Visual positioning and precision mounting system for ultra-small pad micro components

Abstract

The application discloses a super-small-pad micro-component visual positioning and precision mounting system and relates to the field of visual alignment. The system comprises an imaging module, a processing module and the like. The imaging module is used for applying a space-restricted light field to the pad area of a to-be-mounted substrate and the micro-component to be mounted, and arrayed imaging original data of the corresponding area is synchronously collected. The processing module is used for receiving the arrayed imaging original data, performing global normalization processing on the data, and generating a standardized imaging mapping data set. Through coaxial projection and synchronous collection of the space-restricted light field, complete imaging data covering the full depth of field of the pad and the full profile of the component can be obtained in a single exposure period, effectively solving the industry pain points of insufficient depth of field and difficulty in extracting edge features in micro-component mounting.

Description

Technical Field

[0001] This invention relates to the field of visual alignment technology, specifically to a precision mounting system for visual positioning of micro-components with ultra-small pads. Background Technology

[0002] As the electronics manufacturing industry rapidly iterates towards miniaturization and high-density integration, the demand for mounting micro-components with dimensions of 01005 and below, and ultra-small pitch pads, continues to surge. As a core process in SMT and electronic packaging, mounting accuracy, mounting efficiency, and yield directly determine the performance and reliability of end products. Machine vision-based positioning technology has become a core support for overcoming the mounting bottlenecks of such components.

[0003] However, existing placement systems for micro-components with ultra-small pads generally suffer from problems such as visual feature extraction being easily interfered with by pad reflections, insufficient sub-pixel positioning accuracy, and excessive deviations in pad alignment and coaxiality during high-speed placement, making it difficult to adapt to the large-scale precision placement requirements of high-density arrays of ultra-small pads.

[0004] To address this, we propose a precision mounting system for micro-components with ultra-small pads based on visual positioning. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a precision mounting system for visual positioning of micro-components with ultra-small pads, which can effectively solve the problems of the existing technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0007] This invention discloses a precision mounting system for visual positioning of micro-components with ultra-small pads, comprising:

[0008] The imaging module applies a spatial constraint light field to the pad area of ​​the substrate to be mounted and the micro-components to be mounted, and simultaneously acquires array-type imaging raw data of the corresponding area; the processing module receives the array-type imaging raw data, performs global normalization processing on the data, and generates a standardized imaging mapping dataset; the calculation module receives the standardized imaging mapping dataset, extracts the corresponding pose parameters of the pads and micro-components, and generates a pose parameter sequence; the compensation module obtains the pose parameter sequence, calculates the coaxial alignment compensation amount corresponding to the mounting path, and generates a compensated mounting alignment parameter set; the generation module calls the compensated mounting alignment parameter set from the compensation module to generate multi-degree-of-freedom motion control commands for the mounting execution end; the control module receives the multi-degree-of-freedom motion control commands, drives the mounting execution end to complete the bonding operation between the micro-components and the substrate pads, synchronously transmits back real-time status data of the mounting action, and completes the mounting qualification verification;

[0009] The imaging module is interconnected with a processing module via a local area network. The processing module is interconnected with a solution module and a compensation module via a local area network. The solution module and the compensation module are interconnected via a local area network. The compensation module is interconnected with a generation module via a local area network. The generation module is interconnected with a control module via a local area network.

[0010] Furthermore, the imaging module is integrated with a coaxial illumination unit, a spatial light modulation unit, an array-type photosensitive acquisition unit, and a synchronous triggering unit;

[0011] The spatial light modulation unit is used to generate a spatially constrained light field with a gradient distribution along the optical axis. The coaxial illumination unit coaxially projects the spatially constrained light field onto the imaging surface of the substrate pad area and the micro-component to be mounted. The synchronous triggering unit synchronously drives the light field modulation timing of the spatial light modulation unit and the exposure acquisition timing of the array-type photosensitive acquisition unit, so that the array-type photosensitive acquisition unit can acquire array-type imaging raw data covering the full depth range of the pad and the full contour range of the component within a single exposure cycle.

[0012] Furthermore, during the processing module's operation phase, dark level correction and light field response distortion correction are performed on the original array imaging data. Then, a global normalized mapping model is constructed based on the gradient distribution parameters of the spatially constrained light field. The corrected imaging data is then converted into a standardized imaging mapping dataset through the mapping model.

[0013] The global normalized mapping model expression is as follows:

[0014] ;

[0015] In the formula: The normalized imaging mapping value at spatial coordinates (x, y, z); The grayscale values ​​of the original array imaging data after correction at the corresponding coordinates; This is the reference value for the dark level at the corresponding coordinates; This refers to the photosensitive unit response calibration coefficient at the corresponding coordinates; The gradient distribution coefficient of the spatially constrained optical field along the optical axis z-direction; This is the fixed bias compensation value for the system.

[0016] Furthermore, when performing dark level correction and light field response distortion correction on the raw array imaging data, the following conditions must be met:

[0017] Dark level correction is performed when the array-type photosensitive acquisition unit is in a state of complete darkness with no incident light. A preset number of dark field imaging data are acquired, and the grayscale output value of each photosensitive pixel unit under the corresponding exposure parameters is statistically analyzed to obtain the dark level reference value for each pixel unit. During the imaging acquisition process of real-time mounting operation, the grayscale value of the corresponding pixel in the array-type imaging raw data is subtracted from the dark level reference value of that pixel to eliminate the fixed pattern noise caused by the dark current of the photosensitive unit and the system noise floor.

