|Time Zone||Difference from UTC During Standard Time||Difference from UTC During Daylight Time|
|Pacific||-8 hours||-7 hours|
|Mountain||-7 hours||-6 hours|
|Central||-6 hours||-5 hours|
|Eastern||-5 hours||-4 hours|
Once your radio controlled clock has synchronized, it won’t decode the signal from WWVB again for a while. Most clocks only decode the signal once per day, but some do it more often (for example, every 6 hours). Those that decode the signal just once per day usually do it at midnight or in the very early hours of the morning, because the signal is easiest to receive when it is dark at both WWVB and at the site where the clock is located. In between synchronizations, the clocks keep time using their quartz crystal oscillators. A typical quartz crystal found in a radio controlled clock can probably keep time to within 1 second for a few days or longer. Therefore, you shouldn’t notice any error when you look at your clock display, since it will appear to be on the right second, even though it has probably gained or lost a fraction of a second since the last synchronization.
My clock doesn’t synchronize at all
- If your clock uses batteries, check them and replace if necessary.
- If you have a desk top unit, try rotating it 90 degrees. If you have a wall clock try mounting it on a wall perpendicular to the one it is currently on (e.g. if it is on a north-south wall try an east-west wall). The antennas are directional and you might be able to improve the signal strength by turning the antenna.
- Place the clock along a wall or near a window that faces Fort Collins, Colorado.
- Locate the clock at least 1 or 2 meters away from any computer monitors, which can cause interference (some monitors have a scan frequency at or near the WWVB carrier frequency of 60 kHz).
- If nothing else works, take the clock outdoors at night and power it down (remove the batteries or unplug it), then power it up again to force it to look for the WWVB signal. If it works outdoors but not indoors, you probably have a local interference problem inside your house or building. If it doesn’t work outdoors at night, its probably best to return it and try a different model.
- The shielding provided by a metal building might prevent the clock from working. For example, if you live in a mobile home or a house with steel siding, the clock might not work.
- If you think your clock is defective, ask the manufacturer or dealer about obtaining a replacement.
My clock is off by one or more hours
My clock is off by a few minutes or seconds
We switched to Daylight Saving Time, and my clock didn’t change
My clock switched to Daylight Saving Time, but we don't observe DST where I live
108 March 2010 Horological Journal
How Accurate is a Radio Controlled Clock?
by Michael A Lombardi.
The advertisements for radio controlled clocks and
wristwatches often make sensational claims about the
accuracy of the products. One overly enthusiastic writer
penned this memorable bit of ad copy:
While this particular advertisement contained more
hyperbole than most, it was not alone in referring to a radio
controlled clock (RCC) as an atomic clock, a claim made in
nearly all sales pitches and product literature. The claim, of
course, is false. The oscillator found inside an RCC is based
on the mechanical vibrations of a quartz crystal, typically
counting 32,768 vibrations of the crystal to mark one second.
A true atomic clock oscillates based on the energy transitions
of an atom and ‘ticks’ much faster. For example, the second is
defined internationally as the duration of 9,192,631,770
energy transitions of a cesium atom.
Although advertisers are wrong when they call an RCC an
atomic clock, they are correct in stating that an RCC benefits
from atomic timekeeping. An RCC periodically synchronizes
its quartz oscillator to a real atomic clock by receiving a time
signal from one of the radio stations listed in Table 1. Some
RCCs are capable of receiving just one station, and must be
within the coverage area of that station in order to work.
Others, including many wristwatches, are now capable of
receiving all of the stations in Table 1 and will synchronize to
the signal from the nearest station1.
Now that we’ve established that a RCC is not a real atomic
clock, but simply a quartz clock periodically synchronized by
radio, can we determine its true accuracy? The answer is yes,
but at least four questions first need to be answered:
·How accurate is the time kept at the radio station?
·How long does it take for the radio signal to travel from the
station to the RCC?
·When the signal arrives, how accurately is the RCC’s display
·How accurate is the RCC’s quartz crystal oscillator between
We’ll look at each question in turn, using the U.S. radio
station WWVB, Figure 1, as an example.
Q1. How accurate is the time kept at the radio station?
Time signal stations synchronize their clocks to Coordinated
Universal Time (UTC), the international standard for
timekeeping. No clock keeps UTC exactly because UTC is an
average time, calculated with data collected from hundreds of
atomic clocks located around the world. The calculations are
performed by the Bureau International des Poids et Mesures
(BIPM) in France. Laboratories such as the National Institute
of Standards and Technology (NIST) in the United States keep
local versions of UTC that closely agree with the BIPM’s
calculations. The NIST version of UTC, called UTC(NIST), is
generated by averaging an ensemble of cesium beam and
hydrogen maser clocks. The ensemble is periodically
calibrated using a cesium fountain clock called NIST-F1,
which serves as the primary time interval and frequency
standard for the United States2.
