Paster quality closed-loop control method and device based on online three-dimensional X-ray detection

Abstract

The invention provides a patch quality closed-loop control method and device based on online three-dimensional X-ray detection. The method comprises the steps that soldering paste is formed on a current circuit board, and patch is conducted; acquiring three-dimensional body data of soldering paste on the current circuit board; extracting three-dimensional morphological characteristic parameters of the soldering paste according to the three-dimensional volume data; importing the three-dimensional morphological characteristic parameters into a pre-trained hole risk prediction model, and predicting the possibility that the current circuit board generates holes after reflow soldering; and according to the possibility, process parameter adjustment is carried out, new process parameters are adopted to form soldering paste on the circuit board with the preparation time later than that of the current circuit board, surface mounting is carried out, and after reflow soldering is carried out on the circuit board with the preparation time later than that of the current circuit board, a plurality of formed welding spots meet the quality requirement. Through the method, the welding spot defects can be predicted before reflow soldering, and the process parameters are improved, so that the problem of batch welding spot defects is effectively avoided, the yield is improved, and the rework cost is reduced.

Description

Technical Field

[0001] This application belongs to the field of electronic assembly surface mount technology, and particularly relates to a closed-loop control method and device for patch quality based on online three-dimensional X-ray inspection. Background Technology

[0002] In surface mount technology, solder joint voids are one of the key defects affecting product reliability. Voids reduce the mechanical strength, electrical conductivity, and thermal conductivity of solder joints, posing a serious threat to high-reliability fields such as aerospace, automotive electronics, and high-end communication equipment.

[0003] Solder joint quality inspection (including void detection) is generally performed after reflow soldering, using 2D or 3D X-ray inspection equipment to inspect the solder joints formed after the reflow soldering process. However, when inspecting solder joints formed after reflow soldering, if defective products are found, they can only be scrapped or subjected to costly rework. Solder paste quality inspection, on the other hand, is performed after the solder paste has formed on the circuit board, using a solder paste inspection machine to check its quality. However, simply checking the solder paste quality after its formation process cannot accurately reflect the risk of solder joint voids.

[0004] How to provide a method to prevent solder joint voids and avoid the occurrence of a large number of solder joint voids is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] In view of this, this application provides a closed-loop control method and apparatus for patch quality based on online three-dimensional X-ray inspection, which aims to predict the risk of solder joint voids on circuit boards in advance, so as to reduce rework costs.

[0006] In a first aspect, this application provides a closed-loop control method for patch quality based on online three-dimensional X-ray inspection, including: Solder paste is formed on the current circuit board, and components are mounted on the current circuit board according to the solder paste. Before reflow soldering the current circuit board, acquire the three-dimensional volume data of the solder paste on the current circuit board; Extract the three-dimensional morphological feature parameters of the solder paste from the three-dimensional volume data; The three-dimensional morphological feature parameters are imported into a pre-trained hole risk prediction model to predict the likelihood of holes being generated on the current circuit board after reflow soldering. Based on the aforementioned possibilities, process parameters are adjusted, and new process parameters are used to form solder paste and mount it on a circuit board whose preparation time is later than the current circuit board, so that after reflow soldering on the circuit board whose preparation time is later than the current circuit board, several solder joints formed meet the quality requirements.

[0007] Optionally, the three-dimensional morphological feature parameters include: Solder paste volume, solder paste thickness distribution, solder paste shape profile, and alignment offset between component leads and pads.

[0008] Optionally, the step of extracting the three-dimensional morphological feature parameters of the solder paste based on the three-dimensional volume data includes: The three-dimensional volume data is voxelized to identify and segment the three-dimensional region of each solder paste. The volume of each solder paste is calculated based on the number of voxels in the three-dimensional region of each solder paste. Perform a three-dimensional Euclidean distance transformation on the three-dimensional region of each solder paste to obtain a distance map of the three-dimensional region of each solder paste, and then obtain the solder paste thickness distribution. Calculate the three-dimensional moment features of the three-dimensional region of each solder paste to obtain the solder paste shape profile; By using image segmentation or template matching, the positions of component pins and several solder pastes on the circuit board in three-dimensional space are identified respectively. The Euclidean distance and azimuth angle between the three-dimensional area of ​​each solder paste and its corresponding component pin are calculated as the alignment offset between the component pin and the pad.

[0009] Optionally, the step of adjusting process parameters based on the likelihood of voids forming on the current circuit board after reflow soldering includes: When the probability of a hole being generated on the current circuit board after reflow soldering does not meet a preset threshold, a process parameter control command is generated to control one or more process parameters of the pick-and-place machine, including placement pressure, placement height, placement coordinates, squeegee pressure, speed, and demolding parameters. Optionally, the process for determining the quality of process parameter control is as follows: Based on the distribution pattern of the defective solder paste, the root cause of the defect is determined, and the process parameters to be optimized are determined based on the root cause of the defect. The process parameters to be optimized are then optimized to obtain process parameter control instructions.

