Measuring Component Self-Alignment by Automatic Image Processing Method

Martinek, Péter; Villányi, Balázs; Krammer, Olivér

doi:10.1109/isse.2019.8810280

Cited by 2 publications

(2 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several hidden neurons, training algorithms and a values were evaluated. Martinek et al (2019) included several factors, such as component geometry and pad geometry, which were not considered in the previous studies. They obtained insight into these factors and used the significant ones as inputs for their data-driven models like support vector regressor (SVR), neural networks (NN) and random forest regressor (RFR) for self-alignment prediction.…”

Section: Related Workmentioning

confidence: 99%

“…During the reflow operation, the mounted components move toward the most symmetric position, known as self-alignment. This shift occurs due to the restoring forces caused by the surface tension of the molten solder (Martinek et al, 2019). However, self-alignment can also lead to the misalignment of components, resulting in defects such as tombstoning and overhanging (Parviziomran et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Parameter optimization for surface mounter using a self-alignment prediction model

Mistry

Gupta

Jain

et al. 2022

SSMT

View full text Add to dashboard Cite

Purpose The purpose of this paper is to develop a machine learning model that predicts the component self-alignment offsets along the length and width of the component and in the angular direction. To find the best performing model, various algorithms like random forest regressor (RFR), support vector regressor (SVR), neural networks (NN), gradient boost (GB) and K-nearest neighbors (KNN) were performed and analyzed. The models were implemented using input features, which can be categorized as solder paste volume, paste-pad offset, component-pad offset, angular offset and orientation. Design/methodology/approach Surface-mount technology (SMT) is the technology behind the production of printed circuit boards, which is used in several types of commercial equipment such as communication devices, home appliances, medical imaging systems and sensors. In SMT, components undergo movement known as self-alignment during the reflow process. Although self-alignment is used to decrease the misalignment, it may not work for smaller size chipsets. If the solder paste depositions are not well-aligned, the self-alignment might deteriorate the final alignment of the component. Findings It were trained on their targets. Results obtained by each method for each target variable were compared to find the algorithm that gives the best performance. It was found that RFR gives the best performance in case of predicting offsets along the length and width of the component, whereas SVR does so in case of predicting offsets in the angular direction. The scope of this study can be extended to developing this model further to predict defects that can occur during the reflow process. It could also be developed to be used for optimizing the placement process in SMT. Originality/value This paper proposes a predictive model that predicts the component self-alignment offsets along the length and width of component and in the angular direction. To find the best performing model, various algorithms like RFR, SVR, NN, GB and KNN were performed and analyzed for predicting the component self-alignment offsets. This helps to achieve the following research objectives: best machine learning model for prediction of component self-alignment offsets. This model can be used to optimize the mounting process in SMT, which reduces occurrences of defects and making the process more efficient.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%