Visual quality inspection method and system for BGA chip

Abstract

The application relates to the technical field of intelligent detection, and discloses a visual quality inspection method and system for a BGA (Ball Grid Array) chip. The method comprises the following steps: irradiating the BGA chip through a multi-spectrum light source array to obtain an initial reflection image of the BGA chip; performing regional segmentation on the initial reflection image based on the geometric distribution characteristics of the solder balls in the chip to obtain independent imaging areas of the solder balls; collecting multi-view difference images of the same solder ball area in the independent imaging areas and fusing the multi-view difference images into a three-dimensional topology graph; separating a deformation feature map of the solder balls in the three-dimensional topology graph, applying a virtual thermal stress field to the deformation feature map, and obtaining a thermal expansion coefficient distribution cloud map of the solder balls; marking abnormal solder ball coordinates in the BGA chip according to the comparison result of the thermal expansion coefficient distribution cloud map and a standard threshold interval; and generating a quality inspection report of the BGA chip according to the abnormal solder ball coordinates and the thermal expansion coefficient distribution cloud map. The application can improve the accuracy of visual quality inspection of the BGA chip.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection technology, and in particular to a visual quality inspection method and system for BGA chips. Background Technology

[0002] In existing technologies for visual quality inspection of BGA chips, the image acquisition stage often relies on a single light source or fixed wavelength illumination, which makes it difficult to accurately capture the optical response differences between the solder balls and the substrate. This results in blurred boundary transition zones between the solder balls and the substrate, and deviations are prone to occur during region segmentation. It is impossible to accurately separate independent solder ball imaging areas, which leads to errors in the basic data for subsequent quality inspection analysis and affects the overall detection accuracy.

[0003] Meanwhile, existing technologies have limitations in extracting deformation features and identifying anomalies in solder balls. They are mostly based on two-dimensional images, lacking effective fusion of multi-view information and making it difficult to construct accurate three-dimensional topological structures, resulting in incomplete identification of deformation features. Furthermore, the analysis of thermal expansion coefficients does not consider the influence of virtual thermal stress fields, and the use of a uniform threshold to judge anomalies cannot adapt to the differences in characteristics of solder balls in different regions, resulting in a high rate of false positives and false negatives, significantly reducing the efficiency and reliability of quality inspection. Summary of the Invention

[0004] This invention provides a visual quality inspection method and system for BGA chips to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a visual quality inspection method for BGA chips, comprising:

[0006] S1. Illuminate the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip;

[0007] S2. Based on the geometric distribution characteristics of the solder balls in the BGA chip, the initial reflection image is segmented to obtain independent imaging regions of the solder balls;

[0008] S3. Acquire multi-view differential images of the same solder ball region in the independent imaging region, and fuse the multi-view differential images into a three-dimensional topology map;

[0009] S4. Separate the deformation feature map of the solder ball in the three-dimensional topology map, and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball;

[0010] S5. Based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range, mark the coordinates of the abnormal solder balls in the BGA chip;

[0011] S6. Generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

[0012] In a preferred embodiment, the step of illuminating the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip includes:

[0013] Send wavelength switching commands to the multispectral light source array to activate the combined illumination mode of near-infrared and visible light bands;

[0014] Based on the material properties of the substrate in the BGA chip, the incident angle of the combined illumination mode is adjusted to a preset tilt angle range, and the surface reflected light intensity distribution data of the solder balls in the BGA chip is captured.

[0015] By fusing the surface reflected light intensity distribution data, an initial reflection image of the BGA chip is obtained.

[0016] In a preferred embodiment, the step of segmenting the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions for the solder balls includes:

[0017] Identify regions of optical response difference in the initial reflection image, and delineate the boundary transition zone between the solder ball and the substrate based on these regions of optical response difference;

[0018] Based on the preset solder ball spatial distribution template, locate the low reflectivity isolation zone between the solder balls;

[0019] A closed cutting path for generating the initial reflection image is generated along the low-reflectivity isolation zone;

[0020] Based on the closed cutting path and the boundary transition zone, the solder ball imaging block of the initial reflection image is separated to obtain the independent imaging region of the solder ball.

[0021] In a preferred embodiment, acquiring multi-view differential images of the same solder ball region within the independent imaging region includes:

[0022] Within the independent imaging area, all viewing angles remain at a fixed exposure time;

[0023] Simultaneously capture the surface microstructure reflection features of the solder ball from the stated viewpoint;

[0024] Pixel-level difference operations are performed on the reflection features of adjacent viewpoints to obtain the multi-view difference image of the solder ball.

[0025] In a preferred embodiment, fusing the multi-view difference images into a three-dimensional topological map includes:

[0026] Extract the solder ball contour deformation gradient vector of each frame of the multi-view differential image;

[0027] The coefficient of thermal expansion of the solder ball material used to load the BGA chip and the current ambient temperature;

[0028] Based on the solder ball profile deformation gradient vector, the thermal expansion coefficient of the solder ball material, and the current ambient temperature, the three-dimensional topological coordinates of the solder ball are calculated, wherein the calculation formula for the three-dimensional topological coordinates is as follows:

[0029]

[0030] In the formula, The three-dimensional topological coordinates are... The total number of the stated viewpoints. Let be the ordinal number of the stated viewpoint. For the first The weld ball profile deformation gradient vector from each perspective The coefficient of thermal expansion of the solder ball material is... The current ambient temperature, For standard reference temperature, The coordinates of the ideal geometric center of the solder ball;

[0031] A three-dimensional topology map of the solder ball is constructed based on the three-dimensional topological coordinates.

[0032] In a preferred embodiment, separating the deformation feature map of the solder ball in the three-dimensional topology map includes:

[0033] The three-dimensional topology map is spatially registered with the standard solder ball three-dimensional template to obtain the height deviation field of the solder ball.

[0034] Based on the material yield strength threshold of the welding ball, mark the plastic deformation region in the height deviation field;

[0035] The curvature gradient distribution characteristics of the plastic deformation region are extracted to obtain the deformation feature map of the solder ball.

[0036] In a preferred embodiment, applying a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball includes:

[0037] Construct the position vector and normal vector of the solder ball surface points on the deformation feature map;

[0038] A virtual thermal stress field is applied to the deformation feature map based on the position vector and the normal vector.

[0039] Based on the Young's modulus and Poisson's ratio of the virtual thermal stress field, the local thermal expansion coefficient of the solder ball is calculated, wherein the formula for calculating the local thermal expansion coefficient is as follows:

[0040]

[0041] In the formula, The local thermal expansion coefficient is... The nominal coefficient of thermal expansion of the material. The deformation-thermal coupling factor. For gradient operators, This represents the three-dimensional displacement of a point on the surface of the solder ball. Let be the normal vector. Let be the direction vector of the virtual thermal stress field. For Young's modulus, Poisson's ratio, Norm operator;

[0042] By iterating through the local thermal expansion coefficients of the points on the surface of the solder ball, a thermal expansion coefficient distribution cloud map of the solder ball is obtained.

[0043] In a preferred embodiment, marking the coordinates of abnormal solder balls in the BGA chip based on the comparison result between the thermal expansion coefficient distribution cloud map and the standard threshold range includes:

[0044] The thermal expansion coefficient distribution cloud map is divided into a central region, an edge region, and a corner region according to the position of the solder ball array;

[0045] Differentiated threshold ranges are applied to the central region, the edge region, and the corner region;

[0046] Traverse the thermal expansion coefficient distribution cloud map to identify abnormal data points that exceed the differential threshold range;

[0047] For a cluster region consisting of consecutive abnormal data points, determine the physical coordinates of the centroid of the cluster region in the BGA chip;

[0048] The physical coordinates are mapped and associated with the solder ball number to obtain the coordinates of the abnormal solder ball.