[0010] Alternatively, if only a single process parameter needs to be optimized, the process parameter control command can be generated using the following formula:

[0011] In the formula, Indicates the adjustment amount of a certain process parameter; This represents the proportionality coefficient, used to adjust the ratio of... Response speed; This represents the difference between the actual solder paste volume and the target volume. Represents the integral coefficient, used to eliminate Historical accumulation deviations; This represents the historical cumulative sum of deviations of the parameter over several previous production cycles.

[0012] Optionally, if multiple process parameters are to be optimized simultaneously, process parameter control instructions are generated using the following method: Identify the process parameters to be optimized and form a combination of process parameters to be optimized; An optimization algorithm is used to optimize the combination of process parameters to find the combination that minimizes the difference between the actual solder paste volume and the target volume, thereby generating process parameter control instructions.

[0013] Secondly, this application provides a patch quality closed-loop control device based on online three-dimensional X-ray inspection, comprising: The surface mount module forms a number of solder pastes on the current circuit board and performs surface mount on the current circuit board according to the solder pastes; The acquisition module is used to acquire three-dimensional volume data of several solder pastes on the current circuit board before reflow soldering the current circuit board; The extraction module is used to extract the three-dimensional morphological feature parameters of the solder paste based on the three-dimensional volume data. The prediction module is used to import three-dimensional morphological feature parameters into a pre-trained hole risk prediction model to predict the likelihood of holes being generated on the current circuit board after reflow soldering. The adjustment module is used to adjust the process parameters according to the possibilities, and to use the new process parameters to form solder paste and mount it on a circuit board whose preparation time is later than the current circuit board, so that after reflow soldering on the circuit board whose preparation time is later than the current circuit board, the formed solder joints meet the quality requirements.

[0014] Thirdly, this application provides an electronic device, including the patch quality closed-loop control device based on online three-dimensional X-ray inspection as described above.

[0015] Fourthly, this application provides a computer-readable storage medium storing at least one piece of program code, which is executed by a processor to implement the patch quality closed-loop control method based on online three-dimensional X-ray inspection as described in any of the preceding claims.

[0016] The beneficial effects of the technical solution provided in this application include: (1) This application performs solder joint void prediction after surface mount technology (SMT) and before reflow soldering, which can mitigate the risk of solder joint voids in advance. In addition, existing technical solutions generally only consider the influence of the solder paste formation process on solder joint void defects, without considering the influence of the surface mount process on solder joint void defects. Performing three-dimensional inspection of the solder paste after surface mount technology (SMT) can reflect the combined influence of the solder paste formation process and the surface mount process on the three-dimensional morphology of the solder paste. Therefore, when making void risk prediction based on this three-dimensional data, the prediction accuracy will be more accurate.

[0017] (2) Utilizing the CT scanning capability of 3D Automatic X-ray Inspection (AXI), online full inspection of the quality of hidden solder joints such as BGA and QFN after mounting is performed, with better detection accuracy and effect.

[0018] (3) Defect-process parameter closed-loop control: Through a closed-loop process of "data detection-data analysis-defect prediction-parameter adjustment", the production line is equipped with adaptive and self-optimizing process parameter adjustment capabilities, thereby improving the overall intelligence level of the production line. When the production line predicts, based on the three-dimensional data of the solder paste after surface mount, that the solder paste on the current circuit board has a high risk of solder joint voids after reflow soldering, it will intercept the current circuit board and perform process parameter optimization steps on all circuit boards after the current circuit board to improve the solder paste quality of subsequent circuit boards, thereby reducing the risk of solder joint voids in subsequent circuit boards after reflow soldering. In addition, this application can determine the risk of solder joint voids before reflow soldering, thereby intercepting circuit boards with higher risks before reflow soldering, avoiding the problem of increased rework difficulty after continued reflow soldering. When intercepting risky circuit boards before reflow soldering, since the solder paste on the surface of the circuit board has not been converted into solder joints, the rework difficulty will be lower. It is only necessary to remove the solder paste and re-form the solder paste. Furthermore, since this application can automatically optimize process parameters based on the risk of solder joint voids in the previous circuit board, it can also avoid the problem of solder joint voids in a batch of circuit boards, which is conducive to further reducing rework costs. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention 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 invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 A flowchart of a patch quality closed-loop control method based on online three-dimensional X-ray inspection provided in an embodiment of this application; Figure 2 A structural block diagram of a patch quality closed-loop control device based on online three-dimensional X-ray inspection provided in an embodiment of this application; Figure 3 A schematic diagram of a patch quality closed-loop control device based on online three-dimensional X-ray inspection provided in an embodiment of this application; Figure 4 This is a structural block diagram of a data processing and analysis server provided in one embodiment of this application; Figure 5 A schematic diagram of a control method executed by a patch quality closed-loop control device based on online three-dimensional X-ray inspection, provided in an embodiment of this application; Figure 6 This is a structural block diagram of an electronic device provided in an embodiment of this application.

