Semiconductor test system and method

Through real-time monitoring and automated docking technology of semiconductor testing systems, the problems of low testing efficiency and insufficient placement status monitoring in the existing technology are solved, and continuous testing and efficient factory delivery of semiconductors are achieved.

CN118914810AInactive Publication Date: 2024-11-08CHONGQING ACAD OF METROLOGY & QUALITY INST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410965130.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing semiconductor testing devices cannot achieve continuous testing during use, the testing efficiency is low, and the semiconductor placement status cannot be monitored in real time, resulting in incorrect angle or position, which leads to problems such as inability to test or damage to components.

Method used

Adopt a semiconductor testing system, including collecting semiconductor placement data modules, data preprocessing, establishing placement models, training and tuning, real-time monitoring feedback and execution modules, capturing images through cameras, using convolutional neural networks to detect semiconductor placement, and driving them through rotation. and flip components to achieve continuous testing and correct docking of semiconductors.

Benefits of technology

Real-time monitoring of the semiconductor placement status is realized, testing failure or component damage caused by misalignment of angle or position is avoided, testing efficiency and factory rate are improved, and semiconductors can be continuously tested after production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914810A_ABST
    Figure CN118914810A_ABST
Patent Text Reader

Abstract

The invention discloses a semiconductor testing system and method, and belongs to the technical field of semiconductor testing. Comprising the steps of collecting a semiconductor placement data module, collecting placement flatness or angle conditions of a semiconductor in a placement groove, performing cutting, zooming, denoising and brightness adjustment on the semiconductor placement data, ensuring the quality and consistency of the data, dividing a data set to train and optimize a placement model, and applying the trained placement model. The semiconductor placement condition is detected, a real-time image is captured, the image is transmitted to a computer for placement model analysis, rotation is carried out, a probe and semiconductor butt joint test is carried out, and the semiconductor placement condition is monitored. In the testing process, whether the semiconductor is placed in a reverse direction or the semiconductor is not pressed in place and is inclined is determined according to the state after the semiconductor is placed, and the situation that testing cannot be carried out or parts are damaged due to the fact that the angle or the position is incorrect is avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of semiconductor testing technology, and more specifically, to a semiconductor testing system and method. Background Art

[0002] When a semiconductor test device is testing a chip to be tested, it is necessary to connect the chip to be tested with an IC test carrier, usually using a probe to make contact.

[0003] The document with the prior art publication number CN115718245A provides a semiconductor testing system, which includes an IC test carrier, a measurement function board, a first telescopic cylinder, a mounting frame, a fixed tube, a mounting plate, a fine spring and a probe. The top of the IC test carrier is provided with a test slot, the measurement function board is mounted above the IC test carrier, the first telescopic cylinder is fixedly mounted above the IC test carrier, the mounting frame is fixedly connected to the output end of the first telescopic cylinder, the top of the fixed tube is fixedly connected to the mounting frame, the bottom of the fixed tube is provided with a receiving slot, the two ends of the fine spring are fixedly connected to the mounting plate and the fixed tube respectively, the fine spring is located inside the receiving slot, the mounting plate is slidably connected to the fixed tube, the probe is mounted below the mounting plate, and the probe is electrically connected to the measurement function board. The IC test carrier is used to install the chip to be tested, and the measurement function board is used to test the chip to be tested. The first telescopic cylinder can drive the mounting frame to move up and down, and the mounting plate can slide up and down relative to the first telescopic cylinder. The fine spring supports the mounting plate downward, and the probe is installed below the mounting plate.

[0004] Although the above-mentioned prior art solution can achieve the relevant beneficial effects through the structure of the prior art, it still has the following defects: the device cannot continuously test the semiconductor during use, and the test efficiency is low.

[0005] Regarding the above-mentioned related technologies, the inventors believe that in the process of testing semiconductors, it is necessary to conduct continuous testing on the semiconductors after production to improve the testing efficiency and factory yield. At the same time, during the testing process, the status of the semiconductors after placement needs to be monitored in real time to avoid the situation where components cannot be tested or are damaged due to incorrect angles or positions.

[0006] In view of this, we propose a semiconductor testing system. Summary of the invention

[0007] 1. Technical issues to be solved

[0008] The purpose of the present application is to provide a method for a semiconductor testing system, which solves the technical problem that the device in the above-mentioned background technology cannot continuously test the semiconductor during use and the testing efficiency is low, thereby achieving a technical effect.