[0018] The optical field response distortion correction involves placing a standard diffuse reflection calibration board with the same substrate and surface reflectivity as the substrate to be mounted within the focal plane coverage of the imaging field of view. A spatial light modulation unit outputs a gradient distribution spatially constrained light field consistent with the mounting operation conditions. This controls an array-type photosensitive acquisition unit to collect full-field calibration imaging data. Combining the theoretical gradient distribution parameters of the spatially constrained light field, the deviation between the actual light response value and the theoretical standard response value of each photosensitive pixel unit at its corresponding spatial position is calculated, yielding the photosensitive unit response calibration coefficient for each pixel unit. During the real-time mounting operation's imaging acquisition process, the imaging data after dark level correction is divided by the corresponding pixel's photosensitive unit response calibration coefficient to eliminate optical field response distortion caused by non-uniform distribution of the spatially constrained light field, photosensitive unit response consistency deviation, and optical system field of view distortion.

[0019] Furthermore, the pose parameters extracted by the calculation module include the 6-DOF spatial pose parameters of the micro-component to be mounted, the 6-DOF spatial pose parameters of the substrate pads to be mounted, and the one-to-one correspondence matching feature parameters between the pads and the component pins.

[0020] During the solution module's operation phase, edge enhancement and feature segmentation are performed on the standardized imaging mapping dataset. Pad graphic features and pin contact area features are extracted from the segmented pad imaging region, and component contour features and pin electrode features are extracted from the segmented micro-component imaging region.

[0021] The solution module completes the corresponding feature matching between pads and micro-components based on preset feature matching constraint rules:

[0022] The preset feature matching constraint rules include the arrangement spacing matching constraint between the pad array and the component pin array, the coordinate deviation constraint between the centroid of the pad and the centroid of the corresponding pin electrode, the contour overlap constraint between the conductive contact area of ​​the pad and the working surface of the corresponding pin electrode, and the number matching constraint between the pad and the pin.

[0023] After feature matching is completed, the calculation module uses the feature point pairs formed by the matched pad features and the corresponding component pin features as a reference. It calculates the 6-DOF spatial pose parameters of the pads and micro-components through the spatial pose mapping of the feature point pairs. Then, it sorts the pose parameters of all pads and corresponding micro-components according to the preset mounting order to generate a pose parameter sequence.

[0024] Furthermore, when the compensation module calculates the coaxial alignment compensation amount, it includes calculating the theoretical pose difference of the mounting path based on the component pose parameters and pad pose parameters in the pose parameter sequence, and then combining the system coaxiality error parameters and motion axis dynamic response parameters to calculate the full-degree-of-freedom coaxial alignment compensation amount, and finally generating the compensated mounting alignment parameter set.

[0025] The calculation logic for the theoretical pose difference of the mounting path and the coaxial alignment compensation amount of the full degrees of freedom is as follows:

[0026] ;

[0027] In the formula: This represents the 6-DOF theoretical pose difference vector between the component and the pad. The 4×4 homogeneous transformation matrix corresponds to the actual 6-DOF pose of the substrate pads to be mounted. This is a fixed homogeneous transformation matrix for the target pose relative to the pad pose when the micro-component and the pad are precisely bonded; This is the 4×4 homogeneous transformation matrix corresponding to the current actual 6-DOF pose of the micro-component after it has been picked up; A mapping operator from a homogeneous transformation matrix to a 6-DOF pose vector; This is a 6-DOF coaxial alignment compensation vector; The rotation matrix represents the coaxiality deviation between the imaging optical axis and the motion axis of the mounting execution end; The dynamic tracking error vector of the motion axis corresponding to the mounting path; This is the imaging projection error vector during the pose calculation process.

[0028] Furthermore, the generation module executes the multi-degree-of-freedom motion control command flow for the mounting execution end as follows:

[0029] After the compensation of the mounting alignment parameter set is validated and decoupled in multiple dimensions, the parameters are decomposed into target pose parameters corresponding to each degree of freedom of motion of the mounting execution end, key node pose parameters of the entire mounting path, and action timing reference parameters of the entire mounting process. Simultaneously, the state control parameters corresponding to the picking, holding, and releasing of micro-components that match the mounting action are extracted.

[0030] Using the target pose parameters and key node pose parameters obtained from the disassembly as input boundaries, and taking the priority of mounting alignment accuracy and the absence of rigid impact during the movement process as dual constraints, a continuous and smooth position control sequence, speed control sequence and acceleration control sequence are generated for the entire path of the cover element picking stroke, alignment adjustment stroke and final precision mounting stroke. At the same time, for the final precision mounting stroke corresponding to the ultra-small pad, a micro-displacement closed-loop control sub-sequence with micron-level stepping is generated.

[0031] Based on the action timing reference parameters obtained from the disassembly, the above full-path control sequence, micro-displacement closed-loop control sub-sequence and the extracted state control parameters are aligned with the timing of all nodes to generate nozzle vacuum on / off control timing and placement pressure holding control timing corresponding to each node of the placement path, so that the synchronization deviation between motion execution action and component control action is within the preset range.

[0032] Finally, the timing-aligned full-path control sequence, micro-displacement closed-loop control sub-sequence, and synchronous control timing sequence are encapsulated into multi-degree-of-freedom motion control commands that can be directly recognized and executed by the motion control bus of the mounting execution end.

[0033] Furthermore, the control module receives real-time status data from the placement execution end and compares the real-time status data with the target value of the multi-degree-of-freedom motion control command in real time. When the deviation exceeds the preset threshold, it immediately sends a correction trigger signal to the generation module to complete the online correction of the motion control command. Simultaneously, after the placement action is completed, based on the placement imaging data collected by the imaging module, it calculates the placement alignment deviation between the component and the pad. When the alignment deviation is within the preset range, the placement is deemed qualified; otherwise, a placement abnormality feedback signal is generated.

[0034] When the correction trigger signal is sent to the generation module, the generation module runs again to complete the online correction of the motion control command.

