In recent years, neural network based deep learning models has demonstrated high accuracy in object detection and classification in the area of digital image processing. Manufacturing industry has successfully implemented prototypes and small-scale deployment to employ artificial intelligence (AI) models for quality inspection. It has been proven that AI-assisted quality inspection can improve inspection accuracy, operation throughput and efficiency, significantly through those prototypes and small-scale deployment. However, the industry-known challenge of Operational Technology (OT) and Information Technology (IT) integration arises when scaling up AI-assisted quality inspection in manufacturing operation. While model accuracy is the main concern from an inspection point of view, IT implementation has to meet the requirements of high availability, scalability, security and model & device lifecycle management. 

This paper discusses in detail the challenges in large-scale deployment of AI models for quality inspection operation and introduces a framework for large-scale AI-assisted quality inspection in manufacturing environment using edge computing architecture. The framework focuses on IT architectural decisions to fulfill the OT requirement, including user experience in the quality inspection ecosystem. Keyword: Quality Inspection, AI Models, Edge Computing

Blockchain technology has a lot of publicity in the electronics industry because of the way it has been used to address issues with sharing data across distributed networks and is recognized for providing "greater transparency, enhanced security, improved traceability, increased efficiency and speed of transactions, and reduced costs". The challenge for the electronics industry is to set up an electronics supply chain blockchain architecture correctly such that we can leverage those benefits quickly while expanding the blockchain back into sub-tier suppliers. This paper will discuss key blockchain cases within the supply chain, how to determine if a blockchain solution is the right answer for a problem, the key design considerations for a blockchain solution, and the importance of standards. Some of the lessons learned from recent efforts will also be shared, along with discussion some of the key challenges.

Keywords —blockchain, supply chain, electronics industry

Automated test equipment plays a large role in the manufacturing process at the Kansas City National Security Campus (KCNSC). Every test run generates data which is used to determine if a part meets its requirements or to troubleshoot failures if they occur. Among this data are log events which contain unique information about each test run. Ordinarily, these events are rendered to a string and routed to a file, database, or status window. Strings are used because they are easy to create and store, are human-readable, and are capable of encoding many data types. 

When troubleshooting failures, these logs are a critical resource. Log files can be browsed manually to find key information; however, it can quickly become tedious to correlate a specific log event across multiple runs. At the KCNSC, tester software now utilizes an approach known as structured logging. A structured log event message is still defined using a string but one which is annotated to indicate the parts that were derived from data. This enhances each event, allowing it to be treated as a collection of properties rather than a simple string.

Using this approach to define log events sacrifices no flexibility but highly promotes data accessibility. With access to this data, tester teams can quickly find answers to their questions about tester or part performance. For example, one tester exhibited an issue with crosstalk between digitizer channels. Using log aggregation software, the structured logs were queried and used to produce plots indicating on which channels the issue most frequently appeared. This minimized the effort required for inspecting signal routing through the tester. Analysis like this is critical to reducing downtime during troubleshooting and also allows for proactive monitoring in a way which is not feasible with conventional unstructured logs.

Data is the New Oil

Clive Humby, UK Mathematician and architect of Tesco’s Clubcard is widely credited as the first to coin the phrase in 2006: “Data is the new oil. It’s valuable, but if unrefined it cannot really be used. It must be changed into gas, plastic, chemicals, etc. to create a valuable entity that drives profitable activity; so, must data be broken down, analyzed for it to have value.

There is no doubt that we are witnessing an authentic technical revolution. Call it Industry 4.0, Internet of Things, Big Data, or Artificial Intelligence; the proliferation of solutions facilitating Industry 4.0 is accelerating.

Whatever the perspective is with which we look at this fast- paced evolution, there is an asset that is at the center of everything: data. 

However, contrary to what happened in the previous industrial revolutions where manufacturing was at the focus of the revolution, manufacturing has lagged in implementing the base technologies underlying this data transformation. It has been conservative and extremely slow to realize that the application of these technologies is invaluable, perhaps much more than in any other segment. They are refining, but at a much slower pace.

