Chip substrate TGV defect detection positioning device
By using a four-sensor collaborative positioning system and a multi-focal length re-inspection function, the problems of large positioning deviation and low re-inspection efficiency in TGV inspection of chip substrates are solved, achieving high-precision and high-speed defect detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- GUANGDONG WEIHUA INTELLIGENT TECHNOLOGY CO LTD
- Filing Date
- 2025-04-17
- Publication Date
- 2026-04-21
AI Technical Summary
Existing mechanical positioning technology suffers from insufficient motion stability, low multi-axis coordination accuracy, and poor compatibility with optical component installation in TGV inspection of chip substrates, resulting in large positioning deviations, low re-inspection efficiency, and high false detection rate.
A four-sensor collaborative positioning system is adopted, which combines multi-focal length re-inspection function and optical path collaborative structure. Through the optimized layout of the four-sensor collaborative positioning system and optical components, seamless full-board scanning inspection is achieved. The Halcon algorithm and YOLOv8 deep learning model are used for defect classification, and the motion control system ensures positioning accuracy and efficiency.
It significantly improves the accuracy and efficiency of TGV inspection of chip substrates, with a defect classification accuracy of 98%, a false negative rate of 0.1%, and a 3-fold increase in inspection efficiency.
Smart Images
Figure CN224152309U_ABST
Abstract
Description
Technical Field
[0001] This utility model relates to the field of semiconductor packaging inspection equipment technology, specifically to a chip substrate TGV defect detection and positioning device. Background Technology
[0002] In TGV chip carrier manufacturing, through-hole defect detection relies on high-precision positioning devices to ensure accuracy. Existing mechanical positioning technologies suffer from the following problems: Insufficient motion stability: Traditional platforms, due to poor material rigidity or limitations in the driving method, are prone to swaying (>10 arcseconds), causing displacement deviations in the glass substrate during movement; Low multi-axis coordination accuracy: Multi-axis mechanical structures (such as a hybrid drive of lead screw modules and linear motors) struggle to achieve ±1μm repeatability due to asynchronous control; Poor adaptability of optical component installation: The mechanical layout of optical components such as sensors and light sources cannot simultaneously meet the requirements of full-board scanning and multi-aperture synchronous detection.
[0003] To address the aforementioned problems, there is an urgent need for a dedicated mechanical positioning device for TGV inspection of chip substrates, capable of resolving the issue of inability to achieve high-precision positioning. Utility Model Content
[0004] The current optical components suffer from poor adaptability: the mechanical layout of optical components such as sensors and light sources cannot simultaneously meet the requirements of full-board scanning and multi-aperture synchronous detection. This application provides a chip substrate TGV defect detection and positioning device to solve the above problems.
[0005] To achieve the above objectives, this utility model provides the following technical solution:
[0006] This application discloses a TGV defect detection and positioning device for a chip substrate, characterized by comprising: a four-sensor collaborative positioning system, the four-sensor collaborative positioning system comprising: a second sensor: mounted on a first vertical axis module, configured with a lens and a coaxial parallel light source, used to identify reference holes on the glass substrate and establish a mapping relationship between the platform coordinate system and the CAM coordinate system through the Halcon algorithm; and further comprising a first line scan sensor, a second line scan sensor, and a third line scan sensor, respectively mounted on a second vertical axis module, a third vertical axis module, and a fourth vertical axis module, wherein... The first line scan sensor is aligned with the minimum aperture of the through hole, while the second and third line scan sensors are aligned with the upper and lower apertures of the through hole, respectively. The first sensor is mounted on the fifth vertical axis module and is equipped with an adjustable focus lens and a coaxial light source for multi-focal-length secondary imaging of initial defects. The optical path coordination structure includes a reflector, which works with the coaxial light source to achieve shadow-free imaging of the inner wall of the through hole. The motion control system is driven synchronously by a multi-drive motion controller along the X1, X2, Y, Z1, Z2, and Z3 axes to control the glass substrate to move along an arc-shaped trajectory and provides real-time position data feedback through a grating ruler.
[0007] Preferably, the layout of the first line scan sensor, the second line scan sensor, and the third line scan sensor covers the entire glass substrate; the first line scan sensor fits the through-hole profile based on the Canny edge detection algorithm.
[0008] Preferably, the first sensor integrates a YOLOv8 deep learning model and classifies defects by combining Halcon edge detection results; during the re-inspection stage, a servo motor is used to locate the defect coordinates, and the angle of the reflector is dynamically adjusted to eliminate shadow interference.
