Powertech Technology (Taiwan)

Fault Detection and Diagnosis (FDD) in complex industrial systems requires methods that can handle uncertain operating conditions, soft thresholds, evolving sensor behavior, and increasing volumes of heterogeneous data. Traditional model-based or rule-driven approaches offer interpretability but lack adaptability, while purely data-driven and Large Language Model (LLM)-based methods often struggle with consistency, traceability, and causal grounding. Dynamic Master Logic (DML) provides a causal and temporal reasoning structure with fuzzy rules that capture gradual drift, soft limits, and asynchronous sensor signals while preserving traceability and deterministic evidence propagation. Building on this foundation, this paper presents a DML-LLM hybrid architecture that integrates targeted LLM inference to interpret unstructured information such as logs, notes, or retrieved documents under controlled prompts that maintain domain constraints. The combined system integrates Bayesian updating, deterministic routing, and semantic interpretation into a unified FDD pipeline. In a semiconductor manufacturing case study, the proposed framework reduced time to detection (TTD) from 7.4 h to 1.2 h and improved the F1 score from 0.59 to 0.83 when compared with conventional Statistical Process Control (SPC) and Fault Detection and Classification (FDC) workflows. Provenance completeness increased from 18% to 96%, while engineer triage time was reduced from 72 min to 18 min per event. These results demonstrate that the hybrid framework provides a scalable and explainable approach to anomaly detection and fault diagnosis in sensor-rich industrial environments.

We present the latest reliability results from our high power, high efficiency red VCSEL arrays at 660 nm with an output power in the range of 1 to over 100 mW, suitable for consumer application. Manufacturing process was done using high volume 6” GaAs fab. Under pulsed operation conditions (1ms pulse width, 10% duty cycle), the VCSEL arrays are able to produce an optical output power exceeding 100 mW, maintaining stable operation at ambient temperatures up to 50 °C. The combination of a carefully designed epi structure which minimizes stress and thermal impedance and an optimized layout for heat conduction are implemented in our red VCSELs. The optimized structure not only shows good thermal properties, the resulting reliability also is improved significantly. Highly Accelerated Stress Testing (HAST), HighTemperature Operating Life (HTOL), and 85C/85% Damp heat tests have reached over 1000+ hours with minimal drift. We will provide important design optimization results and discuss the influence of various array layouts on the thermal management and overall device performance for consumer and industrial applications.

Burst mode control strategies have been widely adopted in LLC resonant converters due to the significant increase in switching loss at light load. This paper proposes a standby-mode burst-operation control based on bang-bang charge control (BBCC) for half-bridge LLC resonant converters to improve light-load efficiency while preserving a fast transient response. The voltage variation across the resonant capacitor, which is proportional to the per-cycle charge transfer and thus the output power, is used as a control indicator to compare with the feedback voltage. To compensate for the extra output power induced by turn-off delays, a digital/logic compensation block is introduced to cancel the delay-dependent term, thereby restoring the proportionality between the output power and the threshold voltage. To verify the control method, a controller IC implementing the proposed logic was evaluated on a 400 V input, 20 V/12 A (240 W) output half-bridge LLC power board. At 10 W (4.17% load), the efficiency improves from 87.26% (normal mode) to 93.94% (standby mode), i.e., +6.68%. The mode-transition boundaries are set at 10 W (entry) and 30 W (exit), and the transition times are 108 μs (normal mode to standby mode) and 107 μs (standby mode to normal mode), respectively.

Abstract Silicon photonics is increasingly important for advancing optical communication networks. The reliability testing of silicon photonics is a critical issue. This study demonstrates damp heat and temperature cycling reliability tests for silicon photonics, followed by failure analysis (FA).