[0035] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0036] This invention utilizes coaxial projection and synchronous acquisition of spatially constrained light fields to acquire complete imaging data covering the entire depth of field of the pads and the full outline of the components within a single exposure cycle. This effectively addresses industry pain points such as insufficient depth of field and difficulty in edge feature extraction during micro-component mounting. Furthermore, through global normalized imaging data processing, it eliminates imaging distortion caused by uneven light field distribution, photosensitivity deviation, and system noise, further improving the accuracy and stability of pose parameter calculation. Simultaneously, by combining coaxial alignment compensation and full-path smooth motion control, it achieves multi-degree-of-freedom micron-level mounting alignment. Simultaneously, through real-time status feedback during the mounting process and post-mounting qualification verification, it effectively avoids mounting damage to components and substrates, improving the mounting efficiency and yield of micro-components with ultra-small pads. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0038] Figure 1 This is a schematic diagram of a precision mounting system for visual positioning of micro-components with ultra-small pads. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0040] The present invention will be further described below with reference to embodiments.

[0041] Example:

[0042] This embodiment of the ultra-small pad micro-component visual positioning precision mounting system, such as Figure 1 As shown, it includes:

[0043] The imaging module is used to apply a spatially constrained light field to the pad area of ​​the substrate to be mounted and the micro-components to be mounted, and simultaneously acquire array-type imaging raw data of the corresponding area.

[0044] The imaging module integrates a coaxial illumination unit, a spatial light modulation unit, an array-type photosensitive acquisition unit, and a synchronous triggering unit;

[0045] The spatial light modulation unit is used to generate a spatially constrained light field with a gradient distribution along the optical axis. The coaxial illumination unit coaxially projects the spatially constrained light field onto the imaging surface of the substrate pad area and the micro-component to be mounted. The synchronous triggering unit synchronously drives the light field modulation timing of the spatial light modulation unit and the exposure acquisition timing of the array-type photosensitive acquisition unit, so that the array-type photosensitive acquisition unit can acquire array-type imaging raw data covering the full depth range of the pad and the full contour range of the component within a single exposure cycle.

[0046] The processing module is used to receive raw array imaging data, perform global normalization processing on the data, and generate a standardized imaging mapping dataset.

[0047] During the processing module's operation phase, dark level correction and light field response distortion correction are performed on the raw array imaging data. Then, a global normalized mapping model is constructed based on the gradient distribution parameters of the spatially constrained light field. The corrected imaging data is then converted into a standardized imaging mapping dataset through the mapping model.

[0048] The global normalized mapping model expression is:

[0049] ;

[0050] In the formula: The normalized imaging mapping value at spatial coordinates (x, y, z); The grayscale values ​​of the original array imaging data after correction at the corresponding coordinates; This is the reference value for the dark level at the corresponding coordinates; This refers to the photosensitive unit response calibration coefficient at the corresponding coordinates; The gradient distribution coefficient of the spatially constrained optical field along the optical axis z-direction; Set the system's fixed bias compensation value;

[0051] The above formula addresses the issues of grayscale distortion and insufficient feature recognition in imaging data caused by multiple sources of errors, such as dark current noise, inconsistent pixel photoelectric response, uneven axial gradient distribution of spatially constrained light field, and system circuit bias, during panoramic depth imaging of ultra-small pads and micro-components. It takes imaging data in three-dimensional spatial coordinates as the processing object, eliminates fixed-mode noise by subtracting the dark level reference value, and then combines the photosensitive unit response calibration coefficient, the light field axial gradient distribution coefficient, and the system fixed bias compensation value to complete the standardized mapping of imaging data in the entire field of view and panoramic depth range. This allows imaging data collected from different depth positions and different photosensitive units to have a unified quantization reference, fundamentally eliminating the interference of light field non-uniformity and inherent errors in the imaging link on subsequent feature extraction and pose calculation, and ensuring high-fidelity reproduction of imaging features of micron-level ultra-small pads and micro-components.

[0052] in, The inherent grayscale output value of the corresponding spatial coordinate pixel within the standard exposure cycle is determined by calibrating and taking the arithmetic mean of multiple sets of repeated sampling under the rated working conditions of fully shaded and no incident light for the array-type photosensitive acquisition unit.

[0053] The ratio of the single-pixel response grayscale to the average response grayscale of all effective pixels in the array is determined by calibration under full-range illumination of a standard uniform parallel light field. Its value is proportional to the photoelectric conversion quantum efficiency of the corresponding pixel, and its range is constrained to [0.8, 1.2].

[0054] The preset phase modulation parameters of the spatial light modulation unit are determined by fitting and calibrating with the measured values ​​of the light intensity distribution at the corresponding depth of field position. The value is proportional to the light field focusing energy density at the corresponding z-axis position, and the value range is constrained to [0.2,1].

[0055] The reference compensation value used to compensate for the fixed grayscale shift caused by the inherent circuit bias of the imaging link and ambient stray light is calibrated by the arithmetic mean of the global imaging background noise under the state of no imaging target in the whole system. Its value is proportional to the floor noise level of the system imaging link, and the value range is 0.02% to 1% of the full-scale grayscale output value of the photosensitive unit.

[0056] When performing dark level correction and optical field response distortion correction on the raw data of array imaging, the following rules apply:

[0057] Dark level correction is performed during the system initialization phase. With the array-type photosensitive acquisition unit in a state of complete darkness and no incident light, a preset number of dark field imaging data are acquired. The grayscale output value of each photosensitive pixel unit under the corresponding exposure parameters is statistically analyzed to obtain the dark level reference value for each pixel unit. During the imaging acquisition process of real-time mounting operation, the grayscale value of the corresponding pixel in the array-type imaging raw data is subtracted from the dark level reference value of that pixel to eliminate the fixed pattern noise caused by the dark current of the photosensitive unit and the system noise floor.