[0009] Preferably, the data from the second sensor, the first line scan sensor, the second line scan sensor, and the third line scan sensor are fused in real time by an industrial control computer to generate a full-board through-hole size heat map and defect distribution map; the detection results are synchronized to the MES system, and a final report containing defect location, type, and repair coordinates is output.
[0010] Preferably, the motion control system adopts a multi-drive motion controller to synchronously control the co-movement displacement of the X1 axis and the X2 axis, ensuring that the gantry dual-drive structure has no sway in the bow-shaped trajectory; the grating ruler feedback signal is directly embedded in the controller closed loop to compensate for position deviation in real time.
[0011] Preferably, the reflector is fixed on both sides of the stage, and the light path is perpendicular to the inner wall of the through hole; the installation position of the reflector is adjusted by a dynamic calibration module to ensure that there is no overlap or blind spot in the field of view of the three-line scanning sensor.
[0012] Preferably, the second sensor uses a coordinate mapping algorithm based on a reference hole to convert the platform coordinate system into a coordinate system with the second sensor as the reference, and the mapped coordinates correspond one-to-one with the CAM graphic coordinates.
[0013] Technical effects:
[0014] This invention significantly improves the detection performance of TGV on chip substrates through four-sensor collaborative positioning and multi-focal length re-inspection: based on the collaborative work of the second sensor and the three-line scanning sensor, and by optimizing the optical path through the layout of the reflector and coaxial light source, seamless scanning of the glass substrate is supported; in the re-inspection stage, the multi-focal length sensor combined with Halcon edge detection and YOLOv8 algorithm realizes defect classification, shortening the re-inspection time to 30% of the initial inspection, improving the overall detection efficiency by 3 times, the defect classification accuracy rate ≥98%, and the false negative rate ≤0.1%, effectively solving the core problems of large positioning deviation, low re-inspection efficiency and high false positive rate in traditional technology. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a schematic diagram of the TGV defect detection and positioning device for the chip substrate in this application.
[0017] Figure 2 This is a schematic diagram of the TGV defect detection and positioning device system for the chip substrate of this application;
[0018] Figure 3 This is a diagram of the motion axis control scheme for this application.
[0019] In the picture:
[0020] 1. Second sensor; 2. Parallel light source; 3. First linear scan sensor; 4. Second linear scan sensor; 5. Third linear scan sensor; 6. First sensor; 7. Four-sensor cooperative positioning system; 8. Motion control system. Detailed Implementation
[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.
[0022] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.
[0023] Please see Figures 1-3 In existing TGV defect detection for chip substrates, the visual positioning and re-inspection functions have the following technical problems:
[0024] Insufficient positioning accuracy: Traditional single-sensor positioning relies on manual adjustment of the reference hole, which is prone to human error and cannot meet the precise positioning requirements of submicron level through holes;
[0025] Low re-inspection efficiency: Inspection requires offline manual re-testing or multiple scans by a single sensor, which cannot achieve simultaneous acquisition of multiple focal lengths, resulting in a long inspection cycle;
[0026] Poor optical system adaptability: The multi-sensor layout is not optimized for optical path coordination, resulting in overlapping fields of view or blind spots, which affects the full-board scanning efficiency.
[0027] To address the above issues, this utility model provides a TGV glass substrate positioning and re-inspection device based on four-sensor collaboration. The specific technical solution is as follows: A four-sensor collaborative positioning system includes a second sensor and three line scan sensors; wherein the second sensor is used to identify reference holes (P1-P4) on the glass substrate, and is mapped to the platform coordinate system through image coordinate system; the line scan sensor among the three line scan sensors is equipped with a lens and a coaxial light source, and is dedicated to the real-time acquisition of the minimum aperture of the through hole; while the second line scan sensor and the third line scan sensor are respectively aligned with the upper and lower apertures of the through hole, and the optical path is adjusted by a reflector to ensure that there are no blind spots in the full board scanning. A multi-focal-length re-inspection function design includes a first sensor integrated into the Z-axis module, supporting automatic switching between multiple focal lengths for secondary high-precision imaging of suspected defective through-holes. The entire re-inspection process involves: in the initial inspection stage, the first, second, and third line scan sensors complete a full-board "bow-shaped" trajectory scan, with an AI algorithm (Halcon+YOLOv8) marking the defect location in real time; in the re-inspection stage, the first sensor acquires multi-focal-length images of the marked locations, and deep learning methods are used to classify defect types (such as foreign objects, black holes, aperture deviations, etc.). An optical path and motion co-optimization method is also included: 1. Mirror layout: a mirror projects the coaxial parallel light source path vertically onto the inner wall of the through-hole, eliminating shadow interference; 2. Motion control: a multi-drive motion controller synchronously drives six axes (X1 / X2 / Y / Z1 / Z2 / Z3) to ensure that the four sensors are strictly aligned with the scanning path within the "bow-shaped" trajectory.