[0049] In a preferred embodiment, generating a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map includes:

[0050] An anomaly level label for the BGA chip is generated based on the distribution coefficient of thermal expansion corresponding to the coordinates of the abnormal solder balls.

[0051] By superimposing the thermal stress transmission trajectory on the thermal expansion coefficient distribution cloud map, the thermal influence correlation path between abnormal solder balls in the BGA chip is obtained.

[0052] Spatially align the three-dimensional topology map with the thermal expansion coefficient distribution cloud map to obtain the three-dimensional deformation characteristics of the solder ball.

[0053] By integrating and implementing anomaly level labels, implementing thermal impact correlation paths, and implementing three-dimensional deformation features, a quality inspection report for the BGA chip is obtained.

[0054] To address the above problems, the present invention also provides a visual quality inspection system for BGA chips, the system comprising:

[0055] A light source illumination module is used to illuminate the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip;

[0056] The image segmentation module is used to segment the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions of the solder balls;

[0057] The image fusion module is used to acquire multi-view differential images of the same solder ball area in the independent imaging area and fuse the multi-view differential images into a three-dimensional topology map;

[0058] The thermal expansion coefficient distribution cloud map generation module is used to separate the deformation feature map of the solder ball in the three-dimensional topology map and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball.

[0059] An abnormal solder ball monitoring module is used to mark the coordinates of abnormal solder balls in the BGA chip based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range.

[0060] The quality inspection report generation module is used to generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. This invention, through a combination of illumination modes using a multispectral light source array and adjusting the incident angle based on substrate material properties, can accurately capture the intensity distribution of reflected light on the solder ball surface, providing a high-quality initial reflection image for subsequent region segmentation. The region segmentation method based on the geometric distribution characteristics of the solder balls can clearly delineate boundary transition zones and isolation zones, efficiently separating independent imaging regions, laying a solid foundation for subsequent accurate analysis, and significantly improving the accuracy and efficiency of image preprocessing.

[0063] 2. This invention constructs a three-dimensional topology map through multi-view differential image fusion and calculates the thermal expansion coefficient distribution cloud map using a virtual thermal stress field, achieving comprehensive extraction of solder ball deformation characteristics. Simultaneously, it employs differentiated threshold intervals to identify anomalies in solder balls in different regions and integrates multi-dimensional information to generate a quality inspection report. This not only accurately marks the coordinates of abnormal solder balls but also presents the thermal impact correlation path and three-dimensional deformation characteristics, significantly improving the comprehensiveness and reliability of the quality inspection results and effectively enhancing the overall efficiency of BGA chip visual quality inspection. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating a visual quality inspection method for BGA chips according to an embodiment of the present invention.

[0065] Figure 2 This is a functional block diagram of a visual quality inspection system for BGA chips provided in an embodiment of the present invention;

[0066] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0067] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0068] This application provides a visual quality inspection method for BGA chips. The execution subject of the visual quality inspection method for BGA chips includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the visual quality inspection method for BGA chips can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0069] Reference Figure 1 The diagram shown is a flowchart illustrating a visual quality inspection method for a BGA chip according to an embodiment of the present invention. In this embodiment, the visual quality inspection method for the BGA chip includes:

[0070] S1. Illuminate the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip;

[0071] In this embodiment of the invention, obtaining an initial reflection image of the BGA chip by illuminating it with a multispectral light source array includes:

[0072] Send wavelength switching commands to the multispectral light source array to activate the combined illumination mode of near-infrared and visible light bands;

[0073] Based on the material properties of the substrate in the BGA chip, the incident angle of the combined illumination mode is adjusted to a preset tilt angle range, and the surface reflected light intensity distribution data of the solder balls in the BGA chip is captured.

[0074] By fusing the surface reflected light intensity distribution data, an initial reflection image of the BGA chip is obtained.

[0075] Specifically, the control terminal is connected to the control interface of the multispectral light source array via a dedicated data transmission line. The light source control software on the control terminal is then opened. In the software's command input interface, a wavelength switching command containing activation information for the near-infrared and visible light bands is accurately entered. The command must clearly indicate the specific activation identifiers for both bands. Upon receiving the command, the control module of the multispectral light source array immediately parses it, confirms the type of band to be activated, and then sends activation signals to both the near-infrared and visible light band light source groups. Upon receiving the signals, the drive circuits inside both light sources quickly connect the power supply, causing the light sources to emit light at the set power, thereby activating the combined near-infrared and visible light band illumination mode.

[0076] Furthermore, an infrared spectrometer is used to scan and inspect the substrate of the BGA chip. When infrared light emitted by the spectrometer illuminates the substrate surface, some light is absorbed and some is reflected. The instrument receives the reflected light and analyzes its spectral characteristics to obtain information about the substrate material's composition and structure. Based on the detected substrate material properties, the optimal incident angle range for that material, i.e., the preset tilt angle range, is retrieved from the equipment parameter manual. Subsequently, the mechanical adjustment system of the multispectral light source array is operated. Upon receiving a control signal, the stepper motor in this system drives the support of the light source array to rotate via a gear transmission structure. During adjustment, an angle sensor mounted on the support monitors the current angle of the light source array in real time and feeds the angle data back to the control terminal until the angle reaches the preset tilt angle range, at which point the adjustment stops. Next, the lens of the optical inspection device is aimed at the solder ball area in the BGA chip, and the lens focal length is adjusted to make the solder ball surface clearly imaged. The CCD sensor inside the optical inspection device starts to work, receiving the combined light rays reflected back from the surface of the solder ball. Each photosensitive unit on the sensor corresponds to a tiny area on the surface of the solder ball. The photosensitive unit converts the received light signal into an electrical signal. The strength of the electrical signal is proportional to the intensity of the reflected light. The device converts the electrical signal into a digital signal through the internal analog-to-digital converter module, and records these digital signals according to the arrangement order of the photosensitive units to form the surface reflected light intensity distribution data of the solder ball.

[0077] Furthermore, the surface reflection light intensity distribution data stored in the optical inspection device is transmitted to an image processing computer via a high-speed data transmission interface. The image processing software on the computer automatically reads this data. The software first parses the data, identifying the solder ball surface position information corresponding to each light intensity value. This position information corresponds one-to-one with the actual spatial coordinates of the solder balls in the BGA chip. Then, the software converts the light intensity value at each position into the brightness value of an image pixel according to a preset mapping rule; that is, the higher the light intensity value, the higher the corresponding pixel brightness. Finally, the software arranges and combines all the converted pixels sequentially according to the actual arrangement order of the solder balls in the BGA chip, forming a complete two-dimensional pixel matrix. The image presented by this matrix is ​​the initial reflection image of the BGA chip, and the software automatically saves this image in a preset image format.

[0078] In summary, by illuminating the BGA chip with a multispectral light source array to obtain an initial reflection image, a combined illumination mode of near-infrared and visible light bands can be activated. By adjusting the incident angle to a preset range in combination with the substrate material properties, the data on the intensity distribution of reflected light on the solder ball surface can be accurately captured.

