicitly specified.

[0024] As described in the background section of this application, with the rapid development of data centers, high-performance computing, and artificial intelligence applications, the demand for high-speed optical interconnect technology is increasing. Traditional optical engine structures typically use upright VCSEL chips and connect them to the driving circuit via wire bonding.

[0025] However, due to factors such as signal integrity, heat dissipation efficiency, package size, and coupling efficiency, this connection method makes it difficult to simultaneously ensure normal electrical connection and compact chip arrangement density in multi-channel array design, making it unsuitable for large-scale high-speed optical transmission arrays.

[0026] In the upright structure, the heat generated in the light-emitting area needs to be conducted to the heat dissipation substrate through the substrate, resulting in high thermal resistance and low heat dissipation efficiency, which limits the power density of the device. In addition, in the upright structure, the light is emitted from the substrate side, requiring additional optical components for optical path adjustment, which increases system complexity and optical loss, and limits coupling efficiency.

[0027] In related technologies, there is a lack of optical engine architectures built on flip-chip technology, especially a lack of efficient optical coupling schemes under flip-chip structures. Therefore, how to achieve efficient optical coupling schemes under flip-chip architectures has become an urgent technical problem to be solved in this field.

[0028] Firstly, please refer to Figure 1 This application provides a light engine structure, which includes a driver chip, a driver circuit, and a bonding electrode. The driver chip is used to connect to a substrate or PCB board via gold wire bonding. A flip-chip VCSEL array is flip-chip bonded to the bonding electrode of the driver chip and electrically connected to the driver circuit. A lens array is disposed in the light emission path of the flip-chip VCSEL array and is used to collimate or focus the light emitted by the flip-chip VCSEL array.

[0029] The following section will provide an example of a driver chip. For example, the driver chip, as the core control unit of the optical emission module, can be manufactured on a silicon substrate using CMOS technology. The thickness of the driver chip can be 300-500µm, and the specific size is determined according to the scale of the VCSEL array and packaging requirements, with typical sizes ranging from 2mm×2mm to 10mm×10mm.

[0030] For example, the driving circuit includes multiple independent current driving channels, each corresponding to a VCSEL unit. Each driving channel includes a current source, a modulation circuit, and a switching circuit; wherein, the current source is used to provide a stable DC bias current; the modulation circuit is used to convert the input electrical signal into a modulated current; and the switching circuit is used to control the opening and closing of the channel. The operating bandwidth of the driving circuit is greater than 1.5 times the required data transmission rate; for example, for a data transmission rate of 50Gbps, the bandwidth of the driving circuit should be greater than 75GHz.

[0031] For example, the welding electrodes are distributed on the upper surface of the driver chip and can adopt a multi-layer metal structure. A typical structure is as follows: the bottom layer is titanium (Ti) or chromium (Cr) with a thickness of 50-100 nm, used to form good adhesion with the silicon substrate; the middle layer is nickel (Ni) or platinum (Pt) with a thickness of 100-200 nm, serving as a barrier layer to prevent metal diffusion; and the top layer is gold (Au) with a thickness of 1-3 µm, used to form a reliable metallurgical bond with the solder. The diameter of the welding electrodes is 50-150 µm, and the spacing matches the electrode spacing of the VCSEL unit, typically 150-500 µm.

[0032] The following is an example of a flip-chip VCSEL array. For example, a flip-chip VCSEL array can be composed of multiple VCSEL cells arranged in a linear array or a two-dimensional array. The number of VCSEL cells is determined according to application requirements, with typical configurations including 1×4, 1×8, 1×12, 4×4, and 8×8.

[0033] For example, each VCSEL unit can employ a vertical-cavity surface-emitting structure, including a substrate, a lower DBR layer, an active region, an upper DBR layer, and electrodes. The substrate is made of gallium arsenide (GaAs) material with a thickness of 100-300 µm. For an 850 nm wavelength VCSEL, the substrate is transparent to the emission wavelength. The lower DBR layer consists of 20-40 pairs of AlGaAs / GaAs or AlAs / GaAs alternating layers, each pair having a thickness of one-quarter of the emission wavelength and a reflectivity greater than 99.5%. The active region adopts a multi-quantum-well structure, consisting of 3-5 InGaAs or GaAs quantum wells and an AlGaAs barrier layer, with a total thickness of 100-300 nm. The emission wavelength can be adjusted in the range of 780 nm-1100 nm by adjusting the quantum well composition and thickness. The upper DBR layer consists of 15-30 pairs of AlGaAs / GaAs alternating layers, with a reflectivity of approximately 99%. The electrodes include p-type electrodes and n-type electrodes. The p-type electrodes are located on the upper DBR layer, and the n-type electrodes are located on the back side of the substrate or the side of the lower DBR layer.

[0034] In the flip-chip structure, the VCSEL cell is flipped 180 degrees so that the p-type electrode faces downward and connects to the bonding electrode of the driver chip, with the substrate located on the light-emitting side. The substrate has a light-emitting window, which can be achieved by depositing an anti-reflection coating (material such as SiO2 / TiO2) on the substrate surface or by using surface microstructures (such as subwavelength gratings) to reduce Fresnel reflection loss.

[0035] The flip-chip soldering process includes solder bump preparation, alignment, and reflow soldering. Solder bump preparation includes electroplating, vapor deposition, and ball bonding. In electroplating, the bump pattern is defined by photolithography in wafer-level electroplating equipment, and then solder is electroplated onto the electrodes. Commonly used solders include gold-tin alloy (melting point 280°C) and tin-silver-copper alloy (melting point 217°C). The bump diameter is 40-100µm, and the height is 20-60µm. In vapor deposition, a solder layer is vapor-deposited onto the electrodes using a mask, and then reflow soldering forms a hemispherical bump. In ball bonding, pre-made solder balls are placed on the electrodes and fixed by reflow soldering. The diameter of the solder balls can be 50-150µm.

[0036] During the alignment process, a high-precision flip-chip mounter can be used to align the flip-chip VCSEL array with the driver chip. Alignment accuracy requirements: XY direction alignment error less than 5µm, rotational alignment error less than 0.5 degrees. The alignment process can be achieved using a vision alignment system, by identifying alignment marks on the VCSEL cells and the driver chip.

[0037] During reflow soldering, the aligned components are placed in the reflow oven and heated according to a preset temperature profile. Taking gold-tin solder as an example, a typical reflow soldering temperature profile is: room temperature to 150°C, heating rate 2-3°C / second; 150°C to 280°C (peak temperature), heating rate 1-2°C / second; holding at 280°C for 30-60 seconds; then cooling to room temperature at a rate of 2-4°C / second. During reflow soldering, the solder melts and forms intermetallic compounds (such as AuSn, NiSn, etc.) with the welding electrodes, achieving reliable electrical connection and mechanical fixation upon cooling. The shear strength after soldering is greater than 20 g / bump, and the resistivity is less than 10 mΩ.

[0038] The following is an exemplary description of lens arrays. A lens array is positioned along the light output path of the flip-chip VCSEL array to collimate or focus the diverging light emitted by the VCSEL units. The lens array comprises multiple lens units, each corresponding one-to-one with a VCSEL unit in the flip-chip VCSEL array, with a correspondence accuracy requirement better than 10µm.