[0009] 2. Technical solution

[0010] The technical solution of the present application provides a semiconductor testing system, comprising

[0011] A semiconductor placement data collection module collects the placement flatness or angle of the semiconductor inside the placement slot;

[0012] Semiconductor placement data preprocessing, including cropping, scaling, denoising and brightness adjustment of semiconductor placement data to ensure data quality and consistency;

[0013] Establishing a placement model, using semiconductor placement data to establish a placement model;

[0014] The placement model training and tuning module divides the data set to train and optimize the placement model;

[0015] The placement model is saved and applied for monitoring, and the trained placement model is applied to detect the semiconductor placement situation;

[0016] Real-time monitoring feedback module to capture real-time images and transmit graphics to the computer for placement model analysis;

[0017] The execution module is used to perform a rotation test of the probe and the semiconductor docking test.

[0018] By adopting the above technical solution, through monitoring the placement of semiconductors, the state of the semiconductors after placement can be used to determine whether the semiconductor is placed in the opposite direction or the semiconductor is not pressed into place and is tilted during the test, so as to avoid the situation where the angle or position is incorrect and the components cannot be tested or damaged.

[0019] As an optional solution of the technical solution of the present application document, the module for collecting semiconductor placement data is carried out through a camera, and the camera is fixedly arranged on one side of the inner wall of the test box, facing the placement slot directly above.

[0020] As an optional solution to the technical solution of the present application document, the semiconductor placement data is preprocessed, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe is marked, and the situation where the semiconductor is skewed is marked; then, the semiconductor placement data is cropped to remove irrelevant areas in the image, which will help reduce the size of the input data and reduce computing costs, and use the image processing library to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and denoised using median filtering, Gaussian filtering and wavelet transform. The brightness adjustment is used to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of specific tasks, and the data is rotated, flipped, translated, scaled, and random noise is added to allow the model to better generalize to different scenarios and changes.

[0021] As an optional solution of the technical solution of the present application document, the model training and tuning module is placed, and the architecture of the convolutional neural network model is selected to create the model, including the hierarchy, activation function, loss function, etc. The model is trained using a training set. In each training cycle, the model uses a back-propagation algorithm to adjust weights and parameters to minimize the loss function, and the performance of the model is evaluated using a validation set to check whether there is overfitting or underfitting. When tuning the model, regularization terms are added to reduce overfitting, the learning rate is adjusted to speed up or slow down the training process, the training data is expanded to improve the generalization ability of the model, and the hyperparameters of the model, such as the number of layers, the number of neurons, etc., are adjusted. After the model passes the verification, the performance of the model is evaluated using a test set.

[0022] As an optional solution of the technical solution of the present application document, the real-time monitoring feedback module displays the model output results on a display screen on a computing device and feeds them back to the execution module for execution.

[0023] As an optional solution of the technical solution of this application document, the execution module includes

[0024] A receiving block, wherein the receiving block is used to place a plurality of semiconductors;

[0025] A rotation drive, wherein the rotation drive drives the receiving block to rotate and switches different semiconductors for testing;

[0026] A probe, the probe being used for testing a semiconductor;

[0027] The flip assembly is used to flip the probe and dock it to test the semiconductor. After completion, it is flipped and reset to protect the probe and the semiconductor from interference and damage.

[0028] The present application also discloses a method for using the aforementioned semiconductor testing system, comprising the following steps:

[0029] S1. First, the placement flatness or angle of the semiconductor inside the placement groove is collected by a camera;

[0030] S2. At this time, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe is marked, and the skewed situation of the semiconductor is marked; then, the semiconductor placement data is cropped, and irrelevant areas in the image are removed, which will help reduce the size of the input data and reduce the computational cost. The image processing library is used to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and the median filter, Gaussian filter and wavelet transform are used for denoising. The brightness adjustment is used to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of specific tasks. The data is rotated, flipped, translated, scaled, and random noise is added to allow the model to better generalize to different scenarios and changes;

[0031] S3. At this point, select the architecture of the convolutional neural network to create the model, including the hierarchical structure, activation function, loss function, etc. Use the training set to train the model. In each training cycle, the model uses the backpropagation algorithm to adjust the weights and parameters to minimize the loss function. Use the validation set to evaluate the performance of the model and check whether there is overfitting or underfitting. When tuning the model, add regularization terms to reduce overfitting, adjust the learning rate to speed up or slow down the training process, expand the training data to improve the generalization ability of the model, and adjust the model's hyperparameters, such as the number of layers and the number of neurons. After the model passes the verification, use the test set to evaluate the performance of the model.

