PCB Solder Joint Detection Model for Fewer SPI False Positives
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Solder Paste Inspection (SPI) devices have insufficient detection accuracy, leading to unnecessary manual re-judgment of solder joints, as they often incorrectly identify poorly soldered joints, increasing the workload and time required for review.
Innovation Solution
A system and method that builds a detection model based on standard value data to classify solder joints, using data collection, classification, model building, and analysis modules to accurately assess soldering conditions, reducing manual re-judgment by generating classifications and models from first and second solder joint data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional SPI devices are used to detect solder joints, then detection coverage is achieved, but detection accuracy is insufficient leading to false positives
Solution Approach 1:
The patent segments the detection process into multiple stages: initial SPI detection, classification of suspicious solder joints, and secondary verification. This multi-stage segmentation allows the system to focus detailed analysis only on potentially defective joints rather than all solder joints, thereby improving detection accuracy while reducing false positives.
Solution Approach 2:
The patent performs preliminary classification of solder joints using SPI data before conducting detailed verification. By pre-identifying suspicious joints based on extracted features (area, height, offset, etc.) and comparing them against standard values, the system prepares a targeted list for manual or automated verification, improving overall detection efficiency and accuracy.
2Measurement precision
If manual re-judgment is performed on all detected solder joints, then detection accuracy improves, but workload and time consumption increase
Solution Approach 1:
The patent applies partial action by performing manual or detailed automated verification only on solder joints classified as suspicious, rather than all detected joints. The classification module filters out clearly normal joints, so verification resources are concentrated only on cases where false positives are most likely, significantly reducing total verification time while maintaining high accuracy.
Solution Approach 2:
The patent introduces an automated classification module as an intermediary between SPI detection and manual verification. This intermediary processes SPI data to identify suspicious joints, acting as a filter that reduces the volume of work requiring human judgment while preserving detection accuracy through targeted verification.
3Measurement precision
If detection thresholds are lowered to reduce false positives, then detection precision improves, but more joints require verification
Solution Approach 1:
The patent dynamically adjusts detection parameters based on classification results. The system extracts multiple features (area, height, offset, volume) and compares them against standard values with different threshold levels for different feature types. This multi-parameter approach allows precise identification of truly defective joints while maintaining high throughput by using lenient thresholds only for initial screening.
Data Source
AI summary
The present disclosure provides a method and a system for building a detection model based on a standard value to confirm a soldering status. The present disclosure generates the solder joint classification based on the standard value data, and uses the detection model to analyze whether the second solder joint data generated by detecting the PCB has poor soldering conditions after building the detection model based on the first solder joint data corresponding to the solder joint included in the solder joint classification, thereby achieving the technical effect of reducing the number of solder joints that are misjudged as poor soldered and shortening the time required for manual re-judgment.