August 1964 Electronics World
People old and young
enjoy waxing nostalgic about and learning some of the history of early electronics. Electronics World
was published from May 1959 through December 1971. See all
Electronics World articles.
It was not until 1963 that the International Committee of Weights
and Measures (CIPM) adopted the
cesium clock as the world scientific community's standard time
reference. It boasted an accuracy that kept it within 1.1 parts
in 100 billion, meaning it would not gain or lose more than
a second in 3 thousand years. To show how far technology has
advanced since 1963, in April of 2014 the National Institute
of Standards and Technology (NIST)
launched a new atomic clock called
NIST-F2 (also cesium-based)
to serve as a new U.S. civilian time and frequency standard.
NIST-F2 would neither gain nor lose one second in about 300
million years - a factor of 10 thousand.
According to the U.S. Navy's official
website, the Internet time reported on my computer was 1 minute
and 37 seconds behind official U.S. time. This is not surprising
since by default Windows only
re-syncs with the network once every 7 days. That means
my computer clock isn't much more accurate than a cheap old
spring-driven wind-up watch. Going into the clock settings window
and clicking on the "Update" button brought the computer clock
back into sync with the
NIST server clock.
Frequency & Time Standards
By George E. Hudson, Asst. Chief
Radio Standards Physics Division, National Bureau of Standards
The work of the NBS on atomic frequency
standards and their accuracy. Included is a description of the
U.S. Frequency Standard, which can control clocks that could
run 3000 years without gaining or losing over a second.
In recent years, the National Bureau of Standards has
been conducting developmental research on the precise measurement
and generation of frequency and time. For radio communications,
the tracking of satellites, the control of long-range rockets,
and astronomical observations, timing accuracies of one part
in a billion or better will be required in the future.
Many scientific and engineering activities rely on regular
radio transmissions of standard frequencies from Bureau stations
WWV, WWVB, WWVL, and WWVH and from the Navy's NBA to provide
high accuracies. These broadcasts are based, in part, on astronomical
observations related to the earth's rotation as made by the
U. S. Naval Observatory. However, to meet the ever-increasing
need for even greater accuracy, the Bureau has been investigating
atomic frequency standards, which are a thousand times more
precise for time-interval determinations than the earth's rotation.
A Frequency-Time Standard
Let us first consider a device which actually realizes the
limits of accuracy with which we are able to generate and determine
frequencies - and times-for the present. This is the United
States Frequency Standard, illustrated on the cover and in the
various accompanying pictures and diagrams. It is located in
Boulder, Colorado, in the Radio Standards Laboratory of the
Institute for Basic Standards - one of the four Institutes which
now comprise the National Bureau of Standards.
For some time now the primary standard for frequency in the
United States has been an atomic one. Up until now the most
precise such standard has used a cesium beam, undergoing a hyperfine
dipole structure transition using magnetic resonance as its
controlling element. In effect, the cesium atoms are constantly
spinning in such a way as to produce magnetic poles having a
certain orientation. Under certain conditions of excitation,
the atoms appear to suddenly "flip over" or reverse their spin
direction and the magnetic poles reverse.
The cesium beam frequency standard is essentially an atomic
beam spectrometer which emits a signal only when the frequency
of radiation introduced through a waveguide into the cavity
through which the beam is passing is precisely equal to the
resonance transition frequency. This frequency is actually in
the microwave region. The presence or absence of the signal
indicates whether the frequency of the radiation, itself generated
by a quartz-crystal oscillator driving a frequency multiplier
chain, is within the allowable tolerance limits.
In practice, the quartz signal generator is manually or automatically
varied over a narrow band to find the "center" frequency. When
the spectrometer output is at a peak, the signal generator frequency
is known within ±0.1 cps or 1.1 parts in 1011 (a
hundred billion). Automatic equipment is used to control the
signal generator so that the spectrometer output stays at the
maximum. This provides a signal of known and nearly constant
frequency for as long as the device can be kept running. As
the separations of the quantum states of an isolated atom are
fundamental constants of nature, they provide a stable, reproducible
standard of frequency and time interval when the atomic beam
resonance technique is used.