[0021] The attached figures are labeled as follows: 11: Patch module; 12: Acquisition module; 13: Extraction module; 14: Prediction module; 15: Adjustment module; 21: Solder paste printer; 211: Squeegee; 212: Wire mesh; 22: Patch mounter; 221: Robotic arm; 222: Electronic components; 223: Circuit board; 23: Online 3D X-ray inspection equipment; 231: Microfocus X-ray source; 232: High-speed flat panel detector; 233: Precision motion platform; 24: Data processing and analysis server; 241: Feature extraction module; 242: Hole risk prediction model; 243: Decision and control module; 25: Central database; 26: Control communication interface; 27: Conveyor belt; 31: Processor; 32: Memory. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] Soldering components onto a circuit board generally involves the following process: First, solder paste is formed in the soldering area of ​​the components on the circuit board. Then, the components are placed on the solder paste surface of the soldering area by a pick-and-place machine. Finally, the solder paste is transformed into solder joints by reflow soldering, and the components are then soldered onto the circuit board.

[0024] For the soldering quality of components on a circuit board, 2D or 3D inspection equipment is typically used to inspect the morphology of the solder joints after reflow soldering to determine the soldering quality. Alternatively, after solder paste forms on the circuit board, a solder paste inspection machine is used to check the solder paste and determine its quality. However, the surface mount process (component placement process) alters the original form of the solder paste (such as extrusion or offset), and this change is crucial for the formation of voids in the final solder joints. Currently, there is a general lack of effective means to monitor and control the quality of the surface mount process before reflow soldering.

[0025] Based on this, the core concept of this application is to inspect the three-dimensional morphology of the solder paste after surface mount technology (SMT) and before reflow soldering. This allows for prediction of whether solder joint voids will occur after reflow soldering, thus enabling early avoidance of such issues. Furthermore, when a circuit board is detected to have a high probability of solder joint voids, the system automatically identifies the specific process (solder paste formation or component mounting) responsible. Finally, by adjusting the process parameters of the corresponding process, the risk of voids appearing on all subsequent circuit boards can be reduced.

[0026] Figure 1 This is a flowchart illustrating a patch quality closed-loop control method based on online three-dimensional X-ray inspection, provided in one embodiment of this application. See also... Figure 1 ,include: S101. Solder paste is formed on the current circuit board, and the solder paste is applied to the surface of the current circuit board.

[0027] In some examples, the current circuit board includes a circuit board, or other circuit boards that include solder joints.

[0028] In some examples, solder paste can be formed on designated areas of the circuit board surface using a metal mesh and a squeegee. After the metal mesh is attached to the circuit board surface, the solder paste is pressed onto the circuit board surface through the holes in the metal mesh using a squeegee.

[0029] In some examples, a pick-and-place machine can be used to place components onto solder paste in a designated area on the surface of the current circuit board. Then, when a reflow soldering process is performed, the solder paste is converted into solder joints, which solder the components onto the current circuit board.

[0030] S102. Before reflow soldering the current circuit board, obtain the three-dimensional volume data of the solder paste on the current circuit board.

[0031] In some examples, step S102 includes: An online 3D X-ray inspection device is used to perform a computed tomography scan on the current circuit board to obtain high-resolution three-dimensional volume data containing all solder paste.

[0032] S103. Extract the three-dimensional morphological feature parameters of the solder paste based on the three-dimensional volume data.

[0033] In some examples, step S103 includes: S1031. Perform voxelization on the three-dimensional volume data to identify and segment the three-dimensional region of each solder paste.

[0034] In some examples, step S1031 includes: By using a pre-trained deep learning segmentation model, the three-dimensional volume data is processed to accurately identify and segment the three-dimensional region of each solder paste.

[0035] The deep learning segmentation model adopts a 3D U-Net architecture. This model is trained using a large amount of labeled 3D solder paste data, which can effectively distinguish solder paste, component leads, PCB substrate, and pads, and obtain a probability map of each voxel belonging to the solder paste region. Based on this probability map, a binarized 3D solder paste mask is generated by setting a confidence threshold.

[0036] S1032. Calculate the volume of each solder paste based on the number of voxels in the three-dimensional region of each solder paste.

[0037] In some examples, step S1032 includes: The physical volume can be directly calculated by counting all voxels in the segmented 3D region of solder paste and then multiplying them by the physical volume of a single voxel.

[0038] S1033. Perform a three-dimensional Euclidean distance transformation on the three-dimensional region of each solder paste to obtain a distance map of the three-dimensional region of each solder paste, and then obtain the solder paste thickness distribution.

[0039] In some examples, step S1033 includes: A three-dimensional Euclidean distance transformation is performed on the segmented three-dimensional solder paste region. This method can calculate the distance from each voxel within the three-dimensional region of the solder paste to the nearest solder paste boundary. This distance value can be approximated as the local thickness. By analyzing the distance map of the entire three-dimensional region of the solder paste, characteristics such as the thickness distribution, maximum / minimum / average thickness of the solder paste can be obtained.