[0039] The optical parameters of the lens unit are designed based on the light-emitting characteristics of the VCSEL unit and the fiber coupling requirements. The following is an illustrative introduction to the optical parameters of the lens unit. The numerical aperture of the lens unit is larger than the divergence angle of the VCSEL unit, typically ranging from 0.3 to 0.6. The focal length of the lens unit is determined based on the distance from the light-emitting surface of the VCSEL unit to the lens unit and the required beam collimation or focused spot size. The lens unit can be a spherical lens, an aspherical lens, a gradient refractive index (GRIN) lens, or a diffractive optical element (DOE). Spherical lenses are simple to manufacture and low in cost, but suffer from spherical aberration; aspherical lenses can correct spherical aberration and improve optical performance, but are more difficult to manufacture; GRIN lenses are small in size and have low aberrations; DOEs have the advantages of being thin, lightweight, and highly integrable. The substrate material of the lens array can be silicon, glass (such as fused silica), plastics (such as PMMA, PC), or compound semiconductors (such as GaAs). The substrate thickness is determined based on the focal length and mechanical strength requirements, typically ranging from 0.5 to 2 mm.

[0040] In one embodiment, the optical engine structure further includes an optical fiber array, the light input end of which is coupled to the lens array for receiving optical signals processed by the lens array.

[0041] Each VCSEL unit in the flip-chip VCSEL array corresponds one-to-one with each lens unit in the lens array, and each lens unit in the lens array corresponds one-to-one with each fiber unit in the fiber array.

[0042] The input end of the fiber optic array is coupled to a lens array to receive the optical signal processed by the lens array. The fiber optic array consists of multiple fiber optic units, each corresponding one-to-one with a lens unit in the lens array. The fiber optic units can be single-mode or multimode fiber.

[0043] In scenarios where the fiber unit is a single-mode fiber, the core diameter can be 8-10µm and the cladding diameter is 125µm, which is suitable for long-distance, high-speed transmission. The typical model is SMF-28.

[0044] In scenarios where the fiber unit is multimode fiber, the core diameter is 50µm or 62.5µm and the cladding diameter is 125µm. It is suitable for short-distance, high-coupling-efficiency applications, with typical models being OM3 and OM4.

[0045] The fiber optic units can be arranged in a linear array (1×N) or a two-dimensional array (M×N); linear arrays are suitable for one-dimensional optical signal transmission, while two-dimensional arrays are suitable for high-density optical interconnection; the center-to-center pitch of adjacent fiber optic units is determined according to application requirements, and typical values ​​can be 127µm, 250µm or 500µm.

[0046] In a flip-chip VCSEL array, each VCSEL unit corresponds one-to-one with each lens unit in a lens array, and in turn, each lens unit in a lens array corresponds one-to-one with each fiber unit in an fiber array. This one-to-one correspondence ensures that the light emitted by each VCSEL unit is processed by its corresponding lens unit and then efficiently coupled into its corresponding fiber unit.

[0047] In this way, by integrating the VCSEL array, driver chip, lens array, and fiber array into one unit, the optical engine structure is miniaturized and highly integrated, making it suitable for space-constrained applications. The lens array collimates or focuses the emitted light from the VCSEL and precisely couples it with the fiber array, achieving high optical coupling efficiency. The one-to-one optical design ensures efficient transmission of optical signals and achieves low insertion loss. The modular design supports flexible configuration of the number of VCSEL units, which can be expanded to dozens or hundreds of channels according to application requirements, providing excellent scalability.

[0048] In one embodiment, the VCSEL cells in the flip-chip VCSEL array adopt a flip-chip structure, the substrate of the VCSEL cell is located on the light-emitting side, and the substrate is provided with a light-emitting window.

[0049] In a flip-chip structure, the substrate is located on the light-emitting side, and light is emitted from one side of the substrate. In this design, the substrate itself is the light-emitting window, thus eliminating the need for additional etching and simplifying the fabrication of the light-emitting window. The substrate can be made of GaAs material, with a thermal conductivity of approximately 55 W / (m·K). Heat can be rapidly conducted from the active region through the lower DBR layer and the substrate to the driver chip and the heat dissipation substrate, improving heat dissipation efficiency. The flat substrate surface facilitates the direct integration of microlenses or the deposition of anti-reflective coatings, making lens integration easier.

[0050] For example, the light-emitting window can be disposed on the substrate to reduce light reflection loss at the substrate-air interface. The light-emitting window can be implemented by an anti-reflective coating, surface microstructure, or tilted light-emitting surface; wherein, when the anti-reflective coating is deposited on the light-emitting surface of the substrate, the anti-reflective coating can be a single-layer film, a double-layer film, or a multi-layer film.

[0051] In single-layer film applications, a low-refractive-index material (such as SiO2, n≈1.46) with a thickness of one-quarter wavelength can be used, reducing reflectivity to 1%-2%. In double-layer film applications, a combination of high-refractive-index materials (such as TiO2, n≈2.3) and low-refractive-index materials (such as SiO2) can be used, reducing reflectivity to below 0.5%. In multilayer film applications, 3-5 alternating layers of high and low refractive-index materials can be used, reducing reflectivity to below 0.1% and achieving a bandwidth of over 100nm. Anti-reflective coating deposition processes include electron beam evaporation, sputtering, and plasma-enhanced chemical vapor deposition (PECVD).

[0052] In surface microstructure applications, subwavelength gratings or moth-eye structures can be fabricated on the substrate surface to reduce reflection through the gradient refractive index effect. The period of the surface microstructure is smaller than the emission wavelength, and the depth is 0.5-2 times the wavelength. This structure exhibits broadband anti-reflection properties and a large angular tolerance.

[0053] In applications with tilted light-emitting surfaces, the light-emitting surface of the substrate can be processed at a tilt angle (e.g., 8-15 degrees) to deflect the reflected light away from the incident light path and reduce return loss. This method is suitable for applications sensitive to return loss.

[0054] In this way, Fresnel reflection loss can be reduced from about 30% to less than 1% through the light-emitting window, significantly improving the light-emitting efficiency. The substrate is located on the light-emitting side, and the light has a certain waveguide effect when passing through the substrate, which can smooth the beam profile, improve the far-field distribution, and achieve the effect of improving beam quality. The flip-chip structure eliminates the fragile connection points of wire bonding, improves mechanical reliability and vibration resistance, and enhances reliability. The thermal resistance of the flip-chip structure is about 200-500K / W, which is much lower than that of the conventional structure (1000-2000K / W), supporting higher power operation and achieving the effect of reducing thermal resistance.

[0055] In one embodiment, the driver chip further includes a control circuit for independently controlling the drive current of each VCSEL cell in the flip-chip VCSEL array.

[0056] The control circuit is integrated onto the driver chip and works in conjunction with the driver circuit to achieve independent control of each VCSEL unit. The control circuit may include the following functional modules: Module 1: Digital-to-Analog Converter (DAC) Array Each VCSEL unit corresponds to one DAC channel, which converts externally input digital control signals into analog voltage signals. The DAC resolution is determined by the control accuracy requirements, typically 8-12 bits. For a 12-bit DAC, the current regulation accuracy can reach 0.025%.

[0057] The DAC update rate should be greater than the data modulation rate to ensure real-time control. Taking 50Gbps modulation as an example, the DAC update rate is greater than 100MHz.