[0032] S4. At this point, the placement model is saved and applied for monitoring, and the placement model output is displayed on a display screen on a computing device and fed back to the execution module for execution;

[0033] S5. At this time, the semiconductor to be tested is placed into the placement groove of the receiving block and pressed to the bottom, and the external power supply is turned on;

[0034] S6. At this time, the output shaft of the starter motor is rotated to drive the rocker arm to rotate, and the driving wheel on the rocker arm is slidably connected with the driving groove on the rotating block, thereby driving the rotating block to rotate. At this time, the receiving block will be driven to rotate, thereby changing the position of the placement slot, and the placement slot directly below will dock with the probe.

[0035] S7. At this time, when the rocker arm rotates, it will drive the belt rod to rotate, and the belt rod will pull the connecting rod to move, and the connecting rod will drive one end of the flip rod B to swing. At this time, the flip rod A fixed to the flip rod B will swing at an angle under the limit of the limit rod, and the connecting rod will be driven to flip at an angle, and the adaptation groove at one end of the connecting rod will adapt to the upper limit of the adaptation block, and the adaptation block will be at the upper limit of the fixed rod, so that the connecting rod drives the detection block to flip, so that the probe can detect the semiconductor;

[0036] S8. At this time, the receiving block continues to rotate to switch different semiconductors.

[0037] 3. Beneficial effects

[0038] One or more technical solutions provided in the technical solution of this application have at least the following technical effects or advantages:

[0039] 1. This application monitors the placement of semiconductors and determines whether the semiconductor is placed in the wrong direction or tilted due to the incorrect angle or position, so as to avoid the inability to test or damage to components.

[0040] 2. The present application drives the receiving block to rotate through the setting of the rotation drive, and switches different semiconductors for testing, so that in the process of testing the semiconductors, the semiconductors after production can be continuously tested, thereby improving the test efficiency and factory delivery rate.

[0041] 3. This application uses the setting of the flip component to allow the probe to be flipped and docked to test the semiconductor. After completion, it is flipped and reset to protect the probe and semiconductor from interference and damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of the structure of a semiconductor testing system disclosed in a preferred embodiment of the present application;

[0043] Figure 2 A schematic diagram of the overall structure of a semiconductor testing device disclosed in a preferred embodiment of the present application;

[0044] Figure 3 A cross-sectional view of the overall structure of a semiconductor testing device disclosed in a preferred embodiment of the present application;

[0045] Figure 4 A disassembled diagram of the flip assembly structure of a semiconductor testing device disclosed in a preferred embodiment of the present application;

[0046] Figure 5 A schematic diagram of a rotation driving structure of a semiconductor testing device disclosed in a preferred embodiment of the present application;

[0047] Explanation of the numbers in the figure: 1. Test box; 2. Inspection plate; 3. Receiving block; 31. Rocker arm; 33. Driving wheel; 34. Rotating block; 35. Driving slot; 301. Fixed rod; 302. Adaptation block; 303. Connecting rod; 304. Adaptation slot; 305. Flip rod A; 306. Flip rod B; 307. Limit rod; 308. Connecting rod; 309. Belt rod; 4. Placement slot; 5. Detection block; 6. Probe. DETAILED DESCRIPTION

[0048] The present application is further described in detail below in conjunction with the accompanying drawings.

[0049] Reference Figure 1 , the embodiment of the present application provides a semiconductor testing system, a module for collecting semiconductor placement data, and collecting the placement flatness or angle of the semiconductor inside the placement slot 4;

[0050] Semiconductor placement data preprocessing, including cropping, scaling, denoising and brightness adjustment of semiconductor placement data to ensure data quality and consistency;

[0051] Establishing a placement model, using semiconductor placement data to establish a placement model;

[0052] The placement model training and tuning module divides the data set to train and optimize the placement model;

[0053] The placement model is saved and applied for monitoring, and the trained placement model is applied to detect the semiconductor placement situation;

[0054] Real-time monitoring feedback module to capture real-time images and transmit graphics to the computer for placement model analysis;

[0055] The execution module is rotated to conduct a docking test between the probe 6 and the semiconductor.