[0040] S1034. Calculate the three-dimensional moment features of the three-dimensional region of each solder paste to obtain the solder paste shape profile.

[0041] In some examples, step S1034 includes: Calculate the three-dimensional moment characteristics (such as centroid, principal axis direction, moment of inertia, etc.) of the three-dimensional region of the solder paste. These characteristics are invariant to translation, rotation and scaling and can be used to describe and distinguish the overall three-dimensional shape of the solder paste (such as fullness, depression, bridging).

[0042] S1035. By image segmentation or template matching, identify the positions of component pins and several solder pastes on the circuit board in three-dimensional space, calculate the Euclidean distance and azimuth angle between the three-dimensional area of ​​each solder paste and its corresponding component pin, and use it as the alignment offset between the component pin and the pad.

[0043] In some examples, step S1035 includes: By using image segmentation or template matching, the positions of component pins in three-dimensional space and the theoretical positions of circuit board design pads in three-dimensional data are identified respectively. The three-dimensional centroids of the regions where the component pins and design pads are located are calculated. The Euclidean distance and orientation angle between the two centroids in the XY plane (or three-dimensional space) are calculated, and the alignment offset between the component pins and the pads is obtained.

[0044] S104. Import the three-dimensional morphological feature parameters into the pre-trained hole risk prediction model to predict the probability of holes being generated on the current circuit board after reflow soldering.

[0045] In some examples, the hole risk prediction model is a machine learning model, which is trained by constructing a dataset using three-dimensional morphological feature parameters in historical states and hole detection results of several solder points on the current circuit board surface after reflow soldering.

[0046] In some examples, the hole risk prediction model is obtained by performing regression analysis on historical 3D data and actual hole detection results after reflow soldering using machine learning algorithms.

[0047] In some examples, during actual production, the three-dimensional shape data of the circuit board is collected each time it is detected, as well as whether holes appear at the solder joints after reflow soldering. This data serves as a dataset for incremental learning, which facilitates subsequent incremental learning of the hole risk prediction model and improves its predictive ability.

[0048] S105. Based on the aforementioned possibility, adjust the process parameters and use the new process parameters to form solder paste and mount it on a circuit board whose preparation time is later than the current circuit board, so that after reflow soldering is performed on the circuit board whose preparation time is later than the current circuit board, the resulting solder joints meet the quality requirements.

[0049] In some examples, after obtaining the probability value through step S104, the probability value is compared with a preset qualified threshold. If the probability value is lower than the qualified threshold, the current circuit board is allowed to enter the reflow soldering process without modifying the process parameters. If the risk probability value is higher than the qualified threshold, the current circuit board needs to be intercepted and the process parameters need to be corrected to ensure the solder joint quality of circuit boards whose preparation time is later than the current circuit board.

[0050] In some examples, "current board" refers to a board that has completed solder paste preparation and surface mount technology (SMT) and is ready for reflow soldering. A board whose preparation time is later than the current board refers to another board whose defect process is executed later than the solder joint void risk prediction time of the current board.

[0051] Among them, defective processes refer to processes that cause solder joint void risks. For example, if the solder joint void risk of the current circuit board is due to the surface mount technology (SMT) process, then "processing time later than the current circuit board" refers to other circuit boards whose SMT process is later than the solder joint void risk prediction time of the current circuit board.

[0052] In some examples, when the probability of a hole being generated on the current circuit board after reflow soldering does not meet a preset threshold, a process parameter adjustment instruction is generated to adjust one or more process parameters of the pick-and-place machine, including placement pressure, placement height, placement coordinates, squeegee pressure, speed, and demolding parameters.

[0053] In some examples, step S105 includes: S1051. Based on the distribution pattern of the defective solder paste, determine the root cause of the defect, and determine the process parameters to be optimized based on the root cause of the defect.

[0054] In some examples, step S1051 includes: Based on the distribution pattern of the defective solder paste, the root cause of the defect is determined to be either the surface mount process or the solder paste printing process. When the defect is concentrated on a specific component, the root cause is determined to be the surface mount process, and the generated control instructions include adjusting the mounting pressure, mounting height, or mounting coordinates. When the defect is distributed regionally along the squeegee stroke direction, the root cause is determined to be the solder paste printing process, and the generated control instructions include adjusting the squeegee pressure, squeegee speed, or demolding parameters.

[0055] S1052. Optimize the process parameters to be optimized to obtain process parameter control instructions.

[0056] In some examples, the generation process of a single process parameter control command is as follows:

[0057] In the formula, Indicates the adjustment amount of a certain process parameter; This represents the proportionality coefficient, used to adjust the ratio of... Response speed; This represents the difference between the actual solder paste volume and the target volume. Represents the integral coefficient, used to eliminate Historical accumulation deviations; This indicates the deviation of a certain process parameter in the historical production cycle.

[0058] Each process parameter has a preset standard range. If adjusting one process parameter can ensure that the circuit board prepared later than the current circuit board meets the quality requirements, then only one process parameter needs to be adjusted.

