A laser-ultrasound-based BGA solder joint defect detection system and method

Abstract

This invention relates to a laser-ultrasound-based BGA solder joint defect detection system and method, belonging to the field of chip inspection. The system includes a laser-ultrasound emitting device, a laser-ultrasound detection device, a data processing computer, a mobile inspection platform, and a camera. The laser-ultrasound-based BGA solder joint defect detection system of this invention can automatically record data and automatically focus, reducing labor costs. The laser-ultrasound-based detection method of this invention uses machine learning methods to pre-train models on defective and defect-free samples, automatically identifying whether the sample contains defects during the inspection process. Compared with other BGA solder joint detection methods, the method of this invention has the advantages of being non-destructive, fast, accurate, and low-cost.

Description

Technical Field

[0001] This invention belongs to the field of chip inspection, specifically relating to a BGA solder joint defect detection system and method based on laser ultrasound. Background Technology

[0002] As integrated circuits continue to evolve towards higher integration, Ball Grid Array (BGA) has become a mainstream electronic packaging technology. This packaging method can effectively reduce output voltage fluctuations and data transmission latency. The key feature of BGA packaging is the array of solder balls beneath the chip, which significantly reduces device size while increasing the number of pins. Therefore, it is widely used in the integrated packaging of multi-pin chips such as high-performance processors.

[0003] The advancement of BGA packaged devices has significantly driven the development of chip soldering technology, resulting in smaller and more numerous solder balls, more refined soldering processes, and increasingly stringent requirements for solder ball quality assurance. Quality parameters for BGA solder balls include: number of solder balls, solder ball diameter, and solder ball center point. Quality defects arising from parameters not meeting requirements include missing / redundant solder balls, solder ball deformation, incorrect solder ball placement, and solder ball bridging. These defects can cause the chip to malfunction.

[0004] Existing BGA solder joint inspection methods are mainly divided into destructive testing and non-destructive testing. Due to the characteristics of our task, we first exclude destructive testing. In non-destructive testing, the mainstream method is X-ray-based inspection, but the testing equipment is relatively expensive. Existing vision inspection systems for BGA packaged chips have poor inspection accuracy and unsatisfactory inspection results, and cannot effectively determine whether there are defects in the connection between the BGA packaged chip and the PCB board after chip mounting. Currently, the international research focus is on laser-ultrasonic inspection methods. This method is fast, accurate, and low-cost, and is expected to be more widely used.

[0005] Existing non-destructive testing methods for BGA packaged chip solder balls mainly rely on X-ray inspection and machine vision inspection. X-ray inspection equipment is bulky and expensive, but it can inspect the inside of the solder balls. Existing vision inspection systems for BGA packaged chips have poor inspection accuracy and unsatisfactory results, failing to effectively identify defects in the BGA packaged chips and solder balls present during the manufacturing process, and lack automated means for recording and tracking solder ball inspection results.

[0006] For BGA solder joint defects, there are generally two types of detection methods: destructive testing and non-destructive testing.

[0007] Destructive testing methods include:

[0008] 1. Red ink test method

[0009] This method can determine the soldering cracking by observing the staining of solder joints on printed circuit boards (PCBs) and chip components, and can examine whether there are defects such as cold solder joints, false solder joints, and cracks in the soldering of electronic components.

[0010] The principle of the red ink test is based on the penetrability of liquids. The solder joint is placed in a red dye, allowing the dye to penetrate into the cracks of the solder joint. After drying, the solder joint is forcibly separated, and the color of the interface at the crack is observed to determine whether the solder joint has broken.

[0011] 2. Slice analysis method

[0012] This method is an important technique for PCB failure analysis. Cross-section analysis is more time-consuming than red ink testing. The cross-section analysis process includes: sampling, cleaning, vacuum mounting, grinding, polishing, micro-etching, and analysis. The quality of the cross-section directly affects the accuracy of identifying the failure location, thus requiring highly skilled personnel. Based on the cross-section analysis, SEM and EDS can be used for further analysis of the failure causes of solder joints.

[0013] Since chips subjected to destructive testing are unusable, they can only be used for sampling and inspection of BGA solder joints or for fault analysis, and cannot guarantee 100% reliability of the solder joints. Non-destructive testing should be used in such scenarios.

[0014] Non-destructive testing methods include:

[0015] 1. Visual inspection method

[0016] This method involves observing the solder joints with a high-magnification microscope to initially detect any obvious defects from their appearance. However, this method has significant limitations; it can only be used for preliminary assessment and cannot determine whether there are other internal defects or voids on the surface of the solder joint.