[0056] By monitoring the placement of semiconductors, the state of the semiconductors after placement can be used to determine whether the semiconductor is placed in the wrong direction or is not pressed into place or is tilted during the test, so as to avoid the situation where the angle or position is incorrect and the components cannot be tested or are damaged.

[0057] Reference Figure 1 and Figure 2 The embodiment of the present application provides a semiconductor testing system, which collects semiconductor placement data module through a camera, and the camera is fixedly arranged on one side of the inner wall of the test box 1, facing the placement slot 4 directly above.

[0058] Semiconductor placement data preprocessing, marking the semiconductor placement data, marking the wrong direction in which the semiconductor cannot dock with the probe 6, and marking the skewed situation of the semiconductor; then, cropping the semiconductor placement data and removing irrelevant areas in the image will help reduce the size of the input data and reduce the computing cost. Use image processing libraries such as OpenCV to crop images; adjust the images to the same size to ensure the consistency of the input data, use median filtering, Gaussian filtering and wavelet transform for denoising, and adjust the brightness to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of specific tasks. Rotate, flip, translate, scale and add random noise to the data to allow the model to better generalize to different scenarios and changes.

[0059] Place the model training and tuning module, select the convolutional neural network CNN to create the model architecture, including the hierarchy, activation function, loss function, etc. Use the training set to train the model. In each training cycle epoch, the model uses the backpropagation algorithm to adjust the weights and parameters to minimize the loss function. Use the validation set to evaluate the performance of the model and check whether there is overfitting or underfitting. When tuning the model, add regularization terms to reduce overfitting, adjust the learning rate to speed up or slow down the training process, expand the training data to improve the generalization ability of the model, and adjust the model's hyperparameters, such as the number of layers and the number of neurons. After the model passes the verification, use the test set to evaluate the performance of the model.

[0060] The real-time monitoring feedback module displays the model output results on a display screen on a computing device and feeds them back to the execution module for execution.

[0061] The execution module includes

[0062] A receiving block 3, the receiving block 3 is used to place a plurality of semiconductors;

[0063] Rotation drive, the rotation drive drives the receiving block 3 to rotate, and switches different semiconductors for testing;

[0064] Probe 6, probe 6 is used for testing semiconductors;

[0065] The flip assembly is used to flip the probe 6 and dock it to test the semiconductor. After completion, it is flipped and reset to protect the probe 6 and the semiconductor from interference and damage.

[0066] Reference Figure 3 , the present application embodiment provides a semiconductor testing system, comprising

[0067] Test box 1;

[0068] An inspection plate 2, the inspection plate 2 is fixed to one side of the test box 1 by screws;

[0069] The receiving block 3 is rotatably arranged inside the test box 1 by means of rotation driving;

[0070] A placement slot 4, which is provided on the outer wall of the receiving block 3 and is provided in plurality;

[0071] The detection block 5 is rotatably arranged in the test box 1 through the flip assembly, and is located at the lower side of the receiving block 3, and is flipped for detection;

[0072] The probe 6 is arranged on the outer wall of the detection block 5 .

[0073] The placement grooves 4 are arranged in a ring shape with multiple equal intervals.

[0074] And two groups are provided.

[0075] Reference Figure 3 and Figure 5 , the embodiment of the present application provides a semiconductor testing system, the rotation drive includes a rocker arm 31 rotatably arranged on the inner wall of the test box 1, and a driving wheel 33 is fixedly arranged on one side of the rocker arm 31;

[0076] The rocker arm 31 is coaxially fixedly connected to the motor outside the test box 1;

[0077] A rotating block 34 is fixedly provided on one side of the receiving block 3 . The outer wall of the rotating block 34 is provided with a plurality of driving grooves 35 in an annular shape and at equal intervals. The driving grooves 35 are slidably connected to the driving wheel 33 .

[0078] The number of the driving grooves 35 is half the number of the placement grooves 4 .

[0079] By starting the motor output shaft to rotate, the rocker arm 31 is driven to rotate, and the driving wheel 33 on the rocker arm 31 is slidably connected to the driving groove 35 on the rotating block 34, thereby driving the rotating block 34 to rotate. At this time, the receiving block 3 will be driven to rotate, thereby changing the position of the placement groove 4. The placement groove 4 directly below will dock with the probe 6. By setting the rotation drive, the receiving block is driven to rotate and different semiconductors are switched for testing. In the process of testing the semiconductor, the semiconductors after production can be continuously tested, thereby improving the test efficiency and factory delivery rate.