[0059] In other examples, adjusting a single process parameter may fail to meet quality requirements for circuit boards manufactured later than the current circuit board. For instance, the adjusted value of the process parameter calculated using the above formula might cause it to exceed the standard range. In such cases, multiple process parameters are adjusted to ensure that circuit boards manufactured later than the current circuit board meet the quality requirements, while simultaneously ensuring that all process parameters remain within the standard range.

[0060] In some examples, the typical process execution times for circuit boards are adjacent. For instance, after the current circuit board completes the surface mount technology (SMT) process, when it moves to the next process, the next circuit board will also begin its SMT process. When a high risk of solder joint voids is predicted for the current circuit board, the current circuit board will be intercepted, and the process parameters will be determined based on the current circuit board, thereby adjusting the current process parameters used on the production line.

[0061] When adjusting the process parameters of the production line using the above-mentioned process parameter adjustment formula, there is a possibility of repeated adjustments to the process parameters when dealing with two circuit boards with similar preparation times. For example, after the first circuit board undergoes the surface mount process, the risk of solder joint voids on the first circuit board is predicted. If the risk of solder joint voids on the first circuit board is high, a process parameter adjustment instruction will be generated based on the first circuit board.

[0062] Assuming the solder joint void risk on the first circuit board originates from the surface mount technology (SMT) process, if the process parameters for the first circuit board are adjusted while the second circuit board is undergoing SMT, the process parameter adjustment command will not take effect on the second circuit board. The adjustment will only be made after the second circuit board is fully manufactured. Even after the second circuit board has undergone SMT, inspection will still reveal the risk of solder joint voids because it was manufactured using the same SMT process parameters as the first circuit board. If another process parameter adjustment value is obtained based on the second circuit board, it results in two adjustments for the same defect. The first adjustment may have already overcome the defect, but the second adjustment would be a mistake, leading to another defect.

[0063] In some examples, when it is determined that the risk of solder joint voids on the current circuit board is high, after identifying the root cause of the defect (i.e. the defective process) on the current circuit board, all circuit boards that have already undergone the defective process and those that are currently undergoing the defective process are blocked. This avoids predicting the risk of solder joint voids on circuit boards that have already undergone the defective process and those that are currently undergoing the defective process, thereby avoiding the situation of incorrect adjustment of process parameters.

[0064] In some examples, the adjustment process for multiple process parameters is as follows: The first step is to determine the process parameters to be optimized and form a combination of process parameters to be optimized.

[0065] Among them, the value of each process parameter in the combination of process parameters to be optimized , as the optimization target.

[0066] in, This indicates the first combination of process parameters to be optimized. The values ​​of the process parameters, The value range is determined based on the number of process parameters to be optimized.

[0067] In some examples, the process parameters to be optimized include one or more of the following: placement pressure, placement height, placement coordinates, squeegee pressure, speed, and demolding parameters of the pick-and-place machine.

[0068] The second step is to use an optimization algorithm to optimize the combination of process parameters to find the combination of process parameters that minimizes the difference between the actual solder paste volume and the target volume, and then form process parameter control instructions.

[0069] In some examples, the optimization algorithm includes any one of particle swarm optimization, genetic algorithm, or response surface methodology.

[0070] In the process of optimization, the standard range of each process parameter in the combination of process parameters to be optimized is used as a constraint.

[0071] After the adjustment amount of the process parameters is calculated according to the above method, the control command of the process parameters is sent to the control system of the pick-and-place machine or solder paste printer; the pick-and-place machine or solder paste printer adjusts the process parameters of the circuit board whose preparation time is later than the current circuit board according to the control command; the system runs continuously, constantly collecting new three-dimensional data and reflow soldering results, and performs self-learning and optimization on the hole risk prediction model.

[0072] Figure 2 This is a structural block diagram of a patch quality closed-loop control device based on online three-dimensional X-ray inspection, provided as an embodiment of this application. See also... Figure 2 ,include: The surface mount module 11 forms a number of solder pastes on the current circuit board and performs surface mount on the current circuit board according to the solder pastes; The acquisition module 12 is used to acquire three-dimensional volume data of several solder pastes on the current circuit board before reflow soldering the current circuit board; Extraction module 13 is used to extract the three-dimensional morphological feature parameters of solder paste based on the three-dimensional volume data; Prediction module 14 is used to import three-dimensional morphological feature parameters into a pre-trained hole risk prediction model to predict the likelihood of holes being generated on the current circuit board after reflow soldering. The adjustment module 15 is used to adjust the process parameters according to the possibility, and to use the new process parameters to form solder paste and attach it to the circuit board whose preparation time is later than the current circuit board, so that after reflow soldering is performed on the circuit board whose preparation time is later than the current circuit board, the formed solder joints meet the quality requirements.

[0073] It should be noted that, Figure 2 The apparatus shown is modularized according to function. The modules may belong to the same device or apparatus or to different devices or apparatuses. This application does not limit this.