[0058] Optical field response distortion correction is performed during the system calibration phase. A standard diffuse reflection calibration board with the same substrate and surface reflectivity as the substrate to be mounted is placed within the focal plane coverage of the imaging field of view. The spatial light modulation unit outputs a gradient distribution spatial constraint light field consistent with the mounting operation conditions, controlling the array-type photosensitive acquisition unit to acquire full-field calibration imaging data. Combining the theoretical gradient distribution parameters of the spatial constraint light field, the deviation between the actual light response value and the theoretical standard response value of each photosensitive pixel unit at the corresponding spatial position is calculated, obtaining the photosensitive unit response calibration coefficient corresponding to each pixel unit. During the imaging acquisition process of the real-time mounting operation, the imaging data after dark level correction is divided by the photosensitive unit response calibration coefficient of the corresponding pixel to eliminate optical field response distortion caused by non-uniform distribution of the spatial constraint light field, photosensitive unit response consistency deviation, and optical system field of view distortion.

[0059] The solution module is used to receive the standardized imaging mapping dataset, extract the corresponding pose parameters of the pads and micro-components, and generate a pose parameter sequence.

[0060] The pose parameters extracted by the calculation module include the 6-DOF spatial pose parameters of the micro-component to be mounted, the 6-DOF spatial pose parameters of the substrate pads to be mounted, and the one-to-one correspondence matching feature parameters between the pads and the component pins.

[0061] The one-to-one matching feature parameters are the center coordinate deviation between the pad and the corresponding component pin, the contour overlap value, and the electrode contact area overlap value.

[0062] During the solution module's operation phase, edge enhancement and feature segmentation are performed on the standardized imaging mapping dataset. Pad graphic features and pin contact area features are extracted from the segmented pad imaging region, and component contour features and pin electrode features are extracted from the segmented micro-component imaging region.

[0063] The pad pattern features include the outer contour shape, center coordinates, contour boundary direction, and pad array spacing. The pin contact area features include the area, centroid coordinates, edge contour, and surface flatness mapping value of the conductive contact area on the pad used to mate with the component pin.

[0064] The component profile features include the outer profile shape, center coordinates, and body boundary normal vector of the micro-component. The pin electrode features include the number of component pins, the spacing between them, the centroid coordinates of each pin electrode, the profile shape, and the flatness mapping value of the electrode working surface.

[0065] The solution module completes the corresponding feature matching between pads and micro-components based on preset feature matching constraint rules:

[0066] The preset feature matching constraint rules include the spacing matching constraint between the pad array and the component pin array, the coordinate deviation constraint between the centroid of the pad and the centroid of the corresponding pin electrode, the contour overlap constraint between the conductive contact area of ​​the pad and the working surface of the corresponding pin electrode, and the number matching constraint between the pad and the pin.

[0067] After feature matching is completed, the calculation module uses the feature point pairs formed by the matched pad features and the corresponding component pin features as a reference, and calculates the 6-DOF spatial pose parameters of the pads and micro-components through the spatial pose mapping of the feature point pairs. Then, it sorts the pose parameters of all pads and corresponding micro-components according to the preset mounting order to generate a pose parameter sequence.

[0068] In the solution module, when performing edge enhancement and feature segmentation on the standardized imaging mapping dataset, anisotropic diffusion filtering is used to suppress noise in the standardized imaging mapping dataset based on the gray-level gradient distribution of the dataset, preserving the edge gradient information of the pads and component pins. Then, the second-order differential operator is used to perform edge enhancement on the filtered imaging data, amplifying the gray-level gradient differences between the pads, pin electrodes and the imaging background area to obtain the edge-enhanced imaging data. Subsequently, based on the preset gray-level threshold range, the edge-enhanced imaging data is binarized into foreground and background, extracting the foreground region of the pad imaging and the foreground region of the micro-component imaging, respectively, thus completing the feature segmentation process.

[0069] The compensation module is used to obtain the pose parameter sequence, calculate the coaxial alignment compensation amount corresponding to the mounting path, and generate the compensated mounting alignment parameter set.

[0070] When the compensation module calculates the coaxial alignment compensation amount, it calculates the theoretical pose difference of the mounting path based on the component pose parameters and pad pose parameters in the pose parameter sequence, and then combines the system coaxiality error parameters and motion axis dynamic response parameters to calculate the full-degree-of-freedom coaxial alignment compensation amount, and finally generates the compensated mounting alignment parameter set.

[0071] The calculation logic for the theoretical pose difference of the mounting path and the coaxial alignment compensation amount of the full degrees of freedom is as follows:

[0072] ;

[0073] In the formula: This represents the 6-DOF theoretical pose difference vector between the component and the pad. The 4×4 homogeneous transformation matrix corresponds to the actual 6-DOF pose of the substrate pads to be mounted. This is a fixed homogeneous transformation matrix for the target pose relative to the pad pose when the micro-component and the pad are precisely bonded; This is the 4×4 homogeneous transformation matrix corresponding to the current actual 6-DOF pose of the micro-component after it has been picked up; A mapping operator from a homogeneous transformation matrix to a 6-DOF pose vector; This is a 6-DOF coaxial alignment compensation vector; The rotation matrix represents the coaxiality deviation between the imaging optical axis and the motion axis of the mounting execution end; The dynamic tracking error vector of the motion axis corresponding to the mounting path; This represents the imaging projection error vector during the pose calculation process;

[0074] This set of formulas takes the 6-DOF spatial pose of pads and micro-components as its core. First, through the mapping transformation of homogeneous transformation matrix, it accurately calculates the theoretical pose deviation between the component and the corresponding pad, fully covering the alignment requirements of translation and rotation. Then, based on the theoretical deviation, it synchronously couples three types of core system errors: the coaxiality deviation between the imaging optical axis and the motion axis, the dynamic following error of the motion axis, and the imaging projection error of the pose calculation. This completes the closed-loop calculation of the alignment compensation amount of the full degree of freedom, thereby offsetting the impact of inherent and dynamic errors in the imaging-motion link on the mounting accuracy in advance. It ensures the micron-level bonding accuracy between ultra-small pads and micro-components from the control source and avoids the alignment offset caused by multi-axis motion coupling.

[0075] The generation module is used in the self-compensation module to call the compensated mounting alignment parameter set and generate multi-degree-of-freedom motion control instructions for the mounting execution end.

