China Topscom Issured Article On PCB World Magazine About PCB boards industry urgently needs AI and machine learning.China Topscom provide professional Pcb board layout design fabrication, pcb assembly and manufacturing, full turnkey systems integration box build assembly, contract manufacturing service.

 
Now PCB has developed to a new stage. With the introduction of new technologies such as high-density interconnection (HDI) PCB,IC substrate (ICS), the whole production process has changed from manual to fully automated. With the further development of manufacturing technology, the process becomes more and more complex, and defect inspection becomes more and more important and difficult. These fatal defects may lead to the scrapping of the whole PCB board. For the PCB manufacturing industry, opportunities are emerging to use artificial intelligence to (AI) and optimize the production process, and finally optimize the entire PCB manufacturing process.

PCB manufacturing usually relies on experts who have accumulated knowledge for many years, who know and understand every step of the manufacturing process very well, and they know how to use their knowledge to optimize production and increase production. Human limitations (including misoperation and fatigue) hinder efficiency growth, and operator errors or misidentification of PCB defects ("error alarms") may affect yield due to overhandling, or even damage PCB itself. By integrating AI into the manufacturing process (figure 1), machines can add value by taking over certain "learning" tasks, while human experts continue to undertake more complex tasks that require optimization and "training" while thinking and interacting with artificial intelligence systems. The combination of human and artificial intelligence improves the overall efficiency and operation, and is the biggest opportunity for AI system.



Artificial Intelligence and Industry 4.0.
 
The ultimate trend of PCB development is to have a factory with a fully integrated Industry 4.0system, which uses AI technology at the global and manufacturing system levels. The Global level includes all systems in the factory, not just a single manufacturing system. Industry 4.0 provides an automation and data exchange infrastructure that enables real-time production analysis, two-way communication and data sharing, traceability and on-demand data analysis. Within any given plant, AI can improve processes using data obtained from various manufacturing systems and machines, collected through industrial 4. 0 mechanisms (such as traceability, two-way communication). The factory benefits because AI analyzes a large amount of system-wide data to optimize plant settings and achieve the highest levels of productivity and yield. Artificial intelligence analysis and self-learning are under way and are carried out through artificial neural networks. Within a few years, it will eliminate the intervention of manual operators and lead to the establishment of fully automated factories.
 
This new PCB manufacturing model requires all plant systems to be fully connected and AI as a monitoring and decision-making mechanism. Currently, there are proprietary and technical challenges that limit the full automation of PCB factories, but AI has been added to a single system as much as possible, such as automatic optical inspection (AOI) solutions. The advantages of moving production facilities to the global AI model include more reliable notification of PCB defects-"real defects" and a feedback mechanism that identifies the root cause of the problem and then automatically modifies the plant process to eliminate related problem defects.
 
A subset of AI, including machine learning and deep learning, will move the PCB factory towards full automation. The algorithms used in machine learning enable computers to use data and the examples they have experienced and learned from to improve the performance of tasks without explicit programming. In the case of PCB manufacturing, machine learning increases production, improves manufacturing operations and processes, and reduces manual operations, while helping to promote more efficient handling of plant assets, inventory and supply chains.
 
Deep learning takes AI to a more complex level, which is beneficial at the global plant system level. Deep learning is inspired by human brain neurons and the ability of multi-layer artificial neural networks to learn, understand and infer. In PCB factories, software systems can effectively collect data and learn from complex representations of patterns and contexts, and then learning will form the basis for automatic process improvement in PCB manufacturing.
 
The implementation of machine learning and deep learning provides PCB manufacturers with capabilities beyond human understanding; artificial intelligence systems discover new optimization opportunities by digging deeper into places that people do not want to explore. The AI expert system is very efficient, reducing the number of manual experts required and improving efficiency and best practices by using more and more complex parameters to monitor plant systems around the world.
 