[0058] Module 2, Current Driver Array Each DAC channel is followed by a current driver to convert the analog voltage signal into a drive current. The current driver includes: a voltage-to-current conversion circuit to convert the DAC output voltage into a reference current; a current mirror circuit to amplify the reference current into a drive current, with adjustable amplification; and a current switch to control the on / off state of the drive current. The current driver's output current range is 0-20mA, with an adjustment step of 5-20µA, and a bandwidth greater than 10GHz.

[0059] Module 3, Control Logic Unit The control logic unit is responsible for coordinating the working status of each channel and realizing functions such as channel addressing, status management, automatic power control, and temperature compensation.

[0060] In this way, independent current regulation reduces the dynamic range requirements of the receiver and improves channel consistency; independent switch control isolates faulty channels, improves the system's fault tolerance and maintainability, and enhances system reliability; the dynamic channel configuration function allows the system to adjust the number of working channels according to actual needs, optimizes the trade-off between power consumption and bandwidth, and supports engineers to configure flexibly according to their needs; automatic power control and temperature compensation functions ensure that the VCSEL unit operates in the best condition, avoids overdrive and thermal damage, and extends device life.

[0061] In one embodiment, the lens array is fixed above the driver chip by an optical bracket, and a cavity is formed between the optical bracket and the driver chip to accommodate the flip-chip VCSEL array.

[0062] refer to Figure 1 As shown, the optical bracket is used to fix the lens array and support it above the driver chip. The optical bracket includes the following parts: Part 1: Base The base is connected to the substrate or PCB board using methods such as bonding, soldering, or mechanical snap-fit. The height of the base is determined based on the height of the flip-chip VCSEL array and the focal length of the lens array. Let the height of the flip-chip VCSEL array be h1 (typically 150-300µm), and the focal length of the lens array 130 be f (typically 0.5-2mm). Then the base height h2 should satisfy: h2 = f - h1 + Δh, where Δh is the adjustment margin (50-200µm).

[0063] Part Two: Support Wall The support wall connects the base and the top plate, forming the sidewall of the cavity. The support wall can be a continuous ring structure or a column structure; in the continuous ring structure scenario, the support wall surrounds the flip-chip VCSEL array to form a closed or semi-closed cavity, providing good electromagnetic shielding and environmental protection; in the column structure scenario, the support wall consists of 3-4 columns, saving material and facilitating observation and alignment; the thickness of the support wall is determined according to mechanical strength requirements, typically 0.2-1mm. Vent holes can be provided on the support wall to release gases generated during welding or bonding processes.

[0064] Part Three: Top Slab The top plate is used to fix the lens array. The top plate has a window, the size of which is slightly larger than the effective area of ​​the lens array. The lens array can be fixed to the top plate by means of bonding, mechanical clamping, welding, etc.

[0065] A cavity is formed between the optical support and the driver chip to accommodate the flip-chip VCSEL array. The cavity serves functions such as protection, heat dissipation, and alignment reference.

[0066] Specifically, the cavity isolates the flip-chip VCSEL array from the external environment, preventing the effects of dust, moisture, and mechanical damage. The cavity can be filled with an inert gas (such as nitrogen) or dry air, and a desiccant can be added to maintain a dry internal environment. The air or filling gas within the cavity can conduct heat, aiding in the heat dissipation of the flip-chip VCSEL array. Heat dissipation holes or fins can be incorporated into the support walls to enhance convective heat dissipation. The size and shape of the cavity provide an alignment reference for the lens array and fiber array. µm-level alignment accuracy can be achieved by incorporating precision positioning features (such as positioning pins and V-grooves) on the optical support.

[0067] The cavity dimensions are determined based on the dimensions of the flip-chip VCSEL array and the heat dissipation requirements. Assuming the dimensions of the flip-chip VCSEL array are L×W (mm), the internal cavity dimensions should be (L+0.5)×(W+0.5)mm to provide sufficient assembly clearance and heat dissipation space.

[0068] In addition, integrated optical brackets, adjustable optical brackets, flexible optical brackets, and other solutions can be used, which will not be described in detail here.

[0069] In this way, the optical bracket provides a fixed mounting position for the lens array, eliminating the need for complex three-dimensional alignment and reducing assembly time from hours to minutes, thus simplifying the assembly process; the precision-machined positioning features improve alignment accuracy; the cavity protects the flip-chip VCSEL array from environmental influences, enhancing environmental adaptability; in addition, the optical bracket, which uses mechanical snap-fit ​​or adhesive bonding, can be disassembled, facilitating the replacement of damaged lens arrays or flip-chip VCSEL arrays.

[0070] In one embodiment, the light-incident end face of the fiber array and the light-exiting end face of the lens array are precisely aligned using an optical coupling adhesive or a mechanical alignment structure.

[0071] Among them, optical coupling adhesive is used to fill the gap between the light-incident end face of the fiber array and the light-outcident end face of the lens array, reduce Fresnel reflection loss, and fix the relative position of the two.

[0072] Specifically, the refractive index of the optical coupling adhesive should be between that of the lens array material and the fiber cladding, typically 1.4-1.5. For example, for silicon lenses (n≈3.5) and silica fibers (n≈1.46), a coupling adhesive with a refractive index of 1.46 can be selected. The optical coupling adhesive should have a transmittance greater than 95% at the emission wavelength, with no significant absorption or scattering. The optical coupling adhesive can be a UV-curable adhesive (curing time in the seconds range), a thermosetting adhesive (curing temperature 80-150°C), or a two-component adhesive (curing at room temperature). UV-curable adhesives are suitable for heat-sensitive devices, while thermosetting adhesives offer high bonding strength. The coefficient of thermal expansion of the optical coupling adhesive should match that of the lens array 130 and the fiber array 140 to avoid cracking or debonding due to temperature cycling. The glass transition temperature should be higher than the maximum operating temperature (e.g., 150°C).

[0073] The application process of optical coupling adhesive can include surface treatment, adhesive application, alignment, and curing. Surface treatment includes cleaning the light-emitting surfaces of the lens array and the light-receiving surfaces of the fiber array, removing dust, grease, and other contaminants. Adhesive application involves applying an appropriate amount of coupling adhesive to the light-emitting or light-receiving surfaces, with the adhesive layer thickness controlled at 10-50µm. Precise application can be achieved using a dispensing machine or screen printing. Alignment involves aligning the light-receiving surfaces of the fiber array with the light-emitting surfaces of the lens array, ensuring a one-to-one correspondence between each fiber element and each lens element. For UV-curable adhesives, UV light irradiation for 10-60 seconds can be used; for thermosetting adhesives, heating in an oven for 30-60 minutes is sufficient. Optical coupling adhesive eliminates air gaps between the lens array and the fiber array, reducing reflection loss. The cured coupling adhesive bonds the lens array and fiber array together, improving mechanical strength.

[0074] On the other hand, the mechanical alignment structure achieves precise alignment of the fiber array and the lens array through the features of precision machining, without the need for optical coupling adhesive or with only a small amount of adhesive for fixation.

