Real time synchronisation of a free-running atomic clock time base with UTC using GNSS signals for application in experimental physics

TL;DR

This work addresses the need for precise UTC synchronization in experimental physics by real-time correcting a free-running atomic clock time base using GNSS signals. The authors employ CGGTTS-based time differences against UTC(OP), perform online linear fits on the most recent measurements, and continuously extrapolate clock drift to apply corrections with a latency of about $1$ s. Tests with a Rubidium clock and a magnetic Caesium clock show residual differences to UTC(OP) within $±15$ ns and standard deviations around $\sim$ $2$–$3$ ns, confirming feasibility for long-baseline and multi-messenger experiments. The Caesium clock offers superior long-term stability, while the Rubidium clock can be adapted to a semi-free-running mode, suggesting practical implementations for projects like Hyper-Kamiokande where sub-$100$ ns synchronization is essential.

Abstract

We present the results obtained by applying, in real-time, a correction method to precisely synchronize a time base generated from a free-running atomic clock with the Coordinated Universal Time (UTC). The method uses the Global Navigation Satellite System (GNSS) signals to have regular time comparisons between the atomic clock generated time base and the GPS Time, perform linear fits of the measurements and extrapolate a correction to apply to the free-running signal. In this work, we apply for the first time this method in real-time. Two atomic clocks were tested, a low-cost Rubidium clock and a more expensive magnetic Caesium clock. We demonstrate that we can obtain a residual difference between the clock time base and the French official realization of UTC (UTC(OP)) in the range of $\pm 15$ ns with no apparent residual drift.

Figures (4)

• Figure 1: Overlapping Allan standard deviation (OASD) of the Caesium clock frequency signal (green) as measured by the manufacturer, the Rubidium clock PPS signal (blue) and the GPS signals (orange) measured at LPNHE against UTC(OP) using the counter and the GNSS receiver respectcively.
• Figure 2: Time differences between the PPS of the atomic clock (Rb on the left and Cs on the right) and the UTC(OP) measured by the counter. For the Rb clock (left) the counter took one measurement every ten seconds whereas for the Cs clock it took one measurement per second.
• Figure 3: Residual time differences between the PPS of the atomic clock (Rb on the left and Cs on the right) and the UTC(OP) after the correction applied in real time on the measurements of the counter. The x-axis corresponds to the date and time (in Modified Julian Day) at which the correction was applied.
• Figure 4: Overlapping Allan standard deviation (OASD) of the free-running Caesium clock PPS signal (blue) as measured by the counter and the signal corrected in real time (orange) with respect to UTC(OP).