With I4.0 sensors (sensors that can send data from equipment) and systems, data can be created globally throughout the PCB manufacturing process, from simple read and write functions to advanced tracking of process parameters to the smallest PCB unit. Process parameters can include etching, resist development and even the concentration of chemical materials in the manufacturing process. Use deep learning to analyze these types of data to inform optimized manufacturing methods and parameters, identify patterns, and make informed decisions about the changes needed in the process. All of this can be done 24 hours a day, 7 days a week.
 


System-level AI.
 
At the system level, such as in the AOI process, the AI implementation in the PCB manufacturing workshop has had a significant impact on productivity and yield. In this case, machine learning greatly reduces human errors in detecting PCB defects. Examples of PCB defects include short and open circuits, even excessive copper. Automated inspection can detect very small defects, which may not be detected by manual inspection, or may be omitted due to human error, which is the natural result of repetitive work.
 
Without AI, a classic inspection of 100 panels typically reveals 20 to 30 defects per panel, about 75 per cent of which are false alarms. Due to the policy that all defects must be checked manually, the review of false alarms wastes valuable production time and increases the handling of PCB, which may lead to new damage and may affect further errors made by operators during the review process.
 
Through machine learning on AOI systems, such false alarms and maintenance can be greatly reduced (figure 2). Fewer false positives means less processing of PCB boards and improved efficiency. In addition, AI provides consistent (dynamically improved) defect classification without inherent limitations of the operator, thereby providing more reliable results and reducing verification time. According to Orbotech internal research, AI in AOI systems has been found to reduce false positives by up to 90 per cent. What makes AOI unique is that the system collects more data than any other manufacturing solution, which makes it well suited as the first step in AI implementation. At the same time, the AOI room is the most labor-intensive area in the PCB factory, so the use of AI in its process will bring the greatest benefits. For PCB manufacturers, all this means that millions of defects can be identified and classified more accurately, making it possible to increase production and reduce costs.
The following is an example of a system working with a global-level AI:
 
Suppose the AOI system examines 100 panels. At the system level, AI supported by machine learning can filter out false positives, which have been classified by the system. The AI system produces the smartest classification results by evaluating multiple AOI images while leveraging its "panel understanding" (the AOI solution's understanding of the elements on the panel and their appearance). This information is fed into the global AI system, which is powered by deep learning, collects this data from system-level solutions and determines that the real defect identified is a short circuit and requires additional etching time to remove excess copper. The AI system uses data from the system level to make global decisions to adjust panel parameters during etching so that all panels manufactured in the future have fewer, if any, the same type of defects. Ultimately, communication between system-level solutions will further increase and improve AI's global decision-making capabilities.
 

Create an increase in challenges.
 
Although the development of AI is developing rapidly across the industry, the challenges of PCB manufacturing are growing at the same rate, or even faster. For flexible materials and the geometry of reduced alignment, there are two difficult areas of defect detection. Next-generation composites such as liquid crystal polyamide (LCP) and modified polyamide (MPI), pose new challenges for manufacturers, including image acquisition, processing, deformation and finer lines. For example, the more advanced the material used for flexible PCB, the more defects are identified, resulting in more false alarms. The purpose of manufacturers using this complex material is to minimize the processing of panels in the process of determining false alarms. So, Flex PCB (figure 3) is a product type that will likely benefit greatly from AI implementation, as the system will learn to manufacture within a stricter range of parameters.
 
PCB for 5G is another high requirement and is likely to benefit greatly from the expertise supported by artificial intelligence. The HDI PCB required for 5G applications requires finer linewidth, straight sidewall geometry and strict parameters. This makes defect detection more difficult than ever. For human experts, it will be a great challenge to complete defect detection effectively.
 
Considering these and other unknown PCB manufacturing challenges, artificial intelligence-driven factories will be the key to future production. In order to realize the development of AI applications around the world, it takes more time to realize PCB manufacturing, but it is obvious that the implementation of system-level AI has come, laying the foundation for the future of a fully automatic PCB factory.