Integrated circuit patch z-axis compensation method and system for pcb board warpage

Abstract

This application discloses a method and system for Z-axis compensation of integrated circuit surface mount technology (SMT) for PCB board warpage, relating to the field of PCB board surface mount technology. The method includes: acquiring sampling data of the PCB board surface height and PCB board design information, extracting the physical structural features corresponding to each sampling point, and constructing a board surface warpage height model based on the sampling data; calculating the estimated board surface height for the target mounting position, and determining the Z-axis safe switching height of the mounting nozzle based on the thickness information of the component to be mounted and a preset safety margin; controlling the mounting nozzle to descend to the Z-axis safe switching height and then using a cutting force feedback correction mode, adaptively correcting the Z-axis displacement based on the dynamic error between the actual contact force and the target mounting force until the actual contact force falls within the preset error range to complete the mounting. This improves the accuracy of mounting height control, the stability of mounting force, and the consistency of mounting quality under complex warpage conditions.

Description

Technical Field

[0001] This application relates to the field of PCB surface mount technology, and in particular to a method and system for Z-axis compensation of integrated circuit chip mounting for PCB warpage. Background Technology

[0002] As integrated circuit packaging continues to evolve towards miniaturization and high density, the requirements for placement accuracy in surface mount production lines are constantly increasing. Among these requirements, Z-axis height control of the pick-and-place machine is particularly critical. In actual manufacturing, PCB boards often experience varying degrees of overall warping or localized undulations due to processing, transportation, assembly, and heating, causing the actual height of the board surface to deviate from its theoretical position. In such cases, if the placement head still performs the downward pressure action according to the preset height, it can easily lead to abnormal component placement, unbalanced contact, or improper force application, thereby affecting subsequent soldering quality and the overall assembly reliability. This problem is particularly prominent in high-density, fine-pitch component placement scenarios.

[0003] To address placement height deviations caused by PCB warpage, common methods include preset height control, board height detection, external topography modeling, or mechanical support. While these methods can improve placement consistency to some extent, most focus on a single aspect: some rely primarily on static settings, making it difficult to adapt to actual board topography changes; some depend on additional detection or scanning methods, increasing cycle time and system complexity; and others rely on external support or contact detection, limiting their applicability and control precision. Therefore, achieving a balance between addressing localized dynamic warpage, protecting small components during placement, and meeting the demands of high-speed production line operation is often challenging.

[0004] Furthermore, PCB warpage is usually not a simple overall planar offset, but rather has the characteristics of uneven spatial distribution and significant local differences. If the mounting system is adjusted based solely on a limited number of measurement points, fixed compensation values, or hysteresis contact signals, it is often difficult to reflect the true height status at the mounting point in a timely and accurate manner, and it is also difficult to achieve stable mounting while avoiding component damage. Summary of the Invention

[0005] This application provides a method, system, storage medium, computer program product, and electronic device for Z-axis compensation of integrated circuit chip mounting for PCB board warping, in order to at least solve the problem of insufficient mounting control accuracy caused by PCB board warping deformation in the prior art.

[0006] In a first aspect, embodiments of this application provide a method for Z-axis compensation of integrated circuit chip mounting for PCB board warpage. The method includes: acquiring sampling data of the board surface height of the PCB board to be mounted and PCB board design information of the PCB board to be mounted; extracting physical structural features corresponding to the positions of each sampling point on the PCB board to be mounted based on the PCB board design information, and constructing a board surface warpage height model incorporating the physical structural features by combining the board surface height sampling data; calculating the corresponding estimated board surface height using the board surface warpage height model for the target mounting position of the component to be mounted on the PCB board to be mounted, and... Based on the thickness information of the component to be mounted and the preset safety margin, the Z-axis safe switching height of the mounting nozzle is determined; the mounting nozzle is controlled to lower the component to be mounted to the Z-axis safe switching height and switch to force feedback correction mode; in the force feedback correction mode, the actual contact force of the mounting nozzle during the continued downward pressing process is collected in real time, and based on the dynamic error between the actual contact force and the preset target mounting force, adaptive feedback correction is performed on the Z-axis displacement of the mounting nozzle until the actual contact force falls within the preset error range corresponding to the target mounting force, so as to complete the mounting of the component to be mounted.

[0007] Secondly, embodiments of this application provide an integrated circuit Z-axis compensation system for PCB board warpage. The system includes: a data acquisition unit for acquiring sampling data of the board surface height of the PCB board to be mounted and PCB board design information of the PCB board to be mounted; a warpage modeling unit for extracting physical structural features corresponding to the positions of each sampling point on the PCB board to be mounted based on the PCB board design information, and constructing a board surface warpage height model that integrates the physical structural features by combining the board surface height sampling data; and a height switching determination unit for calculating the corresponding estimated board surface warpage height using the board surface warpage height model for the target mounting position of the component to be mounted on the PCB board to be mounted. The system determines the Z-axis safe switching height of the pick-and-place nozzle by taking into account the thickness information of the component to be placed and a preset safety margin. A mode switching control unit controls the pick-and-place nozzle to descend to the Z-axis safe switching height while carrying the component to be placed, and switches to a force feedback correction mode. A force feedback correction unit, in the force feedback correction mode, collects the actual contact force of the pick-and-place nozzle during its continued downward pressure in real time, and performs adaptive feedback correction on the Z-axis displacement of the pick-and-place nozzle based on the dynamic error between the actual contact force and the preset target placement force, until the actual contact force falls within the preset error range corresponding to the target placement force, thereby completing the placement of the component to be placed.

[0008] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the integrated circuit patch Z-axis compensation method for PCB board warping according to any embodiment of the present application.

[0009] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, characterized in that, when the program is executed by a processor, it implements the steps of the integrated circuit patch Z-axis compensation method for PCB board warping according to any embodiment of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the integrated circuit patch Z-axis compensation method for PCB board warping according to any embodiment of this application.

[0011] The integrated circuit patch Z-axis compensation method and system for PCB board warpage provided in this application can achieve at least the following technical effects: (1) By fusing the physical structural features corresponding to the sampling points in the PCB design information with the board height sampling data, a board warpage height model oriented towards the target mounting position is constructed. This design breaks the limitation of relying solely on discrete measurement points for pure geometric interpolation, enabling board height evaluation to deeply integrate the differences in local board structure (such as copper cladding and board thickness), thereby providing a more physically reliable prediction of the actual height state of the target area. Based on this prediction result, and further combined with the thickness information of the components to be mounted and the preset safety margin, the Z-axis safe switching height is determined, and a precise "position-force" mode switching critical point is defined for the downward pressing action of the mounting nozzle. Thus, it not only provides a more accurate initial position basis for subsequent mounting actions that conforms to the actual board state, but also effectively avoids the risk of collision caused by a single static prediction deviation, improving the pertinence and safety of height control under complex board conditions.

[0012] (2) After the placement nozzle descends to the safe switching height of the Z-axis, the force feedback correction mode is switched on, and the actual contact force is collected in real time during the continued pressing process. Based on the dynamic error between the actual contact force and the target placement force, the Z-axis displacement is adaptively corrected. Thus, the execution logic of the placement end is upgraded from rigid blind drop to flexible closed-loop adjustment that is coordinated with the real-time force state. This dynamic process can absorb and compensate for the residual local height prediction error, component manufacturing tolerance, and dynamic deformation at the moment of contact in real time, ensuring that the placement action smoothly converges to the target force state, reducing the risk of cracking or poor soldering of fine-pitch devices in high-speed placement, and improving the stability of the force during the component placement process and the reliability of the final placement result.

[0013] This technical solution constructs a Z-axis compensation mechanism that combines "feedforward physical modeling prediction" and "end-point adaptive force control correction." The former provides the placement head with a safe approach basis to the target position, while the latter dynamically converges and controls the actual placement contact process. The two are seamlessly integrated, jointly achieving precise constraints on both placement height and placement force. As a result, it is possible to better balance placement accuracy, force stability, and execution efficiency during integrated circuit placement, thereby significantly improving the overall placement quality consistency and process robustness under complex PCB warpage deformation conditions. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating an example of an integrated circuit patch Z-axis compensation method for PCB board warpage according to an embodiment of this application is shown. Figure 2 A flowchart illustrating an example of constructing a board warpage height model according to an embodiment of this application is shown. Figure 3 A flowchart illustrating an example of controlling the Z-axis safe switching height of the patch nozzle and switching to a force feedback correction mode in a method according to an embodiment of this application is shown. Figure 4 This paper presents an example of an operation flowchart illustrating adaptive feedback correction of the Z-axis displacement of the patch nozzle based on dynamic error in a method according to an embodiment of this application. Figure 5 A schematic diagram illustrating the system operation principle of an example integrated circuit patch Z-axis compensation method for PCB board warpage according to an embodiment of this application is shown. Figure 6 A comparative simulation diagram showing the results of different methods for constructing a model of the warp height of a board surface is presented. Figure 7 A comparative simulation diagram illustrating the effects of different methods for controlling a patch nozzle to perform a dynamic placement process is shown. Figure 8 A comparative simulation diagram of the mounting pressure tolerance distribution under different control methods is shown. Figure 9 A structural block diagram of an example of an integrated circuit patch Z-axis compensation system for PCB board warping is shown according to an embodiment of this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] It should be noted that in actual production environments, PCB warpage is often not caused by a single factor, but is related to a combination of factors such as soldering thermal stress, mismatch in material thermal expansion coefficients, and local structural weakening. For micro-packaged devices, even a small local height deviation in the mounting area can cause the actual Z-axis position of the pads to deviate from the mounting reference, resulting in significant changes in component release height, contact state, and stress level. Especially under control methods based on static height settings of the theoretical plane, if the actual height of the local board surface is lower than the system estimate, the component may not be able to effectively fit the target area; conversely, if the local board surface height is higher than expected, the mounting head may generate excessive mechanical force at the end when it continues to press down, thus affecting the mounting reliability of micro-devices.

[0018] To address the aforementioned issues, some existing technologies employ board height measurement or shape mapping methods for compensation. For example, some equipment uses laser measurement or visual measurement to obtain height information at several discrete locations on the board surface before mounting, and corrects the mounting height accordingly. Other methods attempt to analyze the board surface condition using three-dimensional shape data obtained from previous inspection processes. These methods can improve the ability to identify board surface undulations to some extent, but the former usually relies on point-by-point scanning or local sampling, which can easily increase the burden of data acquisition and processing, while the latter remains more at the level of front-end inspection or static evaluation, and the results are difficult to apply in a high-frequency, real-time closed-loop manner to the underlying motion control during the mounting execution stage.