[0076] The process of generating multi-degree-of-freedom motion control instructions for the mounting execution end by the generation module is as follows:

[0077] The system performs legality verification and multi-dimensional decoupling on the compensated mounting alignment parameter set. Legality verification checks the dimensional completeness, numerical validity, and timing matching of the compensated mounting alignment parameter set. Parameters that fail the verification trigger an exception feedback process. Multi-dimensional decoupling is based on the axis configuration and kinematic model of the mounting execution end, decoupling the coupled pose parameters into independent control parameters corresponding to each degree of freedom of motion. The parameters are then decomposed into target pose parameters corresponding to each degree of freedom of motion of the mounting execution end, pose parameters of key nodes in the entire mounting path, and action timing reference parameters for the entire mounting process. The system also extracts the state control parameters corresponding to the picking, holding, and releasing of micro-components that match the mounting actions.

[0078] Using the target pose parameters and key node pose parameters obtained from the disassembly as input boundaries, and taking the priority of mounting alignment accuracy and the absence of rigid impact during the movement process as dual constraints, a continuous and smooth position control sequence, speed control sequence and acceleration control sequence are generated for the entire path of the cover element picking stroke, alignment adjustment stroke and final precision mounting stroke. At the same time, for the final precision mounting stroke corresponding to the ultra-small pad, a micro-displacement closed-loop control sub-sequence with micron-level stepping is generated.

[0079] Based on the action timing reference parameters obtained from the disassembly, the above full-path control sequence, micro-displacement closed-loop control sub-sequence and the extracted state control parameters are aligned with the timing of all nodes to generate nozzle vacuum on / off control timing and placement pressure holding control timing corresponding to each node of the placement path, so that the synchronization deviation between motion execution action and component control action is within the preset range.

[0080] Finally, the timing-aligned full-path control sequence, micro-displacement closed-loop control sub-sequence, and synchronous control timing sequence are encapsulated into multi-degree-of-freedom motion control commands that can be directly recognized and executed by the motion control bus of the mounting execution end.

[0081] The control module is used to receive multi-degree-of-freedom motion control commands, drive the placement execution end to complete the bonding operation between the micro-component and the substrate pad, synchronously transmit the real-time status data of the placement action, and complete the placement qualification verification.

[0082] The control module receives real-time status data from the placement execution end and compares the real-time status data with the target value of the multi-degree-of-freedom motion control command in real time. When the deviation exceeds the preset threshold, it immediately sends a correction trigger signal to the generation module to complete the online correction of the motion control command. Simultaneously, after the placement action is completed, based on the placement imaging data collected by the imaging module, it calculates the placement alignment deviation between the component and the pad. When the alignment deviation is within the preset range, the placement is deemed qualified; otherwise, a placement abnormality feedback signal is generated.

[0083] When the correction trigger signal is sent to the generation module, the generation module runs again to complete the online correction of the motion control command;

[0084] The placement execution end is equipped with a multi-axis motion platform, a high-precision nozzle unit, and a pressure sensing unit that are matched with multi-degree-of-freedom motion control commands;

[0085] Among them, the multi-axis motion platform is used to perform alignment motion with translational and rotational degrees of freedom, the high-precision nozzle unit is used to pick up and release micro components, and the pressure sensing unit is used to collect vertical contact pressure data in real time during the mounting process and transmit the pressure data back to the control module in real time. The control module adjusts the motion control parameters of the vertical mounting degree of freedom based on the pressure data to avoid damage to the components and the substrate.

[0086] The imaging module is interconnected with the processing module via a local area network. The processing module is interconnected with the solution module and the compensation module via a local area network. The solution module and the compensation module are interconnected with the generation module via a local area network. The generation module is interconnected with the control module via a local area network.

[0087] In this embodiment, the imaging module applies a spatially constrained light field to the pad area of ​​the substrate to be mounted and the micro-component to be mounted, and simultaneously acquires array-type imaging raw data of the corresponding area. The processing module simultaneously receives the array-type imaging raw data, performs global normalization processing on the data, and generates a standardized imaging mapping dataset. The solution module further receives the standardized imaging mapping dataset, extracts the corresponding pose parameters of the pads and micro-components, and generates a pose parameter sequence. The compensation module acquires the pose parameter sequence in real time, calculates the coaxial alignment compensation amount corresponding to the mounting path, and generates a compensated mounting alignment parameter set. Then, the generation module calls the compensated mounting alignment parameter set from the compensation module to generate multi-degree-of-freedom motion control commands for the mounting execution end. Finally, the control module receives the multi-degree-of-freedom motion control commands and drives the mounting execution end to complete the bonding operation between the micro-component and the substrate pads, and simultaneously transmits back real-time status data of the mounting action to complete the mounting qualification verification.

[0088] In the above embodiments, the system can achieve high-speed and precision mounting of ultra-small pad components in actual operation scenarios of mass production and mounting of micro-components. A single exposure can complete the panoramic depth imaging acquisition of pads and components, greatly reducing the single-station operation cycle. Through precise pose calculation and alignment compensation, the mounting deviation is controlled at the micron level, effectively reducing the defect rate of cold solder joints, misalignment, etc. At the same time, real-time closed-loop management avoids damage to components and substrates, effectively improving the mounting yield and mass production efficiency of micro-assembly production lines.