Figure 2 shows the time difference between UTC(NIST)
and UTC over a one-year period with the data points taken at
five-day intervals3. During the year, UTC(NIST) never varied
from UTC by more than 20 nanoseconds (0.000 000 020 s).
Thus, while there is technically a difference between
UTC(NIST) and UTC, the difference is miniscule and for all
practical purposes can be ignored.
The time signal stations listed in Table 1 are located some
distance away from the timing laboratories that control them,
typically in rural areas where there is enough space for their
antennas. For example, the NIST timing laboratories are in the
city of Boulder, Colorado, and WWVB is located in a rural area
about 78 km away. Therefore, UTC(NIST) in Boulder is not
directly connected to WWVB. Instead, WWVB has its own
Station Call Sign Frequency (kHz) Country Controlling Organization
BPC 68.5 China National Time Service Center (NTSC)
DCF77 77.5 Germany Physikalisch-Technische Bundesantalt
JJY 40, 60 Japan National Institute of Information and
Communications Technology (NICT)
MSF 60 United Kingdom National Physical Laboratory (NPL)
WWVB 60 United States National Institute of Standards and
Table 1. Time Signal Stations used by RCCs.
1. Aerial view of NIST time signal station WWVB.
‘We’re still perfecting Einstein’s theory. We must apologize
that our Atomic Watch loses 1 second every 20,000,000
years. Our scientists are working diligently to correct this
problem…’Horological Journal March 2010 109
clock, actually a group of cesium beam clocks, that are
steered to agree with UTC(NIST) in Boulder by making time
comparisons using satellites4.
Figure 3 shows the difference between UTC(NIST) and
WWVB station time over the same one year interval shown in
Figure 1 (here the data points are at 1-day intervals). Note
that the station clock never varied from UTC(NIST) by more
than 35 nanoseconds (0.000 000 035 s). Again, for all
practical purposes, the differences are so tiny they can be
ignored. In fact, the frequency offset of the station clock is
less than 1 × 10-15. If this frequency offset were held constant,
it would take more than 30 million years before the
accumulated time error reached one second. If the
advertising writer was referring to the station clock, and not
the RCC, you could actually argue that they were being
Q2. How long does it take for the radio signal to travel
from the station to the RCC?
The exceptional accuracy of the station clock becomes a
moot point once you start to consider the problem of path
delay. There is some path delay before the signal even leaves
the radio station. Once the time signal is generated from the
station clock, it passes through a transmission system that
includes the radio transmitters, the antenna feed lines, and the
At WWVB this delay is about 0.000 017 s5 or about one
thousand (103) times larger than the time difference between
the station clock and UTC. Even so, it is still too tiny to matter
to RCC users, so the station does not advance its signal to
compensate for the transmission delay.
Once the signal leaves the transmitter and enters free
space, it travels at the speed of light to the RCC,
approximately 0.000 003 336 seconds per kilometer. If you
know the location of your RCC and the location of the time
signal station, you can calculate the distance between the
transmitter and the receiver, and estimate the amount of this
delay. For example, a 3000 kilometer path would delay the
time signal by about 10 milliseconds (0.01 s). This assumes,
of course, that the signal travels along the ground and covers
the shortest possible distance between the station and the
For the low frequencies used by the time signal stations in
Table 1, a groundwave path can be assumed for short
distances, perhaps up to 1500 km, but at longer distances the
signal might bounce off the ionosphere (skywave) and take
slightly longer to arrive. Even so, it is safe to assume that the
path delay will be less than 20 milliseconds (0.02 s) because
the signal will rarely be usable at distances of more than 5000
km, and any additional delays introduced by skywave will be
relatively small. However, for RCCs located on the east and
west coasts of the United States the path delay is roughly 0.01
s, or about one million times (106) larger than the time
difference between the station clock and UTC.
Q3. When the signal arrives, how accurately is the RCW’s
The time signal stations in Table 1 send information using a
very simple modulation scheme. For example, WWVB
broadcasts a continuous 60 kHz sine wave signal, but drops
the carrier power by about 98% (17 dB) every second,
restoring it to full power a fraction of a second later. This
power drop serves two purposes. Its first purpose is to send
bits of a binary time code, as the length of time that the power
is held low determines whether a bit is a 0, a 1, or a frame
marker. A complete time code is 60 bits long and thus requires
60 s to transmit. The second purpose of the power drop is to
send an on-time marker (OTM) that is synchronized with
UTC(NIST). The OTM is the first 60 kHz cycle that is sent at
reduced power. For example, when the power is held low for
0.2 seconds to signal a 0 bit, 12,000 cycles are transmitted at
low power, but only the first of these reduced power cycles is
the OTM. In theory, an RCC should be able to synchronize to
within one half of the period of 60 kHz, or within ±0.000 008 s
of the OTM.
