The NASA/JPL Deep Space Atomic Clock (DSAC) now in orbital test uses a radically different approach to achieve higher performance with lower size, weight, and power requirements than existing.
The real-time clock (RTC)–as distinct from the processor clock, but often using the same source–is a critical function in many system designs. Depending on the application, it may need to be accurate (in the loose sense of the word) anywhere from a few microseconds or milliseconds to a few seconds per month. Most RTCs begin with a fairly accurate crystal, in some cases with extra stabilizing circuitry, oven-controlled housing, and other enhancement techniques. Solid-state MEMS-based resonators are also increasingly used as alternatives to the venerable crystal, and can offer advantages in cost, packaging, size, power, and performance.
But for truly critical applications, even the precision of a calibrated, ovenized crystal or MEMS unit may be inadequate. The solution may require very large step up to atomic clocks, most of which use hyperfine atomic-level transitions of cesium. This includes the NIST-established US physical official clock (Reference 1) as well as the 2018 world-wide redefinition of Système international (SI – International System of Units) constants, (Reference 2 and Figure 1).
Obviously, cesium-based atomic clocks are amazingly good, but they may not be good enough for truly unique applications such as deep-space navigation. Such navigation uses clocks to determine location and velocity, of course, in addition to gyroscope/accelerometer-based inertial measurement units (IMUs). NASA’s deep-space problem is that the clocks sent on these missions, despite their extreme accuracy, still have errors which are large enough to result in significant course errors over the enormous distances involved (Reference 3). For these missions, the correction technique used is to send a signal to the space vehicle, query its clock, look at the round-trip time, and then send appropriate corrections to the on-board clock/navigation function. Clearly, this is awkward, error prone, and a work-around in the best sense of the phrase.
While a better on-board clock would be the obvious answer, the only ones which until now met spacecraft size, weight, power, and ruggedness mandates were less accurate than needed. In space, even tiny errors in RTC time can result in missing the target by enormous distances, and there’s no “let’s go back and try again.” At best, you can do a md-course correction, but these are tricky to execute, and at some point you run out of fuel if several are required.
Of course, accurate clocks are not needed solely for deep-space missions. One of the enabling technologies for GPS was the development of space-compatible rubidium-based atomic clocks (Reference 4); each GPS satellite has a primary unit and a backup. This allows the Earth-based receivers to be small, low-cost units with basic processor clocks, but no need for an RTC, let alone a super-accurate one. (Note that these GPS clocks need twice-daily corrections, both to correct for their inherent inaccuracies as well as time-dilation effects which are a consequence of Einstein’s theory of general relatively (Reference 5) – but that’s another story for another time).
The need for better deep-space atomic clocks has driven NASA and the Jet Propulsion Laboratory (JPL) to develop the Deep Space Atomic Clock (DSAC). The test unit was launched in June 2019 for a one-year orbital trial; it is the primary payload (one of six) onboard the General Atomics Electromagnetic Systems Orbital Test Bed (OTB) satellite.
This atomic clock does not use cesium or rubidium transitions. Instead, it is based on a very different set of atomic principles and arrangement (Figure 2); you get various levels of detail from broad overview to in-depth technical specifics at the references. Note that the design is based on use of trapped mercury ions. There is no use of lasers, cryogenics, or a microwave cavity; it has low sensitivity to temperature, magnetic field, or voltages variations; and radiation tolerance is similar to GPS clocks. Short term stability (from 1 – 10 sec) is dependent on the local oscillator while the longer-term stability (>10 sec) is determined by its “atomic resonator” ion trap and light system (Figure 3).
The 16-kg unit measures 29 cm × 27 cm × 23 cm (Figure 4) and requires under 50 W; the next-generation version now under development targets a weight of under 10 kg and 30 W dissipation. While there are many metrics used to characterize clock accuracy, the DSAC is 50 times more stable than the atomic clocks currently flown on GPS satellites; Reference 10 has some performance graphs, numbers, and comparisons.
The use of time for navigation is not new. The principle was understood in the 17th and 18th centuries and was first implemented using crude clocks to track the real time at zero meridian while comparing it to local observed solar time. Accuracy improved with development of better mechanical clocks, an effort which cumulated with the famous clock built over many iterations and years by John Harrison, an 18th-century clockmaker who created the first clock sufficiently accurate to be used to determine longitude at sea (every gear tooth was made by hand!). The story of his efforts, frustrations, success, denial of earned reward, and eventual recognition has been told many times, most recently in the 1995 best seller by Dava Sobel (Reference 11) as well as many other articles and books
(Personal note: Despite its popularity, I was very disappointed in the book. Like many writers, she preferred to spend thousands of words and many pages to describe various aspects of the clock and its predecessors, when a drawing or two would have been much clearer. Interestingly, a companion book The Illustrated Longitude, with 180 images of characters, events, instruments, maps and publications was published in 1998 as a supplement to the original book … hmmm, I wonder why that was done.)
Today’s RTCs range from <$1 crystal/MEMS units to $30 over-the counter digital watches of surprising accuracy and stability (my $45, nothing-fancy “sport watch” is good to ~2 sec/month!) and extending to atomic-based units costing thousands of dollars. all with performance which is almost unimaginably and indescribably impressive. John Harrison would likely be overwhelmed by the concept of such accurate clocks — after all, how did he even assess his? (He did it using via known and accepted sunrise/sunset/noon times at defined Earth locations, a real challenge.)
What’s the most accurate RTC you have had to include in a circuit or system? Have you ever had to give your RTC accuracy much thought?