[0079] In summary, by fully utilizing the differences in the optical response of different wavelengths of light to the solder balls and the substrate, the contrast between the two is enhanced, providing a high-quality image foundation for subsequent region segmentation based on geometric distribution features. This effectively improves the clarity of the boundary between the solder balls and the substrate in the initial image, ensures the accuracy of subsequent independent imaging region extraction, guarantees the reliability of quality inspection data from the source, and lays a key foundation for the efficient implementation of the entire quality inspection process.

[0080] S2. Based on the geometric distribution characteristics of the solder balls in the BGA chip, the initial reflection image is segmented to obtain independent imaging regions of the solder balls;

[0081] In this embodiment of the invention, the step of segmenting the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions for the solder balls includes:

[0082] Identify regions of optical response difference in the initial reflection image, and delineate the boundary transition zone between the solder ball and the substrate based on these regions of optical response difference;

[0083] Based on the preset solder ball spatial distribution template, locate the low reflectivity isolation zone between the solder balls;

[0084] A closed cutting path for generating the initial reflection image is generated along the low-reflectivity isolation zone;

[0085] Based on the closed cutting path and the boundary transition zone, the solder ball imaging block of the initial reflection image is separated to obtain the independent imaging region of the solder ball.

[0086] Specifically, the initial reflection image is imported into dedicated image processing software. The software activates a pixel scanning module, reading the brightness information of each pixel in the image point by point from left to right and top to bottom. This brightness information directly corresponds to the optical response intensity at each location in the initial reflection image. The software's built-in adjacent pixel comparison function automatically compares the brightness value of the current pixel with the brightness values ​​of the four pixels to its right, below, and diagonally adjacent. When the difference in brightness values ​​between any two adjacent pixels exceeds a preset fixed threshold, the system immediately marks a difference point at that location. As the scanning continues, these scattered difference points are integrated by the software's region recognition function. When consecutive difference points form a continuous band, the software marks this continuous area as an optical response difference region. Subsequently, the software uses an edge fitting function to connect all the difference points within the optical response difference region into a smooth curve. This curve and the three-pixel-width areas on both sides together constitute the boundary transition zone between the solder ball and the substrate. The inner side of the transition zone corresponds to the edge of the solder ball, and the outer side corresponds to the edge of the substrate.

[0087] Further, a preset solder ball spatial distribution template is opened in the image processing software. This template is a digital model generated based on the BGA chip design drawings, containing the center coordinates, diameter, and spacing parameters of each solder ball. The software activates the template matching function, using the edge contour of the substrate in the initial reflection image as a reference, aligning the coordinate system of the template with the coordinate system of the image to ensure that the solder ball positions in the template roughly coincide with the actual solder ball distribution positions in the image. After alignment, the software selects a detection band with a width equal to the sum of the radii of two solder balls at the corresponding position in the image, according to the direction of the center line connecting adjacent solder balls in the template, and analyzes the reflectivity values ​​of all pixels within the detection band one by one. When the reflectivity value of a continuous area within the detection band is lower than one-third of the reflectivity values ​​of the solder ball areas on both sides, and the error between the length of this area and the corresponding solder ball spacing in the template is within 0.5 pixels, the software marks this area as a low-reflectivity isolation band between solder balls. All such areas that meet the conditions will be completely identified and recorded.

[0088] Furthermore, the image processing software uses an edge detection tool to process the marked low-reflectivity isolation zone. The tool identifies pixels where reflectivity suddenly increases at the edge of the isolation zone; these pixels are the boundaries between the isolation zone and adjacent solder balls, i.e., edge pixels. The software connects these edge pixels sequentially in a clockwise direction, following the principle that adjacent pixels are closest in the image coordinate system, ensuring that the line connecting any two consecutive points does not cross non-isolation zone areas. When connecting to the starting edge pixel, the software automatically checks whether the formed shape is closed. If there is a gap, it fills it by adding adjacent edge pixels. The final continuous and uninterrupted closed line is the closed cutting path of the initial reflective image, and the path perfectly matches the edge contour of the low-reflectivity isolation zone.

[0089] Furthermore, the image processing software simultaneously loads the coordinate data of the closed cutting path and the boundary transition zone. The system first determines the closed cutting path as the outer boundary of the segmentation, which defines the outermost range of a single solder ball imaging block. Then, the edge line of the boundary transition zone closest to the solder ball is set as the inner boundary, which is formed by connecting the pixels in the transition zone closest to the center of the solder ball. The software activates the region segmentation function, cutting the initial reflection image along the outer and inner boundaries. During the cutting process, the substrate imaging portion between the two boundaries is automatically ignored, retaining only the solder ball imaging area enclosed by the inner boundary. After cutting, the software extracts each independent solder ball imaging area from the initial reflection image, removes redundant pixels at the edges of the areas, and arranges and stores them according to their positional order in the original image. These extracted independent areas together constitute the independent imaging area of ​​the solder ball.

[0090] In summary, the initial reflection image is segmented based on the geometric distribution characteristics of solder balls in the BGA chip to obtain independent imaging areas. It can accurately delineate the boundary transition zone between the solder balls and the substrate by identifying regions with differences in optical response. At the same time, it locates the low reflectivity isolation zone according to the preset solder ball spatial distribution template, and then generates a closed cutting path to achieve the separation of the solder ball imaging block.

[0091] In summary, the optical and spatial distribution differences between solder balls and the substrate, as well as between solder balls, can be fully utilized to ensure the independence and integrity of each solder ball imaging area. This effectively avoids mutual interference between adjacent solder ball imaging areas, providing a precise regional basis for the subsequent acquisition of multi-view differential images of the same solder ball area and the construction of a three-dimensional topology map. This significantly improves the targeting and accuracy of subsequent quality inspection analysis and ensures the efficient advancement of the overall quality inspection process.

[0092] S3. Acquire multi-view differential images of the same solder ball region in the independent imaging region, and fuse the multi-view differential images into a three-dimensional topology map;

[0093] In this embodiment of the invention, acquiring multi-view differential images of the same solder ball region within the independent imaging region includes:

[0094] Within the independent imaging area, all viewing angles remain at a fixed exposure time;

[0095] Simultaneously capture the surface microstructure reflection features of the solder ball from the stated viewpoint;

[0096] Pixel-level difference operations are performed on the reflection features of adjacent viewpoints to obtain the multi-view difference image of the solder ball.

[0097] The process of fusing the multi-view difference images into a three-dimensional topological map includes:

[0098] Extract the solder ball contour deformation gradient vector of each frame of the multi-view differential image;

[0099] The coefficient of thermal expansion of the solder ball material used to load the BGA chip and the current ambient temperature;

[0100] Based on the solder ball profile deformation gradient vector, the thermal expansion coefficient of the solder ball material, and the current ambient temperature, the three-dimensional topological coordinates of the solder ball are calculated, wherein the calculation formula for the three-dimensional topological coordinates is as follows:

[0101]

[0102] In the formula, The three-dimensional topological coordinates are... The total number of the stated viewpoints. Let be the ordinal number of the stated viewpoint. For the first The weld ball profile deformation gradient vector from each perspective The coefficient of thermal expansion of the solder ball material is... The current ambient temperature, For standard reference temperature, The coordinates of the ideal geometric center of the solder ball;

[0103] A three-dimensional topology map of the solder ball is constructed based on the three-dimensional topological coordinates.

