Figure 1 - uploaded by Syed usama Khalid Bukhari
Content may be subject to copyright.
Source publication
Development of an automated PCB inspection system as per the need of industry is a challenging task. In this paper a case study is presented, to exhibit, a proposed system for an immigration process of a manual PCB inspection system to an automated PCB inspection system, with a minimal intervention on the existing production flow, for a leading aut...
Contexts in source publication
Context 1
... DQ 200 TCU control unit PCB examination was requested to be reformed, in terms of time interval necessary for the inspection after the board has passed through the soldering process. The workflow is described in figure 1., showing the whole manufacturing process, while the human inspection was presented in figure 2. The procedures of human inspection are: ...
Context 2
... 11 was used for the segmentation of the first part, shown in figure 12(a) with a value of n = 11, as the second part shown in figure 12(b) utilizes a value of n = 12. After obtaining both segmented images, checking the pins employs the eight- neighbor algorithm described above. ...
Context 3
... also can be observed that if comparing to true negatives, detection of true positive is done with more accuracy. The same study was performed for all the 5 group categories and results similar to the ones depicted in table 3., 4., figure 14. and 15. were obtained. ...
Context 4
... software was developed in Matlab 2011, and for time calculation matlab's tic and toc functions were used. Figure 19. The processing time requested for each type of images Figure 19. ...
Similar publications
This article focuses on the application of a vision system for a point to point programing of a robotized soldering station. The station is used in the last stage of production for the soldering of elements, applying Through-Hole Technology (THT). In the soldering station a SCARA-type robot is used. The main aim of the usage of the vision system is...
The idea of increasing the noise immunity of radioelectronic systems with redundancy is presented. Specific technical solutions based on this idea of modal redundancy are described. An estimation of noise immunity improvement was performed by the example of implementation of modal redundancy with the broad-side electromagnetic coupling for a printe...
Citations
... The followed production of boards uses CAD Dip Trace needed for forming g-code. It is strongly advised to follow the steps of editing printed boards [9][10][11][12][13][14][15]. ...
The article presents methods and means of 3D design of printed circuit boards in CAx. Automated placement of elements on the board is implemented by means of API SolidWorks using Visual Studio C#. The API application works by an algorithm that allows you to create a 3D layout of printed circuit boards. Each component of the library contains a conditional graphic notation of the element. With the help of the implemented algorithm, a comprehensive approach is provided, which consists in the fact that already at this stage the preparation of the strategy for tracing the conductors of the future printed circuit board is carried out, the classes of circuits are determined and the necessary technological parameters are set, as well as the data necessary for the preparation of design documentation is generated. After the completion of the work on the input of the scheme, a check is made for the presence of errors and compliance with the specified parameters, and if the test is successful, a list of circuits is generated for transfer to the tracing program. From this moment, any possibility of errors in the subsequent stages of design is excluded. CAD DipTrace was used to trace printed circuit boards and generate g-code. Printed circuit board processing is carried out on a CNC machine-CNC3018 using the Candle program. The printed circuit board tracks are created by forming a groove between the track and the metallized coating of the textolite. With the help of the formed height map, the uniform removal of the metallization layer over the entire area of the textolite is ensured. In addition, holes are drilled for the output elements of the circuit, the printed circuit board is cut along the contour and covered with a layer of tin to prevent oxidation of its metallized coating. The considered CAD methods and tools made it possible to automate the design of the printed circuit board of the FM radio receiver control module. As a result of the performed work, means of automating the design of printed circuit boards were applied and a fully functional printed circuit board with a track width of 0.8 mm was obtained.
... The followed production of boards uses CAD Dip Trace needed for forming g-code. It is strongly advised to follow the steps of editing printed boards [9][10][11][12][13][14][15]. ...