[0017] 2. Machine vision inspection method

[0018] This method is similar to visual inspection, except that the detection subject is changed from a human to a computer. It is mainly divided into 2D detection and 3D detection. 3D detection, in order to reconstruct a 3D image, must rely on structured light to obtain a relatively accurate 3D image reconstruction result. Like visual inspection, this method cannot detect the inside of solder balls or obscured solder joints.

[0019] 2. X-ray detection method

[0020] This method is a non-destructive physical imaging method that uses X-rays to inspect components and detect their internal packaging conditions, such as bubbles, cracks, and abnormal bonding lines.

[0021] 3.2 DX Ray Method

[0022] This method utilizes the recorded changes in light intensity after X-rays penetrate materials of different densities to create a contrasting image, which can be used to analyze problematic areas inside the solder joint. This method can identify BGA soldering defects such as open solder joints and cold solder joints, as well as analyze internal displacement in BGAs and PCBs, and identify defects such as bridging and short circuits.

[0023] 4.3 DX-ray computed tomography (CT scan)

[0024] This method effectively solves the limitations of 2D X-ray methods, can present three-dimensional images, and has high density resolution and high spatial resolution. It can accurately and clearly observe the welding quality and structural defects of BGA solder joints, and can also show the shape, location and size of defects inside the weld.

[0025] In the process of inspecting solder joints using the aforementioned non-destructive testing methods, both visual inspection and 2D X-ray inspection have limitations. 3D X-ray tomography can solve the problem of solder joint defect detection more perfectly, but the testing cost is relatively high.

[0026] 5. Laser Ultrasonic Testing Method (LUI)

[0027] The principle of laser ultrasonic testing for BGA solder joint defects is to focus a pulsed laser beam onto the surface of the chip package under test. Due to the thermoelasticity of the chip package material, minute deformations occur, generating ultrasonic waves. These ultrasonic signals are then collected by a testing device. Because different BGA solder joint defects cause slight differences in vibration, these differences can be detected by the laser ultrasonic testing device. However, this method is currently in the experimental stage, and no commercially available product has yet been released. Summary of the Invention

[0028] (a) Technical problems to be solved

[0029] The technical problem to be solved by the present invention is how to provide a BGA solder joint defect detection system and method based on laser ultrasound, so as to solve the problems existing in BGA solder joint defect detection.

[0030] (II) Technical Solution

[0031] To address the aforementioned technical problems, this invention proposes a laser-ultrasound-based BGA solder joint defect detection system, which includes a laser-ultrasound emitting device, a laser-ultrasound detection device, a data processing computer, a mobile detection platform, and a camera.

[0032] Laser ultrasonic emitting devices are used to generate ultrasonic waves by converting laser thermal energy into vibrations of chip packages, or to convert electrical signals into ultrasonic signals by using piezoelectric conversion devices.

[0033] The laser ultrasound focusing control system connects the laser ultrasound emitting device and the laser ultrasound detection device. It is used to control the laser ultrasound emitting device to move forward or backward in the vertical direction, thereby adjusting the focusing degree of the laser ultrasound.

[0034] The laser ultrasonic testing device connects the laser ultrasonic focusing control system and the data processing computer to detect the laser echo vibration with nanometer-scale amplitude on the surface of BGA packaged chips.

[0035] The mobile testing platform is connected to a data processing computer and uses a fixture to hold the BGA packaged chip under test. It moves in the X and Y directions and has a built-in platform level detection device to ensure the horizontal movement of the system and to detect whether there are strong vibrations in the current environment that could affect the testing experiment.

[0036] The camera connects to a data processing computer, can focus on images, is used for chip character recognition and chip pose acquisition, and has a retractable camera.

[0037] The data processing computer is used to collect, analyze, and record test data, and run chip character recognition algorithms and chip pose control algorithms. During chip pose control, the data processing computer uses image processing algorithms to perform image matching on the target BGA packaged chip within the camera's field of view. After successful matching, it performs further focusing to ensure that the characters are clearly displayed for BGA packaged chip character recognition. During chip character recognition, the camera identifies the chip's identification characters and stores the chip identification information and related test information in the database.

[0038] Furthermore, the laser ultrasonic testing device is a laser interferometer or a laser scanning camera.