Case in point, the term “The Internet of Things” was coined by Kevin Ashton in a presentation to Proctor & Gamble in 1999; and more than 20 years later, manufacturing is still learning the meaning of IoT and trying to devise strategies to take advantage of it.

EMS factories have collected and used machine data for many decades. Over that time, much of the value derived from machine data collection has come from three operational use cases: allowing fewer operators to simultaneously monitor more machines for errors, reducing common operational mistakes through programmatic interlocks, and maintaining traceability records in case of product recalls. There has been significantly less use of machine data for strategic optimization of factory operations, with the notable exception of asset utilization monitoring using simple calculations like Line Utilization and OEE. One of the historical reasons for the absence of large-scale analysis of machine data in the EMS industry has been that it was difficult to interpret machine data absent external context on what intended operation was being performed when the data was collected. More recently however, the advent of big data analysis techniques and machine learning algorithms has largely removed this traditional limitation. In this paper we discuss the difference between tactical and strategic data analysis approaches to the common EMS factory goals of lowering component attrition and increasing line utilization. We show how machine data can provide significant value at the strategic level if it is stored and analyzed in granular detail instead of being pre-aggregated into high level key performance indicators before being analyzed. As the EMS industry looks forward to Industry 4.0, we argue that one of the biggest areas of efficiency gain may come from such strategic data analysis.

Smart Factories require continuous and reliable operation of all equipment. Therefore, equipment maintenance is becoming more and more important. A simple approach for equipment maintenance, such as “let it break – then fix / repair it”, does not fulfill the needs of a continuously and reliably operating Smart Factory: It leads to unexpected, not plannable downtimes of equipment and, thus, impacts production in an unacceptable way. A preventive maintenance approach can avoid this problem of unplanned production downtime due to breaking down machines. A predictive or prescriptive maintenance approach may improve planning of maintenance even further. All these approaches have in common that they need data from operation of equipment. A Smart Factory needs Smart Maintenance, i.e., data driven maintenance.

This presentation provides an overview of an ongoing examination of test methodologies within the ESA, supported by IPC, for the evaluation of stacked and staggered (offset) microvia structures, including air-to-air thermal shock, convection reflow assembly simulation and current induced thermal cycling.

As EMS Providers such as Contract Manufacturers look forward to Industry 4.0, their need for complex data analysis to inform manufacturing decisions takes on even more significance. Managing manufacturing data has become as important to operational excellence as managing the solder paste and electronic components placed on each circuit board. However, while there are many examples of small-scale, operational data collection systems that inform real-time factory operations, there are relatively few examples of large-scale historical machine data collection systems. In this paper we describe one such system and discuss both the advantages it brings as well as the practical implementation challenges to build and maintain it at scale. It is expected that in the years to come many EMS factories will find use for such systems as increasingly proven use cases involving complex machine learning, predictive maintenance and even artificial intelligence become more common in the industry. Each of these advanced Industry 4.0 applications requires as a prerequisite the kind of large, centralized, historical machine dataset that a system like what we describe collects and manages.

The smart factory project for Electronics Manufacturing initiative was developed to create a sandpit to carry out industry 4.0 use cases working with industrial members and partners. The Manufacturing Technology Centre (MTC) based in Coventry UK have carried out a project that outlines the implementation of IPC CFX to demonstrate the connectivity between the machine, the broker and the dashboard. The main objective focuses on the implementation elements for the Adaptor machines and the RabbitMQ server. This includes configuration requirements and user guides around maintaining the delivered implementation.

The paper will review the above example of the successful connecting using IPC CFX, a SMT DEK printer and the Ersa Reflow 10 zone oven which are both classed as legacy machine due to its age and the operating system. The benefits to this is to allow legacy machines used in the electronics manufacturing industry to communicate, transfer operational data for track/trace and monitoring while, in future add-ons, allow them to be agile and have autonomous ability to change parameters based on the live shop floor situation. This will eventually lead to the ‘lights out’ capabilities in the low volume/high mix electronics manufacturing industry.