[0104] Specifically, the independent imaging area of ​​the solder ball is imported into the image acquisition system. A fixed exposure time is preset in the system, which is set and locked through the system control panel to ensure that it will not change due to changes in external light or fluctuations in equipment parameters. The viewing angle switching function of the image acquisition system is activated. The system uses a robotic arm to move the imaging lens sequentially between multiple preset viewing angles. Each viewing angle is aligned with the same solder ball within the independent imaging area. When the lens moves to the designated viewing angle position and stabilizes, the system automatically triggers the exposure mechanism, allowing the lens to complete image acquisition preparation at that viewing angle according to the preset fixed exposure time. After the exposure time ends, the system switches to the next viewing angle. All viewing angles operate according to the same fixed exposure time.

[0105] Furthermore, the image acquisition system establishes a synchronous trigger connection with the multispectral light source array. When the imaging lens reaches the start moment of a fixed exposure time at a certain angle, the system simultaneously sends a light-start signal to the light source array, causing the light source to illuminate the solder ball surface with stable power. At this time, the lens begins to capture images and continuously receives light reflected from the microstructures on the solder ball surface (such as the uneven texture, scratches, oxide layer, etc. on the solder ball surface) within the exposure time. After being focused by the lens, these light rays form a corresponding light signal distribution on the image sensor. The system converts this light signal distribution into digital image data. The brightness and color changes of different pixels in the image directly reflect the reflection characteristics of the solder ball surface microstructure. The reflection characteristics at each angle are stored in real time as independent feature data files.

[0106] Furthermore, the microstructure reflection feature data of the solder ball surface from two adjacent viewpoints are retrieved from the system. The digital images corresponding to the two feature data are aligned in the same coordinate system to ensure that the position of each pixel in the image corresponds one-to-one. The aligned image is then compared pixel by pixel using image processing software. The software reads the brightness value of each pixel in the first viewpoint image, and then reads the brightness value of the pixel at the same position in the second viewpoint image. The brightness difference value of the pixel is obtained by subtracting the brightness value of the first pixel from the latter. All corresponding pixels are processed in the same way, and the brightness difference values ​​of each pixel are combined to form a new digital image. The brightness of the pixels in this image directly reflects the difference in reflection features between adjacent viewpoints. This image is the multi-view difference image of the solder ball, and the software automatically saves the image and marks the corresponding viewpoint information.

[0107] Specifically, multi-view difference images are imported into dedicated contour analysis software. The software performs edge enhancement processing on each frame of the difference image and outlines the complete contour of the solder ball by identifying consecutive pixels with the most dramatic brightness changes in the image. Starting from the starting point along the contour line, two adjacent contour points are selected sequentially, and the direction and length of the line connecting the two points are determined. The curvature trend of the contour is determined based on the change in direction, and the degree of stretching or contraction of the contour is determined based on the change in length. This trend and degree are represented by directional line segments. The direction of each line segment corresponds to the direction of contour deformation, and the length of the line segment corresponds to the magnitude of deformation. These line segments together form the solder ball contour deformation gradient vector of that frame of the difference image. The software stores all vectors in sequence according to the contour position.

[0108] Furthermore, the coefficient of thermal expansion of the solder ball material was retrieved from the BGA chip's material specification sheet. This coefficient, documented in document form, records the proportion of length change of the material per unit temperature change. This coefficient value was then entered into the 3D reconstruction system via a data input interface. Simultaneously, the probe of a temperature sensor was placed close to the working environment of the BGA chip. The sensor measured and displayed the current ambient temperature in real time, and the temperature data was transmitted to the 3D reconstruction system via a data cable. The system automatically read and stored the temperature value, ensuring that both the coefficient of thermal expansion of the solder ball material and the current ambient temperature were accurately loaded.

[0109] Furthermore, the 3D reconstruction system invokes its built-in correlation analysis module. First, it matches the solder ball profile deformation gradient vector with the current ambient temperature, analyzing the differences in the direction and magnitude of profile deformation at different temperatures. Then, combining this with the thermal expansion coefficient of the solder ball material, it determines the proportion of influence of material expansion or contraction caused by temperature changes on the profile deformation. Based on this proportion, the temperature-induced deformation portion of the profile deformation gradient vector is separated, and the remaining portion serves as deformation data reflecting the actual 3D shape of the solder ball. This deformation data is then converted into X, Y, and Z axis coordinates in a spatial coordinate system. Each coordinate value corresponds to a specific spatial location on the solder ball surface. All coordinate values ​​together constitute the 3D topological coordinates of the solder ball, and the system stores these coordinates in relation to their spatial locations.

[0110] Furthermore, the 3D topological coordinates of the solder balls are imported into the 3D modeling software. The software automatically identifies the spatial point corresponding to each coordinate and connects adjacent spatial points sequentially according to the spatial distance between them, forming a continuous wireframe structure. After the wireframe structure covers all 3D topological coordinate points on the surface of the solder balls, the software fills the adjacent wireframes to form a 3D model with surface texture. The surface undulations of the model directly reflect the height changes corresponding to the 3D topological coordinates. At the same time, the software adds appropriate color rendering to the model, so that different height areas present different colors, enhancing the sense of three-dimensionality. The final 3D model is the 3D topological map of the solder balls, which the software saves as a rotatable 3D file format.

[0111] Specifically, the three-dimensional topological coordinates are the calculated results, the total number of viewpoints is the number of viewpoints preset when acquiring multi-view differential images, the ordinal number of the viewpoints is the sequential numbering of each set viewpoint, the solder ball profile deformation gradient vector of the i-th viewpoint comes from the vector extracted from the corresponding frame differential image in the multi-view differential image, the thermal expansion coefficient of the solder ball material comes from the BGA chip solder ball material specification sheet, the current ambient temperature comes from the temperature sensor's measurement result of the environment where the solder ball is located, the standard reference temperature is the reference temperature specified in the solder ball material properties, usually recorded in the material specification sheet, and the ideal geometric center coordinates of the solder ball come from the BGA chip design drawings, which are the theoretical coordinates of the solder ball center.

[0112] Furthermore, the significance of this formula is that by multiplying the gradient vector of the solder ball profile deformation from each viewpoint by the thermal expansion coefficient of the solder ball material, the difference between the current ambient temperature and the standard reference temperature, and then summing all these multiplications, and finally adding the ideal geometric center coordinates of the solder ball, a three-dimensional topological coordinate that can reflect the actual spatial position of the solder ball is obtained. This process comprehensively considers the profile deformation under different viewpoints, the expansion or contraction of the material due to temperature changes, and the theoretical center position of the solder ball, and finally determines the actual three-dimensional spatial coordinates of the solder ball.

[0113] Furthermore, when the current ambient temperature is higher than the standard reference temperature, the difference between the current ambient temperature and the standard reference temperature is positive. If the coefficient of thermal expansion of the solder ball material is positive, the product of the solder ball profile deformation gradient vector at each viewpoint and this difference and the coefficient of thermal expansion is also positive. After accumulation, the result will deviate from the ideal geometric center coordinates of the solder ball and move in the expansion direction. The three-dimensional topological coordinates will reflect the shape of the solder ball expanding due to the increase in temperature. When the current ambient temperature is lower than the standard reference temperature, the difference is negative, and the above product is also negative. After accumulation, the result will move in the contraction direction. The three-dimensional topological coordinates will reflect the contraction shape. At the same time, if the solder ball profile deformation gradient vector at a certain viewpoint points to a specific direction and the value is large, the proportion of that direction will be greater during the accumulation process, and the change of the three-dimensional topological coordinates in that direction will be more significant. Finally, the overall trend of the actual three-dimensional shape change of the solder ball is reflected.