[0019] Furthermore, current technologies also employ mechanical support or end-contact detection to improve mounting height control. For example, in some assembly scenarios, under-board support structures can alleviate board sagging; however, the support position is often limited by board layout, component distribution, and available space, making it difficult to provide continuous and uniform support for large boards or complex local warping areas. Similarly, some solutions attempt to determine contact status by monitoring current changes in actuators or by setting up end-force sensing structures. While these can enhance end-feedback capabilities to some extent, the former often suffers from response lag or excessively high triggering conditions, while the latter easily leads to increased hardware complexity and implementation costs. Therefore, in micro-device mounting scenarios, balancing the accuracy of local warping detection, production line cycle time, and flexible force control during mounting remains a critical issue that urgently needs to be addressed in current technologies.

[0020] It should be understood that the above description of the relevant technologies is intended only to help the public better understand the inventive spirit and motivation of this application, and is not intended to limit this application. Furthermore, the technical solutions described in the above-mentioned relevant technologies are not prior art, and may also be undisclosed technical solutions, such as those under research or in the laboratory stage.

[0021] The technical solutions in this application, including the collection, storage, use, processing, transmission, provision, and disclosure of users' personal information, comply with relevant laws and regulations and do not violate public order and good morals.

[0022] Figure 1 A flowchart illustrating an example of an integrated circuit patch Z-axis compensation method for PCB board warpage according to an embodiment of this application is shown.

[0023] Regarding the execution subject of the method in the embodiments of this application, it can be any controller or processor with computing or processing capabilities, such as a patch control platform controller, which executes the various steps of the method in the embodiments of this application by running a computer program or instructions stored in a storage medium.

[0024] In some examples, it can be integrated into electronic devices or terminals through software, hardware, or a combination of both, and the types of terminals or electronic devices can be diverse. Furthermore, regarding the business application scenarios of this application, it is particularly suitable for high-density surface mount scenarios for micro-packaged devices, such as the automated mounting process of small-sized components like 0201 and 01005.

[0025] like Figure 1As shown, in step S110, the board height sampling data of the PCB board to be mounted and the PCB board design information of the PCB board to be mounted are obtained. The board height sampling data is used to characterize the actual board surface undulation state of the PCB board under the current working conditions, while the PCB board design information is used to characterize the prior attributes of the board at the structural design level. Both serve as the input basis for subsequent board surface warpage modeling.

[0026] In some implementations, before the PCB board enters the placement station or during the placement preparation stage, the system can scan and sample the board surface using a 3D topography inspection device, such as a high-precision laser displacement sensor, a structured light vision system, or a 3D measurement module in an online inspection device, to obtain height discrete point data distributed in different areas of the board surface. This type of height discrete point data can intuitively reflect the actual geometric state of the PCB board under the current physical environment, such as local undulations and overall deformation caused by reflow soldering thermal stress, mechanical clamping stress, or residual stress of the board itself. At the same time, the system can also parse manufacturing engineering files or board-level design files to extract PCB board design information, such as Gerber files, ODB++ files, CAD coordinate data, engineering BOM tables, or other data files that can be used to characterize the structural properties of the board.

[0027] It should be understood that PCB warpage is not entirely a random geometric fluctuation, but is usually strongly correlated with its local material distribution, structurally weakened areas, and differences in board stiffness. Relying solely on limited board surface scanning results can easily lead to interpolation distortion in areas with sparse measurement points or significant structural changes. However, by simultaneously introducing design information containing the original structural properties of the board, the board surface state can be characterized from both actual measurement results and prior structural constraints, providing a solid data foundation for more accurate reconstruction of the board warpage morphology.

[0028] In step S120, the physical structural features corresponding to each sampling point on the PCB board to be mounted are extracted based on the PCB board design information, and the board surface height sampling data are combined to construct a board surface warping height model that integrates the physical structural features.

[0029] Here, physical structural features are used to reflect the differences in local structure, material distribution, or mechanical response in the area where the sampling point is located, so that the model can not only describe "what the current plate height is measured", but also reflect "why this area is more prone to warping or height shift".

[0030] In practical implementation, the system first maps the positions of each height sampling point in the two-dimensional coordinate system of the PCB board. Then, combined with the PCB board design information, it extracts the local structural attributes corresponding to each sampling point. For example, it can extract or calculate local copper distribution characteristics, board thickness structural characteristics, local slotting or segmentation area characteristics, and other structural information that can reflect the board's stiffness and thermal deformation sensitivity, related to the area adjacent to the sampling point. Subsequently, the system jointly processes the aforementioned physical structural characteristics with the actually measured board height sampling data to establish a continuous board warpage height model covering the target mounting area. The board warpage height model can characterize the mapping relationship between the two-dimensional coordinates of the board surface and the local estimated height, thus enabling the system to not only utilize the original sampling point data but also perform height estimation for locations that have not been directly measured.

[0031] In this embodiment, the limitations of relying solely on pure geometric interpolation to construct the plate surface morphology are broken. By combining "measured data" with "structural mechanism", the obtained plate surface warping height model has better adaptability to plate edge areas, internal structural abrupt regions, and areas with significant local stiffness changes. This can reduce the prediction blind zone and local divergence caused by the limited number of measurement points.

[0032] In step S130, for the target mounting position of the component to be mounted on the PCB board, the corresponding estimated board height is calculated using the board warpage height model, and the Z-axis safe switching height of the pick-and-place nozzle is determined by combining the thickness information of the component to be mounted and the preset safety margin.

[0033] Here, the system first determines the target mounting position of the component to be mounted on the PCB board based on the current mounting program, and then calculates the estimated board height corresponding to the target mounting position using the constructed board warpage height model. Based on this, and combined with the thickness information of the component to be mounted and the preset safety margin, the system determines the Z-axis safe switching height corresponding to the control mode switching of the placement nozzle during the mounting press-down process.

[0034] In some implementations, the control system can read the XY target mounting coordinates of the component to be mounted from the production program and input these coordinates into the board warpage height model to obtain a local board height estimate at that location. Then, the system retrieves the component's outline thickness dimensions and, combined with the geometric assembly relationship between the mounting nozzle and the component, calculates the theoretical zero-contact position. Furthermore, a pre-set safety margin can be added beyond this theoretical zero-contact position, such as reserving a small space distance, so that the nozzle switches to the subsequent fine control mode before actually contacting the board surface or before the component is subjected to significant pressure.

[0035] Because the surface mount mechanism itself has mechanical inertia, and the miniature, brittle components are highly sensitive to instantaneous impacts, the system should not directly use the theoretical contact point as the final endpoint for pure position control. By defining a safe switching height along the Z-axis before actual contact occurs, a safe transition range is established during the downward pressure of the nozzle end. This ensures that when the system reaches this height, the bottom of the component is usually not yet in rigid contact with the board surface, thus reserving the necessary physical space and servo response time for subsequent control mode switching.

[0036] In step S140, the placement nozzle is controlled to descend to the safe switching height of the Z-axis carrying the component to be placed, and then switched to the force feedback correction mode.

[0037] Here, the placement nozzle is controlled to carry the component to be placed downward along the Z-axis until it reaches the determined safe switching height of the Z-axis. After the placement nozzle reaches the safe switching height, the control system switches from the previous stage of position trajectory-based pressure control to force feedback correction mode, so as to enter the fine control stage before approaching the actual contact point.

[0038] In practical implementation, the placement nozzle can perform a relatively fast downward pressing action according to conventional position control logic in the first half of its approach to the target placement area to ensure the placement cycle time of the production line. When approaching the safe switching height of the Z-axis, the system can monitor the current Z-axis coordinate of the placement nozzle in real time through position sensors, servo encoders, or other position detection methods; when the coordinate reaches the predetermined switching height, the control system triggers the corresponding mode switching control. In order to reduce the impact disturbance at the moment of switching, the placement nozzle can also adopt a smoother deceleration approach method in the section near the safe switching height, such as limiting the end abruptness (i.e., the rate of change of acceleration) or reducing the instantaneous downward pressing speed, so that the acceleration change of the placement nozzle near the safe switching height of the Z-axis tends to be gentle, thereby improving the dynamic stability at the moment of mode switching.

[0039] In this embodiment, the entire placement and pressing process is divided into two control zones: a "rapid approach stage" and a "fine contact stage." The former helps maintain equipment efficiency, while the latter enhances control flexibility when approaching the board surface. By completing the mode switch at the safe switching height, the nozzle can avoid using a single rigid pressing method at the end contact stage, thereby reducing the risk of component damage caused by position overshoot, local board surface protrusions, or component thickness variations.

[0040] In step S150, under the force feedback correction mode, the actual contact force of the placement nozzle during the continued pressing process is collected in real time, and based on the dynamic error between the actual contact force and the preset target placement force, the Z-axis displacement of the placement nozzle is adaptively corrected until the actual contact force falls within the preset error range corresponding to the target placement force, so as to complete the placement of the component to be placed.

[0041] Here, when the placement nozzle enters the force feedback correction mode, the system collects the actual contact force of the nozzle during its continued downward pressure in real time, and compares this actual contact force with the preset target placement force to obtain dynamic error information reflecting the deviation of the current contact state. Subsequently, based on this dynamic error, the control system performs adaptive feedback correction on the Z-axis displacement of the placement nozzle, so that the nozzle gradually adjusts its end displacement change while continuing to press down, until the actual contact force enters the preset error range corresponding to the target placement force; at this point, it can be determined that the current component has reached the placement contact state that meets the process requirements, and the subsequent release action is executed to complete the placement.

[0042] In some implementations, the system can utilize a force sensing unit located at the placement execution end to collect contact force signals in real time. The force sensing unit can be positioned at the end of the placement nozzle, the placement head, or a location related to the Z-axis execution link to reflect the actual force changes at the nozzle tip during contact. The control system performs a high-frequency comparison of this real-time contact force with a pre-set target placement force for the current component to generate a dynamic error. Based on this dynamic error, the system continuously generates corresponding Z-axis displacement corrections, enabling the nozzle to maintain a gentle downward pressure trend in the initial contact phase. When the contact force increases rapidly, the system suppresses further rigid downward pressure and mitigates end-point impact through dynamic correction. Once the actual contact force stabilizes within the allowable range, the system determines that the component has been properly placed and can further control the nozzle to release the vacuum, thus completing component release.

[0043] This application's embodiments complete the physical closed loop of the Z-axis compensation scheme. Even if the preceding board warpage height model still has slight prediction deviations in local locations, or if different batches of components have manufacturing variations in thickness, the system can still correct the pressing process online based on the actual force state at the end, thereby absorbing the cumulative effects of board surface topography errors, device size errors, and execution errors at the end contact stage. Therefore, it can balance effective bonding of micro-devices with force safety on complex warped board surfaces, improving the consistency, stability, and long-term robustness of the mounting process.

[0044] Regarding the implementation details of data acquisition in step S110, in some examples of embodiments of this application, an adaptive non-uniform topography sampling strategy guided by multi-source design prior features is provided.

[0045] First, the PCB design information of the PCB to be mounted is analyzed to extract the local copper area ratio distribution, component layout density distribution, and the location information of structural weak areas represented by slots and dividing lines on the two-dimensional plane of the PCB to be mounted.

