Asahi Kasei Microdevices’ Current Sensors Integrated Into Microchip’s Arc Fault Detection Design
The reference design uses on-chip machine learning to distinguish dangerous arc faults from noise in EV charging and solar systems.
Microchip Technology has incorporated Asahi Kasei Microdevices’ (AKM) CZ39 and CZ3K series coreless current sensors into a new reference design that utilizes machine learning to detect arc faults. This integration aims to improve safety in DC systems, including EV charging infrastructure, by reducing false positives common in traditional detection methods.
What Happened
The new reference design centers on Microchip’s dsPIC33A digital signal controller, which executes signal-processing and inference algorithms directly on the controller. This architecture allows detection decisions to be made locally without relying on external processing resources. AKM’s CZ39 and CZ3K sensors, which feature a 100 ns response time, provide the high-speed current sensing data required for the system. The San Jose-based engineering team at AKM Semiconductor, AKM’s US subsidiary, collaborated with Microchip on the current-sensing configuration during the development and validation phases.
Why It Matters
Arc faults are a primary cause of electrical fires, but conventional threshold-based detection struggles to differentiate between dangerous faults and ordinary system activity. In DC applications such as EV charging, solar photovoltaics, and energy storage, switching transients from relay bounce, inverter operations, and capacitor inrush generate noise that mimics arc signatures. This often leads to unnecessary system shutdowns or missed faults if thresholds are relaxed. By running an edge machine learning model on the dsPIC33A, the design leverages the controller’s integrated DSP engine to analyze clean sensor data, potentially reducing false triggers and improving detection performance. AKM identifies EV chargers, e-Fuse designs, and smart ignition systems among the key applications for this technology.
The Bottom Line
The arc fault demonstration is available through Microchip’s reference design program. While currently targeted at EV and renewable energy sectors, AKM’s Vice President of Business Development, Chris Baltar, noted the company is exploring adoption for data center applications as AI workloads drive demand for higher-power-density and high-voltage power architectures.
