Machine Yield Assessment Using End-of-Line Defect Attribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional manufacture systems face challenges in assessing the yield rates of individual machines in a product line due to insufficient inspection machines, leading to issues like the 'tunnel effect' and 'latent defects', where it is difficult to determine which machines are responsible for defective products.
Innovation Solution
A yield-rate assessment apparatus and method that calculates bad-piece and good-piece expectation values for each machine based on initial yield rates and detected defects, allowing for the assessment of yield rates without the need for sufficient inspection machines, using a processor and storage system to analyze data and determine the contribution of each machine to product quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only one inspection machine is arranged near the end of the product line, then the device complexity and inspection cost are reduced, but the ability to assess individual machine yield rates deteriorates
Solution Approach 1:
The patent introduces an intermediary computational model that acts as a mediator between the single inspection machine's data and the yield rate assessment of individual machines. The model uses bad-piece expectation values as an intermediate representation to trace defects back to their source machines, enabling precise yield rate assessment without requiring multiple physical inspection machines at each station.
Solution Approach 2:
The patent replaces the mechanical approach of placing multiple physical inspection machines at each manufacturing step with a computational system. The processor uses algorithms to calculate bad-piece expectation values and trace defects back to source machines, substituting physical inspection infrastructure with information processing and mathematical modeling.
2Measurement precision
If more inspection machines are arranged over the product line, then the yield rate assessment capability is improved, but the device complexity and cost increase
Solution Approach 1:
The patent extracts the essential function of multiple inspection machines and consolidates it into a single inspection machine combined with a computational model. By taking out the redundant physical inspection infrastructure and replacing it with information processing, the system achieves the same yield rate assessment capability with fewer physical devices.
Solution Approach 2:
The single inspection machine in the patent serves multiple functions: it inspects final products, provides defect data for the computational model, and enables yield rate assessment for all machines in the product line. This multi-functionality replaces what would traditionally require multiple specialized inspection machines at each manufacturing step.
3Ease of manufacture
If conventional inspection methods are used with insufficient inspection machines, then the inspection cost is reduced, but the ability to determine defect responsibility deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where defect information from the single inspection machine is fed back into the computational model, which then calculates bad-piece expectation values and traces defects back to their source machines. This feedback loop preserves defect responsibility information that would otherwise be lost with conventional inspection methods using insufficient machines.
Solution Approach 2:
The patent performs preliminary computational analysis by calculating bad-piece expectation values for each machine based on the inspection data. This preliminary action of tracing defect origins through mathematical modeling preserves responsibility information before it would be lost, enabling accurate yield rate assessment without requiring multiple inspection machines.
Data Source
AI summary
A yield-rate assessment apparatus for a manufacture system including a plurality of machines, each machine participating in one or more manufacture steps of a batch of products in the manufacture system, performs for each machine: calculating a bad-piece expectation value and a quantity of potential bad pieces at each corresponding manufacture step based on a quantity of bad pieces detected after the last one of the manufacture steps is finished and an initial yield rate of the current machine; calculating a good-piece expectation value based on a quantity of good pieces detected after the last one of the manufacture steps is finished and a summation of all quantities of potential bad pieces calculated for the current machine; and assessing a yield rate according to the good-piece expectation value calculated for the current machine and a summation of the bad-piece expectation value calculated for the current machine at each corresponding step.