Automated optical inspection (AOI) systems are commonly applied for in-line quality control of through hole technology (THT) solder joints. However, these systems usually rely on conventional image processing, often leading to high pseudo error rates and extensive workloads during subsequent manual inspection. Deep learning approaches have shown promise in reducing pseudo errors by classifying solder joint images flagged as defective by AOI systems into two classes: real defects and pseudo errors. Yet, their performance is limited by the availability of sufficient local training data, and while cross-company data sharing could address this issue, it remains constrained due to privacy concerns.
Federated learning (FL) offers a promising solution by enabling collaborative model training across companies without the need to share raw data, thus preserving data privacy. This study evaluates the application of FL for THT-AOI image classification, specifically addressing common FL challenges such as divergence of local model updates and performance decay under non-independent and identically distributed (non-IID) client data. Therefore, local, centralized and federated learning are compared based on real and artificially separated data under IID and non-IID conditions.
Our results demonstrate that convolutional neural network (CNN)-based FL can be effectively applied for THT-AOI image classification, achieving training stability and model performance comparable to centralized learning. By implementing FL, we reduce AOI pseudo errors by an average of 23.6% on an aggregated test set containing heterogeneous data from multiple electronics manufacturers, significantly lowering the need for manual inspection. This represents a 12.1% relative improvement in pseudo error reduction over local learning, with both approaches maintaining a controlled error slip rate of 1.0%.

Manufacturing is getting smarter, with most processes and equipment generating vast volumes of data. Dynamic electronics manufacturers are adopting innovative solutions to increase output while reducing expenses and improving product quality. This research details a robust AI-based system targeting some of the most common issues in the electronics test and measurement industry.
Testing and measurement have multiple phases, each highly complex and requiring significant debugging time. Smarter, robust, and efficient processes are achieved using Artificial Intelligence/Machine Learning (referred as AI/ML) solutions. Additionally, AI/ML can help unearth previously unknown relationships in the data.
The proposed AI and ML approach leverages test logs, systems logs and sensor data. The models process real-time streaming test log data to identify patterns, outliers, anomalies, and problems based on historical data learnings. Companies adopting AI can quickly identify the root cause of various issues, solve them, and integrate those learnings to optimize processes.
AI transformation is a critical element of Industry 4.0 and is still in its early stages of development. It is expected to rapidly grow and disrupt traditional problem-solving methods currently followed in manufacturing. AI use cases currently implemented by the industry have demonstrated tangible value and the ability to be executed at scale. This paper will discuss some of the actual industrial implementations and showcase real benefits achieved.
This paper also provides insights into AI implementation design and the common practical challenges faced by organizations in making such initiatives successful.

The ever-increasing requirements in terms of power density and reliability of power electronic modules make packaging an ever-increasing challenge. Especially when upcoming vertical gallium nitride (GaN) devices are taken into consideration. These devices have promising properties that enable current densities in the range of kA/cm². The handling of this level of current density is representing a big challenge for the adjacent packaging technologies, especially the bonding wires. This work is investigating upon the occurring bond wire temperatures in a setup containing a hypothetical vertical GaN device. It is shown via thermal simulations that the possible amount of thick aluminum bonding wires packaged in standard polymer encapsulation materials (silicone gel: 0,2 W/m∙K) is not capable of carrying the provided amount of electrical current without exceeding the critical recrystallization temperature of aluminum. This represents a significant impact on the reliability of these components. Therefore, this setup is not able to utilize the full potential of vertical GaN devices. Novel ceramic encapsulation materials which have a thermal conductivity of 5 W/m∙K are introduced into the setup. It is shown that these materials are able to cool the wires in a way that they are able to carry the provided amount of electrical current. This enables the use vertical GaN devices to their full potential. This effect is also verified via an experiment conducted on a dedicated test setup. Through the implementation of a temperature sensor and thermographic measurements a correlation between the experiment and the simulation is established.

The presented research is a ground-up investigation of PCBA climatic reliability starting with an insight into surface water layer formation under humid conditions, and progressing to understanding how different forms of ionic process contamination can influence water layer formation and subsequent failure mechanisms. As PCBAs become increasingly miniature and complex, understanding the interplay between these factors on realistic board and component designs is crucial. The impact of no-clean reflow solder flux chemistry and component-specific design on climatic reliability was investigated using DC and AC electrochemical methods, along with Scanning Electron Microscopy (SEM), Energy-Dispersive X-ray Spectroscopy (EDX), Light Optical Microscopy (LOM), and X-ray imaging. An electrochemical model was developed to predict leak current formation and current line movement under condensed droplet conditions and the presence of different surface-level design elements.
Key findings include the necessity of complete droplet coalescence for dendrite formation and the significant role of soldering residues in water layer formation. Comparative analysis of Weak Organic Acid residues revealed Glutaric acid’s higher aggressiveness compared to Adipic acid and its effect on dendrite growth paths in conjunction with connector layout design. The short-term protective role of reflow solder flux encapsulation and eventual longer-term plasticization and release of residues was also examined. The research further explored the interaction between reflow flux and component design parameters under humid conditions, focusing on low-standoff components with small connector pitch distances. Observations highlighted increased leakage currents due to flux residue trapping (improper outgassing), component warpage, and tilting due to thermal gradients flux system choice during reflow, all influencing a variety of humidity-related reliability issues.