[0114] In summary, by acquiring multi-view differential images of the same solder ball region in independent imaging areas and fusing them into a 3D topology map, the surface microstructure reflection features of the solder ball under different viewpoints can be captured synchronously with a fixed exposure time. The reflection differences between adjacent viewpoints can be accurately extracted through pixel-level differential operations, providing rich detailed information for 3D structure reconstruction.

[0115] In summary, by combining the solder ball profile deformation gradient vector, material thermal expansion coefficient, and ambient temperature to calculate three-dimensional topological coordinates, a three-dimensional topological map accurately reflecting the spatial morphology of the solder ball can be constructed. This process fully integrates microscopic features and physical property parameters from multiple perspectives, not only completely preserving the deformation information of the solder ball surface but also achieving a precise conversion from two-dimensional images to three-dimensional structures. This provides comprehensive and detailed data support for subsequent separation of deformation feature maps and analysis of thermal expansion coefficient distribution, significantly improving the ability to capture subtle deformations of the solder ball and the accuracy of three-dimensional morphological characterization.

[0116] S4. Separate the deformation feature map of the solder ball in the three-dimensional topology map, and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball;

[0117] In this embodiment of the invention, separating the deformation feature map of the solder ball in the three-dimensional topology map includes:

[0118] The three-dimensional topology map is spatially registered with the standard solder ball three-dimensional template to obtain the height deviation field of the solder ball.

[0119] Based on the material yield strength threshold of the welding ball, mark the plastic deformation region in the height deviation field;

[0120] The curvature gradient distribution characteristics of the plastic deformation region are extracted to obtain the deformation feature map of the solder ball.

[0121] The step of applying a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball includes:

[0122] Construct the position vector and normal vector of the solder ball surface points on the deformation feature map;

[0123] A virtual thermal stress field is applied to the deformation feature map based on the position vector and the normal vector.

[0124] Based on the Young's modulus and Poisson's ratio of the virtual thermal stress field, the local thermal expansion coefficient of the solder ball is calculated, wherein the formula for calculating the local thermal expansion coefficient is as follows:

[0125]

[0126] In the formula, The local thermal expansion coefficient is... The nominal coefficient of thermal expansion of the material. The deformation-thermal coupling factor. For gradient operators, This represents the three-dimensional displacement of a point on the surface of the solder ball. Let be the normal vector. Let be the direction vector of the virtual thermal stress field. For Young's modulus, Poisson's ratio, Norm operator;

[0127] By iterating through the local thermal expansion coefficients of the points on the surface of the solder ball, a thermal expansion coefficient distribution cloud map of the solder ball is obtained.

[0128] Specifically, the 3D topology diagram of the solder balls and the 3D template of the standard solder balls are imported into the spatial registration software. The standard solder ball 3D template is a theoretical 3D model of the solder balls generated according to the BGA chip design specifications, containing standard height data for each position of the solder balls. The software identifies the geometric feature points (such as vertices and edge midpoints) of the solder balls in the 3D topology diagram and the standard template, aligning their coordinate systems so that the feature points coincide in space. After alignment, the software compares the actual height of each position in the 3D topology diagram with the standard height of the corresponding position in the standard template point by point, calculating the difference between the two. The dataset formed by the spatial distribution of height differences at all positions is the height deviation field of the solder balls. The software visualizes this deviation field in a color-coded form, with different colors representing different height deviation values.

[0129] Furthermore, the material yield strength threshold is obtained from the solder ball material's specification sheet. This threshold is the critical height deviation value corresponding to the maximum stress the material can withstand during plastic deformation. This threshold is input into the deviation field analysis software. The software traverses each position in the height deviation field, reads the height deviation value at that position, and compares it with the material yield strength threshold. When the height deviation value at a certain position exceeds the threshold, it indicates that the deformation at that position has exceeded the elastic range and entered the plastic deformation state. The software will specially mark this position in the height deviation field (e.g., with a red pixel). All marked positions together constitute the plastic deformation region of the solder ball, and the software automatically records the spatial coordinate range of this region.

[0130] Furthermore, the plastic deformation region data of the solder ball is loaded into the deformation feature analysis software. The software first calculates the curvature value of each point within this region. By analyzing the spatial relationship between the points surrounding that point, the degree of curvature of the surface at that point is determined. Then, the software calculates the change in curvature values ​​between adjacent points along the surface of the plastic deformation region, i.e., the magnitude and direction of the increase or decrease in curvature values ​​from one point to adjacent points. This change forms a curvature gradient. The software arranges the curvature gradients of all points according to their spatial position within the plastic deformation region, using different line densities and color shades to represent the magnitude and direction of the gradient. The resulting image is the deformation feature map of the solder ball, which visually demonstrates the distribution pattern of curvature changes within the plastic deformation region.

[0131] Specifically, on the deformation feature map, a three-dimensional Cartesian coordinate system is established with the geometric center of the solder ball as the origin. Each point on the surface of the solder ball is selected one by one, and the coordinate values ​​of the point on the X, Y, and Z axes in the coordinate system are measured to determine the directed line segment from the origin to the point. This directed line segment is the position vector of the surface point. For each surface point, six adjacent surface points are selected, and the position vectors of these six points are substituted into a plane fitting tool. The tool calculates and determines a plane that fits these six points. The directed line segment perpendicular to this plane is the normal vector of the surface point. The position vectors and normal vectors of all surface points are stored according to their corresponding positions on the deformation feature map.

[0132] Furthermore, a virtual stress field generation tool is invoked, importing the position vectors and normal vectors of all solder ball surface points from the deformation feature map. The tool determines the specific spatial location of each point based on the position vector and the orientation of each point's surface based on the normal vector. It simulates the stress distribution generated when heat is transferred across the solder ball surface under preset temperature variations. The stress component along the position vector reflects the tensile or compressive tendency of that point in space, while the stress component along the normal vector reflects the compressive or delamination tendency of that point's surface. The tool then uses these stress components to form continuous stress field data according to the distribution of surface points and superimposes it onto the deformation feature map, completing the loading of the virtual thermal stress field.

[0133] Furthermore, the Young's modulus and Poisson's ratio of the solder ball material are retrieved from the material property database. Young's modulus reflects the material's ability to resist elastic deformation under stress, while Poisson's ratio reflects the relative deformation ratio perpendicular to the direction of stress. The stress value at each solder ball surface point in the virtual thermal stress field is correlated with that point's Young's modulus and Poisson's ratio. By analyzing the relationship between strain and temperature change at that point under stress, the strain is divided by the corresponding temperature change to obtain the proportion of dimensional change at that point due to temperature change, i.e., the local thermal expansion coefficient of that surface point.

[0134] Furthermore, the data visualization software was launched, and the local thermal expansion coefficient data of all points on the solder ball surface were imported into the software. The software arranged these coefficient values ​​according to the spatial position of each point on the deformation characteristic map. Based on the magnitude of the local thermal expansion coefficient, each point was assigned a corresponding color: points with smaller values ​​were represented by blue, points with larger values ​​by red, and points with values ​​in the middle range by a transitional color from green to yellow. The software connected all the colored points to form a continuous color distribution map, which visually displays the differences in local thermal expansion coefficients at different locations on the solder ball surface; this is the thermal expansion coefficient distribution cloud map of the solder ball.