Previously, the most uniform time-intervals available were
those derived from astronomical observations of the rotation
of the earth relative to the fixed stars. This was corrected
to the orbital motion of the earth about the sun. The orbital
motion of the earth at 12 hours (noon, E.T.) on January 0, 1900
(December 31, 1899) is the basis of what is called Ephemeris
Time (E.T.). In 1956, the second of Ephemeris Time was adopted
as the fundamental unit of time by the International Committee
of Weights and Measures (CIPM) and this action was confirmed
by the General Conference on Weights and Measures in 1960.
Steps were taken by the Conference toward the adoption of
an atomic standard for time-interval. In early December of 1963,
the Consultative Committee for the Definition of the Second
of the CIPM met in Paris. This committee recommended, among
other things, that the cesium frequency should serve as an international
provisional standard on which the atomic second of time should
be based. Research, of course, should continue as to the best
atomic standard, but it recommended that the astronomical second
be dropped except as needed for celestial mechanics.
A time scale approximating Ephemeris Time can be made immediately
available by the use of atomic standards, quartz-crystal oscillators,
and counters. In terms of the ephemeris second, the frequency
of the cesium transition has been experimentally determined
to be 9,192,631,770±20 cps. The probable error, ±20 cps (or
2 parts in 109), results from the limitations on
the precision of the astronomical measurements. Consequently,
it is common practice now, and in line with the policy that
the atomic frequency transition studied at the Bureau of Standards
is the primary one for the United States, to take the frequency
of the transition to be exactly at 9,192,631,770 cps (or 9192.63177
To avoid inconsistency, we must regard the error quoted of
2 parts in 109 to be the error in determination of
the ephemeris second in terms of an exactly defined atomic second.
Such a definition is being currently considered. If the cesium
transition is adopted for this purpose, one second of atomic
time will consist of 9,192,631,770 cycles of the electromagnetic
radiation absorbed or emitted by a cesium atom in changing its
state. The physical picture in this particular case is that
the magnetic dipole moment of the cesium atom changes because
of a spin transition.
Measurement of a frequency or a time-interval in terms of
the cesium transition can be made with a typical precision of
1 part in 1012 and is not limited by the instrumental
difficulties involved in astronomical observations. See Fig.
1. Until official action is forthcoming it can not, however,
supplant the present definition of a scale for time based on
the non-uniform apparent motion of the sun.
Fig. 1. Simplified block diagram of the frequency
measuring system which utilizes the cesium beam U.S. Frequency
It seems natural to base the standard of time-interval on
the physical process that provides the most uniform and most
accessible interval. The precision of measurement for the atomic
standards is a hundred times better over a 15-second period
than astronomical measurements made over a period of 3 years.
Also under investigation as a standard of frequency is the
thallium atom, which has certain significant advantages over
the cesium atom in this application. However, thallium may have
some disadvantages in practical use. A thallium beam is now
in operation at the Boulder Laboratories to determine which
of the two atomic systems is more suitable.
At the present time, radio transmissions controlled by the
Bureau's master quartz-crystal oscillators are being monitored
with cesium beam frequency standards. Corrections for the 60
and 20 kc. standard frequency broadcasts from NBS radio station
VVWVB and WWVL at Boulder, Colorado, are being made regularly
and are available on request from the Broadcast Service Section
of the Bureau's Boulder Labs.
The time generation system now employed at the Boulder Laboratories
comprises the most accurate clock that man has ever known. Fig.
2 is a schematic diagram of this system, which generates the
NBS-A time scale. W. Atkinson, L. Fey, and J. Barnes have been
chiefly responsible for its development.
Fig. 2. A simplified block diagram of the
NBS-A atomic time-scale generator. The U.S. Frequency Standard,
shown in Fig. 1, beats against each of the five oscillators
in turn. After a time intercomparison, a weighted average is
Recent comparisons with other time scales - such as those
of Essen in England and particularly Bononomi in Switzerland
- the so-called TA1 time scale - have shown agreement
to within 1 part in 1011 over about a two-year period.
This means that these time scales diverge at the rate of 1 second
in 3000 years.