[0039] This invention also provides a method for detecting BGA solder joint defects based on laser ultrasound, the method comprising the following steps:

[0040] Step S1: Establish a training dataset of qualified and defective BGA solder joints based on laser ultrasonic echo, and train the BGA solder joint laser ultrasonic echo classification model.

[0041] Step S2: After the model training is completed, perform image detection on the chip with BGA solder joints to be tested, automatically focus the camera on the chip's identification characters, and record the chip information to be tested;

[0042] Step S3: Move the chip under test to the position directly opposite the laser ultrasonic transmitter;

[0043] Step S4: Automatically focus the laser ultrasonic transmitter to the optimal detection position; move the laser ultrasonic transmitter vertically until the signal-to-noise ratio of the ultrasonic echo is maximized;

[0044] Step S5: Perform laser ultrasonic echo detection on the BGA solder joints of the chip under test;

[0045] Step S6: Process and analyze the laser ultrasonic echo detection data at each location, and then input them into the BGA solder joint laser ultrasonic echo classification model trained in Step S1 for calculation and classification.

[0046] Step S7: Enter the test results into the database system.

[0047] Furthermore, step S1 specifically includes the following steps:

[0048] S11. First, select several PCB boards with qualified BGA solder joints and typical defective BGA solder joints, place them on the test platform, identify their identification characters and enter them into the system for unified management of test data.

[0049] S12. Perform laser ultrasonic testing on the PCB board and record the ultrasonic echo data of each BGA packaged chip as a qualified ultrasonic echo data set and an unqualified ultrasonic echo data set.

[0050] S13. Using the ultrasonic echo data of the BGA packaged chip as the model input and the defect or integrity status of the BGA solder joint as the classification result, the model is trained to obtain the trained BGA solder joint laser ultrasonic echo classification model.

[0051] Furthermore, the detection process in step S12 includes: the solder joints are an N*N matrix, starting from the upper left corner of the chip and ending at the lower right corner, the entire chip is divided into a (N / 3)*(N / 3) grid, that is, there are 9 solder joints in each grid, the detection platform moves in a serpentine manner along the center position of each grid, and performs ultrasonic echo detection at each position.

[0052] Furthermore, in step S13, the laser ultrasound echo classification model uses a CNN convolutional neural network model.

[0053] Furthermore, in step S13, during the training process of the BGA solder joint laser ultrasonic echo classification model, different kernel functions are selected for optimization and adjustment by comparing the classification results with the type labels corresponding to the echo data. The type labels include cold solder joint, void, crack, bridging, and no defect. The qualified training dataset and the unqualified training dataset are input into the optimized BGA solder joint laser ultrasonic echo classification model for model training. When the training results converge, the optimal echo detection model is obtained.

[0054] Further, step S2 specifically includes: placing the BGA packaged chip under test on a movable detection platform, and the detection platform moving regularly along the X-axis and Y-axis respectively. First, it moves from the origin along the X-axis and then back to the origin. After moving a certain fixed distance on the Y-axis, it moves along the X-axis again until the chip under test is in the camera's field of view. During this process, the data processing computer uses a pre-captured image of the BGA packaged chip as a matching template and continuously uses an image matching algorithm to process the captured images. When the image of the chip under test in the field of view matches the matched image, the movement along the X-axis and Y-axis is stopped.

[0055] Furthermore, in step S2, before identifying the identifier of the chip under test, the clarity of the characters is judged by a gradient evaluation algorithm. The camera focal length is finely adjusted by the control of the data processing computer to achieve the best focusing state. The chip information is collected and entered into the chip defect management database system by an OCR algorithm, recording the current chip manufacturing number and test time information.

[0056] Further, step S6 specifically includes: first performing wavelet transform denoising processing on the ultrasonic echo detection data at each location, and then sending the detection data into the BGA solder joint laser ultrasonic echo classification model in step S1 for calculation and classification to determine whether there are defective solder joints among the 9 solder joints at that location; if there are defects, then performing ultrasonic echo detection on the 9 solder joint locations in this area to determine which solder joint has a defect.

[0057] (III) Beneficial Effects

[0058] This invention proposes a BGA solder joint defect detection system and method based on laser ultrasound. The BGA solder joint defect detection system based on laser ultrasound can automatically record data and automatically focus, reducing labor costs.

[0059] This invention relates to a laser ultrasound-based detection method: using machine learning, a model is pre-trained on both defective and defect-free samples, and during the detection process, it automatically identifies whether the sample contains defects.