[0135] Specifically, the local thermal expansion coefficient is a calculated result, while the nominal thermal expansion coefficient comes from the solder ball material's specification manual and is the nominal value of the material's thermal expansion coefficient under standard conditions. The deformation-thermal coupling factor comes from material mechanics experimental data and reflects the degree of correlation between material deformation and thermal effects; it can be retrieved from relevant material property databases. The gradient operator is a mathematical tool used to describe the spatial rate of change of physical quantities, and in the calculation, it is used to handle the three-dimensional displacement changes of points on the solder ball surface. The three-dimensional displacement of a point on the solder ball surface is obtained by comparing the actual three-dimensional topological coordinates of the point with the coordinates of its ideal geometric center, reflecting the change in the point's position in space. The normal vector comes from the normal vector of the solder ball surface point constructed on the deformation feature map and describes the vertical direction of the surface at that point. The direction vector of the virtual thermal stress field comes from the virtual thermal stress field applied on the deformation feature map and describes the direction of the stress field. Young's modulus and Poisson's ratio come from the solder ball material's property database and reflect the material's ability to resist elastic deformation and the relative deformation ratio in the vertical direction under stress, respectively. The norm operator is a mathematical representation used to calculate the magnitude of a vector.

[0136] Furthermore, the significance of this formula is that, based on the nominal thermal expansion coefficient of the material, by introducing a deformation-thermal coupling factor, and combining the magnitude of the cross product of the three-dimensional displacement gradient, normal vector, and virtual thermal stress field direction vector of the solder ball surface points, as well as the Young's modulus and Poisson's ratio of the material, the basic value is corrected, thereby obtaining a local thermal expansion coefficient that can reflect the actual thermal expansion characteristics of different positions on the solder ball surface under the combined action of local deformation and thermal effects. It comprehensively considers the influence of the material's own properties, local deformation, thermal stress, and other factors on the thermal expansion coefficient.

[0137] Furthermore, when the three-dimensional displacement gradient at a point on the solder ball surface increases, meaning the displacement change around that point becomes more drastic, the numerator in the formula increases, the correction term increases accordingly, and the local thermal expansion coefficient becomes greater than the nominal thermal expansion coefficient of the material, with the difference becoming more significant as the displacement gradient increases. When the magnitude of the cross product between the normal vector and the direction vector of the virtual thermal stress field increases, meaning the perpendicular component of the stress direction to the surface normal is larger, the correction term increases, and the local thermal expansion coefficient also increases. If Young's modulus increases, the denominator increases, the correction term decreases, and the local thermal expansion coefficient becomes closer to the nominal thermal expansion coefficient of the material. When Poisson's ratio increases, the denominator... As the value of the coefficient of thermal expansion decreases, the correction term increases, leading to a greater local coefficient of thermal expansion. When the deformation-thermal coupling factor increases, the influence of the correction term intensifies, and the difference between the local coefficient of thermal expansion and the material's nominal coefficient of thermal expansion becomes more pronounced.

[0138] In summary, by separating the deformation feature map of the solder ball in the 3D topology map and applying a virtual thermal stress field to it to obtain the thermal expansion coefficient distribution cloud map, the height deviation field of the solder ball can be accurately obtained through spatial registration with the standard solder ball 3D template. Combined with the material yield strength threshold to mark the plastic deformation area, the curvature gradient distribution features can be extracted, thus achieving accurate separation and quantitative characterization of the solder ball deformation features.

[0139] In summary, by constructing position and normal vectors on the deformation feature map and loading a virtual thermal stress field, and calculating the local thermal expansion coefficient using parameters such as Young's modulus and Poisson's ratio, the resulting thermal expansion coefficient distribution cloud map can intuitively and comprehensively reflect the spatial distribution differences in the thermal expansion properties of the solder ball surface. This process deeply couples deformation characteristics with thermophysical properties, preserving subtle deformation features while achieving accurate calculation of the thermal expansion coefficient. This provides a quantitative basis based on material properties for subsequent identification of abnormal solder balls, significantly improving the sensitivity and analytical depth for potential defects in solder balls.

[0140] S5. Based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range, mark the coordinates of the abnormal solder balls in the BGA chip;

[0141] In this embodiment of the invention, marking the coordinates of abnormal solder balls in the BGA chip based on the comparison result between the thermal expansion coefficient distribution cloud map and the standard threshold range includes:

[0142] The thermal expansion coefficient distribution cloud map is divided into a central region, an edge region, and a corner region according to the position of the solder ball array;

[0143] Differentiated threshold ranges are applied to the central region, the edge region, and the corner region;

[0144] Traverse the thermal expansion coefficient distribution cloud map to identify abnormal data points that exceed the differential threshold range;

[0145] For a cluster region consisting of consecutive abnormal data points, determine the physical coordinates of the centroid of the cluster region in the BGA chip;

[0146] The physical coordinates are mapped and associated with the solder ball number to obtain the coordinates of the abnormal solder ball.

[0147] Specifically, the image processing software corresponding to the thermal expansion coefficient distribution cloud map is opened. The software automatically reads the arrangement coordinates of the solder ball array in the cloud map. These coordinates come from the design drawings of the BGA chip and record the row and column positions of each solder ball in the array. Using the geometric center of the solder ball array as the reference point, the area containing the 3×3 solder balls closest to the reference point is defined as the central region; the area containing the solder balls located in the outermost rows and columns of the array (excluding the four corner positions) is defined as the edge region; and the area containing the individual solder balls in the four corners of the array is defined as the corner region. The boundaries of each region are marked using the region marking tool in the software to ensure that each region covers the corresponding solder ball positions without overlap.

[0148] Furthermore, the normal fluctuation range of the coefficient of thermal expansion for different regions was retrieved from the quality inspection standard document for BGA chips. The central region, due to its more uniform temperature distribution, had a narrower normal range; the edge regions, significantly affected by ambient temperature fluctuations, had a wider normal range; and the corner regions, due to complex stress conditions, had a normal range between the center and the edge. These ranges constitute the differentiated threshold ranges. Using the software's data association function, the threshold range for the central region was loaded into the marked area of ​​the cloud map. Similarly, the threshold ranges for the edge and corner regions were loaded into their corresponding marked areas, allowing the coefficient of thermal expansion data for each region to be compared with its corresponding threshold range.

[0149] Furthermore, the software's automatic traversal function is activated. Starting from the top left corner of the thermal expansion coefficient distribution cloud map, the thermal expansion coefficient values ​​of each solder ball surface point are read one by one in a left-to-right, top-to-bottom order. For each point, the software automatically identifies its region (center, edge, or corner) and compares the thermal expansion coefficient of that point with the differential threshold range of its region. When the value is higher than the upper limit of the range or lower than the lower limit, the software marks the point as an abnormal data point. All abnormal data points are displayed on the cloud map in a special color.

[0150] Furthermore, the software performs connectivity analysis on the marked abnormal data points. By checking the spatial distance between adjacent abnormal points, consecutive abnormal points with a distance less than one-third of the diameter of a single solder ball are grouped into the same cluster region. For each cluster region, the software reads the X and Y coordinate values ​​of all abnormal data points in the BGA chip coordinate system within that region, calculates the arithmetic mean of these coordinate values, and the resulting average is the centroid coordinate of that cluster region. This coordinate directly corresponds to the physical location on the BGA chip, and the software stores the centroid coordinates with micrometer-level precision.