[0080] Reference Figure 3 and Figure 4 , the embodiment of the present application provides a semiconductor testing system, the flip assembly includes two fixing rods 301 symmetrically fixedly arranged on the inner wall of the test box 1, and an adaption block 302 is arranged at one end of the fixing rod 301 through a pin shaft rotation;

[0081] A connecting rod 303 is provided on one side of the fixing rod 301, an adapting groove 304 is provided in the middle of the connecting rod 303, and the adapting block 302 is provided in an I-shaped structure, and the adapting block 302 is slidably connected with the adapting groove 304;

[0082] One end of the connecting rod 303 is connected and fixed to the inspection plate 2;

[0083] A flip rod A305 is rotatably provided on the outer side of the connecting rod 303 through a pin shaft, a flip rod B306 is fixedly provided on one end of the flip rod A305 through a fixed shaft, a limit rod 307 is rotatably provided on the fixed shaft connecting the flip rod B306 and the flip rod A305, and the limit rod 307 is fixedly provided on the inner wall of the test box 1;

[0084] A connecting rod 308 is rotatably provided at one end of the flip rod B306 via a pin shaft, and a belt rod 309 is rotatably provided at one end of the connecting rod 308 via a pin shaft. The belt rod 309 is rotatably connected to the outer wall of the rocker arm 31 via a pin shaft.

[0085] When the rocker arm 31 rotates, it will drive the belt rod 309 to rotate. At this time, the belt rod 309 will pull the connecting rod 308 to move. At this time, the connecting rod 308 will drive one end of the flip rod B306 to swing. At this time, the flip rod A305 fixed to the flip rod B306 will swing at an angle under the limit of the limit rod 307. At this time, it will drive the connecting rod 303 to flip at an angle, and the adaptation groove 304 at one end of the connecting rod 303 will adapt to the upper limit of the adaptation block 302, and the adaptation block 302 will be upper limit on the fixed rod 301, so that the connecting rod 303 drives the detection block 5 to flip, so that the probe 6 can detect the semiconductor. Through the setting of the flip component, the probe can be flipped and docked to test the semiconductor. After completion, it can be flipped and reset to protect the probe and the semiconductor from interference and damage.

[0086] The present invention provides a semiconductor testing system, comprising the following steps:

[0087] S1. First, the placement flatness or angle of the semiconductor inside the placement groove 4 is collected by a camera;

[0088] S2. At this time, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe 6 is marked, and the situation where the semiconductor is skewed is marked; then, the semiconductor placement data is cropped, and irrelevant areas in the image are removed, which will help reduce the size of the input data and reduce the computing cost. Image processing libraries such as OpenCV are used to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and median filtering, Gaussian filtering and wavelet transform are used for denoising. Brightness adjustment is used to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of a specific task. The data is rotated, flipped, translated, scaled, and random noise is added to allow the model to better generalize to different scenarios and changes;

[0089] S3. At this point, select the convolutional neural network CNN to create the model architecture, including the hierarchical structure, activation function, loss function, etc. Use the training set to train the model. In each training cycle epoch, the model uses the backpropagation algorithm to adjust the weights and parameters to minimize the loss function. Use the validation set to evaluate the performance of the model and check whether there is overfitting or underfitting. When tuning the model, add regularization terms to reduce overfitting, adjust the learning rate to speed up or slow down the training process, expand the training data to improve the generalization ability of the model, and adjust the model's hyperparameters, such as the number of layers and the number of neurons. After the model passes the verification, use the test set to evaluate the performance of the model.

[0090] S4. At this point, the placement model is saved and applied for monitoring, and the placement model output is displayed on a display screen on a computing device and fed back to the execution module for execution;

[0091] S5. At this time, the semiconductor to be tested is placed in the placement groove 4 of the receiving block 3 and pressed to the bottom, and the external power supply is turned on;

[0092] S6. At this time, the output shaft of the starter motor is rotated to drive the rocker arm 31 to rotate, and the driving wheel 33 on the rocker arm 31 is slidably connected with the driving groove 35 on the rotating block 34, thereby driving the rotating block 34 to rotate an angle, which will drive the receiving block 3 to rotate an angle, thereby changing the position of the placement groove 4, and the placement groove 4 directly below will dock with the probe 6.