[0046] In practical implementation, the system can extract data from each conductive layer by parsing Gerber, ODB++, or related CAD manufacturing engineering files to calculate the local copper coverage area ratio within a specified region. Simultaneously, it can extract data from assembly layers, silkscreen layers, or related layout files to calculate the component layout density. For physical boundaries formed by V-cuts, milled grooves, or panel dividing lines, the system can also extract their geometric coordinate information as location information for structurally weak areas.

[0047] It should be noted that the warping deformation of PCB boards caused by factors such as thermal cycling, assembly stress, or local structural weakening is usually strongly correlated with material thermal expansion mismatch, local stiffness changes, and structural boundary effects. Therefore, pre-extracting three types of physical structural attributes—local copper distribution, layout density, and structurally weak areas—can provide prior design constraints for subsequently identifying potential areas prone to significant warping.

[0048] Next, the two-dimensional plane of the PCB board to be mounted is divided into multiple basic evaluation areas. The gradient magnitude of the local copper area ratio, the component layout density, and the Euclidean distance to the nearest structurally weak area are extracted at the center coordinates of each basic evaluation area.

[0049] In some examples, the gradient magnitude of the local copper area ratio can be approximated by combining the central difference method with the copper area ratio of adjacent regions. Specifically, it can be calculated as follows: Equation (1) In the formula, This represents the ratio of the local copper cladding area at the corresponding coordinates. and denoted as the basic span of the basic evaluation area in the X-axis and Y-axis directions of the two-dimensional plane, respectively; in equation (1), the gradient of the change of the local copper area ratio between adjacent areas is obtained by the central difference algorithm, so as to quantify the non-uniformity of the local copper distribution on the board surface in space, thereby quantifying the drastic change of the local copper distribution on the board surface in space.

[0050] Next, after normalizing the gradient magnitude, component layout density, and exponential distance attenuation term constructed based on Euclidean distance and preset spatial attenuation constant, the normalized terms are weighted and summed using preset copper gradient weights, layout density weights, and structural weak area weights to calculate the warpage sensitivity index of each basic evaluation region.

[0051] For example, the warpage sensitivity index can be calculated using the following formula. : Equation (2) In the formula, This represents the normalized value of the gradient magnitude. This represents the normalized value of the component layout density. Representing coordinates Euclidean distance to the nearest structurally weak area This represents the distance attenuation term after normalization. , , These are the preset copper cladding gradient weights, layout density weights, and structurally weak area weights, respectively. This is a preset spatial attenuation constant used to control the nonlinear attenuation rate of the structural weakening effect with increasing distance; through the design of this exponential term, the closer to the structurally weak region (i.e., The smaller the position, the higher its sensitivity index to warping. The higher the contribution weight, the better.

[0052] It should be noted that areas with more drastic changes in copper cladding distribution are more prone to local deformation due to uneven thermal stress distribution; areas with denser component layouts typically have more complex heat capacity and residual stress distributions; and locations closer to structurally weak areas are usually more significantly affected by local stiffness reduction. To reduce the impact of dimensional differences between different characteristic quantities on the calculation results, the above normalization process can employ methods such as maximum-minimum value normalization to uniformly map each characteristic quantity to a dimensionless interval. By weighted fusion of the above factors, a comprehensive quantitative characterization of the warp sensitivity of different areas of the board surface can be formed.

[0053] Then, an adaptive mesh subdivision operation is performed based on the warp sensitivity index of each basic evaluation region: for basic evaluation regions with a warp sensitivity index higher than a preset sensitivity threshold, progressive spatial subdivision is performed to generate a non-uniform sampling mesh, so that the mesh node distribution density in the basic evaluation region where progressive spatial subdivision has been performed is greater than the mesh node distribution density in the basic evaluation region where progressive spatial subdivision has not been performed.

[0054] In some implementations, the adaptive subdivision process can be implemented using a quadtree segmentation algorithm. When the warpage sensitivity index corresponding to a certain basic evaluation region exceeds a preset sensitivity threshold, the system can further divide the region into multiple smaller sub-regions and continue to determine whether further subdivision is needed until a preset subdivision termination condition is reached, such as reaching a set maximum subdivision depth or the local sensitivity no longer meeting the conditions for further subdivision. In this way, the system can automatically generate denser sampling nodes in regions with high deformation risk and complex structural changes, while maintaining a sparser sampling density in relatively flat regions or regions with small structural changes, thereby achieving on-demand allocation of detection resources.

[0055] Then, the 3D scanning system is controlled to perform fixed-point height measurement on the PCB board to be mounted according to each grid node of the non-uniform sampling grid generated after subdivision, so as to obtain the height value corresponding to each grid node and combine them to generate board surface height sampling data.

[0056] In some implementations, the underlying control system can send the node coordinates corresponding to the aforementioned non-uniform sampling grid to a 3D scanning system, such as a laser displacement sensor, a 3D topography probe, or other devices capable of performing spatial height measurement. This allows the system to perform point-to-point ranging at specified node positions, thereby obtaining the actual board height value corresponding to each node. In this way, the system can perform more comprehensive height acquisition of potentially critical warping areas while maintaining low measurement overhead, without performing a global high-density scan of the entire PCB board, thus balancing sampling efficiency and sampling accuracy.

[0057] In this embodiment, by introducing a non-uniform topography sampling mechanism guided by prior design features, the system can estimate the warpage sensitivity of different areas of the PCB before formal mounting, based on prior information such as the local copper distribution, component layout, and structurally weak areas, and dynamically adjust the sampling grid density accordingly. This improves the sampling coverage of complex warpage areas while avoiding excessive detection resources in relatively flat areas, thus providing high-quality data input for building a board warpage height model that balances accuracy and efficiency.

[0058] Figure 2 A flowchart illustrating an example of constructing a board warpage height model according to an embodiment of this application is shown. Here, physical structural features include local copper area ratio, normalized laminate thickness parameters, and normalized relative distance from the geometric center of the PCB to be mounted.

[0059] like Figure 2 As shown, in step S210, for each sampling point on the PCB board to be mounted, the local copper area ratio, normalized laminate thickness parameters, and normalized relative distance corresponding to the sampling point are extracted based on the PCB board design information.

[0060] In practical implementation, the system can parse manufacturing engineering documents to target the first... The physical structural features of the local area are extracted from each sampling point location. Among them, the local copper coverage area ratio The material distribution ratio of the metal layer to the insulating substrate in this region is used to characterize the thermal stress distribution characteristics of this region under thermal cycling; laminate thickness parameters. To characterize the local bending resistance of the region, in some embodiments, the laminate thickness parameter can be obtained from the original thickness value after normalization and risk-oriented mapping, so that areas with thinner local laminate thickness and more prone to flexural deformation correspond to higher risk characterization values; normalized relative distance This parameter characterizes the degree to which the sampling point deviates from the geometric center of the PCB board. Since board warpage often exhibits a certain boundary accumulation effect in engineering, this parameter helps reflect the macroscopic deformation location characteristics of the sampling point. By performing normalization processing on the above parameters, the dimensional differences between different physical quantities can be eliminated, providing a unified data foundation for subsequent weight fusion.

[0061] In step S220, the normalized relative distance, local copper area ratio, and normalized laminate thickness parameters are multiplied by their corresponding empirical coefficients and then linearly summed. A basic constant is added to the summation result to calculate the physical weight value corresponding to each sampling point location. The empirical coefficients are used to characterize the influence of relative distance, copper distribution, and board thickness parameters on warping deformation, respectively.

[0062] For example, the following formula is used to calculate the first... The physical weight value corresponding to each sampling point location : Equation (3) In the formula, Indicates the local copper cladding area ratio, Represents the normalized laminate thickness parameters and Represents the normalized relative distance; All are non-negative empirical coefficients, used to characterize the influence weights of relative distance, copper cladding distribution, and board thickness parameters on warping deformation, respectively.

[0063] In Equation (3), an adaptive weight allocation mechanism that reflects the sensitivity of local deformation is constructed. The basic constant "1" in the equation is used to ensure that each effective sampling point participates in the subsequent fitting with the basic weight; the additional linear superposition term is used to introduce structural risk gain on top of the basic weight. Since the empirical coefficients are all non-negative, when a sampling point is closer to the edge of the board, the local copper cladding area ratio is larger (or the copper cladding is denser), or the local board thickness risk characterization value is higher, its corresponding physical weight value will increase accordingly. In this way, the subsequent fitting process can pay more attention to the height observation value of potentially high-risk deformation areas, thereby reducing the possibility of complex deformation areas being over-smoothed.

[0064] In step S230, an initial set of polynomials of different orders in a two-dimensional coordinate system is constructed, and the initial set of polynomials is orthogonalized to generate a set of orthogonal polynomial basis functions.

[0065] In practical implementation, if a conventional standard polynomial basis is directly adopted, for example... However, as the order increases, the numerical stability of coefficient calculation during the fitting process may decrease. To improve this, this embodiment first constructs an initial set of polynomials, and then uses the Gram-Schmidt orthogonalization method or other similar numerical processing methods to orthogonalize the initial set of polynomials, obtaining a set of orthogonal polynomial basis functions with good numerical properties on a given set of sampling points. Thus, by using orthogonal basis functions, the correlation between higher-order terms can be reduced, improving the stability of the subsequent solution process.

[0066] In step S240, the height sampling values ​​corresponding to each sampling point position in the board height sampling data are extracted. Based on the principle of weighted least squares, the optimization objective is to minimize the weighted sum of squared residuals between the height sampling values ​​at each sampling point position and the predicted values ​​of the linear combination of the orthogonal polynomial basis function set. Combined with the boundary constraint penalty term used to characterize the deformation continuity and displacement limitation of the PCB board to be mounted at the physical support boundary, the objective solution function is constructed. The weights of each term in the weighted sum of squared residuals are the physical weight values ​​corresponding to the sampling point positions.

[0067] For example, the objective function can be expressed as follows: Equation (4) In the formula, Solve for the objective function. The total number of sampling points. The number of basis functions of the orthogonal polynomial. The first in the set of basis functions of orthogonal polynomials basis functions For its two-dimensional coordinates The function value at that point, For the first The coordinates of each sampling point. The vector of fitting coefficients to be solved The first in One coefficient, As a penalty factor, This is a boundary constraint penalty term used to characterize the deformation continuity and displacement limitation of the PCB board to be mounted at the physical support boundary.

[0068] Equation (4) above constructs a weighted optimization problem with constraints. The first term is the weighted sum of squared residuals, which assigns different costs to the fitting error of different sampling points by the physical weight values ​​obtained in step S220. The second term is the boundary constraint penalty term, which introduces the mechanical constraints of the pick-and-place machine track clamping device or other boundary support conditions on the PCB board boundary area into the fitting process.