[0074] Figure 3 This is a schematic diagram of a patch quality closed-loop control device based on online three-dimensional X-ray inspection, provided in an embodiment of this application. See also... Figure 3 ,include: Solder paste printer 21 is used to form solder paste on circuit boards.

[0075] The pick and place machine 22 is used to perform pick and place operations on circuit boards.

[0076] An online 3D X-ray inspection device 23 is used to inspect the three-dimensional volume data of solder paste on a circuit board; it is located behind the pick-and-place machine 22 and includes an X-ray emitter, a detector, a precision motion platform for carrying and rotating the PCB board, and a computer for three-dimensional image reconstruction.

[0077] The data processing and analysis server 24 is communicatively connected to the online 3D X-ray inspection device 23.

[0078] The central database 25 is used to store all detection data, characteristic parameters, risk probabilities, and control records.

[0079] The control communication interface 26 connects the data processing and analysis server and the control system of the pick-and-place machine / solder paste printer.

[0080] Conveyor belt 27 is used to transport circuit boards.

[0081] In some examples, the solder paste printer 21 includes: A squeegee 211 and a wire mesh 212 are used, with the wire mesh 212 serving as a template for solder paste printing. After the wire mesh is placed on the surface of the electrical components, the movement of the squeegee 211 on the surface of the wire mesh 212 fills the electrical component surface at the bottom of the opening of the wire mesh 212 with solder paste. Then, the wire mesh is removed, and solder paste is formed on the surface of the circuit board.

[0082] In some examples, the pick-and-place machine 22 includes: Robotic arm 221 is used to attach components 222 onto circuit board 223.

[0083] In some examples, the online 3D X-ray inspection device 23 includes: The microfocus X-ray source 231 emits a cone-shaped X-ray beam to penetrate the weld.

[0084] The high-speed flat panel detector 232 receives attenuated X-rays and directly converts them into high-quality digital image signals.

[0085] The precision motion platform 233 drives the PCB board to rotate.

[0086] See Figure 4 In some examples, the data processing and analysis server 24 includes: Feature extraction module 241 is used to quantify the three-dimensional morphological feature parameters of solder paste from three-dimensional volume data; A hole risk prediction model 242 is used to calculate the risk probability based on characteristic parameters; The decision and control module 243 is used to generate process parameter control instructions.

[0087] In some examples, the hole risk prediction model 242 is a machine learning model trained based on the XGBoost algorithm.

[0088] In some examples, the control communication interface 26 uses Modbus TCP or SECS / GEM communication protocols to interact with the pick-and-place machine 22 and the solder paste printer 21.

[0089] See Figure 5 This is a control method executed by a patch quality closed-loop control device based on online three-dimensional X-ray inspection, provided in an embodiment of this application.

[0090] This application is combined with Figure 1 The method described, and Figure 3 The device described herein provides some specific embodiments.

[0091] Example 1: After the placement machine completes the placement of a circuit board containing the main processor BGA and several surface-mount components, it is conveyed by a conveyor belt into the shielded inspection chamber of the online 3D automated X-ray inspection (AXI) equipment.

[0092] A precision motion platform grips the PCB and begins to rotate. A micro-focus X-ray source and a high-speed flat panel detector work synchronously, completing CT data acquisition of the entire board within 120 seconds. An industrial computer then uses a filtered back-projection algorithm to reconstruct high-resolution three-dimensional volumetric data of the entire PCB and transmits it in real-time to a data processing and analysis server.

[0093] The feature extraction module in the server is activated to perform automated analysis of the 3D volume data. For the main processor's BGA module, an image segmentation algorithm is used to accurately identify the solder paste area beneath each hidden solder ball and calculate its volume value. Simultaneously, a cloud map of the solder paste thickness distribution across the entire BGA area is generated.

[0094] For chip component modules, measure the height and volume of the solder at both ends of the component and determine whether its three-dimensional profile is a standard crescent shape.

[0095] All extracted feature parameters (including the volume of each solder ball in the BGA and the volume of the solder ends of the chip components) are input into the void risk prediction model in real time. The model performs parallel calculations for each solder paste and outputs its corresponding void risk probability value.

[0096] The model calculation results show that the solder paste of BGA number A17 located in the edge area of ​​the PCB board has a volume of only 65% ​​of the standard value. The model calculates that the probability of a hole risk is as high as 85%, which is far beyond the 50% action threshold.

[0097] The decision and control module then initiated the cause analysis logic. It retrieved the volume data of all solder paste for the BGA and found that the volumes of the four corner solder pastes (including A17) were significantly lower than normal, while the volume of the solder paste in the central area was normal. This pattern ruled out the possibility of abnormal global parameters of the solder paste printer, pinpointing the root cause to "poor coplanarity of the BGA components after mounting, resulting in insufficient force at the corners, and the solder paste failing to fully adhere and form."