[0075] Mechanical alignment structures include V-groove alignment, locating pin alignment, precision guide rail alignment, and MT (Mechanical Transfer) connector alignment. In V-groove alignment applications, V-grooves are machined on the optical support or independent substrate, and the fiber optic unit is placed within the V-groove and positioned by the sidewalls of the groove. The included angle of the V-groove is typically 60 degrees or 90 degrees, and the depth is determined by the fiber diameter (e.g., for a 125µm diameter fiber, a 60-degree V-groove depth is 108µm). The machining accuracy requirements for V-grooves are: groove width tolerance ±1µm, groove spacing tolerance ±1µm, and surface roughness Ra < 0.1µm. In locating pin alignment applications, locating pins are set on the lens array, and locating holes are set on the substrate of the fiber optic array. Alignment is achieved through the engagement of the pins and holes. The locating pin diameter is 0.5-1mm, and the locating hole diameter is 2-5µm larger than the locating pin to achieve a clearance fit. In precision guide rail alignment applications, precision guide rails (such as T-slots or dovetail slots) are set on the optical support, and sliders are set on the substrate of the fiber array. Alignment is achieved through the cooperation of the guide rail and slider, and the straightness of the guide rail is better than 1µm / 10mm. In MT connector alignment applications, the fiber array uses MT connectors, and the ferrule of the MT connector has precision guide pin holes. Corresponding guide pins are set on the substrate of the lens array, and alignment is achieved through the cooperation of the guide pins and guide pin holes. The alignment accuracy of the MT connector can reach ±1µm, and the mating life is greater than 500 cycles.

[0076] In addition, coupling can also be achieved through laser welding, optical contact, elastic coupling, etc., which will not be elaborated here.

[0077] In this way, optical coupling adhesive or mechanical alignment structures can improve coupling efficiency and reduce reflection loss; mechanical alignment structures avoid problems such as adhesive layer aging and cracking, thus enhancing reliability; pluggable structures such as MT connectors facilitate the replacement and maintenance of fiber optic arrays.

[0078] In one embodiment, the fiber units in the fiber array are arranged in a linear array or a two-dimensional array.

[0079] The parameters of the linear array include the number of optical fibers, fiber spacing, and arrangement accuracy. The number of optical fibers can be determined according to application requirements, with typical values ​​being 4, 8, 12, 16, and 24. The center-to-center spacing between adjacent fiber units is typically 127µm (0.5mm specification), 250µm (1mm specification), and 500µm (2mm specification). A 127µm spacing is suitable for high-density applications, while a 500µm spacing is suitable for low-crosstalk applications. The positional deviation of each fiber unit should be less than 1µm, and the angular deviation should be less than 0.1 degrees. Application scenarios for linear arrays include parallel optical interconnection, wavelength division multiplexing, and redundancy backup, which will not be elaborated here.

[0080] Two-dimensional arrays (M×N) arrange fiber units in a matrix, suitable for higher density applications. The parameters of a two-dimensional array include the array size M×N, row spacing and column spacing, and arrangement. Typical values ​​for the array size M×N are 4×4, 8×8, and 12×12. The row spacing and column spacing can be the same (square array) or different (rectangular array), with typical values ​​of 250µm or 500µm. The arrangement can be a square grid arrangement, a hexagonal close arrangement, or other optimized arrangements. Application scenarios for two-dimensional arrays include inter-chip optical interconnects, optical imaging, and optical switches, which will not be elaborated here.

[0081] In addition, fiber optic units can also be arranged in the form of ring arrays, irregular arrays, reconfigurable arrays, etc., which will not be elaborated here.

[0082] In this way, linear arrays are suitable for one-dimensional signal transmission, while two-dimensional arrays are suitable for high-density integration, and can be flexibly selected according to application requirements; two-dimensional arrays can arrange more fiber units in the same area, improving space utilization; and reasonable design of fiber spacing and arrangement can reduce crosstalk.

[0083] In one embodiment, each lens unit in the lens array is a convex lens, a microlens, or an integrated optical lens.

[0084] In addition, concave lenses, cylindrical lenses, prisms, and other types of lenses can also be used.

[0085] In one embodiment, a heat dissipation substrate is further included, on which the driver chip is mounted, and the heat dissipation substrate is provided with pins or pads that are electrically connected to the driver chip.

[0086] The heat dissipation substrate supports the driver chip and provides heat dissipation channels and electrical connection interfaces. The heat dissipation substrate has a structure consisting of a bottom layer, a middle layer, and a top layer. The bottom layer is the main body of the heat dissipation substrate, providing mechanical support and heat conduction. The material can be ceramic, metal, or silicon, and the thickness is 0.5-2mm. The middle layer is a metal wiring layer used for electrical signal transmission. It is prepared using thin film technology (sputtering, electroplating) or thick film technology (screen printing), and the line width / spacing is 10-100µm. The top layer is the pad or pin connected to the driver chip. The pad material is gold, nickel-gold, or nickel-palladium-gold, and the diameter is 50-200µm.

[0087] The heat sink substrate has pins or pads for connecting to external circuits. The pins can be in the form of a pin grid array (PGA), ball grid array (BGA), or planar grid array (LGA). The pads are connected to the corresponding electrodes of the driver chip through internal wiring.

[0088] In one embodiment, the heat dissipation substrate is a ceramic substrate, a metal substrate, or a silicon substrate.

[0089] Among them, ceramic substrates can be made of ceramic materials such as alumina (Al2O3), aluminum nitride (AlN) or beryllium oxide (BeO); metal substrates can be made of copper, aluminum or copper-molybdenum-copper (CMC) composite materials; silicon substrates are made of single-crystal silicon materials and can be fabricated with microstructures through MEMS processes.

[0090] In addition, diamond substrates, composite substrates, and liquid-cooled substrates can also be used, which will not be described in detail here.

[0091] In this way, the heat dissipation substrate quickly conducts the heat from the driver chip and the flip-chip VCSEL array to the external environment, thereby reducing thermal resistance; the good heat dissipation performance allows the module to operate at higher ambient temperatures, increasing the operating temperature range; by selecting appropriate substrate materials (such as aluminum nitride and silicon), thermal stress can be eliminated, solder joint fatigue failure can be prevented, and long-term reliability can be improved; the heat dissipation substrate provides standardized electrical connection interfaces (such as BGA and LGA), which facilitates integration with external circuit boards and simplifies system integration.

[0092] In some embodiments, please refer to Figure 2 and Figure 3 This application also provides a communication device, including: a host computer having a motherboard; and a light engine structure as described in the above embodiments, the light engine structure being connected to the motherboard.

[0093] The substrate has a connection end, and the light engine structure also includes a connection seat. The connection seat has a mounting groove, the substrate is housed in the mounting groove, and the connection end is electrically connected to the connection seat. The light engine structure is connected to an external motherboard through the connection seat.

[0094] In some embodiments, the inner wall shape of the mounting groove can be adapted to the outer contour of the substrate, serving as a limit and guide to prevent displacement or loosening of the substrate under vibration generated by high-speed signal transmission. The connector, as an intermediate transition component, not only achieves mechanical fixation between the optical engine structure and the external motherboard but also undertakes the function of transitioning electrical signals. By embedding the substrate into the mounting groove, the path length of signal transmission from the optical engine structure to the external motherboard can be shortened, reducing signal reflection and attenuation, which is crucial for maintaining the integrity of high-speed optical communication signals.