[0069] In some preferred embodiments of this application, it is assumed that the PCB board has a surface mount track. There are several boundary support points, with coordinates as follows: The corresponding reference displacement is Then the boundary constraint penalty term It can be constructed using the following formula: Equation (5) in, The reference value can be set according to the boundary support conditions; in some embodiments, it can be taken as the zero displacement reference. An appropriate penalty factor can be set. This allows the fitted surface to better meet the boundary support conditions during the solution process, thereby reducing the possibility of abnormal extension or unreasonable deviation of the surface in the edge region.

[0070] In step S250, the fitting coefficient vector of each orthogonal polynomial basis function is determined by solving the minimum value of the target solution function. Then, the fitting coefficient vector and the set of orthogonal polynomial basis functions are combined to construct a board warpage height model with continuous surface characteristics, which is used to output the estimated board height of the PCB board to be mounted at the two-dimensional coordinates of the target mounting position.

[0071] For example, the surface equation of the board warpage height model can be expressed by the following formula: Equation (6) In the formula, This indicates the PCB board to be mounted in any two-dimensional coordinate system. The estimated height of the board surface at that location For the first basis functions in two-dimensional coordinates The function value at that location.

[0072] In some implementations, the corresponding normal equation system can be established by setting the partial derivatives of the target solution function with respect to each coefficient to zero, and then solving for the fitted coefficient vector. Because orthogonal polynomial basis functions are used in the preceding steps, the solution process typically exhibits good numerical stability. Finally, the obtained fitting coefficient vector is linearly combined with the set of orthogonal polynomial basis functions to form a board surface warping height model describing the continuous morphology of the board surface. Thus, discrete, noisy, and spatially unevenly distributed height sampling points are transformed into a globally continuous two-dimensional surface representation with good local fidelity, enabling the pick-and-place machine to quickly query and predict the board surface height at any target mounting position.

[0073] By introducing physical structural features to construct adaptive fitting weights and combining orthogonal polynomial basis functions with boundary constraint penalty terms to complete the board surface fitting, the constructed board warpage height model can achieve good fitting accuracy in complex local deformation areas, while maintaining good physical consistency at the board boundaries and support areas. This provides a more stable foundation for subsequent high-precision Z-axis height compensation and contact control during the mounting process.

[0074] Figure 3 A flowchart illustrating an example of controlling the Z-axis safe switching height of the patch nozzle and switching to a force feedback correction mode in a method according to an embodiment of this application is shown.

[0075] like Figure 3 As shown, in step S310, the coordinates of the target mounting position on the two-dimensional plane are substituted into the board warping height model to solve and output the corresponding estimated board height.

[0076] In practice, the system first parses the process program corresponding to the current mounting task and extracts the target mounting position coordinates of the component to be mounted in the PCB board design coordinate system. Subsequently, the control system will assign coordinates Substitute the previously constructed board warpage height model. In the process, the estimated height of the plate at the target location is obtained. Therefore, by mapping the discrete two-dimensional mounting task positions to a continuous surface model that represents the actual three-dimensional deformation state of the PCB board, each component to be mounted can obtain a local board height reference corresponding to its target mounting position, instead of relying on a uniform theoretical height setting for the entire board.

[0077] In step S320, under a unified Z-axis reference coordinate system, the geometric length of the currently loaded pick-and-place nozzle is obtained, and the estimated board height, the thickness information of the component to be placed, and the geometric length of the pick-and-place nozzle are linearly superimposed in the vertical direction to obtain the theoretical zero-contact height of the component to be placed at the target placement position.

[0078] For example, theoretical zero-contact height It can be calculated using the following formula: Equation (7) In the formula, Indicates the theoretical zero-contact height. Indicates the estimated board height. This indicates the thickness information of the component to be mounted. This indicates the geometric length of the currently loaded patch nozzle. This indicates the target mounting position coordinates of the component to be mounted on a two-dimensional plane.

[0079] In practical implementation, to ensure the physical consistency of the calculation results, all the above parameters can be calibrated and uniformly mapped to the same Z-axis reference coordinate system of the pick-and-place machine. For example, the machine guide surface, the machine body reference surface, or other preset reference planes can be used as a unified height reference. Component thickness information The nozzle geometry can be obtained from component package databases, production process parameters, or visual measurements before mounting; This can include the nominal length of the nozzle and the assembly deviation compensation value of the nozzle replacement mechanism. Through the above linear superposition, the absolute Z-axis position of the placement head can be obtained when the bottom surface of the component and the estimated board surface are in zero-pressure contact.

[0080] In step S330, in the unified Z-axis reference coordinate system, when the Z-axis coordinate increases in the direction away from the plate surface, a preset safety margin is added to the theoretical zero contact height to calculate the safe switching height of the Z-axis.

[0081] For example, Z-axis safe switching height It can be calculated using the following formula: Equation (8) In the formula, Indicates the safe switching height along the Z-axis. This indicates a preset safety margin.

[0082] This application's embodiments reserve a safe transition range beyond the theoretical zero-contact height. Since the preceding board surface warpage modeling may still be affected by residual fitting errors, batch variations in component thickness, and thermal drift during mechanical transmission, directly using the theoretical zero-contact height as the high-speed pressing endpoint could easily lead to end-point collision risks when the actual height of the local board surface exceeds the estimated height. Because in a unified Z-axis reference coordinate system, the Z-axis coordinate increases along the direction away from the board surface, therefore... On the basis of adding safety margin This is equivalent to setting the termination point of the high-speed position control stage a short distance above the theoretical contact position, thereby ensuring that the placement nozzle reaches the target position. At this time, a small gap is usually maintained between the bottom of the component and the board surface, providing physical buffer space for subsequent control mode switching.

[0083] In step S340, an asymmetric S-shaped motion speed curve is constructed for the patch nozzle to descend from the initial preparation position to the safe switching height of the Z-axis, so that the patch nozzle exhibits a smooth deceleration state that limits abrupt changes in the section near the safe switching height of the Z-axis.

[0084] In some preferred embodiments, to balance placement cycle time and end-approach stability, the bottom-layer motion planner may not use a fully symmetrical trapezoidal velocity curve or a symmetrical S-curve, but instead generate an asymmetrical S-curve motion velocity curve according to different stages of the pressing process. Among them, in the vicinity Within the deceleration zone, the abruptness can be constrained to make the acceleration change at the end of the downward pressure smoother. For example, the kinematic constraints for this zone can be expressed as:

[0085] Equation (9) In the formula, For the patch nozzle The actual Z-axis position at that moment. For instantaneous acceleration, This is the maximum permissible jerk threshold for the system.

[0086] Through the Limiting the acceleration at the end of deceleration can mitigate the sudden acceleration change, reduce the likelihood of mechanical chatter and residual vibration in the patch nozzle as it approaches the target position, and ensure that it reaches the target position. This allows for a lower and smoother instantaneous velocity, thus providing more favorable initial dynamic conditions for the subsequent switching process from position control to force feedback control.

[0087] In step S350, when the servo control system detects that the real-time Z-axis coordinate of the patch nozzle has reached the safe switching height of the Z-axis, it triggers a hard real-time interrupt to automatically freeze the integral state quantity in the pure position servo control loop, and smoothly switches to the force feedback control loop corresponding to the force feedback correction mode, so as to reduce the impact disturbance at the moment of control mode switching.

[0088] In practical implementation, during the low-speed deceleration phase of position servo control, a slight deviation between the actual position and the theoretical trajectory may cause the controller's integral term to continuously accumulate errors. If this integral state is not processed at the moment of switching to force feedback control mode, it may cause a momentary output jump during the switch, resulting in slight end-effector jitter or impact. Therefore, in this embodiment, the step involves detecting when the real-time Z-axis coordinate of the patch nozzle reaches... When a hard real-time interrupt is triggered, the integral state variables in the pure position servo control loop are frozen, held, or cleared, and the force feedback control loop is switched in a disturbance-free manner. This hard real-time interrupt can be implemented by a high-speed real-time control unit, a field-programmable gate array, or other control modules with fast response capabilities. This allows the control system to achieve a smooth transition from position control to force feedback control in the critical stage before contact occurs, reducing the risk of mechanical disturbances and output abrupt changes during control mode switching, thus helping to maintain the continuity of force and motion stability as the nozzle tip approaches the plate surface.

[0089] This application's embodiments combine the predicted local board height at the target mounting location with the geometric dimensions of the component and nozzle to determine the theoretical zero-contact height and its corresponding safe switching height. Furthermore, by integrating asymmetric S-shaped motion planning and integral state freezing mode switching, more stable execution conditions can be provided for the transition from the position control stage to the force feedback correction stage while ensuring mounting efficiency. This helps reduce the impact and disturbance of the mounting nozzle when approaching the board surface, improving the mounting safety and stability of micro-packaged devices on complex warped board surfaces.

[0090] In some examples of embodiments of this application, during the process of controlling the patch nozzle to lower the component to be patched to the safe switching height of the Z-axis in step S140, a multi-axis attitude and coordinate correction step can be further performed.

[0091] In specific implementation scenarios, severe local warping can not only cause an absolute height shift in the board surface but also potentially create a significant spatial tilt near the target mounting location. If the pick-and-place nozzle maintains a fixed vertical posture during downward pressure, the bottom surface of the component to be mounted may initially make localized contact with the board surface from one edge, leading to localized stress concentration and further affecting the uniformity of stress distribution and soldering stability after component mounting. Therefore, in some embodiments, posture correction and horizontal coordinate correction can be introduced simultaneously at the end stage near the board surface to improve mounting consistency under complex warping conditions.

[0092] First, the surface equations based on the constructed plate warpage height model. Determine the target mounting position In a two-dimensional plane, with respect to the coordinate variables of each plane (e.g., ... and The partial derivative value of ).

[0093] For example, the local geometric slope along the X and Y directions at the target mounting location can be directly extracted using the previously constructed continuous surface model, thus providing basic gradient information to characterize the tilt trend of the board surface near that location. Compared with simply approximating the local slope through discrete neighborhood points, directly obtaining the partial derivatives based on the continuous surface equations allows for more convenient acquisition of local differential information at the target mounting location, provided the model is known.

[0094] Next, using the obtained partial derivative values ​​and the preset vertical direction components, a three-dimensional local normal vector at the target mounting position is constructed and normalized in the spatial coordinate system.

[0095] For example, the three-dimensional local normal vector at the target mounting location can be calculated using the following formula. : Equation (10) In the formula, For three-dimensional local normal vectors, and The equations of the surfaces are respectively In coordinates Regarding coordinate variables and The partial derivative value, the constant "1" represents the preset vertical direction component.

[0096] In the above equation (10), the plate surface can be regarded as an implicit function surface. The gradient vector at the corresponding point represents the normal direction at that point. By normalizing the gradient vector, a unit normal vector that reflects only the direction attribute can be obtained. This is used to describe the spatial orientation of a local panel surface at the target mounting location.