[0060] Compared with other BGA solder joint detection methods, the method of the present invention has the characteristics of being non-destructive, fast, accurate, and low-cost. Attached Figure Description

[0061] Figure 1 This is a structural diagram of the BGA solder joint defect detection system based on laser ultrasound of the present invention. Detailed Implementation

[0062] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0063] This invention utilizes the principle of laser ultrasonic inspection (LUI) to fully leverage the advantages of LUI, such as fast detection speed, high detection efficiency, and low cost, to realize a usable and easy-to-use detection device and corresponding detection method.

[0064] To achieve the above objectives, the present invention provides the following technical solution:

[0065] like Figure 1 As shown, a BGA solder joint defect detection system based on laser ultrasound includes:

[0066] Laser ultrasonic emitting device, laser ultrasonic testing device, data processing computer, mobile testing platform, camera, etc.

[0067] Laser ultrasonic transmitter: used to generate ultrasonic waves by converting laser thermal energy into vibrations of chip packages, or to convert electrical signals into ultrasonic signals through a piezoelectric converter.

[0068] Laser ultrasound focusing control system: Connects the laser ultrasound emitting device and the laser ultrasound detection device, and can control the laser ultrasound emitting device to move forward or backward in the vertical direction, thereby adjusting the degree of laser ultrasound focusing.

[0069] Laser ultrasonic testing device: Connects to a laser ultrasonic focusing control system and a data processing computer, used to detect the nanometer-scale amplitude vibration of the BGA packaged chip surface, hereinafter referred to as laser echo, and can use laser interferometers, laser scanning cameras, etc.

[0070] Mobile testing platform: Connects to a data processing computer and holds the BGA packaged chip under test with a fixture. It can move in the X and Y directions. It has a built-in platform level detection device to ensure the horizontal movement of the system and can detect whether there are strong vibrations in the current environment that can affect the testing experiment.

[0071] Camera: Connects to a data processing computer, can focus on images, is used for chip character recognition and chip pose acquisition, and has a camera that can be extended and controlled.

[0072] Data processing computer: Used to collect, analyze, and record test data, and run chip character recognition algorithms and chip pose control algorithms; during chip pose control, the data processing computer uses image processing algorithms to perform image matching on the target BGA packaged chip within the camera's field of view. After successful matching, further focusing is performed to ensure that the characters can be clearly displayed for BGA packaged chip character recognition; during chip character recognition, the camera identifies the chip identification characters and stores the chip identification information and related test information in the database.

[0073] The laser-ultrasound-based BGA solder joint defect detection method of the present invention includes the following steps:

[0074] Step S1: Establish a training dataset of qualified and defective BGA solder joints based on laser ultrasonic echo, and train the BGA solder joint laser ultrasonic echo classification model.

[0075] S11. First, select several PCB boards with qualified BGA solder joints and typical defective BGA solder joints, place them on the test platform, identify their identification characters and enter them into the system for unified management of test data.

[0076] S12. Perform laser ultrasonic testing on the PCB board and record the ultrasonic echo data of each BGA packaged chip as a qualified ultrasonic echo data set and an unqualified ultrasonic echo data set.

[0077] The testing process is as follows: taking an N*N matrix of solder joints as an example, starting from the top left corner of the chip and ending at the bottom right corner, the entire chip is divided into a (N / 3)*(N / 3) (rounded up) grid, meaning each grid contains 9 solder joints. The testing platform moves in a serpentine pattern along the center of each grid. Ultrasonic echo testing is performed at each location.

[0078] S13. Input the qualified and unqualified ultrasonic echo data datasets into the laser ultrasonic echo classification model (e.g., using a CNN convolutional neural network model or other machine learning models suitable for data classification) to train the model. Use the ultrasonic echo data of BGA packaged chips as the model input and the defect or integrity status of BGA solder joints as the classification result to train the model and obtain the trained BGA solder joint laser ultrasonic echo classification model.

[0079] Preferably, during the training process of the BGA solder joint laser ultrasonic echo classification model, different kernel functions are selected for optimization and adjustment by comparing the classification results with the type labels corresponding to the echo data (i.e., defects of BGA packaged chips, such as cold solder joints, voids, cracks, bridging, etc., as well as defect-free states).

[0080] Preferably, the qualified training dataset and the unqualified training dataset are input into the optimized BGA solder joint laser ultrasonic echo classification model for model training. When the training results converge, the optimal echo detection model is obtained.

