CESIUM ATOMIC CLOCKS
Only in the modern era of atomic clocks has timekeeping technology provided sufficient
accuracy to allow the successful construction of the Navstar Global Positioning System.
The evenly spaced timing pulses coming down from each Navstar satellite are
generated by an atomic clock that contains no gears or cogs. It's extraordinary
timekeeping abilities arise from the quantum mechanical behavior of certain specific
atoms (cesium, rubidium, hydrogen), which tend to have a single outer-shell electron.
Cesium atoms can exist in either of two principal states. In the high-energy state, the
spin axis of the lone outer-shell electron is parallel to the spin axis of the atom's
nucleus. In the low-energy state, the electron spins in a anti-parallel direction. For
cesium, the energy difference between the two spin states corresponds to an
electromagnetic frequency of 9,192,631,770 cycles per second. Thus, when a cloud of
cesium gas is struck by radio wave oscillating near that particular frequency, some of
the low-energy atoms will absorb one quantum of energy and, consequently, their outershell
electron will flip over and begin spinning in the opposite direction. The closer the
trigger frequency can be adjusted to 9,192,631,770 cycles per second, the more lowenergy
electrons will reverse their direction of spin.
The heart of the cesium atomic clock is a voltage-controlled crystal oscillator - a small
vibrating slab of quartz similar to the one that hums inside a digital watch. Small
variations in the voltage feeding a voltage-controlled crystal oscillator create
corresponding variations in its oscillation frequency. Any necessary adjustments are
handled by a feedback control loop consisting of a cesium atomic clock wrapped
around the quartz crystal oscillator.
A schematic diagram of the cesium atomic clocks carried onboard the GPS satellites is
sketched in Figure 1. First solid cesium is vaporized at 100 degrees Centigrade and
then it is routed through a collimator to form a steady stream of cesium gas, which, in its
natural state, consists of an equal mixture of high-energy and low-energy atoms.
FIGURE 1. The low-energy atoms floating around inside the resonating chamber of
this cesium atomic clock are hit with a radio wave as close as possible to 9,192,631,770
oscillations per second. Depending on the accuracy of that trigger frequency, larger or
smaller numbers of low-energy atoms will absorb one quanta of energy to become highenergy
atoms - which are subsequently converted into cesium ions by the hot-wire
ionizer (bottom right). The resulting ion current automatically adjusts the frequency of
the quartz crystal oscillator, which, in turn, creates more timing pulses and precisely
controlled electromagnetic waves.
A selector magnet is then used to separate the cesium atoms into two separate
streams. The high-energy atoms are discarded, the low-energy atoms are deflected into
a resonating cavity with precisely machined dimensions were they are hit with radio
waves generated by a voltage-controlled crystal oscillator coupled to a solid-state
frequency multiplier circuit. The closer the trigger frequency is to 9,192,631,770
oscillations per second, the more outer shell electrons will be inverted to produce highenergy
When the atoms emerge from the resonating cavity, they are again sorted by a selector
magnet into two separate streams. This time the low-energy atoms are discarded. The
high-energy atoms are deflected onto a hot-wire ionizer, which strips off their outer-shell
electrons to produce a stream of cesium ions. The resulting current is then routed into a
feedback control loop connected to the voltage controlled crystal oscillator whose
oscillation frequency is constantly adjusted to produce new radio waves.
By adjusting the frequency to maximize the ion current and dithering the oscillator to
make its frequency straddle the desired value of 9,192,631,770 oscillations per second,
the frequency stability of the quartz crystal oscillator can be maintained within one part
in 5 billion. Thus, the feedback control loop just described stabilizes the frequency of the
quartz crystal by a factor of 10,000 or so, compared with a free-running quartz crystal
with similar design characteristics.
RUBIDIUM ATOMIC CLOCKS
The rubidium atomic clocks carried on board the GPS satellites are, in many respects,
similar to the cesium atomic clocks, but there are also important differences in their
design. For one thing, the rubidium atoms are not used up while the device is keeping
time. Instead, the atoms reside permanently inside the resonating chamber. The
sensing mechanisms that monitor and adjust the clocks stability are also based on
distinctly different scientific principles.