[0097] Then, decompose the three-dimensional local normal vectors. The direction cosine component in the spatial coordinate system is used to calculate the spatial tilt angle of the PCB board to be mounted at the target mounting position due to warping deformation. The spatial tilt angle includes roll angle component and pitch angle component. For example, let the components of the three-dimensional local normal vector in the preset spatial rectangular coordinate system be... Under a predefined coordinate system, the spatial tilt attitude angle can be obtained through the following inverse trigonometric function relationship:

[0098] Equation (11) In the formula, Represents the roll angle component. Represents the pitch angle component. , and These are the three-dimensional local normal vectors. Components along the X, Y, and Z axes. In this way, the system can convert the local normal information of the target mounting position into an attitude adjustment amount that the mounting actuator can call upon.

[0099] Subsequently, a multi-axis collaborative attitude adjustment command is generated to control the multi-degree-of-freedom servo actuator of the pick-and-place machine to dynamically adjust the spatial attitude of the pick-and-place nozzle according to the spatial tilt attitude angle, so that the central axis of the pick-and-place nozzle is aligned with the three-dimensional local normal vector in the same direction.

[0100] In practical implementation, the placement actuator can be a placement head mechanism with multi-degree-of-freedom attitude adjustment capabilities. The control system drives the actuator to adjust the spatial attitude of the placement nozzle based on the calculated roll and pitch components. This allows the approach attitude of the placement nozzle and the component to be placed to better conform to the spatial tilt of the local board surface, thereby reducing the potential problems of one-sided initial contact and localized force concentration during fixed vertical downward pressure.

[0101] Furthermore, based on the spatial tilt angle, the estimated board height, and the equivalent geometric length between the rotation center and the mounting reference point, the geometric offset of the component to be mounted on the horizontal projection plane is calculated. The geometric offset is then used as a feedforward compensation parameter and superimposed on the XY axis horizontal positioning command of the pick-and-place machine to correct the horizontal mounting coordinate deviation caused by the local warping and tilting of the PCB board to be mounted.

[0102] In actual mechanism operation, when the mounting head deflects, its end component typically experiences a certain geometric offset in the horizontal plane relative to its original target position. To achieve precise alignment, under a preset coordinate definition, this geometric offset can be quantified through the following spatial projection relationship:

[0103] Equation (12) In the formula, and These represent the geometric offsets of the components to be mounted along the X and Y axes on the horizontal projection plane, respectively. This represents the equivalent geometric length between the rotation center of the surface mount nozzle and the mounting reference point on the bottom surface of the component. and These are the roll and pitch components, respectively. Subsequently, the control system uses these geometric offsets as position feedforward compensation and adds them to the original XY-axis horizontal positioning command to reduce the horizontal positioning deviation caused by local tilt and attitude deflection.

[0104] In this embodiment, by utilizing the local differential information of the board warp height model to extract the normal direction at the target mounting position, and further generating attitude adjustment and horizontal compensation amounts, the system can simultaneously perform collaborative correction of the end mounting attitude and horizontal coordinates in addition to Z-axis height compensation. This helps improve the alignment consistency between the component to be mounted and the locally tilted pads, reduces the mounting risks caused by unilateral stress concentration and associated horizontal offset, and thus enhances the stability and reliability of micro-device mounting under complex warp conditions.

[0105] Figure 4 A flowchart illustrating an example of adaptive feedback correction of the Z-axis displacement of the patch nozzle based on dynamic error in a method according to an embodiment of this application is shown.

[0106] like Figure 4 As shown, in step S410, the dynamic error is defined as the difference between the target mounting force and the actual contact force collected at the current moment.

[0107] For example, at the current moment Dynamic error is defined as ,in, For target mounting force, For the current moment The actual contact force collected.

[0108] In specific implementation scenarios, once the control system switches to force feedback correction mode, the mechanical sensing device located at the end of the mounting head or on the Z-axis drive link can continuously collect the current actual contact force. Target mounting force This can be preset based on the package type of the component to be mounted, the state of the bottom solder paste, or relevant process parameters. The dynamic error is defined as the difference between the target mounting force and the current actual contact force. The system can quantify the deviation between the current contact state and the ideal bonding state in real time. In the initial contact stage, the actual contact force is usually small, and the dynamic error is positive and relatively large; as the nozzle continues to press down and move the component closer to the pad and solder paste area, the actual contact force gradually increases, the dynamic error decreases accordingly and tends to converge.

[0109] In step S420, the initial moment of switching to the force feedback correction mode is set as the integration starting point. Based on the nonlinear adaptive feedback control law, the required real-time Z-axis displacement adjustment within the current servo interpolation cycle is calculated.

[0110] In some implementations, the control components of the nonlinear adaptive feedback control law include a proportional adjustment term, an integral adjustment term, and a second-order nonlinear derivative adjustment term. The proportional adjustment term is calculated based on the dynamic error at the current moment and the proportional gain coefficient. The integral adjustment term is calculated based on the time integral of the dynamic error from the integration start point to the current moment and the integral gain coefficient. The second-order nonlinear derivative adjustment term is calculated based on the product of the absolute value of the rate of change of the dynamic error and the rate of change of the dynamic error itself, combined with the derivative adjustment gain coefficient.

[0111] For example, the required real-time Z-axis displacement adjustment within the current servo interpolation cycle is calculated using the following formula. : Equation (13) In the formula, This is the proportional gain coefficient. This is the integral gain coefficient. For integration time variable, For being in Dynamic error at any given time, This is a second-order nonlinear differential adjustment term constructed based on the rate of change of dynamic error. This is the differential adjustment gain coefficient.

[0112] To facilitate implementation in a digital servo controller, in a preferred embodiment of this application, the aforementioned continuous control law can be discretized according to the interpolation period of the servo system. Let... The underlying servo interpolation period is defined, and the discrete-time index corresponding to the initial moment of the cutting force feedback correction mode is defined as follows: Then, after entering this mode, on the [number]th [day]... Real-time Z-axis displacement adjustment during each sampling period It can be obtained using the following formula:

[0113] Equation (14) In the formula, and They represent the current number of times. Each sampling period and the previous one The dynamic error is calculated from each sampling period. Among them, the preceding dynamic error at the initial time... Default initialization is (That is, the error rate of change is defined as zero at the initial moment to avoid numerical values ​​going out of bounds).

[0114] Here, the proportional adjustment term provides a basic correction corresponding to the current error amplitude, the integral adjustment term compensates for the accumulated error since the self-feedback correction mode was activated to reduce steady-state deviation, and the quadratic nonlinear differential adjustment term utilizes the quadratic effect of the dynamic error rate of change to dynamically characterize the degree of force change at the instant the nozzle contacts the PCB board to be mounted, and is used to nonlinearly enhance the characterization of the degree of change of the dynamic error rate of change; when the actual contact force suddenly increases, causing the dynamic error to shrink sharply, the quadratic nonlinear differential adjustment term amplifies the dynamic error rate of change by square the magnitude, and realizes the nonlinear response to the intensity of the force change; when the impact stress increases sharply at the moment of contact, this term can generate a reverse braking displacement bias that is much larger than that of the conventional linear differential term, thereby realizing the instantaneous convergence of the pressing action.

[0115] In step S430, the real-time Z-axis displacement adjustment amount is... As a positional offset command characterizing the pressing depth, it acts on the Z-axis pressing motion trajectory of the placement nozzle within each bottom-level servo interpolation cycle (e.g., under the coordinate definition where the Z-axis coordinate increases away from the board surface, this adjustment is subtracted from the current target coordinate), driving the placement nozzle to contact the PCB board to be placed with an adaptively decaying pressing speed until the actual contact force at the current moment. It falls within the preset error range to complete the adaptive feedback correction of the Z-axis displacement of the patch nozzle.

[0116] In practical implementation, the above calculations yielded... Or discretized It does not need to directly replace the original absolute position command, but can be injected into the trajectory generator of the current cycle of the Z-axis in real time as a high-frequency dynamic position offset along the downward direction.

[0117] In this way, the downward trajectory of the placement nozzle can be continuously corrected as the actual contact force changes: when the contact force increases rapidly, the single-cycle downward displacement is appropriately reduced, making the nozzle exhibit a gentler approach characteristic; when the contact force change tends to be stable, the system maintains a small correction amount, so that the overall contact process gradually converges to the target placement force range. The control system can continuously execute the above high-frequency correction process until it detects that the actual contact force in several consecutive servo interpolation cycles has stably fallen within the tolerance band corresponding to the target placement force. At this point, it can be determined that the component has reached the bonding state that meets the process requirements, and further component release control can be executed.

[0118] In this way, the fixed position is no longer used as the sole basis for the final placement, but the actual stress state is used as the final convergence target. Even if there are still local residual errors in the prediction of the preceding board height, or if there are certain discrepancies in the component thickness and solder paste state, the system can still compensate for the end contact online through the above-mentioned dynamic displacement correction process, thereby reducing the risks of over-punch contact, sudden stress changes, and placement instability, and improving the placement consistency and reliability of micro-packaged devices under complex board conditions.

[0119] In this embodiment, by constructing a nonlinear adaptive feedback control law comprising proportional, integral, and quadratic nonlinear derivative terms, and applying it to the Z-axis pressing trajectory as a high-frequency position bias, the system can maintain early approach efficiency while providing a more sensitive dynamic response to sudden force changes during the contact phase. This helps to balance operational efficiency, contact flexibility, and force safety during the mounting process of miniature, brittle packaged devices.

[0120] In some examples of embodiments of this application, to further improve the model adaptability and prediction accuracy of the system during long-term operation, a data closed-loop self-learning update step for process parameters can also be performed. In this embodiment, the process parameters are preferably reflected as empirical coefficient updates used to calculate physical weight values.

[0121] In a business context, even PCB boards from different batches with the same design structure may exhibit subtle differences in local stiffness or thermal expansion response due to supplier variations, differences in material glass transition temperatures, changes in storage temperature and humidity, or fluctuations in processing conditions. If the board warpage height model uses fixed parameters for a long period, the model prediction results may gradually drift systematically with batch changes. Therefore, in some embodiments, closed-loop updates of the preceding model parameters can be performed between batches based on actual mounting results and subsequent inspection results.

[0122] Specifically, firstly, after completing the mounting operation of the PCB boards to be mounted in the current production batch, the steady-state Z-axis coordinates of each target mounting position when the actual contact force falls within the preset error range are extracted. Then, combined with the thickness information of the components to be mounted and the geometric length of the mounting nozzle, the actual physical board height at each target mounting position is calculated.

[0123] For example, the first can be calculated by the following reverse relationship. The actual physical board height at the target mounting location : Equation (15) In the formula, Indicates the target mounting location The actual absolute Z-axis coordinate of the patch nozzle when the actual contact force meets the process requirements and reaches a steady state; This indicates the thickness information of the component to be mounted; This indicates the geometric length of the patch nozzle.

