Tape-and-Reel Die Image Review for Defect Rebinning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tape and reel machines fail to accurately detect die defects such as missing solder points and cracks in integrated circuits, leading to the packaging and shipment of damaged dies, which can cause further issues in downstream processes.
Innovation Solution
Implementing machine learning systems trained on specific defect datasets to analyze images of dies in real-time, using Hough transformation, position alignment, and clustering algorithms to identify missing solder points, and employing neural networks to detect cracks, generating alerts for defective dies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional inspection methods are used in tape and reel machines, then the device complexity remains low, but the measurement precision for detecting die defects is insufficient
Solution Approach 1:
The patent replaces traditional mechanical inspection methods with machine learning-based image analysis systems. Neural networks and deep learning models process images of dies to detect defects such as missing solder points and cracks, substituting mechanical inspection with intelligent computational systems that achieve higher measurement precision.
Solution Approach 2:
The patent introduces image processing algorithms and machine learning models as intermediary systems between the camera and defect detection. These intermediaries process visual data through multiple processing stages including image acquisition, preprocessing, feature extraction, and classification, enabling accurate defect detection without direct mechanical interaction.
2Reliability
If machine learning systems are implemented to detect die defects, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent implements feedback mechanisms where the machine learning system continuously learns from detected defects and adjusts its detection criteria. The system provides feedback to improve its accuracy over time, enhancing reliability while managing complexity through adaptive learning rather than requiring overly complex static systems.
Solution Approach 2:
The patent segments the defect detection process into multiple independent modules: image acquisition, image preprocessing, feature extraction, and defect classification. This segmentation allows each component to be optimized independently, improving overall reliability while keeping individual module complexities manageable.
3Productivity
If real-time image analysis is performed on all dies, then the productivity is maintained, but the loss of time for processing increases
Solution Approach 1:
The patent applies partial action by processing only a subset of images through comprehensive analysis while using faster, less computationally intensive methods for other images. The system prioritizes processing based on risk assessment, applying full analysis only when necessary, thus maintaining productivity while reducing overall processing time.
Solution Approach 2:
The patent performs preliminary actions by pre-processing images and extracting features before final defect detection. This preliminary processing prepares the data in advance, allowing the main detection algorithm to operate faster and reducing overall processing time while maintaining accuracy.
Data Source
AI summary
Systems, methods, and computer program products for identifying defective dies at a die processing service with machine learning are provided. A training dataset comprising training images taken by one or more cameras at a tape and reel machine is provided to train a machine learning system. The training images include images of dies having integrated circuits. Positions of solder points are determined in each training image. The positions of solder points in each training image are aligned with positions of solder points in other training images to generate aligned positions of the solder points. The aligned positions are clustered into multiple clusters. A centroid position for each cluster is determined, where the centroid positions correspond to locations of the solder points across all images. The centroid positions are transmitted to the machine learning system in a production environment and are used to identify images with defective dies.