Test Results of IEEE 1588v2 Network Synchronization Holdover Performance using Various Types of Reference Oscillators

This paper explores holdover in an IEEE 1588v2 application. Lab measurements were taken on a reference design configured as a telecom boundary clock (T-BC) comparing the short- and long-term holdover performance of the node when using a TCXO, OCXO, and age-compensated DOCXO, as its reference oscillator. Measurements were taken both with, and without the assistance of a physical layer clock. The results show that when a physical layer clock (e.g. SyncE) is provided, the node meets the ITU-T Recommendation (G.8273.2) [1] for short-term holdover. For this paper, long-term holdover (24-hour) performance is based on the end-to-end LTE-TDD small cell mobile network requirements. When a physical layer clock is provided, the long-term holdover performance is highly dependent on the average drift rate of the PTP clock at the time when this clock is lost and much less dependent on the stability of the local reference oscillator. However, for long-term holdover, when a physical layer clock is not provided, the type of reference oscillator used becomes extremely important and only by employing an age-compensated oscillator was the 24-hour phase error requirement successfully achieved.