[0098] Based on this analysis, the decision and control module system calculates the negative... This means instructing the pick-and-place machine to increase the pressure depth at the solder paste location (e.g., -0.02mm) to squeeze out more solder paste.

[0099] A precise control command is generated: "For the main processor BGA at PCB design reference U1, during subsequent mounting, increase the mounting height compensation value of its four corners (corresponding positions A17, B01, H01, H17) by -0.02mm (i.e., lower it by 0.02mm) to apply slight additional pressure." This command is sent in real-time to the pick-and-place machine's control system via the control communication interface using the Modbus TCP protocol.

[0100] The pick-and-place machine control system receives and confirms the instruction. When placing the next PCB of the same model, the pick-and-place machine automatically calls new parameters when executing the BGA placement program for the U1 position, causing its placement head to perform a deeper pressing action of 0.02mm at the four corners.

[0101] The PCB board, after process adjustments, flows through this system again. A new round of CT scans and feature extraction shows that the solder paste volume in the BGA corners has increased to 92% of the standard value. The hole risk prediction model calculates that its risk probability has decreased to 20%, below the warning threshold. This complete "detection-adjustment-verification" data chain is automatically stored in the central database as a dataset for incremental learning, used for further optimization of the hole risk prediction model.

[0102] Example 2: The pick-and-place machine completes the mounting of a new PCB board containing multiple QFN packaged chips and a dense array of surface-mount components. The PCB board is then transported to an in-line 3D AXI device.

[0103] The 3D AXI equipment performs a standard 120-second CT scan and 3D reconstruction process, sending high-quality 3D volume data to the data processing and analysis server.

[0104] The feature extraction module scans and analyzes the solder paste across the entire board. The module quickly identifies a systematic anomaly: the side pads of multiple QFN chips located in the same scraper travel direction on the PCB board, as well as the solder ends of surface-mount components in this area, have solder paste volumes that are generally lower than the standard value of 70%-80%.

[0105] Three-dimensional contour analysis revealed that the solder paste exhibited irregular "concave" shapes, rather than full cubes or arches. Furthermore, the distribution of defects showed a clear locational correlation, all appearing at the end of the scraper stroke, while the solder paste volume and shape were normal at the beginning of the stroke.

[0106] The void risk prediction model was used to calculate the average void risk probability of this batch of solder paste, and the results showed that it reached 65%.

[0107] The decision and control module was triggered to perform a root cause analysis. It compared historical cases in the central database. Based on the key pattern that "defects have a positional correlation in the scraper stroke direction," the module ruled out the possibility of improper pick-and-place machine parameter settings, because placement problems are usually component-specific rather than location-specific.

[0108] The decision module determined that the root cause was most likely in the solder paste printing process. At the end of the long stroke of the squeegee, insufficient squeegee pressure, excessive speed, or insufficient solder paste rolling may have caused the stencil openings to not be completely filled, resulting in insufficient solder paste volume and poor shape.

[0109] The decision and control module system calculates a positive value. This means instructing the printing press to increase the doctor blade pressure in the corresponding stroke segment to promote solder paste filling.

[0110] Based on this diagnostic conclusion, the decision and control module generates a new control instruction for the solder paste printer: "For all production tasks involving this PCB model, adjust the dynamic parameters of the solder paste printer's 'squeegee B' during its left-to-right stroke: In the last 1 / 3 of the stroke, linearly increase the squeegee pressure from the standard 8.0 kgf to 9.0 kgf, while simultaneously reducing the printing speed from 60 mm / s to 55 mm / s to ensure sufficient solder paste filling at the end."

[0111] The instruction is sent in real time to the control system of the fully automatic solder paste printer upstream via the control communication interface.

[0112] After receiving the instruction, the solder paste printer control system immediately updates the production program for the PCB board. When the next blank PCB board enters the printer, the printer strictly adheres to the new parameters: as the squeegee travels to the latter part of its path, the pressure automatically increases and the speed automatically decreases, thereby effectively squeezing more solder paste into the stencil openings.

[0113] After parameter adjustments, the printed and mounted PCB boards flow through this system again. A new round of CT scans and feature extraction shows that the solder paste volume in the previously insufficient areas of QFN and chip component terminals has recovered to 95%-100% of the standard value. The three-dimensional shape of the solder paste has also become fuller and more regular.

[0114] After the hole risk prediction model was reassessed, the average risk probability in the area has dropped significantly to 15%, far below the warning threshold of 30%.

[0115] The data from this successful "detection-diagnosis-control-verification" process was fully recorded in the central database, providing case studies and model optimization data for handling similar defects in the future.

[0116] Figure 6 This is a structural block diagram of an electronic device provided according to an embodiment of this application. See also... Figure 6 Electronic devices may include Figure 2The aforementioned patch quality closed-loop control device based on online three-dimensional X-ray inspection. Typically, the electronic device includes a processor 31 and a memory 32. The processor 31 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 31 can be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 31 may also include a main processor and a coprocessor. The main processor is used to process data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor used to process data in the standby state. The memory 32 may include one or more computer-readable storage media, which may be non-transitory. The memory 32 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage medium in memory 32 is used to store at least one instruction, which is executed by processor 31 to implement the patch quality closed-loop control method based on online three-dimensional X-ray inspection provided by the method embodiments of this application.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these 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 the present invention.