As the rubidium atoms linger inside the resonating chamber, they are impacted with
electromagnetic waves whose oscillation frequencies are as close as possible to
6,834,682,613 oscillations per second (see Figure 2). As the transmission frequency is
adjusted closer and closer to that precise target value, larger numbers of rubidium
atoms will absorb exactly one quanta of energy. When they do, their spin-states
automatically reverse to convert them from low-energy to high-energy atoms.
FIGURE 2. Unlike the cesium atomic clock, the atoms in a rubidium atomic clock
remain always in the gaseous state. The trigger frequencies for the two devices are also
different. For a rubidium atomic clock the trigger frequency is 6,834,682,613 oscillations
per second. When the rubidium atoms inside the resonating cavity are hit with a trigger
frequency as close as possible to that value, larger numbers of them are converted from
low-energy atoms two high-energy atoms - that is, the spin axis of their lone outer shell
electron is parallel to the spin axis of the nucleus. Successful inversions are monitored
by shining a rubidium lamp through the resonating cavity. When larger numbers of
rubidium atoms have a been converted to the high-energy state, the gaseous cesium in
the resonating cavity is more opaque to rubidium light.
The rubidium atomic clock converges toward the desired frequency through a feedback
control loop whose status is continuously evaluated by shining the beam of rubidium
lamp through the resonating chamber. The gas inside the chamber becomes more or
less opaque to rubidium light, depending on how many of the rubidium atoms inside
have been successfully inverted. The intensity of the rubidium light passing through the
chamber is measured by a photo detector, similar to the electric eye in a digital camera.
The output from the photo detector is fed into a set of solid-state integrated circuits
rigged to make subtle and continuous adjustments to the frequency of the voltagecontrolled
crystal oscillator. Pulses from the crystal oscillator, which vibrates at 5 million
oscillations per second, are used in generating the evenly spaced C/A- and P-code
pulses broadcast by the satellites. A portion of the output from the voltage-controlled
crystal oscillator is also fed into a set of frequency multiplier circuits which generate the
desired 6,834,682,613 oscillation-per-second frequency, which is, in turn, routed into the
atomic clock's resonating chamber.
DEVELOPING ATOMIC CLOCKS LIGHT ENOUGH TO TRAVEL INTO
When the architecture for the Navstar navigation system was first being selected, many
experts argued convincingly that the atomic clocks should remain firmly planted on the
ground. The C/A- and P-code pulse trains, they believed, should be sent up to the
satellites through radio links for rebroadcast back down to the users down below. This
contention position was quite defensible because all available atomic clocks were big
and heavy, power-hungry, an extremely temperamental.
The best available cesium atomic clocks operated by the National Bureau of Standards,
for instance, were larger than a household deep-freeze, and they had to be tended by a
fretful army of highly trained technicians. However, emerging technology soon produced
much smaller and far more dependable atomic clocks. After years of intellectual
struggle, the cesium and rubidium atomic clocks on board the Navstar satellites have
turned out to be surprisingly small and compact. They also consume moderate
quantities of electricity and can operate for several years without failure. The rubidium
clocks carried aboard the Navstar satellites are roughly the same size as a car battery.
Each one weighs about fifteen pounds. The cesium atomic clocks are a little bigger.
They weigh thirty pounds each.
The earliest Navstar GPS performance specifications called for atomic clocks with
fractional frequency stabilities of one part in 1 trillion. The fractional frequency stability of
an atomic clock can be defined as the one sigma error pulse to pulse divided by the
duration between pulses. An atomic clock with a fractional frequency stability of one part
in 1 trillion is capable of keeping time to within one second over at interval of 30,000
Although this performance specification may seem rather stringent, the first few
spaceborne atomic clocks were two to five times more stable than required.
Consequently, the specification goal was eventually raised to two parts in 10 trillion.
The Navstar clocks have turned out to be surprisingly accurate and stable, but clock
reliability problems plagued the first few GPS satellites. On the average, only five on orbit months went by before a satellite component failure occurred. Almost always it was
an atomic clock component that failed. With intense design efforts, these problems were
eventually brought under control so that, today, the probability that at least one of the
four atomic clocks on the Block II satellite will still be operating at the end of its 7.5 year
mission is estimated to be 99.44 percent.