[0095] In some embodiments, the connector may be made of engineering plastic or ceramic material with good insulation properties to isolate electromagnetic interference between adjacent channels; in other embodiments, a shielding layer or grounding structure may be integrated inside the connector to further improve noise immunity.

[0096] In some embodiments, the connector may be provided with electrical connection points, specifically through, for example, elastic contacts on the connector base, or by direct welding between the connector and the connector base. When elastic contacts are used for connection, the connector base may be provided with spring pins or cantilever beam contact pads, which abut against the pads on the substrate connector. This non-permanent connection method allows the optical engine structure to be disassembled and replaced without damaging the substrate, facilitating subsequent maintenance, upgrades, or troubleshooting. The compression of the elastic contacts can be adjusted within a certain range to ensure the stability of the contact pressure, thereby ensuring low contact resistance and high conductivity reliability. Optionally, the surface of the elastic contacts may be gold-plated or palladium-plated to prevent oxidation and adapt to the skin effect of high-frequency signals. In contrast, if a direct welding method is used, such as fusing the solder balls of the substrate connector to the pads in the connector base via reflow soldering, lower parasitic inductance and a stronger mechanical bonding force can be obtained, suitable for applications with extremely high vibration tolerance requirements or where frequent plugging and unplugging is not required. However, welding is irreversible once completed. If the optical engine structure malfunctions, the entire connector may need to be replaced or complex repairs may be required. Therefore, the choice between flexible contact and welding fixation can be flexibly configured based on the actual usage environment, maintenance strategy, and cost considerations of the optical module.

[0097] In some embodiments, the connector has a positioning protrusion, and the positioning protrusion is at least partially located in the mounting groove. The substrate has a positioning groove, and the positioning protrusion engages with the positioning groove.

[0098] This positioning structure employs a "convex-concave" interlocking mechanical fit, where the shape of the positioning protrusion (such as cylindrical, square, or chamfered prism) precisely matches the inner cavity shape of the positioning groove. When the substrate is placed in the mounting groove, the operator or automated equipment simply aligns the positioning protrusion with the positioning groove and inserts it, using physical positioning to quickly determine the unique and correct position of the substrate within the mounting groove. This design not only restricts the substrate's translational freedom in the horizontal plane, preventing lateral or forward / backward displacement, but also limits the substrate's rotation around the vertical axis to a certain extent, ensuring precise alignment between the electrical connection points on the substrate and the elastic contacts or pads on the connector.

[0099] By designing a positioning protrusion that engages with a positioning groove, the substrate can be positioned during installation, significantly reducing reliance on the visual alignment accuracy of the assembly equipment and improving production assembly efficiency and yield. During the operation of the optical engine structure, this tight snap-fit ​​effectively resists minor displacements caused by external vibrations or thermal expansion and contraction, maintaining the stability of the electrical connection and preventing signal interruptions or increased bit error rates due to poor contact. Furthermore, the positioning protrusion can be designed with a foolproof structure (e.g., using an asymmetrical layout or a specific angled cut edge). If the substrate is placed in the wrong orientation or is of the wrong type, it will not be able to be fully embedded in the mounting groove, thus physically eliminating the risk of incorrect installation and protecting the expensive optical engine structure and substrate from damage. In some high-reliability embodiments, a small amount of thermally conductive adhesive or fixing adhesive can be filled between the positioning protrusion and the positioning groove to further enhance the heat dissipation path or mechanical locking force while achieving precise positioning.

[0100] In some embodiments, the connector is detachably connected to the motherboard.

[0101] This modular design integrates the substrate, optical engine structure, and connector into a single, independent optical engine structure. During system assembly, this integrated module can be quickly installed onto the motherboard at a designated location, like a plug-in, eliminating the need for complex micro-assembly or soldering operations on the motherboard. This significantly simplifies the production line process and shortens the manufacturing cycle. More importantly, this discrete architecture greatly facilitates later maintenance: when a channel in the optical engine structure fails or requires a bandwidth upgrade, technicians simply need to remove the module from the motherboard and replace it with a new one, without needing to rework or scrap the motherboard. This significantly reduces system maintenance costs and improves equipment availability.

[0102] Specifically, the motherboard has an interface section, and the connector can be detachably connected to the motherboard and electrically connected to the interface section to form a near-packaged optical NPO architecture. The interface section is configured to be close to the switching chip on the motherboard so that the electrical connection path length between the optical engine structure and the switching chip is less than a preset threshold.

[0103] The preset threshold can range from 5mm to 50mm, with specific intermediate values ​​being 10mm, 20mm, or 30mm. The interface section is located on the motherboard within a fan-shaped area with a radius of 50mm around the switching chip. Preferably, the interface section and the switching chip are directly connected through multi-layer blind vias inside the motherboard, avoiding long-distance traces through the backplane or intermediate adapter board.

[0104] In traditional NPOs, optical modules are typically fixed directly to the motherboard via methods such as soldering. While this permanent connection method offers low contact resistance and good high-frequency characteristics, if the laser or detector in the optical engine structure fails, maintenance requires removing the entire module from the motherboard using a hot air gun or reflow soldering equipment. The high temperatures generated during this process can easily damage other precision components on the motherboard, and the solder pads are difficult to clean after removal, often necessitating the replacement of the entire motherboard, resulting in high maintenance costs and long maintenance cycles. In contrast, the optical engine structure in this embodiment features a detachable connector to the motherboard, facilitating replacement. The connector and interface section utilize a board-to-board connector, spring-loaded contacts, or edge-locking mechanism for mechanical locking and electrical conduction. When maintenance is required, operators simply need to unlock the mechanical latch of the connector or loosen the fixing screws to pull the optical engine structure perpendicular to the motherboard. This eliminates the need for heating or damaging the motherboard's circuitry. After replacing the module, it can be reinserted to complete the repair, significantly improving the maintainability and availability of the data center system.

[0105] Specifically, the contacts of the connector are set to contact and conduct with the interface section. The interface section is correspondingly set as gold finger pads or elastic contact pads on the printed circuit board, and its width matches the contact spacing of the connector. During assembly, when the connector is inserted into the interface section, the contacts undergo elastic deformation to generate positive pressure, which ensures the reliability of the electrical connection, prevents instantaneous circuit breakage due to vibration, and avoids excessive pressure that could cause plastic deformation or premature wear of the contacts.

[0106] After high-speed electrical signals are emitted from the switching chip, they are directly transmitted to the interface section via short traces inside the motherboard, and then directly to the drive circuit of the optical engine structure through the contacts of the connector, without any intermediate cable adapters or connectors. This direct-connect architecture eliminates the signal attenuation and reflection problems caused by long-distance copper cables or printed circuit board traces between the motherboard and the front panel in traditional pluggable optical module solutions.

[0107] Specifically, the connector is connected to the motherboard via a detachable structure. The detachable structure includes a positioning hole and a positioning part. The positioning hole and the positioning part are detachably connected. Either the positioning hole or the positioning part is located on the connector, and the other positioning hole or the positioning part is located on the motherboard.