[0124] In this embodiment, the steady-state contact result achieved by the preceding force feedback correction mode is used to estimate the actual height of the current batch of boards. When the actual contact force meets the process requirements and enters a stable range, it can be considered that a stable contact relationship satisfying the mounting requirements has been formed between the component bottom surface and the local board surface. At this time, the corresponding Z-axis coordinate, after deducting the component thickness and nozzle geometry, can be used as the inverse result of the actual board height at the target location. Thus, a posteriori correction information on the morphology of the current batch of boards can be obtained based on the actual mounting process itself.

[0125] Next, the deformation prediction residual between the estimated board height output by the board warpage height model for each target mounting position and the corresponding actual physical board height is calculated.

[0126] For example, deformation prediction residuals It can be calculated using the following formula: Equation (16) In the formula, For the board warpage height model, target the mounting position. The output is the estimated board height. By calculating this residual, the system can quantify the direction and magnitude of the prediction deviation of the model at different target mounting positions in the current batch, providing an error information basis for subsequent parameter correction.

[0127] Then, obtain the automatic optical inspection results output by the automatic optical inspection equipment for the current production batch, extract the mounting defect distribution data that characterizes the abnormality of component suspension or cracking, and map the mounting defect distribution data to the corresponding spatial penalty weight. In actual production lines, automated optical inspection equipment is typically used after the surface mount technology (SMT) process. Different types of height deviations caused by overestimating or underestimating model predictions can manifest as insufficient mount contact, overload, or abnormal positioning, among other defect tendencies. To incorporate defect information into the parameter update process, the following spatial penalty weight mapping relationship can be constructed:

[0128] Equation (17) In the formula, Indicates the space penalty weight at the target mounting location; This represents the quantified value of the mounting defect severity at this location extracted from the AOI results. For example, it can be assigned values ​​according to levels such as no defects, minor abnormalities, and obvious abnormalities. This is a preset defect sensitivity gain factor. The constant "1" ensures that the residuals in defect-free areas can also participate in subsequent model updates with basic weights. This method allows locations with significant mounting anomalies to receive greater attention during parameter correction.

[0129] Subsequently, by combining the deformation prediction residuals and spatial penalty weights at each target mounting location, a weighted loss function is constructed to evaluate the prediction accuracy of the board warp height model.

[0130] For example, for the current production batch containing Target mounting location, weighted loss function It can be represented as: Equation (18) In the formula, This represents the vector of empirical coefficients to be updated, which are empirical coefficients that respectively characterize the influence of relative distance, copper cladding distribution, and board thickness parameters on warpage deformation; This represents the deformation prediction residual affected by the current empirical coefficient vector.

[0131] In the loss function shown in equation (18), the ordinary residual and the severity of the defect are jointly weighted, so that the position related to obvious mounting anomalies has a greater impact in subsequent updates. The control system applies a recursive least squares algorithm (or other nonlinear iterative optimization algorithms, such as the Gauss-Newton method, gradient descent method, etc.) to iteratively solve the above weighted loss function containing nonlinear mapping relationships, so as to update the empirical coefficients when calculating the physical weight values ​​online. The updated empirical coefficients are then applied to the construction process of the board warpage height model for subsequent production batches of PCB boards to achieve systematic deviation compensation for the fluctuations in stiffness and thermal expansion rate of different batches of boards.

[0132] Here, optimization algorithms such as recursive least squares can utilize the newly added residual samples and spatial penalty weight information of the current batch to perform batch-to-batch recursive correction of the empirical coefficient vector based on the calculated Jacobian matrix or error gradient, without needing to save the original data of all historical batches. After completing the parameter update for the current batch, the system can automatically send the updated empirical coefficients to the model building module of subsequent batches for recalculating the physical weight values ​​of each sampling point. Through this continuous batch-to-batch recursive approach, the model can gradually adjust its internal parameters according to the differences in sheet materials, environmental changes, and process drift during the production process, thereby reducing the systematic prediction bias caused by long-term parameter solidification.

[0133] Furthermore, a recursive least squares algorithm is applied to iteratively solve the weighted loss function to update the empirical coefficients used in calculating the physical weights online. The updated empirical coefficients are then applied to the construction process of the board warpage height model for subsequent production batches of PCB boards to achieve systematic deviation compensation for the fluctuations in stiffness and thermal expansion rate of different batches of boards.

[0134] Here, the recursive least squares algorithm can utilize the newly added residual samples and spatial penalty weight information of the current batch to perform batch-by-batch recursive correction of the empirical coefficient vector without saving the original data of all historical batches. After completing the parameter update for the current batch, the system can automatically send the updated empirical coefficients to the model building module of subsequent batches for recalculating the physical weight values ​​of each sampling point. Through this continuous batch-by-batch recursion, the model can gradually adjust its internal parameters as the production process changes due to variations in sheet materials, environmental changes, and process drift, thereby reducing the systematic prediction bias caused by long-term parameter fixation.

[0135] In this embodiment, by combining the actual board height information obtained from force feedback steady-state coordinate inversion with the downstream automatic optical inspection results, a data closed-loop mechanism for updating model parameters is constructed. This allows the system to continuously correct the empirical coefficients in the board warpage height model. This helps improve the model's adaptability to different batches of boards, different environmental conditions, and long-term process drift, reducing the cumulative impact of local prediction errors on mounting quality, thereby improving mounting stability and yield under long-term operating conditions.

[0136] Figure 5 The diagram illustrates an example of a Z-axis compensation method for integrated circuit patching for PCB warpage according to an embodiment of this application. The compensation mechanism can be mainly divided into four highly collaborative and interactive logical modules in the system architecture: input data area, core processing / model area, scheduling and control / constraint area, and output and result area.

[0137] like Figure 5 As shown, in the feedforward prediction and path planning process, the system first collects multi-source board height data (such as discrete point clouds obtained by sparse laser or structured light scanning) through the input data area, and simultaneously parses the design file to extract local copper distribution, board thickness, and structural information. This multi-source heterogeneous data is input into the core processing / model area to perform warping modeling that integrates physical prior information; subsequently, the system calculates the predicted Z-height based on the generated continuous board height model, and plans the initial Z-axis landing point and safe descent path of the patch nozzle accordingly, thereby establishing a highly reliable spatial benchmark for the downward pressing action.

[0138] In the dynamic control and macroscopic closed-loop path, when the system descends along a predetermined safe path and triggers specific conditions, the scheduling and control / constraint area will issue a mode switching command (starting from the safe descent input force feedback). At this time, the core processing area, combined with the real-time contact force signals transmitted from sensors such as the nozzle strain gauge, calculates the dynamic adjustment amount of the Z-axis at high frequency in the adaptive force-feedback correction module. Simultaneously, the multi-axis attitude collaborative planning module, combined with the central scheduling logic, outputs corresponding attitude adjustment signals (for X / Y / (Axis). The aforementioned control commands work together on the multi-degree-of-freedom actuators in the output and result areas, ensuring the precision and smoothness of the patch placement action. Furthermore, the patch placement results and steady-state data feedback obtained from the output area are fed into a long-cycle model iteration and parameter self-learning loop, which feeds back to the warp modeling module through a closed loop to continuously optimize the underlying empirical coefficients, thereby achieving systematic adaptive compensation for fluctuations in sheet stiffness and thermal expansion across batches.

[0139] To fully verify the effectiveness of the weighted orthogonal polynomial surface fitting and adaptive force feedback correction collaborative compensation mechanism proposed in this application embodiment under complex working conditions, this application embodiment constructs a board-level simulation environment containing physical properties to compare the differences in mounting accuracy, contact force impact, and process stability of different control strategies. Specifically, this embodiment constructs a board-level simulation environment with a size of... A high-density PCB virtual model is used as the simulation object. To highly replicate the real production environment, a ground truth surface model of this virtual model is employed. It is designed to consist of three superimposed parts: First, a macroscopic warping mode combining saddle-shaped and torsional warping is introduced based on the mismatch of thermal expansion coefficients, with a maximum warping amount set at 1.2 mm to simulate extreme conditions exceeding the IPC standard limit; second, a nonlinear sinking feature is introduced in locally dense copper areas based on the actual copper foil distribution density map to simulate the local stress release phenomenon after reflow soldering; finally, the standard deviation is superimposed. Gaussian white noise is used to simulate the measurement uncertainty of a real laser sensor.

[0140] In this simulation environment, three different Z-axis control strategies were set up for lateral comparison. The first was a traditional plane fitting scheme as the benchmark group, which used a conventional bilinear plane fitting algorithm to calculate the overall plate height based only on finite corner point data; the second was a global spline interpolation scheme as the control group, which used... The cubic spline interpolation algorithm for grid sampling aims to simulate the fitting method commonly used in high-end optical inspection equipment. However, this method is prone to producing divergent Runge phenomena at the boundaries of the plates. The third method is the comprehensive compensation scheme proposed in the embodiments of this application, which adopts a weighted orthogonal polynomial surface fitting model that integrates physical prior constraints. The weight coefficients in the model... All are dynamically generated based on the imported PCB design files (such as copper area ratio and board thickness), and an adaptive force feedback control loop is incorporated at the execution end.

[0141] Furthermore, unlike traditional evaluation methods that only perform static height subtraction, this embodiment establishes a complete closed-loop control model for the Z-axis servo motor in a dynamic simulation platform. This dynamic simulation model integrates a mass-spring-damping system characterizing the physical properties of the chip nozzle, and specifically introduces a pressure sensor feedback response loop in the comprehensive compensation scheme corresponding to this embodiment. Therefore, this simulation environment can not only verify the accuracy of static surface prediction, but also realistically reproduce and output the dynamic force-displacement response curve of the chip head at the instant of contact with the PCB board surface, thus providing an extremely reliable dynamic verification basis for evaluating the stress stability and safety of micro-components under complex warping conditions.

[0142] Figure 6This diagram illustrates a comparative simulation of different methods for constructing a slab warpage height model. The simulation uses three sets of three-dimensional spatial residual distribution topographic maps to visually compare the differences in slab height reconstruction accuracy between different prediction methods. In the figure, the Z-axis represents the residual magnitude between the predicted height and the ground truth height (unit: ...). The color mapping bar at the bottom is used to visually represent the polarity of the error: the blue area represents negative error (i.e. the estimated height is too low, which can easily lead to small components being suspended or having poor soldering in actual placement), the red area represents positive error (i.e. the estimated height is too high, which can easily lead to overload or even cracking of components during placement), and the white area represents high precision zero error.

[0143] like Figure 6 As shown, in the comparison group, scheme A (using conventional spline interpolation) performed poorly on the complex warped surface, exhibiting a significant "bowl-shaped" large-scale divergence error distribution, with the maximum deviation reaching as high as [missing information]. This level of error, if directly applied to the mounting of extremely small miniature packages such as 0201 or 01005, will lead to severe defects such as misplacement or missing components. While Solution B (using conventional weighted interpolation) improves accuracy in the inner center area of ​​the PCB, the lack of effective constraints on physical boundaries results in extremely strong edge oscillation effects (i.e., Runge phenomenon) at the PCB edges and corners, causing a sudden jump in edge error. .