One of the time scales that is computed at the Naval Observatory
from astronomical observations is designated as UT-2. It is
an approximation to Ephemeris Time and represents an average
scale obtained by correcting the earth's
period of rotation on its axis. After correcting for longitude
effects on the astronomical observations which yield UT-0 directly,
one obtains UT-1. Then correcting for seasonal variations in
rotation speed one obtains UT-2. The U. S. Naval Observatory
obtains a composite "atomic" time scale - known as A1, by taking
a weighted average of various atomic standards from the United
States and European laboratories. The average is obtained from
broadcast transmissions. However, propagation effects lead to
many technical questions which remain to be investigated: The
Naval Observatory publishes the difference between A1 and UT-2
Scientists are shown above pouring liquid
nitrogen (at a temperature of -320°
F) from a Dewar into one of the cold traps in the NBS
"atomic clock" at the Boulder Laboratories. The nitrogen lowers
the temperature in the evacuated tube and causes stray gas molecules
and other impurities to condense around the cold end of the
trap. This reduces the chance of collision between the cesium
atoms and molecules of air, thus improving the vacuum and increasing
On the basis of the NBS-A atomic scale which has been kept
continuously since the Fall of 1957, it is known that the UT-2
scale, which really is simply an average representation of the
earth's rotation, shows that the earth has slowed down by about
2.9 seconds in this time. It is clear that the uniform atomic
time scale NBS-A generated at the Boulder Laboratories and based
on the atomic frequency standard must be heavily relied on for
the serious business of accurate time-keeping. It is also clear
that the United States is maintaining its eminence in this field.
Principles of Time Keeping
The principle of keeping time with an atomic clock is simple.
If one has a continuous record of the frequency, f, of any oscillator
and a concurrent record of the number of cycles of vibration,
or phase of the oscillator, then the time interval between two
given phases can be calculated. If one plots the quantity 1/2πf
versus the phase, the time interval is just the area underneath
the curve between the two phase values.
Now to do this one must have a continuously running oscillator,
or several oscillators (for redundancy and to increase statistical
reliability) and measure their frequencies periodically by comparison
with the atomic standard. Then one must also count the cycles
generated by the oscillators with a suitable counting circuit.
The NBS-A time scale is generated essentially in just this way.
To account for the tiny variations in frequency of the oscillators
used (four quartz ones and one rubidium one) intensive studies
have been and are being made of their physical and their statistical
The result is a clock whose accuracy in keeping time is only
limited by the accuracy of the U.S. Frequency Standard itself.
There is not one but five complete time keeping systems in the
fail-safe setup shown in Fig. 2. A weighted average of the five
leads to the NBS-A atomic time scale.
Signals from this system control the broadcast frequencies
and time furnished by the Bureau, under the direction of Mr.
D. Andrews. A recent innovation utilizing a phase-lock system
insures that the time signals emitted by WWV are in phase with
the NBS-A scale to within 25 microseconds. It is only at certain
intervals, as the earth gradually changes its rate of rotation,
that shifts must be broadcast so that all the signals simultaneously
are kept in agreement with UT-2 furnished by the Naval Observatory.
Between times, the rate of time ticks is atomically controlled.
Thus the Bureau of Standards must maintain two time scales simultaneously
under its NBS-A system.
An NBS scientist is shown retarding by 0.1
second on April 1, 1964 the clock which maintains Universal
Time (UT) at the Boulder Laboratories.
This 100-millisecond adjustment was necessary due to changes
in speed of rotation of the earth, as determined by astronomical
observation. Center clock (being adjusted)
is kept in close agreement with UT-2
(GMT). The top and bottom clocks, which are controlled
by the U.S. Frequency Standard and which are not adjusted, maintain
atomic time and were in agreement with UT-2 on Jan. 1, 1958,
as determined by the U.S. Naval Observatory. Since that time,
the clock maintaining UT has lost about 2.9 seconds relative
to atomic time. Since UT-2 is determined by the earth's rotation
(which is slowing down), the center
clock must be retarded periodically and progressively loses
time (150 x 10-10 sec./sec.