[0108] The positioning holes and positioning parts work together to form a high-precision mechanical guiding system. During installation, the positioning parts (such as cylindrical pins, tapered pins, or chamfered protrusions) are first inserted into the positioning holes. The interference fit or clearance fit between the two guides the connector to its accurate placement, ensuring precise alignment between the optical engine structure and the electrical interfaces on the motherboard (such as gold finger sockets or board-to-board connectors), preventing pin bending or poor contact due to misalignment. This positioning method not only restricts the module's horizontal freedom but also withstands a certain amount of shear force, preventing the module from loosening during device operation vibrations. Depending on design requirements, the positioning holes can be located on the bottom or side of the connector, while the positioning parts are correspondingly located on the motherboard, and vice versa. This flexible layout allows engineers to optimize the structural design based on the motherboard's space constraints and wiring configuration.

[0109] In other embodiments, the connector and the motherboard can also be combined using detachable connection methods such as screw fixing or plug-in connection. If screw fixing is used, through holes can be made on the flange edge or specific mounting ears of the connector, and screws can be used to lock it onto the threaded post or nut seat of the motherboard. This method provides the strongest mechanical bonding force and excellent shock resistance, making it particularly suitable for high-vibration environments such as automotive and aerospace applications. However, it requires tools for disassembly and assembly, and is relatively slow. If plug-in connection (such as board-to-board connectors or edge slot connectors) is used, true "tool-free" hot-swapping functionality can be achieved: the bottom of the connector integrates a male connector, which can be directly inserted into the female connector on the motherboard, or a flexible snap-fit ​​structure can be used to lock the module into the motherboard slot. Although the mechanical strength of plug-in connection is slightly lower than screw fixing, its extreme convenience makes it the preferred solution for scenarios requiring frequent hardware replacement, such as data center servers and switches. Furthermore, multiple methods can be combined for hybrid fixing, such as initial alignment using positioning holes and positioning parts, followed by final locking with screws or snap-fits, to balance installation accuracy, connection reliability, and maintenance convenience.

[0110] In some embodiments, the connector is provided with first elastic contacts, through which the connector is electrically connected to the connection end. These first elastic contacts typically employ spring pins, cantilever beam springs, or crown spring structures, and are installed on the inner wall or bottom of the mounting groove of the connector, forming pressure contact with the solder pads or conductive contacts on the substrate connection end. When the substrate is inserted into the mounting groove, the first elastic contacts are compressed to generate a preload, ensuring stable electrical conduction under the low contact resistance required for high-speed signal transmission. The greatest advantage of this non-permanent connection method is its ease of disassembly and assembly: when maintaining or replacing the optical engine structure, there is no need to use a soldering iron for desoldering, avoiding thermal shock and potential damage to the precision optical engine structure on the substrate caused by high temperatures. It also eliminates the risk of solder pad detachment due to repeated soldering, greatly improving the maintainability and service life of the module.

[0111] In this embodiment, electrical connection between the first elastic contact and the substrate is achieved simply by contacting the first elastic contact at the connection end. This facilitates easy assembly and disassembly, allowing for fully automated assembly of the substrate and connector, and providing some tolerance for positional tolerances. The deformation stroke of the elastic contact can compensate for minor deviations in substrate thickness or mounting position, ensuring reliable contact. Furthermore, for high-speed signal transmission, the design of the first elastic contact typically considers impedance matching. By controlling the contact's length, diameter, and the surrounding dielectric environment, parasitic inductance and capacitance are minimized, thereby reducing signal reflection and insertion loss, and ensuring the integrity of 400G, 800G, and even higher speed signals.

[0112] Specifically, the connector is provided with a second elastic contact, and the connector is electrically connected to the motherboard through the second elastic contact.

[0113] The second flexible contact is typically located at the bottom or side interface of the connector, used for flexible crimping with the corresponding pads, gold fingers, or sockets on the motherboard. This design transforms the connection between the connector and the motherboard into a detachable, flexible contact mode, thus creating a fully solderless connection architecture from "motherboard-connector-substrate". When the entire optical engine structure needs to be replaced, simply release the mechanical locking (such as screws or clips) to directly remove the connector from the motherboard, and the second flexible contact will disconnect. The entire process is quick, clean, and requires no special tools.

[0114] This design facilitates the easy assembly and disassembly of the connector and motherboard, enabling true modular hot-swappable or cold-swappable functionality. In applications such as data centers or communication base stations, this design allows for the individual replacement of faulty optical modules without interrupting system power or disassembling other peripheral components, significantly reducing service interruption time. Simultaneously, the dual-ended elastic contacts (first and second elastic contacts) also serve as stress buffers: when the device is subjected to external vibration or impact, the elastic contacts can absorb some mechanical energy, preventing substrate breakage or solder joint cracking caused by rigid connections, further improving the operational reliability of the optical engine structure in harsh environments. To optimize high-frequency performance, the second elastic contact also undergoes impedance control and shielding to ensure continuous and stable transmission characteristics throughout the signal path from the motherboard to the optical engine structure.

[0115] Specifically, the connector can be an LGA socket. The use of an LGA socket leverages its high-density planar contact characteristics to accommodate the multi-channel signal transmission requirements of the optical engine structure. In this configuration, the connector integrates an array of flexible contact terminals that correspond one-to-one with the pad positions on the substrate connection end. These terminals are typically encapsulated within an insulating substrate, with only the contact surface exposed. When the substrate is inserted into the mounting slot and pressed firmly, the conductive pads on the bottom of the substrate form surface contact with the flexible terminals within the LGA socket, enabling conductive communication. This connection method avoids the risk of pins bending and breaking as with traditional pin-type connectors. Furthermore, because the LGA socket has no protruding pins, its overall height is lower, which helps reduce the space occupied by the optical engine structure on the motherboard and meets the strict vertical height limitations of rack-mounted equipment.

[0116] In some embodiments, the resilient terminals within the LGA socket can be designed as crown springs, cantilever beams, or hyperboloid structures to provide stable contact pressure and low contact resistance. For 400G, 800G, and even higher-speed optical communication signals, the LGA socket layout can optimize the arrangement of signal and ground pins, for example, by using staggered arrangements or surrounding the ground pins around high-speed signal pairs, to create a good return path and reduce crosstalk and electromagnetic radiation. Furthermore, as a standardized interface component, the LGA socket can be easily sourced from the supply chain, reducing the mold-making costs and manufacturing cycle of custom connectors. If different substrate sizes or pin definitions are required, only the corresponding LGA socket needs to be replaced without redesigning the entire connector's mechanical structure, facilitating the product series expansion of optical engine structures. In certain high-reliability applications, the LGA socket can also be equipped with locking latches or clamping mechanisms to apply additional clamping force after the substrate is installed, preventing momentary disconnection of contact due to equipment vibration and ensuring continuous signal transmission.

[0117] Furthermore, the light engine structure also includes a heat dissipation structure, which is connected to the substrate and thermally coupled to the light engine structure.

[0118] This heat dissipation structure can be made of materials with high thermal conductivity, such as oxygen-free copper, aluminum alloy, silicon carbide, or diamond composite materials. Its form can be designed as a vapor chamber, microchannel liquid cooling plate, or heat sink with dense fins, all closely attached to the chip surface. Since optical engine structures (especially high-speed laser chips and detector chips) generate significant heat during operation and are extremely sensitive to temperature (higher temperatures lead to wavelength drift, increased threshold current, and shortened lifespan), the heat dissipation structure directly covers or closely adheres to the heat-generating surface of the optical engine structure. This allows for the rapid conduction of Joule heat generated by the chip to the substrate with extremely low thermal resistance, and then conduction through the substrate to the connector or external environment, thereby controlling the chip junction temperature within a safe operating range.