[0081] Step S2: After the model training is completed, image detection is performed on the chip with BGA solder joints to be tested. The camera is automatically focused on the chip's identification characters, and the chip information is recorded.

[0082] The BGA packaged chip under test is placed on a movable testing platform. The testing platform moves systematically along the X and Y axes, first from the origin, then translating along the X-axis and returning to the origin. After moving a fixed distance along the Y-axis, it moves along the X-axis again until the chip under test is within the camera's field of view. During this process, the data processing computer uses pre-captured images of the BGA packaged chip as matching templates, continuously processing the captured images using image matching algorithms. Movement along the X and Y axes stops when the image of the chip under test in the field of view matches the matched image.

[0083] Before identifying the identifier of the chip under test, the clarity of the characters is judged by algorithms such as gradient evaluation. The camera focus is finely adjusted by the control of the data processing computer to achieve the best focusing state. The chip information is collected and entered into the chip defect management database system through the OCR algorithm, recording information such as the current chip serial number and test time.

[0084] Step S3: Move the chip under test to the position directly opposite the laser ultrasonic transmitter.

[0085] Calculate the absolute position of the chip from the previous step, then obtain the target position by projecting the relative position of the laser ultrasonic transmitter and the visual camera onto the plane. Calculate the distance the detection platform should move along the X and Y axes, and then move the chip under test to that position.

[0086] Step S4: Automatically focus the laser ultrasonic emitting device to the optimal detection position.

[0087] Move the laser ultrasonic transmitter vertically until the signal-to-noise ratio of the ultrasonic echo is maximized.

[0088] Step S5: Perform laser ultrasonic echo detection on the BGA solder joints of the chip under test.

[0089] Following the laser ultrasonic echo detection process in step S1, laser ultrasonic echo acquisition is performed on the chip under test.

[0090] Step S6: Process and analyze the laser ultrasonic echo detection data from various locations, and then input them into the BGA solder joint laser ultrasonic echo classification model trained in Step S1 for calculation and classification.

[0091] The ultrasonic echo detection data from each location are first processed using wavelet transform for noise reduction. Then, the detection data is fed into the BGA solder joint laser ultrasonic echo classification model in step S1 for calculation and classification to determine whether there are defective solder joints among the nine solder joints at that location. If a defect is found, ultrasonic echo detection is then performed on each of the nine solder joints in that area to determine which solder joint has the defect.

[0092] Step S7: Enter the test results into the database system.

[0093] The laser-ultrasound-based BGA solder joint defect detection system of the present invention can automatically record data and automatically focus, reducing labor costs.

[0094] This invention relates to a laser ultrasound-based detection method: using machine learning, a model is pre-trained on both defective and defect-free samples, and during the detection process, it automatically identifies whether the sample contains defects.

[0095] Compared with other BGA solder joint detection methods, the method of the present invention has the characteristics of being non-destructive, fast, accurate, and low-cost.