[0093] S7. At this time, when the rocker arm 31 rotates, it will drive the belt rod 309 to rotate, and the belt rod 309 will pull the connecting rod 308 to move. At this time, the connecting rod 308 will drive one end of the flip rod B306 to swing. At this time, the flip rod A305 fixed to the flip rod B306 will swing at an angle under the limit of the limit rod 307, and will drive the connecting rod 303 to flip at an angle, and the adaptation groove 304 at one end of the connecting rod 303 will adapt to the upper limit of the adaptation block 302, and the adaptation block 302 will be upper limit of the fixed rod 301, so that the connecting rod 303 drives the detection block 5 to flip, so that the probe 6 can detect the semiconductor;

[0094] S8. At this time, the receiving block 3 continues to rotate to switch different semiconductors.

[0095] When the semiconductor test system is needed, first, the camera is used to collect the placement flatness or angle of the semiconductor inside the placement slot 4. At this time, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe 6 is marked, and the skewed situation of the semiconductor is marked; then, the semiconductor placement data is cropped to remove irrelevant areas in the image, which will help reduce the size of the input data and reduce the computing cost. Image processing libraries such as OpenCV are used to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and median filtering, Gaussian filtering and wavelet transform are used for denoising. The brightness adjustment is used to standardize the brightness and contrast of the image. Converting images to different color spaces may help improve the performance of specific tasks. Perform operations such as rotating, flipping, translating, scaling, and adding random noise to the data to allow the model to better generalize to different scenarios and changes. Select a convolutional neural network (CNN) to create the model's architecture, including hierarchy, activation function, loss function, etc. Use the training set to train the model. In each training cycle epoch, the model uses the backpropagation algorithm to adjust weights and parameters to minimize the loss function. Use the validation set to evaluate the performance of the model and check for overfitting or underfitting. When tuning the model, add regularization terms to reduce overfitting, adjust the learning rate to speed up or slow down the training process, and expand the training data to improve the model. The generalization ability of the model is adjusted, such as the number of layers, the number of neurons, etc. After the model is verified, the test set is used to evaluate the performance of the model. At this time, the model is placed and saved and applied for monitoring. The output result of the placement model is displayed on a display screen on a computing device, and fed back to the execution module for execution. At this time, the semiconductor to be tested is placed in the placement slot 4 of the receiving block 3 and pressed to the bottom. At this time, the rocker arm 31 is driven to rotate by starting the motor output shaft, and the driving wheel 33 on the rocker arm 31 is slidably connected with the driving slot 35 on the rotating block 34, thereby driving the rotating block 34 to rotate at an angle, which will drive the receiving block 3 to rotate at an angle, thereby changing the position of the placement slot 4, and placing it directly below. The slot 4 will dock with the probe 6. At this time, when the rocker arm 31 rotates, it will drive the belt rod 309 to rotate. At this time, the belt rod 309 will pull the connecting rod 308 to move. At this time, the connecting rod 308 will drive one end of the flip rod B306 to swing. At this time, the flip rod A305 fixed to the flip rod B306 will swing at an angle under the limit of the limit rod 307. At this time, it will drive the connecting rod 303 to flip at an angle, and the adaptation slot 304 at one end of the connecting rod 303 will adapt to the upper limit of the adaptation block 302, and the adaptation block 302 will be upper limit on the fixed rod 301, so that the connecting rod 303 drives the detection block 5 to flip, so that the probe 6 can detect the semiconductor. At this time, the receiving block 3 continues to rotate to switch different semiconductors.

Claims

1. A semiconductor testing system, Features: Include: A semiconductor placement data collection module collects the placement flatness or angle of the semiconductor inside the placement slot; Semiconductor placement data preprocessing, including cropping, scaling, denoising and brightness adjustment of semiconductor placement data to ensure data quality and consistency; Establishing a placement model, using semiconductor placement data to establish a placement model; The placement model training and tuning module divides the data set to train and optimize the placement model; The placement model is saved and applied for monitoring, and the trained placement model is applied to detect the semiconductor placement situation; Real-time monitoring feedback module to capture real-time images and transmit graphics to the computer for placement model analysis; Execute the module to perform a rotation test of the probe and the semiconductor docking test.

2. The semiconductor test system according to claim 1, Features: The module for collecting semiconductor placement data is performed by a camera, and the camera is fixedly arranged on one side of the inner wall of the test box, facing the placement slot directly above.

