A Study on the Automatic Calibration Function of RF Amplifiers Using Artificial Neural Networks

Abstract

In this study, power amplifiers are designed and manufactured based on the frequency specifications by customers. After learning appropriate compensation values according to the function and performance degradation factors of the manufactured power amplifier using an artificial neural network, the power amplifier itself can maintain an optimal performance when its function and performance degrade. The artificial neural networks are applied to power amplifiers using the STM32F series of microcontrollers, which are being widely used for industrial purposes in recent years. Hence, after manufacturing the power amplifiers, the optimal state is maintained without additional tuning by workers, as well as changes in the external environment and aging of electronics parts. When the performance of the power amplifier is degraded owing to other factors, the performance of the power amplifier can be improved by itself.