[0089] See the application examples of the system in the above embodiments:

[0090] Mounting process of 01005 multilayer ceramic capacitors (MLCCs) on the main control board of the flagship noise-canceling TWS earphones:

[0091] The component itself measures only 0.4mm × 0.2mm, corresponding to substrate pad sizes of 200μm × 120μm. A single main control board can mount over 200 of these same components. Mass production requires mounting alignment accuracy within ±7μm and a batch yield of no less than 99.95%. This system completes the entire process, and the specific implementation is as follows:

[0092] After the system powers on and completes the full-link initialization, the TWS earphone main control board to be mounted is positioned to the mounting station via the conveyor line. The 01005 micro-components to be mounted are simultaneously sent to the pick-up station, and the system starts the mounting process. First, the imaging module completes the acquisition of raw imaging data. The coaxial illumination unit, spatial light modulation unit, array photosensitive acquisition unit, and synchronous triggering unit integrated in the imaging module work together. The spatial light modulation unit generates a spatially constrained light field with a gradient distribution along the optical axis. The coaxial illumination unit projects this light field coaxially onto the substrate pad area and the imaging surface of the micro-component to be mounted. The synchronous triggering unit precisely matches the light field modulation timing with the exposure acquisition timing of the array photosensitive acquisition unit. Within a single exposure cycle, array imaging raw data covering the full depth range of the pads and the full contour range of the component are acquired synchronously, and the data is transmitted to the processing module in real time.

[0093] After receiving the raw data from the array imaging, the processing module first performs dark level correction and light field response distortion correction to eliminate imaging errors caused by dark current of the photosensitive unit, system noise floor, non-uniform distribution of the light field, and consistency deviation of the photosensitive unit response. Then, based on the gradient distribution parameters of the spatially constrained light field, it constructs a global normalized mapping model to complete the standardization transformation of the corrected imaging data. Finally, it outputs a standardized imaging mapping dataset. The grayscale deviation of the dataset is controlled within 0.5% of the full scale of the photosensitive unit, providing a stable and consistent imaging basis for subsequent pose calculation.

[0094] After receiving the standardized imaging mapping dataset, the solution module first performs edge enhancement and feature segmentation processing. It suppresses imaging noise through anisotropic diffusion filtering, fully preserves the edge gradient information of pads and component pins, and then enhances edge features through second-order differential operators to amplify the grayscale difference between the effective imaging area and the background area. After binarization segmentation with a preset grayscale threshold, the foreground areas of pad imaging and micro-component imaging are extracted respectively. Subsequently, core features such as the outer contour shape, center coordinates, array spacing, centroid coordinates and contour of the conductive contact area of ​​the pins are extracted from the pad imaging area. Core features such as the outer contour, center coordinates, number of pin electrodes, spacing, centroid coordinates of each pin, and contour of the working surface are extracted from the component imaging area. According to the preset rules such as matching the pad and component pin array spacing, centroid coordinate deviation constraint, contour overlap constraint, and quantity matching, the feature matching of the pad and component pin is completed. Then, based on the matched feature point pairs, the 6-DOF spatial pose parameters of the pad and the micro-component are calculated. After being sorted according to the preset mounting order, the pose parameter sequence is generated and output to the compensation module.

[0095] After the compensation module obtains the pose parameter sequence, it first calculates the theoretical pose difference between the component and the corresponding pad with 6 degrees of freedom. Then, it combines the pre-calibrated coaxiality error parameters, motion axis dynamic response parameters, and imaging projection error data to complete the calculation of the full-degree-of-freedom coaxial alignment compensation amount. Finally, it generates a set of compensated mounting alignment parameters. This set of parameters includes all alignment deviations caused by system coaxiality deviation, motion following error, and projection error. The theoretical alignment accuracy after compensation can reach within ±3μm, which fully meets the accuracy requirements of this mounting. The parameter set is synchronously transmitted to the generation module.

[0096] After the generation module calls the compensation and mounting alignment parameter set, it completes the generation of multi-degree-of-freedom motion control commands according to the process. The specific implementation process is as follows:

[0097] The first step is to complete the parameter validity verification and multi-dimensional decoupling. In the validity verification stage, the completeness of the dimensions, the validity of the values, and the timing matching of each parameter are checked one by one to confirm that the parameters meet the axis control requirements of the mounting execution end and that there are no issues such as excessive values, missing dimensions, or timing misalignments. Failure to pass the verification will directly trigger the exception feedback process. In the multi-dimensional decoupling stage, based on the axis configuration and kinematic model of the 6-axis motion platform of the mounting execution end, the coupled 6-DOF pose parameters are decoupled into independent control parameters corresponding to the X, Y, and Z translational axes and the A, B, and C rotational axes, respectively.

[0098] The second step is to complete the decomposition of control parameters and the extraction of state parameters. The decoupled independent control parameters are decomposed into target pose parameters corresponding to each degree of freedom of motion of the placement execution end, pose parameters of key nodes in the entire placement path, and action timing reference parameters of the entire placement process. Simultaneously, the nozzle vacuum state control parameters corresponding to the pickup, holding, and release of micro-components that match the placement action are extracted.

[0099] The third step is to generate a full-path motion control sequence. Using the target pose parameters and key node pose parameters obtained from the disassembly as input boundaries, and with the dual constraints of prioritizing mounting alignment accuracy and ensuring no rigid impact during the motion process, a continuous and smooth position control sequence, speed control sequence, and acceleration control sequence are generated for the entire path, including the component pickup stroke, alignment adjustment stroke, and final precision mounting stroke. For the final precision mounting stroke corresponding to the ultra-small pads in this case, an additional micro-displacement closed-loop control sub-sequence with micron-level steps is generated to ensure the vertical contact accuracy and pressure controllability of the final mounting stroke.

[0100] The fourth step is to complete the full-process timing alignment and instruction encapsulation. Based on the action timing reference parameters obtained from the disassembly, the full-path control sequence and micro-displacement closed-loop control sub-sequence are aligned with the extracted nozzle status control parameters at all nodes. This generates nozzle vacuum on / off control timing and placement holding pressure control timing that correspond one-to-one with each node of the placement path, controlling the synchronization deviation between motion execution actions and component pick-up and drop control actions to within 1ms. Finally, the timing-aligned full-path control sequence, micro-displacement closed-loop control sub-sequence, and synchronization control timing are encapsulated into multi-degree-of-freedom motion control instructions that can be directly recognized and executed by the placement execution end motion control bus, and then sent to the control module.

