Timekeeping with a Chip Scale Atomic Clock in GPS Denied Environments

Systems that rely on accurate timekeeping commonly use a GPS navigation system in combination with an external oscillator. A typical oscillator with short-term accuracy can take advantage of the long-term accuracy of a GPS receiver in order to align its timekeeping with the GPS signal. A quartz or atomic oscillator left to run freely will drift away from UTC time unless paired with a GPS receiver to form a GPS Disciplined Oscillator (GPSDO) [1]. In environments where GPS signals are unreliable or jammed, the time-keeping ability of the external oscillator is essential in maintaining accurate timekeeping. Use of an atomic clock that holds a high level of accuracy and stability is key in improving timekeeping ability in these zones with limited GPS signal. In situations that require low SWaP components and robust timekeeping, a Chip Scale Atomic Clock (CSAC) will be able to meet these constraints even in the event of a loss of GPS signal. Without connection to an atomic external oscillator the receiver must rely on the built-in internal Temperature Compensated Crystal Oscillator (TCXO) which holds a significantly lower level of timekeeping ability. Timekeeping is vital to maintaining accurate positing, which makes it crucial to maintain accurate timing for alternates and backups to GPS [2]. This paper investigates the timekeeping ability of the CSAC as well as conditions that may result in altered or inaccurate time solutions from the Microsemi CSAC SA.65 in comparison with the earlier SA.45s version. This paper will study the Microsemi SA.65 CSAC and a U-Blox ZED-F9T GPS receiver. The ZED-F9T receiver will source the 1 Pulse Per Second (1PPS) signal into the CSAC, which will be used to discipline the CSAC as well be a reference to the CSAC while free running. The ZED-F9T is utilized to provide an accurate reference frequency to within 5 parts per billion [3] as well as a time pulse jitter of +/-4 ns and a time mark resolution of 8 ns [4]. The SA.65 is the latest CSAC oscillator that is commercially available from Microsemi. The SA.65 accepts a 1PPS input to synchronize the output to within 100 ns of a reference clock. It can also discipline its phase and frequency to within 1 ns and 1 × 10^-12 of the reference clock, respectively [5]. The CSAC can be disciplined to UTC time via the 1PPS input to simulate environments where there is no interference of GPS signals. This is simulated by using the CSAC’s 1PPS Disciplining Mode. This mode is present in the SA.65 and the SA.45s and utilizes an internal algorithm that increases the accuracy of a TCXO clock by a magnitude of 4 or 5. This algorithm works in conjunction with the physics package in the CSAC which uses the hyperfine frequency of cesium atoms to continuously compare and correct the TCXO sensitivity and stability [5]. This algorithm takes the error between the 1PPS in and 1PPS out from the CSAC and alters the frequency of the clock to align the 1PPS out with the 1PPS in. Turning off the disciplining mode enables the CSAC to run freely without altering the frequency of the clock. This mode simulates the system entering a region where there is limited GPS signal and where environmental factors can affect the CSAC’s timekeeping ability. Some factors include temperature fluctuations or magnetic field disturbance, which will not be discussed in this paper [6]. Changing the CSAC’s internal disciplining algorithm can also affect its behavior in these environments. Simulation of these environments and CSAC behaviors during its undisciplined operation gives insight into how well the CSAC can keep track of time without the influence of GPS. This is especially important in cases where the time sensitivity of a system requires extreme precision. The CSAC will also be compared to two other types of high-performance oscillators: an OCXO and TCXO. These oscillators will be compared within the context of the Open-Source GPS Software Defined Radio [7]. Each oscillator will provide the reference frequency to a USRP receiver and the data will be used to find a clock solution and propagate this solution to predict the time 180, 300, 660, 1380, 2940, and 5940 seconds in the future. This paper intends to predict errors in the SA.65 CSAC’s timekeeping ability as well as characterize its behavior with respect to other oscillators. The SA.65 is expected to produce the best results as it is a newer chip scale atomic oscillator model; however, this paper attempts to quantify any errors with the SA.65 in the absence of GPS signal input.