Adaptive temperature compensation of GPS disciplined quartz and rubidium oscillators

An algorithm for performing adaptive temperature and ageing compensation of GPS disciplined oscillators is presented. The algorithm is based on a recursive implementation of linear regression, so it is computationally efficient. In addition, essentially unlimited equivalent averaging times are realizable using this approach while storage requirements remain small and constant. Implementation of time varying averaging times is made simple as well. All of these attributes are desirable for operation on relatively weak, embedded processors. A simple semiconductor ambient temperature sensor and A/D converter mounted on the system printed circuit board are the only additional hardware required for implementation of this algorithm. Data is presented showing the performance of the algorithm with a variety of classes of oscillators, including an inexpensive TCXO, a high-stability OCXO and an inexpensive Rb oscillator. The algorithm improves the performance of all of these oscillators relative to algorithms which employ either no compensation or only ageing compensation while operating in holdover.