The article presents methods and means of 3D design of printed circuit boards in CAx. Automated placement of elements on the board is implemented by means of API SolidWorks using Visual Studio C#. The API application works by an algorithm that allows you to create a 3D layout of printed circuit boards. Each component of the library contains a conditional graphic notation of the element. With the help of the implemented algorithm, a comprehensive approach is provided, which consists in the fact that already at this stage the preparation of the strategy for tracing the conductors of the future printed circuit board is carried out, the classes of circuits are determined and the necessary technological parameters are set, as well as the data necessary for the preparation of design documentation is generated. After the completion of the work on the input of the scheme, a check is made for the presence of errors and compliance with the specified parameters, and if the test is successful, a list of circuits is generated for transfer to the tracing program. From this moment, any possibility of errors in the subsequent stages of design is excluded. CAD DipTrace was used to trace printed circuit boards and generate g-code. Printed circuit board processing is carried out on a CNC machine - CNC3018 using the Candle program. The printed circuit board tracks are created by forming a groove between the track and the metallized coating of the textolite. With the help of the formed height map, the uniform removal of the metallization layer over the entire area of the textolite is ensured. In addition, holes are drilled for the output elements of the circuit, the printed circuit board is cut along the contour and covered with a layer of tin to prevent oxidation of its metallized coating. The considered CAD methods and tools made it possible to automate the design of the printed circuit board of the FM radio receiver control module. As a result of the performed work, means of automating the design of printed circuit boards were applied and a fully functional printed circuit board with a track width of 0.8 mm was obtained.
... While a case study reports on the transition from a manual PCB inspection system to an AOI system (Bukhari et al., 2017), another study presents an AOI system for detecting microdefects in PCBs, focusing on copper leakage at the edges of the PCB assembly (Parakontan & Sawangsri, 2019). ...
Remarkable advances in Deep Learning, a subfield of Artificial Intelligence (AI), have attracted considerable attention in recent years. One prominent example is DeepMind, a company working on the development of a general-purpose AI. After AI systems outperformed professional players in games such as Go and chess, DeepMind recently achieved another breakthrough in predicting protein folding. Using Deep Learning, protein structures can now be predicted with over 90 percent accuracy, replacing laboratory experiments for the first time in history. Becoming aware of the successful application of Deep Learning, numerous industrial companies started pilot projects to gain insights. Manufacturing companies, in particular, are faced with the question of how Deep Learning can be leveraged to realize a competitive advantage and what challenges need to be considered. In production environments, quality control is a core task often relying on visual techniques. One of the world's leading German multinational automotive suppliers has been using Automatic Optical Inspection (AOI) for quality assurance in electronics production for decades. Since Computer Vision with Deep Learning has the potential to improve visual quality inspection, the company intends to support its AOI systems with suitable Deep Learning approaches. In this context, the present dissertation aims to contribute to research and gather general knowledge about the application of Deep Learning in an AOI environment. To this end, several studies are conducted. Extensive structured literature reviews form the foundation for selected Deep Learning experiments, which represent the main focus of this thesis. The experiments are based on a dataset provided by the company, containing images of Printed Circuit Boards (PCBs) captured by an AOI camera. The characteristics of the real-world dataset affect both, the experimental design as well as the results. Contributing to debates on architecture selection and Transfer Learning for operational use, the experiments provide significant insights into factors influencing the performance of deep neural networks in machine vision tasks for defect detection on PCBs. Despite recent advances in Computer Vision through Vision Transformers, the results for the case at hand show that the inductive bias inherent in established Convolutional Neural Networks (CNNs) is better suited for inspection tasks on compartmentalized PCBs and that Transfer Learning can accelerate the training-to-production cycle. All studies reveal in different ways that Deep Learning can make a substantial contribution to the industry in the field of optical inspection.
... Besides bare PCB, the entire assembly sequence can be automatic optical inspected, i.e. paste printing, component placement and soldering [1]. This test is also related as Automatic Visual Inspection (AVI) for non-contact inspection [5] proposed in a fully automated system to reduce the time for quality assurance operations. ...
