A novel method for aging estimation of crystal oscillators

The frequency aging of a quartz crystal oscillator is an important characteristic of frequency stability. Aging estimation for frequency correction of a precision crystal oscillator is a valid approach in improving oscillator performance. This paper presents an estimation and prediction model in which the frequency and frequency aging are described by a family of shifted logarithmic functions. Weights of logarithmic function are estimated recursively based upon the present data inputs and the last data outputs to avoid massive data storage. Outlier points in frequency measurements are filtered using a robust estimation computation. Finally, the frequency of an oscillator is predicted based upon the best estimated weights of logarithmic functions.