[0119] Compared to traditional wire bonding solutions for chips, this solution improves the arrangement of the optical engine structure by allowing for additional heat dissipation structures to enhance its heat dissipation. In traditional flip-chip or wire bonding packages, chips are often encased in molding compound or limited by the space of the lead frame, making it difficult to directly mount an efficient external heat sink on top of the chip. This results in heat dissipation primarily through a single path via the bottom substrate, creating a significant heat dissipation bottleneck. However, the advanced integration architecture used in this embodiment (such as wafer-level bonding or Flip-Chip combined with TSV technology) fully exposes or makes the top surface of the optical engine structure easily accessible, breaking the space limitations of traditional packaging. This layout allows engineers to directly "tailor-make" a heat dissipation structure directly above the chip, creating a bidirectional three-dimensional heat dissipation channel: "chip top -> heat dissipation structure -> environment" and "chip bottom -> substrate -> environment." This bidirectional heat dissipation path significantly reduces overall thermal resistance and increases the heat dissipation power density per unit volume. This enables the optical engine structure to support stable operation at higher power consumption and higher speeds (such as 800G / 1.6T) within a smaller size, effectively solving the thermal management challenges brought about by high-density integration.

[0120] Furthermore, the thermal coupling interface between the heat dissipation structure and the light engine structure is typically filled with high-performance thermally conductive interface materials, such as thermally conductive gels, phase change materials, or indium foil, to fill the voids in the microscopic surface, eliminate the air insulation layer, and further improve heat transfer efficiency. In some high-end embodiments, the heat dissipation structure can also be integrated with the connector, utilizing the connector as a large heat sink, and even designing air ducts or water-cooling channels on its surface to achieve active heat dissipation, ensuring the performance consistency and long-term reliability of the light engine structure under long-term high-load operation.

[0121] Specifically, the heat dissipation structure includes a heat sink and a heat dissipation pad. The heat sink is at least partially attached to the substrate, and the heat dissipation pad is located on the side of the heat sink away from the substrate.

[0122] This layered composite structure ingeniously combines structural support and thermal management functions. The heat sink, serving as the core framework, has its bottom surface precisely machined to achieve a large-area, tight fit with the upper surface of the substrate. This fit not only establishes an efficient lateral heat diffusion path from the light engine structure through the substrate to the heat sink, utilizing the heat sink's large surface area to quickly and evenly distribute heat from localized hotspots, preventing heat accumulation beneath the chip; simultaneously, the heat sink acts like a "reinforcing rib" or "cover plate" covering the substrate, significantly improving the overall mechanical rigidity and bending resistance of the light engine structure. This effectively prevents warping and deformation problems common in large-size substrates or thinner designs, thus protecting the delicate internal micro-circuits and light engine structure from mechanical stress damage.

[0123] The placement of heat sinks in conjunction with the substrate facilitates heat dissipation from the light engine structure and enhances its overall strength. The heat dissipation pad, located on top of the heat sink (the side furthest from the substrate), is typically made of thermally conductive silicone, graphite sheets, or phase change materials with high compressibility resilience. Its primary function is to fill the minute gaps between the top of the heat sink and the outer casing, duct walls, or other heat dissipation components. Due to machining and assembly tolerances, it's difficult to achieve perfect flatness between the heat sink surface and the external contact surface. The heat dissipation pad eliminates these air gaps (air is a poor conductor of heat) through its own deformation, ensuring unimpeded heat transfer from the heat sink to the external environment. Furthermore, the soft heat dissipation pad also acts as a cushioning and shock absorber, absorbing external impacts and preventing damage to components caused by hard contact.

[0124] In one embodiment, the heat sink can be made of copper. Copper (especially oxygen-free copper) is chosen as the material for the heat sink due to its superior thermophysical properties: copper has a thermal conductivity of approximately 380-400 W / (m·K), far exceeding that of aluminum alloys (approximately 200 W / (m·K)) and other common structural materials. This means that copper heat sinks can conduct the high heat generated by the light engine structure to the entire module surface at an extremely fast speed, significantly reducing thermal resistance, making it particularly suitable for light engine structures with extremely high power density. Although copper has a higher density, increasing module weight, its excellent thermal conductivity and good ductility (facilitating the fabrication of complex bonding surfaces) make it the preferred choice for high-performance heat dissipation solutions. To balance weight and cost, other embodiments may also employ copper-aluminum composite materials (such as copper base and aluminum fins) or copper plating on aluminum surfaces to achieve the best balance between thermal conductivity and lightweight design. Furthermore, the surface of copper heat sinks is typically nickel-plated to prevent oxidation and improve interfacial adhesion with the heatsink.

[0125] In one embodiment, the heat dissipation structure has a connection hole, and the light engine structure also includes a coupling element that passes through the connection hole.

[0126] By incorporating a coupling element, the optical engine structure can emit or receive laser light. As a medium for optical signal transmission, the coupling element efficiently guides the high-speed modulated optical signal generated by the optical engine structure into an external fiber optic network, or precisely focuses externally input optical signals onto the photosensitive area of ​​the optical engine structure. During optical coupling, the lens structure or end-face shape within the coupling element can collimate, focus, or match the mode field of the beam to improve coupling efficiency and reduce insertion loss. Furthermore, the coupling element also provides isolation and protection, preventing damage to the chip from dust, moisture, or mechanical contact. Optionally, a transparent thermally conductive adhesive or cured UV adhesive can be filled between the coupling element and the connection hole, achieving both optical path sealing and heat transfer from the coupling element to the heat dissipation structure, thus improving local heat dissipation performance.

[0127] Specifically, the coupling element includes a connecting housing and an optical fiber, with the optical fiber located inside the connecting housing and the connecting housing passing through the connecting hole.

[0128] The connecting housing is typically made of metal, ceramic, or high-strength engineering plastic. Its outer diameter forms a clearance fit or interference fit with the inner diameter of the connecting hole, facilitating alignment during assembly and ensuring structural stability after assembly. The optical fiber is fixedly encapsulated within the central through-hole or pre-formed groove of the connecting housing, and its end is ground and polished to form a planar, spherical, or lens-shaped end face. In some embodiments, a microlens array or aspherical lens can also be integrated inside the connecting housing, with the end face of the optical fiber located near the focal plane of the lens. This lens system shapes the laser beam to accommodate external transmission fibers with different core diameters or numerical apertures. The design of the connecting housing passing through the connecting hole ensures that the stress point of the optical fiber is located outside the housing, preventing installation stress from being directly transferred to the fragile optical fiber and improving the reliability of the optical path connection.

[0129] The present invention also provides a communication device, which includes a host and an optical engine structure in any of the above embodiments; the host is provided with a motherboard; the optical engine structure is connected to the motherboard.

