Failure Region Estimation of the Linear Voltage Regulator Using Model-Based Approach

Abstract

Linear voltage regulator (LVR) stability in electronic devices depends on the LVR's internal circuit design and output capacitor, with the specific equivalent series resistance (ESR) values being influenced by the LVR's output terminal. ESR variation due to temperature change and aging factors creates an LVR failure region in the ESR tunnel graph. However, the LVR failure region estimation process is being conducted manually, which consumes high manufacturing time and requires high skill expertise. Furthermore, LVRs differ in their exact models, and manufacturers hardly know the exact internal parametric values of these manufactured LVRs because of the manufacturing variation factor. The objective of this study is to develop a model-based failure region estimation (FRE) method to ensure the efficiency and effectiveness of the FRE process. In this study, the FRE process was initially manually conducted and used as the benchmark. Then, the manually obtained FRE process data were used in the model-based FRE method by integrating the system identification method into the neural network. Finally, the two methods (manual FRE and model-based FRE) were compared. The effectiveness of the model-based FRE was high, with error values approaching zero and an approximate accuracy of 100%. As for efficiency, the model-based FRE was 43.75% faster than the manual FRE. These outcomes indicate that the model-based FRE is suitable for LVR manufacturing. Therefore, LVR manufacturers can test each LVR failure region with better efficiency and effectiveness by adopting the developed FRE method.