[0151] Furthermore, the solder ball number and its corresponding physical coordinate range are obtained from the solder ball layout diagram of the BGA chip. Each solder ball number in the layout diagram corresponds to a square region with its center as the reference and a side length equal to the solder ball diameter. The software matches the centroid physical coordinates of the cluster region with the solder ball coordinate range in the layout diagram. When the centroid coordinates fall within the coordinate range of a certain solder ball, the solder ball number corresponding to that centroid is determined, and the centroid physical coordinates are bound and stored with that solder ball number to form a key-value pair. All the bound physical coordinates together constitute the abnormal solder ball coordinates.

[0152] In summary, by marking the coordinates of abnormal solder balls based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range, the cloud map can be divided into central, edge, and corner regions according to the position of the solder ball array. Different threshold ranges are applied to different regions, enabling targeted judgment of solder balls at different positions.

[0153] In summary, this method identifies anomalous data points exceeding the differential threshold range by traversing the cloud map. It determines the centroid physical coordinates of clusters of consecutive anomalous data points and associates them with solder ball numbers. This not only accurately locates the specific position of the anomalous solder balls but also fully considers the characteristic differences of solder balls in different regions. It effectively avoids misjudgments or omissions caused by uniform threshold judgments, providing accurate anomalous location information for the subsequent generation of comprehensive quality inspection reports, and significantly improving the accuracy and reliability of anomalous solder ball identification.

[0154] S6. Generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

[0155] In this embodiment of the invention, generating a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map includes:

[0156] An anomaly level label for the BGA chip is generated based on the distribution coefficient of thermal expansion corresponding to the coordinates of the abnormal solder balls.

[0157] By superimposing the thermal stress transmission trajectory on the thermal expansion coefficient distribution cloud map, the thermal influence correlation path between abnormal solder balls in the BGA chip is obtained.

[0158] Spatially align the three-dimensional topology map with the thermal expansion coefficient distribution cloud map to obtain the three-dimensional deformation characteristics of the solder ball.

[0159] By integrating and implementing anomaly level labels, implementing thermal impact correlation paths, and implementing three-dimensional deformation features, a quality inspection report for the BGA chip is obtained.

[0160] Specifically, the distribution coefficients of thermal expansion are extracted from the coordinates of the abnormal solder balls. These coefficients reflect the specific values ​​of the abnormal solder balls in terms of thermal expansion characteristics. These coefficients are then compared with the ranges of thermal expansion coefficients corresponding to different abnormality levels specified in the BGA chip quality inspection standard. When the coefficient deviates slightly from the standard range, a "minor abnormality" level label is generated; when the coefficient deviates moderately, a "moderate abnormality" level label is generated; and when the coefficient far exceeds the standard range, a "serious abnormality" level label is generated. Each abnormal solder ball coordinate corresponds to a unique abnormality level label, and the label and coordinate are bound together using a data association tool.

[0161] Furthermore, thermal stress transmission trajectories are extracted from the analysis data of the virtual thermal stress field. These trajectories record the path and direction of thermal stress transmission between solder balls. The order of stress transmission and the positions of solder balls along the path can be determined by the node coordinates on the trajectory. The thermal expansion coefficient distribution cloud map is loaded into the trajectory visualization software. Based on the node coordinates of the trajectory, the software overlays the thermal stress transmission trajectories as continuous lines onto the cloud map. The thickness of the lines represents the intensity of stress transmission, and the color represents the order of transmission. The linear network connecting abnormal solder balls formed after overlay is the thermal influence correlation path between abnormal solder balls in the BGA chip.

[0162] Furthermore, a unified three-dimensional spatial coordinate system is established with the geometric center of the BGA chip substrate as the common origin. Both the three-dimensional topology map and the thermal expansion coefficient distribution cloud map are based on this coordinate system for coordinate transformation, ensuring that their spatial coordinate systems are consistent. By identifying feature points such as vertices and edge midpoints of the solder balls in the three-dimensional topology map and corresponding feature points of the solder balls in the thermal expansion coefficient distribution cloud map, these feature points are matched and aligned one by one, making the two images completely overlap in spatial position. After alignment, three-dimensional morphological information such as height changes and shape distortion of the solder balls is extracted from the three-dimensional topology map. Combined with the differences in thermal expansion characteristics at corresponding positions in the thermal expansion coefficient distribution cloud map, these features together constitute the three-dimensional deformation characteristics of the solder balls. These features include comprehensive changes in the spatial morphology and thermal expansion characteristics of the solder balls.

[0163] Furthermore, the implementation of anomaly level labels, thermal impact correlation paths, and 3D deformation features are imported into the report generation software. The software integrates this data according to a preset quality inspection report template. The first part of the template is a list of abnormal solder balls, listing the number, anomaly level label, and corresponding physical coordinates of each abnormal solder ball; the second part is a thermal impact analysis, which uses a diagram of the thermal impact correlation path to show the heat transfer relationship and impact range between abnormal solder balls; the third part is a 3D deformation description, which combines a 3D model of real-time 3D deformation features with textual descriptions to illustrate the specific morphological changes of the abnormal solder balls. After all parts are integrated, the software automatically generates a complete document containing charts and textual descriptions. This document is the quality inspection report for the BGA chip, and all data and charts in the report correspond one-to-one with the products generated in the previous steps.

[0164] In summary, the quality inspection report for BGA chips is generated based on the coordinates of abnormal solder balls and the distribution cloud map of the coefficient of thermal expansion. It can generate anomaly level labels based on the distribution coefficient of thermal expansion corresponding to the coordinates of abnormal solder balls, clearly defining the severity of defects of different solder balls. At the same time, the thermal stress transmission trajectory is superimposed on the cloud map, intuitively presenting the thermal impact correlation path between abnormal solder balls and revealing the potential transmission law of defects.

[0165] In summary, by spatially aligning the 3D topology map with the thermal expansion coefficient distribution cloud map, the 3D deformation characteristics of the solder balls can be fully presented, enabling multi-dimensional visualization of defects. This quality inspection report, which integrates anomaly level, thermal impact path, and 3D deformation characteristics, not only comprehensively covers key quality inspection information but also provides accurate and systematic basis for subsequent analysis and decision-making. It significantly improves the practicality and guidance of the quality inspection results, effectively supporting the efficient implementation of BGA chip quality control.

[0166] like Figure 2 The diagram shown is a functional block diagram of a visual quality inspection system for BGA chips provided in an embodiment of the present invention.

[0167] The visual quality inspection system 100 for BGA chips described in this invention can be installed in electronic devices. Depending on the functions implemented, the visual quality inspection system 100 for BGA chips may include a light source illumination module 101, an image segmentation module 102, an image fusion module 103, a thermal expansion coefficient distribution cloud map generation module 104, an abnormal solder ball monitoring module 105, and a quality inspection report generation module 106. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and which are stored in the memory of the electronic device.

[0168] In this embodiment, the functions of each module / unit are as follows:

[0169] The light source illumination module 101 is used to illuminate the BGA chip through a multispectral light source array to obtain an initial reflection image of the BGA chip.

[0170] The image segmentation module 102 is used to segment the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions of the solder balls.

[0171] The image fusion module 103 is used to acquire multi-view differential images of the same solder ball area in the independent imaging area, and fuse the multi-view differential images into a three-dimensional topology map;

[0172] The thermal expansion coefficient distribution cloud map generation module 104 is used to separate the deformation feature map of the solder ball in the three-dimensional topology map and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball.

[0173] The abnormal solder ball monitoring module 105 is used to mark the coordinates of abnormal solder balls in the BGA chip based on the comparison result between the thermal expansion coefficient distribution cloud map and the standard threshold range.