[0130] Specifically, the motherboard typically has connector slots or solder pads that match the optical engine structure. The optical engine structure achieves mechanical fixation and electrical connection with the motherboard through methods such as gold finger insertion, board-to-board connector crimping, or direct surface mount soldering. This connection method allows the optical engine structure to directly connect to the host's data exchange chip or network processor, establishing a high-speed data transmission and reception channel. In some embodiments, the motherboard also integrates power management circuitry and signal conditioning circuitry to provide a stable operating voltage for the optical engine structure and to perform impedance matching and noise filtering on input and output signals to ensure the stability of the communication link. The communication equipment can be network devices requiring high-bandwidth data exchange, such as data center switches, routers, servers, base station gateways, or optical transmission terminals. Their internal casings are typically designed with airflow channels or heat sinks to cooperate with the optical engine structure's own heat dissipation structure for system-level thermal management.

[0131] Furthermore, the light engine structure is detachably connected to the motherboard.

[0132] This detachable connection can be implemented by inserting the board edge gold fingers into a slot on the motherboard, or by using intermediate connectors such as board-to-board connectors or spring-loaded contacts to achieve electrical conductivity and mechanical fixation. In some embodiments, the housing of the optical engine structure may also be provided with a locking mechanism, such as a pull ring, a clip, or a screw fixing hole, which cooperates with the corresponding structure on the host chassis or motherboard to prevent accidental loosening of the device during vibration or handling. When maintenance is required, the operator only needs to release the locking mechanism and pull out the optical engine structure axially to separate it from the motherboard without using soldering tools or disassembling the internal structure of the entire communication device. Optionally, the connector interface on the motherboard can be designed to support hot-swapping, that is, while the communication device is powered on, the optical engine structure can be directly inserted or removed. The system can automatically identify the access or removal status of the module and perform corresponding configuration resets or link switching, thereby ensuring uninterrupted service operation.

[0133] The detachable connection between the optical engine structure and the motherboard facilitates the assembly and subsequent disassembly and maintenance of the optical engine structure. During assembly, the detachable design allows the optical engine structure to be produced and tested as an independent standard component, and then installed onto the motherboard after passing the test. This simplifies the overall assembly process of the host machine, reduces reliance on precision alignment equipment on the production line, and improves production efficiency and yield. If a defect is found in the optical engine structure during host assembly, a new module can be directly replaced without scrapping the entire motherboard or performing complex rework, effectively reducing manufacturing costs. In the later maintenance phase, when the optical engine structure reaches the end of its service life, experiences performance degradation, or suffers hardware failure, maintenance personnel can quickly locate and replace the faulty module, significantly shortening the mean time to repair (MTBL) and improving the availability of the communication equipment. Furthermore, with the iterative upgrades of optical communication technology, users can flexibly replace optical engine structures with different rates, transmission distances, or wavelengths according to actual needs without replacing the entire host machine or motherboard. This modular design gives the communication equipment excellent scalability and technical compatibility.

[0134] The present invention also provides a data center system, which includes a server rack and the communication equipment in any of the above embodiments, wherein the communication equipment is disposed in the server rack.

[0135] Specifically, the cabinet can accommodate multiple vertically arranged mounting positions, where communication equipment is secured via rails, trays, or brackets. The cabinet not only provides physical support and protection for the communication equipment but also creates a system-level cooling duct and fiber optic cabling management space. Fiber optic cable trays, cable management rings, and cable routing holes can be pre-installed on the cabinet's internal back panel or side walls. These cabling structures correspond to the fiber optic exit positions of the optical engine structure on the communication equipment, allowing couplers leading from the optical engine structure to directly and via a short path into the cabinet's cabling channels without requiring long-distance copper cable splicing within the cabinet.

[0136] Because the communication equipment adopts a near-package optical NPO architecture, the electrical connection path between the switching chip on the motherboard and the optical engine structure is limited to the board level. This means that high-frequency electrical signals do not need to be transmitted to the front panel of the rack and converted into optical signals through backplane connectors as in traditional architectures. Therefore, the space utilization inside the rack is significantly optimized in this data center system, eliminating a large amount of space occupation and reducing the risk of cable tangling and airflow resistance inside the rack. The airflow design of the rack can be optimized for the heat dissipation structure of the communication equipment. For example, air guides can be installed at the corresponding positions of the heat dissipation pads of the communication equipment, allowing cool air to flow directly over the surface of the heat sink. The negative pressure generated by the rack fan assembly quickly expels heat from the rack, thereby supporting the deployment of equipment with higher power density.

[0137] In some embodiments, a data center system may include multiple racks arranged side-by-side, with communication devices within each rack interconnected via a main fiber optic patch panel at the top or bottom of the rack. Because individual communication devices integrate high-density NPO optical interfaces, the entire data center system can achieve a greater level of data exchange capacity within a limited footprint. This NPO-based data center system effectively addresses the bottlenecks in energy consumption, heat dissipation, and signal integrity faced by hyperscale data centers by placing the optical engine structure as close as possible to the switching chip and combining rack-level optical cabling optimization.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these all fall within the protection scope of the embodiments of this application. Therefore, the protection scope of the patent for the embodiments of this application should be determined by the appended claims.

Claims

1. A light engine structure, characterized in that, include: A driver chip, comprising a driver circuit and a welding electrode, is used to connect to a substrate or PCB board via gold wire bonding. A flip-chip VCSEL array is flip-chip soldered onto the soldering electrode of the driver chip and electrically connected to the driver circuit. A lens array is disposed on the light output path of the flip-chip VCSEL array to collimate or focus the light emitted by the flip-chip VCSEL array.

2. The light engine structure according to claim 1, characterized in that, The optical engine structure also includes an optical fiber array, the light input end of which is coupled to the lens array for receiving the optical signal processed by the lens array; Each VCSEL unit in the flip-chip VCSEL array corresponds one-to-one with each lens unit in the lens array, and each lens unit in the lens array corresponds one-to-one with each fiber unit in the fiber array.

3. The light engine structure according to claim 1, characterized in that, The VCSEL cells in the flip-chip VCSEL array adopt a flip-chip structure, and the substrate of the VCSEL cell is located on the light-emitting side, and the substrate is provided with a light-emitting window.

4. The light engine structure according to claim 3, characterized in that, The driver chip also includes a control circuit, which is used to independently control the drive current of each VCSEL unit in the flip-chip VCSEL array.

5. The light engine structure according to claim 1, characterized in that, The lens array is fixed above the driver chip by an optical bracket, and a cavity is formed between the optical bracket and the driver chip to accommodate the flip-chip VCSEL array.

6. The light engine structure according to claim 2, characterized in that, The light-incident end face of the fiber array and the light-exiting end face of the lens array are precisely aligned using optical coupling adhesive or a mechanical alignment structure.

7. The light engine structure according to claim 2, characterized in that, The fiber units in the fiber array are arranged in a linear array or a two-dimensional array.

8. The light engine structure according to claim 1, characterized in that, Each lens unit in the lens array is a convex lens, a microlens, or an integrated optical lens.

9. The light engine structure according to claim 1, characterized in that, It also includes a heat dissipation substrate, on which the driver chip is mounted, and the heat dissipation substrate has pins or pads that are electrically connected to the driver chip.

10. A communication device, characterized in that, include: The host computer is equipped with a motherboard; as well as, The light engine structure as described in any one of claims 1-9, wherein the light engine structure is connected to the motherboard.

11. The communication device according to claim 10, characterized in that, The light engine structure is detachably connected to the motherboard.

12. A data center system, characterized in that, include: Server racks; and, The communication device as described in claim 10 or 11, wherein the communication device is disposed in the cabinet.

Images