[0144] In contrast, Scheme C (using the fusion of physical weighting and boundary constraint algorithm proposed in this application) generates the smoothest and flattest residual error surface. This is because the scheme innovatively introduces physical weights calculated based on prior structural features such as local copper distribution and board thickness, and strictly superimposes the constraint penalty equations of the physical support boundaries. The model not only effectively suppresses edge divergence but also accurately captures nonlinear sinking or bulging caused by local physical material differences. Experimental results show that even under extreme warping conditions exceeding the IPC standard limit, the residual error of Scheme C is still strictly converged and controlled within approximately [value missing]. Within the extremely small fluctuation band. This fully verifies that the board warpage height model constructed in this invention can provide an extremely reliable and high-precision Z-axis prediction benchmark for micro-pitch devices.

[0145] Figure 7This diagram illustrates a comparative simulation of different methods for controlling a patch nozzle during a dynamic placement process. The diagram, using a dual Y-axis time-domain plot and a force-velocity phase trajectory plot (top right inset), demonstrates the difference in dynamic response between pure position control and the integrated control method (WOPFWC + AFFC) of this application embodiment during a typical placement action (including Z-axis pressing, contact, and holding pressure stages). In the diagram, the left Y-axis represents the Z-axis pressing displacement of the patch nozzle, the right Y-axis represents its actual contact force, and the horizontal axis represents time.

[0146] like Figure 7 In the comparative experiment shown, when using the feedback-free "pure position control" scheme, the patch nozzle descends at high speed blindly according to a preset fixed height. Due to model prediction errors caused by local board warping, the nozzle... Premature rigid contact with the PCB surface caused the instantaneous impact force to surge to 6.5N, far exceeding the 3N physical damage threshold set for micro-components (such as 0201 packages), accompanied by severe stress oscillations. In contrast, when using the comprehensive control method of this application embodiment (i.e., combining physical weighted surface fitting and adaptive force feedback correction), the system smoothly switches to force-resistance control mode after reaching the safe switching height. As shown by the solid curve in the figure, when the sensor detects a tiny initial contact force of 0.2N, the system immediately triggers dynamic feedback adjustment, and the Z-axis pressing speed rapidly and smoothly generates decay braking, achieving a "soft landing" of the micro-component on the complex warped board surface.

[0147] Combination Figure 7 As further illustrated in the force-velocity phase trajectory illustration in the upper right corner, the method of this embodiment can guide the system to converge smoothly to the target contact point, avoiding phase trajectory oscillations that diverge without feedback control. At the end of the mounting process, the final steady-state mounting force of this embodiment is precisely and without overshoot. Within the target control band. Although the introduction of the adaptive force feedback correction algorithm introduces an additional adjustment time of approximately 12ms, resulting in a slight increase in the local cycle time of a single placement, this mechanism fundamentally eliminates the risks of component cracking and poor soldering due to poor contact. Considering the significant improvement in overall line yield and the substantial reduction in rework costs, this technical trade-off of extremely low time loss for extremely high placement force safety has extremely high process value and application advantages in actual high-density miniaturized packaging production lines.

[0148] Figure 8This diagram illustrates a comparative simulation of the mounting pressure tolerance distribution under different control methods. The figure visually demonstrates the statistical results of the overall process capability of different control strategies in 500 Monte Carlo random mounting simulations using a histogram of mounting force distribution and its fitted normal distribution curve. The horizontal axis represents the actual mounting force (in N) when the nozzle contacts the component, and the vertical axis represents the frequency. To clarify the safe process window, two vertical red dashed lines mark the set process thresholds: 0.5 N on the left is the lower limit of force (below this value there is a risk of poor contact or cold solder joints), and 3.0 N on the right is the upper limit of force (above this value there is a risk of micro-cracks causing micro-components to fracture).

[0149] like Figure 8 In the comparative experiment, the mounting pressure distribution of scheme A (baseline group) using the traditional algorithm exhibited a broad, divergent, and highly irregular bimodal characteristic. Due to the failure to effectively compensate for local height prediction errors caused by complex warping and the lack of a flexible force control mechanism, this scheme resulted in approximately... The force exerted on the mounting point exceeded the aforementioned safety specifications. This means that in actual high-volume, high-density mounting production, this solution will face extremely high yield losses and rework risks.

[0150] In contrast, the mounting pressure distribution using the integrated control method of this application embodiment (i.e., scheme C) exhibits an extremely sharp and highly convergent standard normal distribution characteristic. Based on the seamless synergy of the aforementioned high-precision physical weighted surface fitting and adaptive force feedback correction mechanism, the method of this embodiment successfully and accurately limits the vast majority of mounting forces within the safe target range, and its process capability index (… It reached an excellent level of 1.67 (typically) (This can be considered as sufficient process capability). This statistical data fully demonstrates that the Z-axis collaborative compensation mechanism proposed in this invention can effectively absorb and resist physical disturbances caused by complex board surface deformation, thereby significantly improving the process consistency, stress safety, and long-term yield stability of ultra-small and micro devices in mass assembly production.

[0151] Experimental data analysis shows that the WOPFWC (Weighted Orthogonal Polynomial Fitting with Warpage Constraints) and AFFC (Adaptive Force Feedback Control) collaborative compensation method proposed in this application, through the organic combination of physical statistical modeling and closed-loop force control, effectively solves the engineering challenges of global topography prediction and local random error compensation. In simulating severe warping (… Under extreme operating conditions, compared to traditional planar assumptions or single geometric fitting schemes, the method in this embodiment improves the accuracy of mounting height prediction by approximately 60% and significantly reduces component stress overshoot by over 75%. Although simulation studies reveal that when dealing with extremely high-frequency surface textures (such as densely packed micro-hole regions with spatial undulation wavelengths less than 5 mm), the model may exhibit a slight smoothing effect due to limitations in sampling grid density. However, this characteristic can be optimized in practical applications by flexibly balancing sampling density and production cycle time according to board complexity. Experimental results fully demonstrate the significant advantages of this invention in maintaining high-precision mounting consistency and stress safety, especially showing strong applicability in mounting scenarios for high-density, fine-pitch devices and flexible / large-size boards.

[0152] In this embodiment, a complete Z-axis collaborative compensation system with self-learning and evolutionary capabilities is constructed by combining innovative technologies such as multi-source design information analysis, non-uniform mesh adaptive sampling, orthogonal basis function surface reconstruction, and smooth mode switching of the underlying servo drive. First, by integrating prior information on copper distribution and material structure, the height model possesses inherent physical credibility. Second, through deep collaboration between force feedback and the prediction model, the defects of static preset height control or single optical schemes ignoring dynamic contact forces are fundamentally overcome. Furthermore, the synchronously introduced spatial multi-axis attitude correction and parameter closed-loop self-learning mechanism based on AOI feedback enable the system to automatically adapt to material drift in different batches of substrates while ensuring high-fidelity 3D bonding. Thus, without significantly increasing equipment hardware costs, the overall process stability and long-term operating yield of the SMT production line are greatly improved, providing key technical support for the next generation of high-end integrated circuit placement systems.

[0153] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of combined actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Secondly, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0154] Figure 9 A structural block diagram of an example of an integrated circuit patch Z-axis compensation system for PCB board warping is shown according to an embodiment of this application.

[0155] like Figure 9 As shown, the integrated circuit patch Z-axis compensation system 900 for PCB board warping includes a data acquisition unit 910, a warping modeling unit 920, a switching height determination unit 930, a mode switching control unit 940, and a force feedback correction unit 950.

[0156] The data acquisition unit 910 is used to acquire the board height sampling data of the PCB board to be mounted and the PCB board design information of the PCB board to be mounted.

[0157] The warpage modeling unit 920 is used to extract the physical structural features corresponding to each sampling point position on the PCB board to be mounted based on the PCB board design information, and to construct a board warpage height model that integrates the physical structural features by combining the board height sampling data.

[0158] The switching height determination unit 930 is used to calculate the corresponding estimated board height using the board warpage height model for the target mounting position of the component to be mounted on the PCB board, and to determine the Z-axis safe switching height of the pick-and-place nozzle by combining the thickness information of the component to be mounted and the preset safety margin.

[0159] The mode switching control unit 940 is used to control the placement nozzle to carry the component to be placed down to the safe switching height of the Z-axis and switch to the force feedback correction mode.

[0160] The force feedback correction unit 950 is used to collect the actual contact force of the patch nozzle during the continued pressing process in the force feedback correction mode, and perform adaptive feedback correction on the Z-axis displacement of the patch nozzle based on the dynamic error between the actual contact force and the preset target mounting force, until the actual contact force falls into the preset error range corresponding to the target mounting force, so as to complete the mounting of the component to be mounted.

[0161] In some embodiments, this application provides a non-volatile computer-readable storage medium storing one or more programs including execution instructions. The execution instructions can be read and executed by an electronic device (including but not limited to a computer, server, or network device) to perform the steps of any of the above-described integrated circuit patch Z-axis compensation methods for PCB board warping.

[0162] In some embodiments, this application also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of any of the above-described integrated circuit patch Z-axis compensation methods for PCB board warping.

[0163] In some embodiments, this application also provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform steps of an integrated circuit patch Z-axis compensation method for PCB board warping.

[0164] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0165] The electronic devices in this application can exist in various forms, including but not limited to: mobile communication devices, ultra-mobile personal computer devices, portable entertainment devices, or other airborne electronic devices with data interaction functions.

[0166] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0168] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for Z-axis compensation of integrated circuit patch panels for PCB board warpage, characterized in that, The method includes: Obtain the board height sampling data of the PCB board to be mounted and the PCB board design information of the PCB board to be mounted; Based on the PCB design information, the physical structural features corresponding to each sampling point on the PCB to be mounted are extracted, and combined with the board height sampling data, a board warpage height model that integrates the physical structural features is constructed. For the target mounting position of the component to be mounted on the PCB board, the corresponding estimated board height is calculated using the board warpage height model, and the Z-axis safe switching height of the pick-and-place nozzle is determined by combining the thickness information of the component to be mounted and the preset safety margin. Control the placement nozzle to lower the component to be placed to the safe switching height of the Z-axis, and switch to force feedback correction mode; In the force feedback correction mode, the actual contact force of the patch nozzle during the continued pressing process is collected in real time, and based on the dynamic error between the actual contact force and the preset target mounting force, the Z-axis displacement of the patch nozzle is adaptively corrected until the actual contact force falls within the preset error range corresponding to the target mounting force, so as to complete the mounting of the component to be mounted.