[0028] The through-hole contour is fitted based on the Canny edge detection algorithm.
[0029] The first sensor integrates the YOLOv8 deep learning model and combines Halcon edge detection results to classify defects; during the re-inspection stage, the servo motor is used to locate the defect coordinates and the angle of the reflector is dynamically adjusted to eliminate shadow interference.
[0030] The data from the second sensor, the first line scan sensor, the second line scan sensor, and the third line scan sensor are fused in real time by an industrial control computer to generate a heat map of the through-hole dimensions and a defect distribution map of the entire board; the detection results are synchronized to the MES system, and a final report containing the defect location, type, and repair coordinates is output.
[0031] The motion control system adopts a multi-drive motion controller to synchronously control the co-movement displacement of the X1 axis and the X2 axis, ensuring that the gantry dual-drive structure has no sway in the bow-shaped trajectory; the grating ruler feedback signal is directly embedded in the controller closed loop to compensate for position deviation in real time.
[0032] The reflectors are fixed on both sides of the stage, and the light path is perpendicular to the inner wall of the through hole. The installation position of the reflectors is adjusted by the dynamic calibration module to ensure that there is no overlap or blind spot in the field of view of the three-line scanning sensor.
[0033] The positioning machine uses a coordinate mapping algorithm based on a reference hole to convert the platform coordinate system into a coordinate system with the positioning machine as the reference. The mapped coordinates correspond one-to-one with the coordinates of the CAM graphics.
[0034] The four-sensor collaborative positioning function steps are as follows: Install and assemble the second sensor; fix the second sensor to the Z1 axis module, drive vertical movement through a servo motor, project a coaxial light source vertically onto the surface of the glass substrate, adjust the light path through a reflector to eliminate shadow interference; perform reference hole identification and coordinate mapping. After the glass substrate is loaded, the motion platform returns to zero, the Y-axis drives the stage to move to the initial position, and the second sensor moves along the X1 / X2 axis to the CAM preset reference hole position (P1-P4). According to the relationship between the CAM coordinate system and the platform coordinate system, the platform coordinates read above are converted into reference coordinates with the second sensor as the reference. The converted coordinates realize the one-to-one mapping of the coordinates of the holes on the glass substrate and the coordinates of the holes in the CAM graphic.
[0035] Step 2: Perform synchronous calibration of the line scan sensors. Install the first line scan sensor on the Z2 axis, align it with the minimum aperture of the through hole, and adjust the optical path via a reflector to ensure perpendicular incidence on the inner wall. Install the second and third line scan sensors on the Z1 and Z3 axes respectively to detect the upper and lower apertures of the through hole. Dynamically calibrate the reflector angle to avoid overlapping fields of view or blind spots. Next, perform motion trajectory alignment. Drive the X1 / X2 / Y axes through the multi-drive motion controller to ensure that the fields of view of the three line scan sensors are strictly aligned with the arc-shaped scanning path, and the grating ruler provides real-time feedback of position data.
[0036] Full-board scanning and initial defect inspection implementation steps: Step 1: Perform "bow-shaped" trajectory motion and image acquisition. On the Y-axis, on the X-axis, and the X1 / X2 axes, the images are laterally displaced according to the "bow-shaped" trajectory. The three-line scanning sensors synchronously acquire images according to the encoder signal to ensure that the images are continuous and without discontinuities. Step 2: Data fusion and initial inspection report generation. The aperture data of the first line scanning sensor-3 is associated with the positioning coordinates to generate a full-board through-hole size heat map. The physical coordinates of the defects are stored in the database through platform coordinate system mapping for re-inspection.
[0037] The entire device uses three sensors to collect data simultaneously. The following situations will occur in the initial and final states:
[0038] The first line scan sensor reaches the glass substrate and takes a picture, while the second and third line scan sensors are located outside the glass substrate and do not take a picture.
[0039] The first and second line scan sensors reach the glass substrate to capture images, while the third line scan sensor is located outside the glass substrate and does not capture images.
[0040] The three line scan sensors have entered synchronous acquisition mode;
[0041] The first line scan sensor completes image acquisition and leaves the glass substrate position, while the second and third line scan sensors are still acquiring images and are still at the glass substrate position.