Printed circuit boards (PCBs) for automotive and electrical products are becoming smaller, higher density components, and embedded component technology is progressing. Components mounted on these PCBs generate heat, mostly transferred from the components to the PCBs. Therefore, it is necessary that not only the heat on the component surfaces be dissipated into the atmosphere or into the heat sink, but also the heat within the PCB, and the thermal design of mounted PCB has become more significant in recent years. In addition, to ensure high reliability and safety of PCBs, insulation materials for PCBs that can handle high voltages of 1000V, or more are required. Conventional heat-dissipating materials include TIM sheets and heat-dissipating grease, which are relay materials used to dissipate heat from the component surface to the heat sink, not the PCB, and have thermal conductivity but no insulation properties for consideration. As a result of intensive research, a new material with excellent thermal conductivity and high insulation properties for heat dissipation from PCBs was developed. The cured film has a thermal conductivity of 3 W/mK and a dielectric breakdown voltage of 6 kV/100μm. The appearance of the coating film is glossy, with a surface roughness (Ra) of 1.0μm. Furthermore, the simulation and demonstration of heat dissipation when this coating film was formed on a PCB and heat-generating components were mounted on it showed excellent and high heat dissipation as predicted.

The increasing complexity of printed circuit board assemblies (PCBA) necessitates more efficient and comprehensive testing methods to meet stringent quality and production targets. This paper presents a multi-test approach, integrating In-Circuit Testing (ICT), Functional Circuit Testing (FCT), Boundary Scan (BScan), LED Test and Flashing, to optimize both test time and takt time in high-volume and high-mix manufacturing environments. By consolidating these critical test methodologies into a single, synchronized process, the multi-test approach eliminates the need for separate testing stages, significantly reducing overall test time.
Through parallel execution of multiple tests, the takt time is minimized, leading to increased throughput. The paper examines the correlation between reduced test cycle times and the improvement in production line efficiency, highlighting the impact on key performance indicators such as first-pass yield and defect coverage. The expected outcome of this approach is a marked reduction in bottlenecks typically associated with sequential testing processes, as well as improved resource utilization. This methodology provides a scalable solution for modern PCBA testing, offering manufacturers a means to achieve faster time-to-market while maintaining stringent quality control.

By comparing the design fine milling allowance of different milling tools and the chip removal ability of milling cutters, it is concluded that the design fine milling allowance of tools is 0 mm, which can solve the problem of residual glue on the concave edge plate surface when combined with double edge and double groove chip removal milling cutter. By comparing the pore diameter of aluminum sheet with different plug holes and the fluffy treatment of residual glue, it is concluded that the pore diameter of aluminum sheet with plug holes to reduce the concave edge hole and the chemical glue removal method can solve the problem of residual glue on the concave edge plate. By comparing the width of the residual cover and whether the core plate pre-cutting treatment is carried out, it is concluded that increasing the width of the residual cover to cover the stress points with the core plate pre-cutting can solve the problems of cover crack and burr.

In addition to the requirements of good bending performance, in the specific installation process, the flexible plate also needs to have good flexibility and tear resistance to facilitate assembly. In order to meet the requirements of such application scenario, the influences of polyimide type, different copper foil and coating structure on the flexibility and tear resistance of the flexible plate were studied. It is found that the thickness of polyimide and covering film is the main influence factor on the softness of flexible plate, but the copper foil is not the main influence factor. The thickness of copper foil and polyimide has great influence on tear resistance. This study provides a reference for the industry to design flexible board products with high flexibility and tear resistance.

Corrosion issues of electronic devices under humid conditions heavily rely on the cleanliness of Printed Circuit Board Assemblies (PCBAs). Use of no-clean flux during diverse soldering techniques such as wave, selective, and reflow processes significantly influence PCBA cleanliness. Residual flux on PCBA poses a threat, causing corrosion failure modes under humid conditions. All flux residues are hygroscopic and, therefore attract moisture on the PCBA surface at different humidity levels depending on the Critical Relative Humidity (CRH) of the ionic component of the residue. Hence, it can determine the humidity boundary at which water film forms over surface. Another contribution of residue is to the conductivity of the water film due to the dissolution of ionic residues. Although all solder processes can generate residues, use of liquid fluxes for wave and selective soldering processes make it more adverse from corrosion robustness point of view. Therefore, liquid flux chemistry, especially the ionic part of the flux composition, plays a big role in determining the humidity boundary for the PCBA. This paper presents an analysis of large number of PCBA cleanliness-related performance data and effects on corrosion reliability within the framework of descriptive analytics to understand how activator chemistry and chemistry of the flux system play a huge role in determining humidity boundary for PCBA. Additionally, data analyzed using decision tree machine learning algorithm to understand significant factors and their levels.