Advancements in high-performance timing for long term underwater experiments: A comparison of chip scale atomic clocks to traditional microprocessor-compensated crystal oscillators

We discuss the merits and performance of two commercially available time bases - a microprocessor-compensated crystal oscillator designed and manufactured by Seascan, Inc. of Falmouth, MA, and a chip-scale atomic clock (CSAC), designed and manufactured by Symmetricom, Inc. of San Jose, CA - for use in autonomous underwater instrumentation. Over the last ~10 years, the WHOI Ocean Bottom Seismology group has used the Seascan time base in its fleet of ~100 Ocean-Bottom Seismographs (OBS), and has acquired a record of Seascan performance amounting to over 183 cumulative seafloor years and over 25 years of cumulative test time in a temperature-chamber across 120 clocks. These data show that the Seascan clocks perform within their quoted specification in almost all cases, and if carefully selected can provide performance significantly exceeding their specifications, all at a very modest 5 mW of power consumption. Our testing has also shown some unexpected effects in certain units that may undermine timing quality if not understood and controlled. Within the last year or so, chip-scale atomic clocks have become available, and we have installed these time bases in 15 OBS. These CSACs offer two orders of magnitude improvement in precision, at the cost of significantly greater power consumption (~120 mW). We have deployed thirteen instruments on the ocean floor for a 6-month experiment using CSACs as the time base. In addition we have tested 4 more units in the lab. Based on the performance that we have seen in the field and in the lab, both time bases are worthy of consideration in new ocean instrumentation. As expected, the CSACs show better stability, lower aging, and better temperature response, at the expense of higher power consumption. This paper will present a detailed comparison of both clocks, showing the tradeoffs that need to be considered when selecting a time base. We will also look at techniques available to improve the aggregate performance of the Seascan time bases to well above their nominal specifications.