[0042] The first and second line scan sensors have completed image acquisition and moved away from the glass substrate position, while the third line scan sensor is still acquiring images and is still at the glass substrate position.
[0043] All three line scan sensors have completed image acquisition;
[0044] Multi-focal-length re-inspection and defect classification implementation steps 1: First sensor intervention and multi-focal-length acquisition. The first sensor is installed on the Z4 axis module, equipped with a lens and coaxial light source. The servo motor drives the first sensor to quickly locate the defect position marked in the initial inspection. The multi-focal-length image acquisition process is as follows: Surface defect detection: First depth of field, focusing on the surface of the through hole to detect foreign object residue; Inner wall defect detection: Second depth of field, adjusting the optical path through a reflector to acquire inner wall residue or black holes; Bottom defect detection: Third depth of field, focusing on the bottom of the through hole to verify the hole closure.
[0045] Unless otherwise specified, the equipment components involved in the above embodiments are all conventional equipment components, and the connection methods and control methods involved are all conventional connection methods and control methods unless otherwise specified.
[0046] The present invention has been described in detail above with reference to the embodiments. However, those skilled in the art will understand that, without departing from the spirit of the present invention, various specific parameters in the above embodiments can be changed to form multiple specific embodiments, all of which are common variations of the present invention, and will not be described in detail here.
Claims
1. A chip substrate TGV defect detection positioning device, characterized in that, include: The four-sensor collaborative positioning system includes: a second sensor mounted on the first vertical axis module, equipped with a lens and a coaxial parallel light source, used to identify reference holes on the glass substrate and establish a mapping relationship between the platform coordinate system and the CAM coordinate system through the Halcon algorithm; it also includes a first line scan sensor, a second line scan sensor, and a third line scan sensor, respectively mounted on the second, third, and fourth vertical axis modules, wherein the first line scan sensor is aligned with the minimum aperture of the through hole, and the second and third line scan sensors are aligned with the upper and lower apertures of the through hole, respectively; a first sensor mounted on the fifth vertical axis module, equipped with an adjustable focus lens and a coaxial light source, used for multi-focal-length secondary imaging of initial inspection defects; an optical path collaborative structure, including a reflector, which works with the coaxial light source to achieve shadow-free imaging of the inner wall of the through hole; and a motion control system: a multi-drive motion controller synchronously drives the X1, X2, Y, Z1, Z2, and Z3 axes to control the glass substrate to move along an arc-shaped trajectory, and provides real-time position data feedback through a grating ruler.
2. The chip substrate TGV defect detection and positioning device according to claim 1, characterized in that, The layout of the first, second, and third line scan sensors covers the entire glass substrate; the first line scan sensor fits the through-hole profile based on the Canny edge detection algorithm.
3. The chip substrate TGV defect detection and positioning device according to claim 1, wherein, The first sensor integrates the YOLOv8 deep learning model and combines Halcon edge detection results to classify defects; during the re-inspection stage, the servo motor is used to locate the defect coordinates and the angle of the reflector is dynamically adjusted to eliminate shadow interference.
4. The chip substrate TGV defect detection and positioning device of claim 1, wherein, The data from the second sensor, the first line scan sensor, the second line scan sensor, and the third line scan sensor are fused in real time by an industrial control computer to generate a heat map of the through-hole dimensions and a defect distribution map of the entire board; the detection results are synchronized to the MES system, and a final report containing the defect location, type, and repair coordinates is output.
5. The chip substrate TGV defect detection and positioning device according to claim 1, wherein, The motion control system adopts a multi-drive motion controller to synchronously control the co-movement displacement of the X1 axis and the X2 axis, ensuring that the gantry dual-drive structure has no sway in the bow-shaped trajectory; the grating ruler feedback signal is directly embedded in the controller closed loop to compensate for position deviation in real time.
6. The chip substrate TGV defect detection and positioning device according to claim 1, wherein, The reflectors are fixed on both sides of the stage, and the light path is perpendicular to the inner wall of the through hole. The installation position of the reflectors is adjusted by the dynamic calibration module to ensure that there is no overlap or blind spot in the field of view of the three-line scanning sensor.
7. The chip substrate TGV defect detection and positioning device according to claim 1, wherein, The second sensor uses a coordinate mapping algorithm based on a reference hole to convert the platform coordinate system into a coordinate system with the second sensor as the reference. The mapped coordinates correspond one-to-one with the coordinates of the CAM graphics.