[0101] After receiving multi-degree-of-freedom motion control commands, the control module drives the multi-axis motion platform and high-precision nozzle unit of the placement execution end to complete the corresponding placement actions. Simultaneously, the pressure sensing unit of the placement execution end collects vertical contact pressure data in real time during the placement process and transmits this pressure data back in real time. Based on the real-time pressure data, the control module dynamically adjusts the motion control parameters of the vertical placement degree of freedom to prevent damage to the components and substrate due to pressure. During the operation, the control module synchronously transmits real-time status data of the placement actions and compares the real-time status data with the target value of the control commands. When the deviation exceeds a preset threshold, a correction trigger signal is immediately sent to the generation module to complete the online correction of the motion control commands.

[0102] After a single placement operation is completed, the imaging module acquires the imaging data after placement a second time, calculates the actual placement alignment deviation between the component and the pad, and the test shows that the alignment deviation between the component and the pad in this placement is 4.5μm, which is within the preset ±7μm acceptable range. The system determines that the placement is qualified and completes the single placement operation.

[0103] In this mass production verification application, the system continuously completed the placement of 1.5 million 01005-sized micro-components, with an overall placement yield of 99.98% and a single-station placement cycle of 0.18 seconds per component. This fully meets the high-density, high-precision, and high-stability mass production placement requirements of TWS earphone main control boards, and solves the industry pain points of traditional placement equipment in the placement of 01005 and smaller ultra-micro components, such as misalignment, poor soldering, component damage, and substandard mass production yield.

[0104] In summary, this system, through coaxial projection and synchronous acquisition of spatially constrained light fields, can acquire complete imaging data covering the entire depth of field of the pads and the full outline of the components within a single exposure cycle. This effectively solves the industry pain points of insufficient depth of field and difficulty in edge feature extraction in micro-component mounting. Furthermore, through global normalized imaging data processing, it eliminates imaging distortion caused by uneven light field distribution, photosensitivity deviation, and system noise, further improving the accuracy and stability of pose parameter calculation. At the same time, combined with coaxial alignment compensation and full-path smooth motion control, it completes multi-degree-of-freedom micron-level mounting alignment. Simultaneously, through real-time status feedback during the mounting process and post-mounting qualification verification, it effectively avoids mounting damage to components and substrates, improving the mounting efficiency and yield of micro-components with ultra-small pads.

[0105] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A precision mounting system for ultra-small pad micro-components based on visual positioning, characterized in that, include: The imaging module is used to apply a spatially constrained light field to the pad area of ​​the substrate to be mounted and the micro-components to be mounted, and simultaneously acquire array-type imaging raw data of the corresponding area. The processing module is used to receive raw array imaging data, perform global normalization processing on the data, and generate a standardized imaging mapping dataset. The solution module is used to receive the standardized imaging mapping dataset, extract the corresponding pose parameters of the pads and micro-components, and generate a pose parameter sequence. The compensation module is used to obtain the pose parameter sequence, calculate the coaxial alignment compensation amount corresponding to the mounting path, and generate the compensated mounting alignment parameter set. The generation module is used in the self-compensation module to call the compensated mounting alignment parameter set and generate multi-degree-of-freedom motion control instructions for the mounting execution end. The control module is used to receive multi-degree-of-freedom motion control commands, drive the placement execution end to complete the bonding operation between the micro-component and the substrate pad, synchronously transmit the real-time status data of the placement action, and complete the placement qualification verification.

2. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, The imaging module is integrated with a coaxial illumination unit, a spatial light modulation unit, an array-type photosensitive acquisition unit, and a synchronous triggering unit. The spatial light modulation unit is used to generate a spatially constrained light field with a gradient distribution along the optical axis. The coaxial illumination unit coaxially projects the spatially constrained light field onto the imaging surface of the substrate pad area and the micro-component to be mounted. The synchronous triggering unit synchronously drives the light field modulation timing of the spatial light modulation unit and the exposure acquisition timing of the array-type photosensitive acquisition unit, so that the array-type photosensitive acquisition unit can acquire array-type imaging raw data covering the full depth range of the pad and the full contour range of the component within a single exposure cycle.

3. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, During the operation phase of the processing module, dark level correction and light field response distortion correction are performed on the raw array imaging data. Then, a global normalized mapping model is constructed based on the gradient distribution parameters of the spatially constrained light field. The corrected imaging data is converted into a standardized imaging mapping dataset through the mapping model. The global normalized mapping model expression is as follows: ； In the formula: The normalized imaging mapping value at spatial coordinates (x, y, z); The grayscale values ​​of the original array imaging data after correction at the corresponding coordinates; This is the reference value for the dark level at the corresponding coordinates; This refers to the photosensitive unit response calibration coefficient at the corresponding coordinates; The gradient distribution coefficient of the spatially constrained optical field along the optical axis z-direction; This is the fixed bias compensation value for the system.

4. The ultra-small pad micro-component visual positioning precision mounting system according to claim 3, characterized in that, When performing dark level correction and light field response distortion correction on the raw data of array imaging, the following rules apply: Dark level correction is performed when the array-type photosensitive acquisition unit is in a state of complete darkness with no incident light. A preset number of dark field imaging data are acquired, and the grayscale output value of each photosensitive pixel unit under the corresponding exposure parameters is statistically analyzed to obtain the dark level reference value for each pixel unit. During the imaging acquisition process of real-time mounting operation, the grayscale value of the corresponding pixel in the array-type imaging raw data is subtracted from the dark level reference value of that pixel to eliminate the fixed pattern noise caused by the dark current of the photosensitive unit and the system noise floor. The optical field response distortion correction involves placing a standard diffuse reflection calibration board with the same substrate and surface reflectivity as the substrate to be mounted within the focal plane coverage of the imaging field of view. A spatial light modulation unit outputs a gradient distribution spatially constrained light field consistent with the mounting operation conditions. This controls an array-type photosensitive acquisition unit to collect full-field calibration imaging data. Combining the theoretical gradient distribution parameters of the spatially constrained light field, the deviation between the actual light response value and the theoretical standard response value of each photosensitive pixel unit at its corresponding spatial position is calculated, yielding the photosensitive unit response calibration coefficient for each pixel unit. During the real-time mounting operation's imaging acquisition process, the imaging data after dark level correction is divided by the corresponding pixel's photosensitive unit response calibration coefficient to eliminate optical field response distortion caused by non-uniform distribution of the spatially constrained light field, photosensitive unit response consistency deviation, and optical system field of view distortion.

5. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, The pose parameters extracted by the calculation module include the 6-DOF spatial pose parameters of the micro-component to be mounted, the 6-DOF spatial pose parameters of the substrate pads to be mounted, and the one-to-one correspondence matching feature parameters between the pads and the component pins. During the solution module's operation phase, edge enhancement and feature segmentation are performed on the standardized imaging mapping dataset. Pad graphic features and pin contact area features are extracted from the segmented pad imaging region, and component contour features and pin electrode features are extracted from the segmented micro-component imaging region. The solution module completes the corresponding feature matching between pads and micro-components based on preset feature matching constraint rules: The preset feature matching constraint rules include the arrangement spacing matching constraint between the pad array and the component pin array, the coordinate deviation constraint between the centroid of the pad and the centroid of the corresponding pin electrode, the contour overlap constraint between the conductive contact area of ​​the pad and the working surface of the corresponding pin electrode, and the number matching constraint between the pad and the pin. After feature matching is completed, the calculation module uses the feature point pairs formed by the matched pad features and the corresponding component pin features as a reference. It calculates the 6-DOF spatial pose parameters of the pads and micro-components through the spatial pose mapping of the feature point pairs. Then, it sorts the pose parameters of all pads and corresponding micro-components according to the preset mounting order to generate a pose parameter sequence.

6. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, When the compensation module calculates the coaxial alignment compensation amount, it calculates the theoretical pose difference of the mounting path based on the component pose parameters and pad pose parameters in the pose parameter sequence, and then calculates the full-degree-of-freedom coaxial alignment compensation amount by combining the system coaxiality error parameters and motion axis dynamic response parameters, and finally generates the compensated mounting alignment parameter set. The calculation logic for the theoretical pose difference of the mounting path and the coaxial alignment compensation amount of the full degrees of freedom is as follows: ； In the formula: This represents the 6-DOF theoretical pose difference vector between the component and the pad. The 4×4 homogeneous transformation matrix corresponds to the actual 6-DOF pose of the substrate pads to be mounted. This is a fixed homogeneous transformation matrix for the target pose relative to the pad pose when the micro-component and the pad are precisely bonded; This is the 4×4 homogeneous transformation matrix corresponding to the current actual 6-DOF pose of the micro-component after it has been picked up; A mapping operator from a homogeneous transformation matrix to a 6-DOF pose vector; This is a 6-DOF coaxial alignment compensation vector; The rotation matrix represents the coaxiality deviation between the imaging optical axis and the motion axis of the mounting execution end; The dynamic tracking error vector of the motion axis corresponding to the mounting path; This is the imaging projection error vector during the pose calculation process.

7. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, The process of generating multi-degree-of-freedom motion control commands for the mounting execution end by the generation module is as follows: After the compensation of the mounting alignment parameter set is validated and decoupled in multiple dimensions, the parameters are decomposed into target pose parameters corresponding to each degree of freedom of motion of the mounting execution end, key node pose parameters of the entire mounting path, and action timing reference parameters of the entire mounting process. Simultaneously, the state control parameters corresponding to the picking, holding, and releasing of micro-components that match the mounting action are extracted. Using the target pose parameters and key node pose parameters obtained from the disassembly as input boundaries, and taking the priority of mounting alignment accuracy and the absence of rigid impact during the movement process as dual constraints, a continuous and smooth position control sequence, speed control sequence and acceleration control sequence are generated for the entire path of the cover element picking stroke, alignment adjustment stroke and final precision mounting stroke. At the same time, for the final precision mounting stroke corresponding to the ultra-small pad, a micro-displacement closed-loop control sub-sequence with micron-level stepping is generated. Based on the action timing reference parameters obtained from the disassembly, the above full-path control sequence, micro-displacement closed-loop control sub-sequence and the extracted state control parameters are aligned with the timing of all nodes to generate nozzle vacuum on / off control timing and placement pressure holding control timing corresponding to each node of the placement path, so that the synchronization deviation between motion execution action and component control action is within the preset range. Finally, the timing-aligned full-path control sequence, micro-displacement closed-loop control sub-sequence, and synchronous control timing sequence are encapsulated into multi-degree-of-freedom motion control commands that can be directly recognized and executed by the motion control bus of the mounting execution end.

8. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, The control module receives real-time status data from the placement execution end and compares the real-time status data with the target value of the multi-degree-of-freedom motion control command in real time. When the deviation exceeds the preset threshold, it immediately sends a correction trigger signal to the generation module to complete the online correction of the motion control command. Simultaneously, after the placement action is completed, based on the placement imaging data collected by the imaging module, it calculates the placement alignment deviation between the component and the pad. When the alignment deviation is within the preset range, the placement is deemed qualified; otherwise, a placement abnormality feedback signal is generated. When the correction trigger signal is sent to the generation module, the generation module runs again to complete the online correction of the motion control command.

9. The ultra-small pad micro-component visual positioning precision mounting system according to claim 1, characterized in that, The imaging module is interconnected with a processing module via a local area network. The processing module is interconnected with a solution module and a compensation module via a local area network. The solution module and the compensation module are interconnected via a local area network. The compensation module is interconnected with a generation module via a local area network. The generation module is interconnected with a control module via a local area network.

Images