Atomic Beam Spectrometer
The atomic beam spectrometer, which is the heart of our frequency
and time measuring system, is shown in cross-section in one
of the photos. Neutral cesium atoms effuse from the oven and
pass through the non-uniform magnetic field of the deflecting
magnet. As the atoms act as miniature bar magnets (they are
said to have a magnetic dipole moment), a transverse force will
act upon them in this non-uniform field. The magnitude and direction
of this force depends upon which of the energy states the particular
atom is in. Of all the atoms effusing from the oven, suppose
those with nuclear and electron spins in the same direction,
say up, have their trajectories bent toward the axis. Atoms
on the other side of the center line with spins in opposite
directions, with the net being down, will have their trajectories
bent toward the axis also. Note that atoms with upward spins
and those with downward spins experience forces in opposite
directions. The upward-spin atoms and the downward-spin atoms
will cross the axis at the collimator slit, pass through the
slit, and enter the region of the B deflecting magnet.
AAs the B magnetic field is exactly like that of the A magnet,
it exerts the same transverse force on the atoms as does the
A magnet field. The upward-spin atoms will experience a downward
force as before and the downward-spin atoms will be acted on
by an upward force, as before. As the two sets of atoms are
now on opposite sides of the center line of the evacuated tube
(compared to their previous positions), these forces make their
trajectories more divergent.
However, if a radiation field is applied at just the proper
frequency (which matches the energy separation of the two quantum
states), transitions between the two states will occur in the
region between the A and B magnets. The spins will be reversed
in the two beams, and a quantum of energy will be either emitted
or absorbed. Since the sign of the magnetic moment has changed
in moving from the A magnet to the B magnet, the force on these
atoms will reverse its direction and the atoms will be re-focused
onto the axis at the detector. Thus, as the exciting radiation
is swept in frequency, the detected signal will increase and
reach a maximum at the critical frequency, and then decrease
as the radiation frequency is varied on either side.
We have tried to furnish some notion of the intense activity
which is now current at the Boulder Radio Standards Laboratory
of the National Bureau of Standards' Institute of Basic Standards.
Highly accurate atomic standards exist at the Bureau-but they
are constantly being improved and modified as the state of the
A hydrogen maser is under development and will soon be ready
for comparison with the cesium beam frequency standard. Commercial
varieties are now available which hold considerable promise
of being at least competitive with the presently used standard.
The adoption of the definition of the atomic second as the
international standard of time seems to be in the immediate
offing. The NBS atomic frequency standard itself controls the
rate of the very accurate atomic time scale and the broadcast
information, whereby industry and the scientific community
can feel confident that it is receiving the best and most up-to-date
services available in the area of frequency and time.
The author wishes to express his thanks for and acknowledgement
of the contributions made in the writing of this article to
his associates at NBS (BL).
Cut-away scale drawing of the NBS cesium beam frequency standard,
or "atomic clock," shows the location of the cesium oven at
the left end where the cesium atoms are emitted; the deflecting
magnets at each end which bend the beam and direct it through
the collimator slit to the detector; the microwave cavity and
waveguide, which introduce the microwave frequency; the three
liquid nitrogen cold traps used to remove, by condensation,
impurities not removed by the vacuum pump; the magnetic shields,
designed to eliminate the influence of terrestrial and other
outs ide magnetic fields; the hot-wire detector at far right
end where the beam impacts and is detected. This equipment is
essentially an atomic-beam spectrometer excited by a crystal
oscillator driving a frequency-multiplier chain. The radio
waves, oscillating a t a frequency close to the vibration-per-second
rate of the cesium atoms (9192.63177 me.I, are tuned slightly
so that they match the atomic rate. When they match, the radio
waves are absorbed; the radio frequency is then the same as
that of the cesium atoms-and a near absolute standard is available.
The device can measure an unknown frequency with an accuracy
of 1.1 parts in 1011 (one hundred billion). This
is equivalent to measuring the width of the United States with
less error than the thickness of this page. Its stability is
even higher-at least a few parts in 1012 per year-and at the
present time these figures represent one of the best achievements
in this field in the entire world.
Atomic Beam Frequency Standard (see magazine
Posted March 4, 2015