Circuit Assembly Yield Prediction via Defect Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional yield prediction techniques for circuit assembly manufacturing fail to account for various factors such as component attributes, manufacturing processes, and defect frequencies, leading to unreliable results and inability to accurately predict production yields.
Innovation Solution
A system and method for predicting production yield by mapping component attributes to possible defects and manufacturing processes, allowing for 'what-if' analyses of different form factor and manufacturing process options, and providing targeted yield predictions based on design specifications and screening capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional yield prediction techniques are used, then the prediction process is simple, but the prediction accuracy is poor
Solution Approach 1:
The patent segments the yield prediction process into distinct modules: defect data collection module, defect frequency calculation module, yield prediction module, and feedback module. Each module handles specific aspects of the prediction process, allowing complex calculations to be broken down into manageable segments that can be executed systematically
Solution Approach 2:
The patent implements a feedback mechanism where actual factory yield data and test results are continuously fed back into the prediction system. This feedback loop allows the system to refine and adjust defect frequency data and prediction models, improving accuracy over time while managing complexity through iterative optimization
2Reliability
If component attributes and manufacturing processes are considered in detail, then prediction reliability improves, but calculation complexity increases
Solution Approach 1:
The patent applies local quality by associating specific defect types with particular component attributes and manufacturing processes. Instead of treating all components uniformly, the system identifies and analyzes defect patterns specific to each component type, attribute combination, and process step, allowing reliable predictions without overwhelming complexity through targeted analysis
Solution Approach 2:
The system manages complexity by dynamically adjusting calculation parameters based on the specific design specification and component mix. The defect frequency data and prediction models adapt their granularity and detail level according to the particular assembly being analyzed, maintaining reliability while avoiding unnecessary computational burden
3Measurement precision
If defect data is collected and analyzed for each component, then prediction accuracy improves, but data processing time increases
Solution Approach 1:
The patent implements preliminary action by pre-collecting and organizing defect data from manufacturing processes and testing, storing it in structured formats with associated component attributes and process parameters. This pre-processed defect frequency data is readily available when yield predictions are needed, eliminating the need for time-consuming data collection and analysis during the prediction process itself
Data Source
AI summary
Embodiments of the invention include systems and methods for automatically predicting production yield for a circuit assembly according to attributes of its components and defect data mapped thereto. Embodiments receive a proposed design specification for a circuit assembly, including bill of materials (BOM) and schematic data, at a yield prediction environment. The yield prediction environment maps a set of attributes to each component in the BOM and maps a set of possible defects to each component according to its attributes. Defects may be further mapped to a manufacturing process assigned to populate each component in the circuit assembly. The defects are associated with predicted frequencies of occurrence, which can be used to roll up a yield prediction for the circuit assembly. Embodiments further allow “what-if” analyses to be performed so that different yield prediction results can be compared according to different form factor options and/or different manufacturing process options.