[0174] The quality inspection report generation module 106 is used to generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

[0175] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0176] The modules described as separate components may or may not be physically separate. The components shown as modules 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.

[0177] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0178] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0179] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0180] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A visual quality inspection method for BGA chips, characterized in that, The method includes: S1. Illuminate the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip; S2. Based on the geometric distribution characteristics of the solder balls in the BGA chip, the initial reflection image is segmented to obtain independent imaging regions of the solder balls; S3. Acquire multi-view differential images of the same solder ball region in the independent imaging region, and fuse the multi-view differential images into a three-dimensional topology map; S4. Separate the deformation feature map of the solder ball in the three-dimensional topology map, and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball; S5. Based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range, mark the coordinates of the abnormal solder balls in the BGA chip; S6. Generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

2. The visual quality inspection method for BGA chips as described in claim 1, characterized in that, The step of illuminating the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip includes: Send wavelength switching commands to the multispectral light source array to activate the combined illumination mode of near-infrared and visible light bands; Based on the material properties of the substrate in the BGA chip, the incident angle of the combined illumination mode is adjusted to a preset tilt angle range, and the surface reflected light intensity distribution data of the solder balls in the BGA chip is captured. By fusing the surface reflected light intensity distribution data, an initial reflection image of the BGA chip is obtained.

3. The visual quality inspection method for BGA chips as described in claim 2, characterized in that, The method of segmenting the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions for the solder balls includes: Identify regions of optical response difference in the initial reflection image, and delineate the boundary transition zone between the solder ball and the substrate based on these regions of optical response difference; Based on the preset solder ball spatial distribution template, locate the low reflectivity isolation zone between the solder balls; A closed cutting path for generating the initial reflection image is generated along the low-reflectivity isolation zone; Based on the closed cutting path and the boundary transition zone, the solder ball imaging block of the initial reflection image is separated to obtain the independent imaging region of the solder ball.

4. The visual quality inspection method for BGA chips as described in claim 1, characterized in that, The acquisition of multi-view differential images of the same solder ball region within the independent imaging region includes: Within the independent imaging area, all viewing angles remain at a fixed exposure time; Simultaneously capture the surface microstructure reflection features of the solder ball from the stated viewpoint; Pixel-level difference operations are performed on the reflection features of adjacent viewpoints to obtain the multi-view difference image of the solder ball.

5. The visual quality inspection method for BGA chips as described in claim 4, characterized in that, The process of fusing the multi-view difference images into a three-dimensional topological map includes: Extract the solder ball contour deformation gradient vector of each frame of the multi-view differential image; The coefficient of thermal expansion of the solder ball material used to load the BGA chip and the current ambient temperature; Based on the solder ball profile deformation gradient vector, the thermal expansion coefficient of the solder ball material, and the current ambient temperature, the three-dimensional topological coordinates of the solder ball are calculated, wherein the calculation formula for the three-dimensional topological coordinates is as follows: ； In the formula, The three-dimensional topological coordinates are... The total number of the stated viewpoints. Let be the ordinal number of the stated viewpoint. For the first The weld ball profile deformation gradient vector from each perspective The coefficient of thermal expansion of the solder ball material is... The current ambient temperature, For standard reference temperature, The coordinates of the ideal geometric center of the solder ball; A three-dimensional topology map of the solder ball is constructed based on the three-dimensional topological coordinates.

6. The visual quality inspection method for BGA chips as described in claim 1, characterized in that, The separation of the deformation feature map of the solder ball in the three-dimensional topology map includes: The three-dimensional topology map is spatially registered with the standard solder ball three-dimensional template to obtain the height deviation field of the solder ball. Based on the material yield strength threshold of the welding ball, mark the plastic deformation region in the height deviation field; The curvature gradient distribution characteristics of the plastic deformation region are extracted to obtain the deformation feature map of the solder ball.

7. The visual quality inspection method for BGA chips as described in claim 6, characterized in that, The step of applying a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball includes: Construct the position vector and normal vector of the solder ball surface points on the deformation feature map; A virtual thermal stress field is applied to the deformation feature map based on the position vector and the normal vector. Based on the Young's modulus and Poisson's ratio of the virtual thermal stress field, the local thermal expansion coefficient of the solder ball is calculated, wherein the formula for calculating the local thermal expansion coefficient is as follows: ； In the formula, The local thermal expansion coefficient is... The nominal coefficient of thermal expansion of the material. The deformation-thermal coupling factor. For gradient operators, This represents the three-dimensional displacement of a point on the surface of the solder ball. Let be the normal vector. Let be the direction vector of the virtual thermal stress field. For Young's modulus, Poisson's ratio, Norm operator; By iterating through the local thermal expansion coefficients of the points on the surface of the solder ball, a thermal expansion coefficient distribution cloud map of the solder ball is obtained.

8. The visual quality inspection method for BGA chips as described in claim 1, characterized in that, The step of marking the coordinates of abnormal solder balls in the BGA chip based on the comparison result between the thermal expansion coefficient distribution cloud map and the standard threshold range includes: The thermal expansion coefficient distribution cloud map is divided into a central region, an edge region, and a corner region according to the position of the solder ball array; Differentiated threshold ranges are applied to the central region, the edge region, and the corner region; Traverse the thermal expansion coefficient distribution cloud map to identify abnormal data points that exceed the differential threshold range; For a cluster region consisting of consecutive abnormal data points, determine the physical coordinates of the centroid of the cluster region in the BGA chip; The physical coordinates are mapped and associated with the solder ball number to obtain the coordinates of the abnormal solder ball.

9. The visual quality inspection method for BGA chips as described in claim 1, characterized in that, The step of generating a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map includes: Based on the distribution coefficient of thermal expansion corresponding to the coordinates of the abnormal solder balls, an abnormality level label is generated for the BGA chip; By superimposing the thermal stress transmission trajectory on the thermal expansion coefficient distribution cloud map, the thermal influence correlation path between abnormal solder balls in the BGA chip is obtained. Spatially align the three-dimensional topology map with the thermal expansion coefficient distribution cloud map to obtain the three-dimensional deformation characteristics of the solder ball. By integrating and implementing anomaly level labels, implementing thermal impact correlation paths, and implementing three-dimensional deformation features, a quality inspection report for the BGA chip is obtained.

10. A visual quality inspection system for BGA chips, characterized in that, The system includes: A light source illumination module is used to illuminate the BGA chip with a multispectral light source array to obtain an initial reflection image of the BGA chip; The image segmentation module is used to segment the initial reflection image based on the geometric distribution characteristics of the solder balls in the BGA chip to obtain independent imaging regions of the solder balls; The image fusion module is used to acquire multi-view differential images of the same solder ball area in the independent imaging area and fuse the multi-view differential images into a three-dimensional topology map; The thermal expansion coefficient distribution cloud map generation module is used to separate the deformation feature map of the solder ball in the three-dimensional topology map and apply a virtual thermal stress field to the deformation feature map to obtain the thermal expansion coefficient distribution cloud map of the solder ball. An abnormal solder ball monitoring module is used to mark the coordinates of abnormal solder balls in the BGA chip based on the comparison results between the thermal expansion coefficient distribution cloud map and the standard threshold range. The quality inspection report generation module is used to generate a quality inspection report for the BGA chip based on the coordinates of the abnormal solder balls and the thermal expansion coefficient distribution cloud map.

Images