3. The semiconductor test system according to claim 1, Features: The semiconductor placement data is preprocessed, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe is marked, and the situation where the semiconductor is skewed is marked; then, the semiconductor placement data is cropped, and irrelevant areas in the image are removed, which will help reduce the size of the input data and reduce the computing cost. The image processing library is used to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and denoised using median filtering, Gaussian filtering and wavelet transform. The brightness adjustment is used to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of a specific task. The data is rotated, flipped, translated, scaled, and random noise is added to allow the model to better generalize to different scenarios and changes.

4. The semiconductor test system according to claim 1, Features: The model training and tuning module is placed, and the architecture of the convolutional neural network model is selected to create the model, including the hierarchical structure, activation function, loss function, etc. The model is trained using the training set. In each training cycle, the model uses the back propagation algorithm to adjust weights and parameters to minimize the loss function. The performance of the model is evaluated using the validation set to check whether there is overfitting or underfitting. When tuning the model, regularization terms are added to reduce overfitting, the learning rate is adjusted to speed up or slow down the training process, the training data is expanded to improve the generalization ability of the model, and the hyperparameters of the model, such as the number of layers and the number of neurons, are adjusted. After the model passes the verification, the performance of the model is evaluated using the test set.

5. The semiconductor test system according to claim 1, Features: The real-time monitoring feedback module displays the model output results on a display screen on a computing device and feeds them back to the execution module for execution.

6. The semiconductor test system according to claim 1, Features: The execution module includes A receiving block, wherein the receiving block is used to place a plurality of semiconductors; A rotation drive, wherein the rotation drive drives the receiving block to rotate and switches different semiconductors for testing; A probe, the probe being used for testing a semiconductor; The flip assembly is used to flip the probe and dock it to test the semiconductor. After completion, it is flipped and reset to protect the probe and the semiconductor from interference and damage.

7. A method for using a semiconductor testing system according to any one of claims 1 to 6, It is characterized in that The following steps are involved: S1. First, the placement flatness or angle of the semiconductor inside the placement groove is collected by a camera; S2. At this time, the semiconductor placement data is marked, the wrong direction in which the semiconductor cannot dock with the probe is marked, and the skewed situation of the semiconductor is marked; then, the semiconductor placement data is cropped, and irrelevant areas in the image are removed, which will help reduce the size of the input data and reduce the computational cost. The image processing library is used to crop the image; the image is adjusted to the same size to ensure the consistency of the input data, and the median filter, Gaussian filter and wavelet transform are used for denoising. The brightness adjustment is used to standardize the brightness and contrast of the image. Converting the image to a different color space may help improve the performance of specific tasks. The data is rotated, flipped, translated, scaled, and random noise is added to allow the model to better generalize to different scenarios and changes; S3. At this point, select the architecture of the convolutional neural network to create the model, including the hierarchical structure, activation function, loss function, etc. Use the training set to train the model. In each training cycle, the model uses the backpropagation algorithm to adjust the weights and parameters to minimize the loss function. Use the validation set to evaluate the performance of the model and check whether there is overfitting or underfitting. When tuning the model, add regularization terms to reduce overfitting, adjust the learning rate to speed up or slow down the training process, expand the training data to improve the generalization ability of the model, and adjust the model's hyperparameters, such as the number of layers and the number of neurons. After the model passes the verification, use the test set to evaluate the performance of the model. S4. At this point, the placement model is saved and applied for monitoring, and the placement model output is displayed on a display screen on a computing device and fed back to the execution module for execution; S5. At this time, the semiconductor to be tested is placed into the placement groove of the receiving block and pressed to the bottom, and the external power supply is turned on; S6. At this time, the output shaft of the starter motor is rotated to drive the rocker arm to rotate, and the driving wheel on the rocker arm is slidably connected with the driving groove on the rotating block, thereby driving the rotating block to rotate. At this time, the receiving block will be driven to rotate, thereby changing the position of the placement slot, and the placement slot directly below will dock with the probe. S7. At this time, when the rocker arm rotates, it will drive the belt rod to rotate, and the belt rod will pull the connecting rod to move, and the connecting rod will drive one end of the flip rod B to swing. At this time, the flip rod A fixed to the flip rod B will swing at an angle under the limit of the limit rod, and the connecting rod will be driven to flip at an angle, and the adaptation groove at one end of the connecting rod will adapt to the upper limit of the adaptation block, and the adaptation block will be at the upper limit of the fixed rod, so that the connecting rod drives the detection block to flip, so that the probe can detect the semiconductor; S8. At this time, the receiving block continues to rotate to switch different semiconductors.