Claims

1. A closed-loop control method for patch quality based on online three-dimensional X-ray inspection, characterized in that, include: Solder paste is formed on the current circuit board, and components are mounted on the current circuit board according to the solder paste. Before reflow soldering the current circuit board, acquire the three-dimensional volume data of the solder paste on the current circuit board; Extract the three-dimensional morphological feature parameters of the solder paste from the three-dimensional volume data; The three-dimensional morphological feature parameters are imported into a pre-trained hole risk prediction model to predict the likelihood of holes being generated on the current circuit board after reflow soldering. Based on the aforementioned possibilities, process parameters are adjusted, and new process parameters are used to form solder paste and mount it on a circuit board whose preparation time is later than the current circuit board, so that after reflow soldering on the circuit board whose preparation time is later than the current circuit board, several solder joints formed meet the quality requirements.

2. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 1, characterized in that, Three-dimensional morphological feature parameters include: Solder paste volume, solder paste thickness distribution, solder paste shape profile, and alignment offset between component leads and pads.

3. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 1, characterized in that, The steps for extracting the three-dimensional morphological feature parameters of solder paste from three-dimensional volume data include: The three-dimensional volume data is voxelized to identify and segment the three-dimensional region of each solder paste. The volume of each solder paste is calculated based on the number of voxels in the three-dimensional region of each solder paste. Perform a three-dimensional Euclidean distance transformation on the three-dimensional region of each solder paste to obtain a distance map of the three-dimensional region of each solder paste, and then obtain the solder paste thickness distribution. Calculate the three-dimensional moment features of the three-dimensional region of each solder paste to obtain the solder paste shape profile; By using image segmentation or template matching, the positions of component pins and several solder pastes on the circuit board in three-dimensional space are identified respectively. The Euclidean distance and azimuth angle between the three-dimensional area of ​​each solder paste and its corresponding component pin are calculated as the alignment offset between the component pin and the pad.

4. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 3, characterized in that, Based on the likelihood of voids forming on the current circuit board after reflow soldering, the steps for adjusting process parameters include: When the probability of a hole being generated on the current circuit board after reflow soldering does not meet a preset threshold, a process parameter control command is generated to control one or more process parameters of the pick-and-place machine, including placement pressure, placement height, placement coordinates, squeegee pressure, speed, and demolding parameters.

5. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 1, characterized in that, The process for determining the quality of process parameter control is as follows: Based on the distribution pattern of the defective solder paste, the root cause of the defect is determined, and the process parameters to be optimized are determined based on the root cause of the defect. The process parameters to be optimized are then optimized to obtain process parameter control instructions.

6. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 5, characterized in that, If only a single process parameter needs to be optimized, the process parameter control command can be generated using the following formula: In the formula, Indicates the adjustment amount of a certain process parameter; This represents the proportionality coefficient, used to adjust the ratio of... Response speed; This represents the difference between the actual solder paste volume and the target volume. Represents the integral coefficient, used to eliminate Historical accumulation deviations; This indicates the deviation of a certain process parameter in the historical production cycle.

7. The patch quality closed-loop control method based on online three-dimensional X-ray inspection according to claim 5, characterized in that, If multiple process parameters are to be optimized simultaneously, process parameter control instructions can be generated using the following method: Identify the process parameters to be optimized and form a combination of process parameters to be optimized; An optimization algorithm is used to optimize the combination of process parameters to find the combination that minimizes the difference between the actual solder paste volume and the target volume, thereby generating process parameter control instructions.

8. A patch quality closed-loop control device based on online three-dimensional X-ray inspection, characterized in that, include: The surface mount module forms a number of solder pastes on the current circuit board and performs surface mount on the current circuit board according to the solder pastes; The acquisition module is used to acquire three-dimensional volume data of several solder pastes on the current circuit board before reflow soldering the current circuit board; The extraction module is used to extract the three-dimensional morphological feature parameters of the solder paste based on the three-dimensional volume data. The prediction module is used to import three-dimensional morphological feature parameters into a pre-trained hole risk prediction model to predict the likelihood of holes being generated on the current circuit board after reflow soldering. The adjustment module is used to adjust the process parameters according to the possibilities, and to use the new process parameters to form solder paste and mount it on a circuit board whose preparation time is later than the current circuit board, so that after reflow soldering on the circuit board whose preparation time is later than the current circuit board, the formed solder joints meet the quality requirements.

9. An electronic device, characterized in that, Includes the patch quality closed-loop control device based on online three-dimensional X-ray inspection as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one piece of program code, which is executed by a processor to implement the patch quality closed-loop control method based on online three-dimensional X-ray inspection as described in any one of claims 1 to 7.

Images