This study explores the correlation between metabolic syndrome and night-shift work in Taiwan by analyzing employees' annual health check data (March 2019). More than 9000 records were enrolled. Through the analysis of the annual health check data of a semiconductor factory's day-shift and night-shift workers' metabolic syndrome indicators, the relationship between night-shift work and metabolic syndrome was identified. According to the definition of long-term night work labor by the Ministry of Labor of Taiwan, this study divides all employees of the factory into long-term night work and non-long-term night work employees. Abnormal waist circumference (WC) was defined as WC ≥ 90 cm in males and WC ≥ 80 cm in females. Abnormal blood pressure was defined as systolic blood pressure ≥ 130 mm Hg, diastolic blood pressure ≥ 85 mm Hg, or taking medication for hypertension. Abnormal fasting blood sugar was defined as ≥100 mg/dL or taking medication for diabetes. Elevated triglyceride (TG) was defined as TG ≥ 150 mg/dL or taking TG-lowering medication. Abnormal high-density lipoprotein (HDL) was defined as HDL < 40 mg/dL in males and HDL < 50 mg/dL in females. A total of 9322 workers from the factory were recruited: 4039 workers were night-shift employees, and 5283 workers were non-night-shift employees. Night work status increased the risk of poor blood pressure (systolic pressure + 3.1%, diastolic pressure + 1.7%) and elevated TG (+5.7%). Night-shift workers had a higher body mass index and abdominal circumference than day-shift workers. Night-shift workers had a 3.6% higher risk of obesity/overweight. Logistic regression analysis demonstrated that, after adjusting for age, sex, and body mass index, night-shift work was significantly associated with a higher risk of metabolic syndrome (odds ratio = 1.17, 95% confidence interval 1.02-1.35). This study supports the hypothesis that circadian rhythm disruption plays a role in the development of metabolic syndrome. It is recommended that business units implement targeted health promotion strategies, including regular health screening, nutrition guidance, onsite exercise support, and psychosocial well-being programs. Future research should include longitudinal designs and sleep quality assessments to clarify causality.

This article presents an advanced radar system that utilizes common Wi-Fi signals for human imaging and Doppler detection. The system employs a time-division multiplexing (TDM) multiple-input-multiple-output (MIMO) radar configuration, synchronizing a local oscillator with the transmitted Wi-Fi signal through injection locking to detect Doppler phase shifts in the returning Wi-Fi echoes. A significant challenge in human imaging is the varying radar cross sections (RCSs) between the torso and limbs, which can obscure limb images. To address this issue, this system incorporates a dc nulling filter in its digital signal processor as an RCS equalizer. This filter enhances rapidly moving limb images, while maintaining the clarity of quasi-stationary torso images. The experimental results reveal that the system produces high-resolution images that distinctly depict limbs, facilitating the recognition of human postures and activities. Additionally, it effectively captures posture images in various scenarios, including standing, sitting, falling, and lying down. These captured images are essential for both vital sign monitoring and fall detection.

Energy supply on high mountains remains an open issue since grid connection is not feasible. In the past, diesel generators with lead–acid battery energy storage systems (ESSs) were applied in most cases. Recently, photovoltaic (PV) systems with lithium-ion (Li-ion) battery ESSs have become suitable for solving this problem in a greener way. In 2016, an off-grid PV system with a Li-ion battery ESS was installed in Paiyun Lodge on Mt. Jade (the highest lodge in Taiwan). After operating for more than 7 years, the aging of the whole electric power system became a critical issue for its long-term usage. In this work, a method is established for analyzing the massive energy data (over 7 million rows), such as daily operation patterns, as well as the C-rate, temperature, and accumulated energy distributions, and estimating the health of the Li-ion battery system. A completed electric power improvement project dealing with power system aging is reported. Based on the long-term usage experience, a simple cost analysis model comparing lead–acid and Li-ion battery systems is built, revealing that expensive Li-ion batteries can compete with cheap lead–acid batteries for long-term usage on high mountains. This case study can provide engineers and researchers with a fundamental understanding of the long-term usage of off-grid PV ESSs and engineering on high mountains.

In this work, we investigated the stability of a <inline-formula> <tex-math notation="LaTeX">${p}$ </tex-math></inline-formula>-GaN gate with high electron mobility transistors (HEMTs) including an internal integrated gate circuit. A circuit was designed to improve <inline-formula> <tex-math notation="LaTeX">${p}$ </tex-math></inline-formula>-GaN gate stability by using capacitance to release the hole into the <inline-formula> <tex-math notation="LaTeX">${p}$ </tex-math></inline-formula>-GaN layer to mitigate the threshold voltage shift. Through pulse I-V measurement and positive bias temperature instability (PBTI) test, the carrier transporting behavior in the gate region achieved dynamic equilibrium at 5 V gate bias. The positive gate shift <inline-formula> <tex-math notation="LaTeX">$(\Delta V_{\mathrm{ TH}})$ </tex-math></inline-formula> of 0.4 V is observed with increasing voltage from 3 V to 8 V; <inline-formula> <tex-math notation="LaTeX">$\Delta V_{\mathrm{ TH}}$ </tex-math></inline-formula> initially drops smoothly after release stresses by external capacitance discharge. Finally, integrated passive components and <inline-formula> <tex-math notation="LaTeX">${p}$ </tex-math></inline-formula>-GaN gate HEMT circuit are recommended to mitigate the <inline-formula> <tex-math notation="LaTeX">$V_{\mathrm{ TH}}$ </tex-math></inline-formula> instability for E-mode HEMT.