Prediction of Component Shifts in Pick and Place Process of Surface Mount Technology Using Support Vector Regression

Abstract

In pick and place (P&P) process of surface mount technology (SMT) the placed component can shift from its ideal (or designed) position on the wet solder paste. The solder paste with some fluid properties could slump and the unbalance between different sides of solder paste can lead to other forces on the components as well. Though the shifts are usually considered to be negligible and can be made up to some extent by the following self-alignment during the process of soldering reflow, it should be attracted attention as its importance for addressing the quality of the printed circuit board (PCB) in SMT. To minimize or control the component shifts, whose relationship with the characteristics of the solder paste (e.g., offset, volume) should be studied initially. In this paper, we design a comprehensive experiment and collect the data from a state-of-the-art SMT assembly line. Then we use support vector regression (SVR) model to predict the component shifts based on different situations of solder paste and placement settings. Also, two kernel functions, linear (SVR-Linear) and radial basis function (SVR-RBF), are employed. The achieved results indicate that the component shift in P&P process is significant, and the SVR model is highly qualified for the forecast of the component shifts. Particularly, the SVR-RBF model outperforms the SVR-Linear model considering the prediction error.

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

“…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. Cao et al (2019) used SVR to predict the component shifts during mounting in the SMT process. After implementing the two kernel functions, it was found that the SVR-RBF model was more effective for predictions compared to the SVR-Linear model.…”

“…resistors, capacitors, transistors and integrated circuits. Every slot contains just one reel, whereas reels might occupy one or more slots (Cao et al , 2019). The machine contains a robot arm that has one or more heads.…”

“…The pre-AOI machine is located before the reflow oven and checks for defects on the PCBs after the components are mounted. The post-AOI machine checks for defects after reflow soldering (Cao et al , 2019). During the reflow operation, the mounted components move toward the most symmetric position, known as self-alignment.…”

Enhancing EfficientNet-YOLOv4 for Integrated Circuit Detection on Printed Circuit Board (PCB)

“…Accurate component detection plays a pivotal role in automating PCB production monitoring, specifically in addressing critical manufacturing defects such as component shifts or missing parts within surface mount technology (SMT) pick-and-place processes [2], [5]. Strengthening automated PCB inspection tools is imperative to tackle these issues effectively.…”