[0096] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting BGA solder joint defects based on laser ultrasound, characterized in that, This method utilizes a laser-ultrasound-based BGA solder joint defect detection system, which includes a laser-ultrasound emitting device, a laser-ultrasound detection device, a data processing computer, a mobile detection platform, and a camera. Laser ultrasonic emitting devices are used to generate ultrasonic waves by converting laser thermal energy into vibrations of chip packages, or to convert electrical signals into ultrasonic signals by using piezoelectric conversion devices. The laser ultrasound focusing control system connects the laser ultrasound emitting device and the laser ultrasound detection device. It is used to control the laser ultrasound emitting device to move forward or backward in the vertical direction, thereby adjusting the focusing degree of the laser ultrasound. The laser ultrasonic testing device connects the laser ultrasonic focusing control system and the data processing computer to detect the laser echo vibration with nanometer-scale amplitude on the surface of BGA packaged chips. The mobile testing platform is connected to a data processing computer and uses a fixture to hold the BGA packaged chip under test. It moves in the X and Y directions and has a built-in platform level detection device to ensure the horizontal movement of the system and to detect whether there are strong vibrations in the current environment that could affect the testing experiment. The camera connects to a data processing computer, can focus on images, is used for chip character recognition and chip pose acquisition, and has a retractable camera. The data processing computer is used to collect, analyze, and record test data, and run chip character recognition algorithms and chip pose control algorithms. During chip pose control, the data processing computer uses image processing algorithms to perform image matching on the target BGA-packaged chip within the camera's field of view. After successful matching, further focusing is performed to ensure clear character display for BGA-packaged chip character recognition. During chip character recognition, the camera identifies the chip's identification characters, and the chip's identification information and related test information are stored in the database. The method includes the following steps: Step S1: Establish a training dataset of qualified and defective BGA solder joints based on laser ultrasonic echo, and train the BGA solder joint laser ultrasonic echo classification model. Step S2: After the model training is completed, perform image detection on the chip with BGA solder joints to be tested, automatically focus the camera on the chip's identification characters, and record the chip information to be tested; Step S3: Move the chip under test to the position directly opposite the laser ultrasonic transmitter; Step S4: Automatically focus the laser ultrasonic transmitter to the optimal detection position; move the laser ultrasonic transmitter vertically until the signal-to-noise ratio of the ultrasonic echo is maximized; Step S5: Perform laser ultrasonic echo detection on the BGA solder joints of the chip under test; Step S6: Process and analyze the laser ultrasonic echo detection data at each location, and then input them into the BGA solder joint laser ultrasonic echo classification model trained in Step S1 for calculation and classification; Step S7: Enter the test results into the database system; in, During the training of the BGA solder joint laser ultrasonic echo classification model, different kernel functions were selected for optimization by comparing the classification results with the type labels corresponding to the echo data. The type labels included cold solder joint, void, crack, bridging, and no defect. The qualified training dataset and the unqualified training dataset were input into the optimized BGA solder joint laser ultrasonic echo classification model for model training. When the training results converged, the optimal echo detection model was obtained.

2. The BGA solder joint defect detection method based on laser ultrasound as described in claim 1, characterized in that, Step S1 specifically includes the following steps: S11. First, select several PCB boards with qualified BGA solder joints and typical defective BGA solder joints, place them on the test platform, identify their identification characters and enter them into the system for unified management of test data. S12. Perform laser ultrasonic testing on the PCB board and record the ultrasonic echo data of each BGA packaged chip as a qualified ultrasonic echo data set and an unqualified ultrasonic echo data set. S13. Using the ultrasonic echo data of the BGA packaged chip as the model input and the defect or integrity status of the BGA solder joint as the classification result, the model is trained to obtain the trained BGA solder joint laser ultrasonic echo classification model.

3. The BGA solder joint defect detection method based on laser ultrasound as described in claim 2, characterized in that, The detection process in step S12 includes: the solder joints are an N*N matrix, starting from the upper left corner of the chip and ending at the lower right corner, the entire chip is divided into (N / 3)*(N / 3) grids, that is, there are 9 solder joints in each grid. The detection platform moves in a serpentine manner along the center position of each grid and performs ultrasonic echo detection at each position.

4. The BGA solder joint defect detection method based on laser ultrasound as described in claim 2, characterized in that, In step S13, the laser ultrasonic echo classification model uses a CNN convolutional neural network model.

5. The BGA solder joint defect detection method based on laser ultrasound as described in claim 1, characterized in that, Step S2 specifically includes: placing the BGA packaged chip under test on a movable detection platform, and moving the detection platform regularly along the X-axis and Y-axis respectively. First, it moves from the origin along the X-axis and then back to the origin. After moving a certain fixed distance on the Y-axis, it moves along the X-axis again until the chip under test is in the camera's field of view. During this process, the data processing computer uses a pre-captured image of the BGA packaged chip as a matching template and continuously uses an image matching algorithm to process the captured images. When the image of the chip under test in the field of view matches the matched image, the movement along the X-axis and Y-axis is stopped.

6. The BGA solder joint defect detection method based on laser ultrasound as described in claim 5, characterized in that, In step S2, before identifying the identifier of the chip under test, the clarity of the characters is judged by a gradient evaluation algorithm. The camera focal length is finely adjusted by the control of the data processing computer to achieve the best focusing state. The chip information is collected and entered into the chip defect management database system by the OCR algorithm, recording the current chip manufacturing number and test time information.

7. The BGA solder joint defect detection method based on laser ultrasound as described in claim 1, characterized in that, Step S6 specifically includes: first, performing wavelet transform denoising on the ultrasonic echo detection data at each location, and then sending the detection data into the BGA solder joint laser ultrasonic echo classification model in step S1 for calculation and classification to determine whether there are defective solder joints among the 9 solder joints at that location; if there are defects, then performing ultrasonic echo detection on the 9 solder joint locations in this area to determine which solder joint has a defect.

Images