2. The method according to claim 1, characterized in that, The acquisition of the board surface height sampling data of the PCB board to be mounted and the PCB board design information of the PCB board to be mounted includes: The PCB design information of the PCB to be mounted is analyzed to extract the local copper area ratio distribution, component layout density distribution, and structural weak area location information represented by slots and dividing lines on the two-dimensional plane of the PCB to be mounted. The two-dimensional plane of the PCB board to be mounted is divided into multiple basic evaluation areas. The gradient magnitude of the local copper area ratio, the component layout density, and the Euclidean distance to the nearest structural weak area are extracted at the center coordinates of each basic evaluation area. After normalizing the gradient magnitude, the component layout density, and the exponential distance attenuation term constructed based on the Euclidean distance and the preset spatial attenuation constant, the normalized terms are weighted and summed using preset copper cladding gradient weights, layout density weights, and structurally weak area weights to calculate the warpage sensitivity index of each basic evaluation area. Based on the calculated warp sensitivity index of each basic evaluation region, an adaptive mesh subdivision operation is performed: for the basic evaluation region where the warp sensitivity index is higher than a preset sensitivity threshold, a progressive spatial subdivision is performed to generate a non-uniform sampling mesh, so that the mesh node distribution density in the basic evaluation region where the progressive spatial subdivision has been performed is greater than the mesh node distribution density in the basic evaluation region where the progressive spatial subdivision has not been performed. The control system performs fixed-point height measurement on the PCB board to be mounted according to each grid node of the non-uniform sampling grid generated after subdivision, so as to obtain the height value corresponding to each grid node and combine them to generate board surface height sampling data.

3. The method according to claim 1, characterized in that, The physical structural features include the local copper cladding area ratio, the normalized laminate thickness parameters, and the normalized relative distance from the geometric center of the PCB board to be mounted. The step of extracting the physical structural features corresponding to each sampling point position on the PCB board to be mounted based on the PCB board design information, and constructing a board warpage height model that integrates the physical structural features by combining the board surface height sampling data, includes: For each sampling point location on the PCB board to be mounted, the local copper area ratio, the normalized laminate thickness parameter, and the normalized relative distance corresponding to the sampling point location are extracted based on the PCB board design information. The normalized relative distance, the local copper area ratio, and the normalized laminate thickness parameter are multiplied by their respective empirical coefficients and then linearly summed. A basic constant is added to the summation result to calculate the physical weight value corresponding to each sampling point position. The empirical coefficients are used to characterize the influence of relative distance, copper distribution, and board thickness parameters on warpage deformation, respectively. Construct an initial set of polynomials of different orders in a two-dimensional coordinate system, and perform orthogonalization on the initial set of polynomials to generate a set of orthogonal polynomial basis functions; The height sampling values ​​corresponding to each sampling point position in the board height sampling data are extracted. Based on the principle of weighted least squares, the optimization objective is to minimize the weighted sum of squared residuals between the height sampling values ​​at each sampling point position and the predicted values ​​of the linear combination of the orthogonal polynomial basis function set. Combined with the boundary constraint penalty term used to characterize the deformation continuity and displacement limitation of the PCB board to be mounted at the physical support boundary, the objective solution function is constructed. The weights of each term in the weighted sum of squared residuals are the physical weight values ​​corresponding to the sampling point positions. By solving for the minimum value of the target solution function, the fitting coefficient vector of each orthogonal polynomial basis function is determined. Then, the fitting coefficient vector and the set of orthogonal polynomial basis functions are combined to construct a board warpage height model with continuous surface characteristics, which is used to output the estimated board height of the PCB board to be mounted at the two-dimensional coordinates of the target mounting position.

4. The method according to claim 3, characterized in that, The step of calculating the corresponding estimated board height using the board warpage height model, and determining the Z-axis safe switching height of the placement nozzle by combining the thickness information of the component to be mounted and the preset safety margin, includes: Substitute the coordinates of the target mounting position on the two-dimensional plane into the board warpage height model to solve and output the corresponding estimated board height. Under a unified Z-axis reference coordinate system, the geometric length of the currently loaded patch nozzle is obtained, and the estimated board height, the thickness information of the component to be mounted, and the geometric length of the patch nozzle are linearly superimposed in the vertical direction to obtain the theoretical zero-contact height of the component to be mounted at the target mounting position. In the unified Z-axis reference coordinate system, when the Z-axis coordinate increases in the direction away from the plate surface, the preset safety margin is added to the theoretical zero contact height to calculate the safe switching height of the Z-axis. The step of controlling the placement nozzle to lower the component to be placed to the safe switching height along the Z-axis and switching to the force feedback correction mode includes: An asymmetric S-shaped motion speed curve is constructed for the patch nozzle as it descends from the initial preparation position to the safe switching height of the Z-axis, so that the patch nozzle exhibits a smooth deceleration state with limited abrupt changes in the section near the safe switching height of the Z-axis. When the servo control system detects that the real-time Z-axis coordinate of the patch nozzle has reached the safe switching height of the Z-axis, it triggers a hard real-time interrupt to automatically freeze the integral state quantity in the pure position servo control loop and smoothly switches to the force feedback control loop corresponding to the force feedback correction mode to reduce the impact disturbance at the moment of control mode switching.

5. The method according to claim 4, characterized in that, During the process of controlling the placement nozzle to lower the component to be placed to the safe switching height along the Z-axis, the method further includes a multi-axis attitude and coordinate correction step, specifically including: Based on the surface equation of the constructed board warpage height model, the partial derivative values ​​of the target mounting position with respect to each plane coordinate variable in the two-dimensional plane are obtained; Using the obtained partial derivative values ​​and the preset vertical direction components, a three-dimensional local normal vector at the target mounting position is constructed and normalized in the spatial coordinate system. The direction cosine component of the three-dimensional local normal vector in the spatial coordinate system is decomposed, and the spatial tilt attitude angle of the PCB board to be mounted at the target mounting position due to warping deformation is calculated. The spatial tilt attitude angle includes roll angle component and pitch angle component. Generate multi-axis collaborative attitude adjustment commands to control the multi-degree-of-freedom servo actuator of the pick-and-place machine to dynamically adjust the spatial attitude of the pick-and-place nozzle according to the spatial tilt attitude angle, so that the central axis of the pick-and-place nozzle is aligned with the three-dimensional local normal vector in the same direction and collinearity. Based on the spatial tilt angle, the estimated board height, and the equivalent geometric length between the rotation center and the mounting reference point, the geometric offset of the component to be mounted on the horizontal projection plane is calculated. The geometric offset is then used as a feedforward compensation parameter and superimposed on the XY axis horizontal positioning command of the pick-and-place machine to correct the horizontal mounting coordinate deviation caused by the local warping and tilting of the PCB board to be mounted.

6. The method according to claim 1, characterized in that, The adaptive feedback correction of the Z-axis displacement of the patch nozzle based on the dynamic error between the actual contact force and the preset target mounting force, until the actual contact force falls within the preset error range corresponding to the target mounting force, includes: The dynamic error is defined as the difference between the target mounting force and the actual contact force collected at the current moment; Let the initial moment of switching to the force feedback correction mode be the starting point of integration. Based on the nonlinear adaptive feedback control law, calculate the real-time Z-axis displacement adjustment required in the current servo interpolation cycle. The control components of the nonlinear adaptive feedback control law include a proportional adjustment term, an integral adjustment term, and a second-order nonlinear derivative adjustment term. The proportional adjustment term is calculated based on the dynamic error and proportional gain coefficient at the current moment; the integral adjustment term is calculated based on the time integral of the dynamic error from the integration start point to the current moment and the integral gain coefficient; and the second-order nonlinear differential adjustment term is calculated based on the product of the absolute value of the dynamic error change rate and the dynamic error change rate itself, combined with the differential adjustment gain coefficient. The quadratic nonlinear differential adjustment term utilizes the quadratic effect of the dynamic error change rate to dynamically characterize the degree of force change at the instant the pick-and-place nozzle contacts the PCB board to be mounted. When the actual contact force undergoes a sharp increase, causing the dynamic error to decrease drastically, the quadratic nonlinear differential adjustment term generates a dynamic correction component to suppress under-pressure overshoot in a manner that increases with the magnitude of the dynamic error change rate, thereby mitigating the local force impact caused by under-pressure inertia. The real-time Z-axis displacement adjustment is used as a position offset command and applied to the Z-axis downward motion trajectory of the placement nozzle in each underlying servo interpolation cycle. This drives the placement nozzle to contact the PCB board to be placed with an adaptively decaying downward pressing speed until the actual contact force at the current moment falls within the preset error range, thereby completing the adaptive feedback correction of the Z-axis displacement of the placement nozzle.

7. The method according to claim 3, characterized in that, The method also includes a data closed-loop self-learning update step for process parameters, specifically including: After completing the mounting operation of the PCB board to be mounted in the current production batch, the steady-state Z-axis coordinates of each target mounting position when the actual contact force falls within the preset error range are extracted, and the actual physical board height at each target mounting position is calculated by combining the thickness information of the component to be mounted and the geometric length of the mounting nozzle. Calculate the deformation prediction residual between the estimated board height output by the board warpage height model for each of the target mounting positions and the corresponding actual physical board height; Obtain the automatic optical inspection results output by the automatic optical inspection equipment for the current production batch, extract the mounting defect distribution data that characterizes the abnormality of component suspension or cracking, and map the mounting defect distribution data to the corresponding spatial penalty weight. By combining the deformation prediction residuals at each of the target mounting positions with the spatial penalty weights, a weighted loss function is constructed to evaluate the prediction accuracy of the board warp height model. The weighted loss function is iteratively solved using a recursive least squares algorithm to update the empirical coefficients used to calculate the physical weight values ​​online. The updated empirical coefficients are then applied to the construction process of the board warpage height model for subsequent production batches of PCB boards to achieve systematic deviation compensation for the fluctuations in stiffness and thermal expansion rate of different batches of boards.

8. A Z-axis compensation system for integrated circuit patch panels to address PCB board warpage, characterized in that, The system includes: The data acquisition unit is used to acquire the board surface height sampling data of the PCB board to be mounted and the PCB board design information of the PCB board to be mounted. The warpage modeling unit is used to extract the physical structural features corresponding to each sampling point position on the PCB board to be mounted based on the PCB board design information, and to construct a board warpage height model that integrates the physical structural features by combining the board height sampling data. The switching height determination unit is used to calculate the corresponding estimated board height using the board warpage height model for the target mounting position of the component to be mounted on the PCB board, and to determine the Z-axis safe switching height of the pick-and-place nozzle by combining the thickness information of the component to be mounted and the preset safety margin. The mode switching control unit is used to control the placement nozzle to carry the component to be placed down to the safe switching height of the Z-axis and switch to the force feedback correction mode; The force feedback correction unit is used to collect the actual contact force of the patch nozzle during the continued pressing process in the force feedback correction mode, and perform adaptive feedback correction on the Z-axis displacement of the patch nozzle based on the dynamic error between the actual contact force and the preset target mounting force, until the actual contact force falls within the preset error range corresponding to the target mounting force, so as to complete the mounting of the component to be mounted.

Images