Substrate Inspection Compensation Matrix Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional substrate inspection methods face challenges in generating reliable compensation matrices due to warpage or distortion in printed circuit boards, which affects the accuracy of feature object positioning within the field of view.
Innovation Solution
A method involving the selection of randomly predetermined feature objects within the field of view, generation of multiple compensation matrices based on offset values, and repeated iterations to maximize the number of feature objects with offset values below a predetermined reference value, ultimately generating a third compensation matrix for improved reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a compensation matrix is generated using feature objects in the FOV, then the position of pads can be compensated, but the reliability of the compensation matrix is reduced due to substrate warpage or distortion
Solution Approach 1:
The patent applies preliminary action by pre-selecting N1 feature objects that are randomly predetermined within the FOV before generating the compensation matrix. This preliminary selection ensures that only suitable feature objects are used, improving the reliability of the compensation matrix while maintaining position compensation accuracy
Solution Approach 2:
The patent implements feedback by repeatedly performing the compensation matrix generation process N2 times and evaluating the offset values of feature objects. The system uses this feedback information to determine the number of feature objects with offset values less than a reference value, and adjusts the compensation matrix generation accordingly to maximize reliability
2Reliability
If multiple feature objects are selected randomly in the FOV, then the compensation matrix can be generated, but the time required for substrate inspection increases
Solution Approach 1:
The patent applies periodic action by repeating the compensation matrix generation process N2 times with predetermined iterations. This periodic repetition allows the system to evaluate multiple compensation matrices and select the one with maximum reliability, while the predetermined iteration count prevents excessive time consumption
Solution Approach 2:
The patent uses parameter changes by adjusting the number of feature objects N1 and the number of repetitions N2 as controllable parameters. These parameters can be optimized to achieve the desired balance between compensation matrix reliability and inspection time, allowing flexible adaptation to different inspection requirements
Data Source
AI summary
The present invention relates to a method for generating a compensation matrix during a substrate inspection. The method comprises the steps of: selecting information of N1 (N1≥2) feature objects which are randomly predetermined within a field of view (FOV) on a substrate; generating a first compensation matrix on the basis of information of the feature objects which are extracted on the substrate; comparing an offset value of each of all the feature objects with a predetermined reference value by applying all the feature objects within the FOV to the compensation matrix to count the number of the feature objects of which the offset value of the each of all the feature objects is less than the predetermined reference value; and repeatedly performing the above steps N2 times (N2≥1), and generating a second compensation matrix using information of the feature objects which have the offset value which is less than the predetermined reference value, in case the number of the counted feature objects is the maximum.