In practice, there are other problems. One is that the quartz
crystal oscillator inside an RCC runs at a frequency of about
half of the incoming radio signal (32.768 kHz), and thus even if
the correct OTM were found, the quartz crystal could still be
synchronized only to within half the period of its ow
frequency, or to within about ±0.000 015 s.
2. UTC(NIST) Time Scale compared to Coordinated Universal
3. WWVB Station Clock compared to UTC(NIST) Time Scale.
4. WWVB OTM as seen on an oscilloscope.110 March 2010 Horological Journal
As it turns out, that doesn’t matter because it is very difficult
for the RCC to find the correct OTM. The WWVB waveform
(Figure 4) has a long exponential decay that is related to the
period of the antenna bandwidth. This makes it very difficult to
determine exactly where the carrier power drop began.The
actual OTM is located in the flat part of the waveform, before a
noticeable drop in amplitude can be seen on an oscilloscope.
Finding the OTM becomes even more difficult when the signal
is weak or noisy, or when both groundwave and skywave
signals are received. Therefore, the OTM synchronization
accuracy is probably limited to about 1 millisecond (0.001 s),
although the actual accuracy will depend upon the quality of
both the received signal and the RCC’s digital signal
processing (DSP) firmware.
Processing delays also occur while the RCC’s display is
being synchronized to the right time. These delays include
DSP software delays, the time required to retrieve the data
from the microprocessor unit and to process and output the
data, and the response time of the stepping motor used by
analog clocks or the LCD display used by digital clocks. The
processing delays can exceed 100 milliseconds (0.1 s), but
the RCC manufacturer normally takes them into account, and
advances the display to compensate. Even so, the amount of
delay compensation will not be perfectly estimated, and a
synchronization error of 10 milliseconds (0.01 s) is probably
Table 2 summarizes the accuracies discussed in the
answers to the first three questions. Based on this analysis, it
seems reasonable to expect that an RCC will be accurate to
within 30 milliseconds (0.03 s) at the time of synchronization,
with the path delay and synchronization errors the only two
factors that really matter. However, a much larger time error is
likely to accumulate during the interval between
synchronizations, as will be seen in the answer to question 4.
Q4. How accurate is the RCC’s quartz crystal oscillator
Some RCCs schedule only one synchronization attempt per
day, at 2 am, for example. If the synchronization attempt fails,
they will wait 24 hours before trying again. Others are
designed to schedule multiple attempts (at 2, 3, 4, and 5 am,
for example). Some RCCs will attempt synchronization during
each of their scheduled times, synchronizing again at 3 a.m.
even if the 2 am attempt was successful. Other RCCs will skip
the remaining attempts on the schedule after synchronization
is achieved and wait until the next day to try again. For these
reasons, the interval between synchronizations is typically
either 24 hours, or just a few hours less than 24 hours.
It might seem reasonable to receive the radio signal more
often because the signals are always being broadcast, and a
synchronization attempt could be made at any time. However,
the number of attempts is limited for several reasons. One
reason is that the signal is much stronger at night and
synchronization attempts during daytime hours are far more
likely to fail. A second reason is that many RCCs are battery
powered, and fewer synchronization attempts means longer
battery life. The most important reason is simply that one
synchronization per day is usually all that is needed to keep
an RCC accurate to within a fraction of a second of UTC.
NIST has published guidelines recommending that RCCs
keep time between synchronizations to within ±0.5 s of
UTC(NIST). If this requirement is met, the time displayed on
an RCC will always be correct when rounded to the nearest
second6. At least one manufacturer of RCCs specifies the
accuracy of their quartz crystals as ±15 s per month, which is
essentially the same thing as ±0.5 s per day. This type of
accuracy is now commonplace in low cost quartz oscillators7.
It translates to a frequency offset of about 6 × 10-6, more than
one billion (109) times less accurate than the station clock!
While ±0.5 s per day is a reasonable benchmark,
synchronization to within ±0.2 s is more desirable and
achieved by many RCCs. An 0.2 s error is unlikely to be
noticed by the human eye if an RCC is checked against an
independent time reference, whereas a 0.5 s error might be.
Of course, an accumulated time error of 0.2 s between
synchronizations is still much larger than the other factors
listed in Table 2. This means that an RCC will be most
accurate immediately after a successful synchronization and
will become less accurate from that point forward until the next
To demonstrate this, the accumulated time error of an
analog radio controlled watch was measured between
synchronizations. A sensor was used, Figure 5, that could
record the beat rate of the watch’s stepping motor pulses. At
the time of synchronization, it is assumed that the time error
was 0, although in reality the synchronization accuracy is
limited by the factors shown in Table 2. The assumption had
to be made because the second hand stops during
synchronization attempts, which in turns stops the sensor from
collecting data. However, since the watch was operating near
WWVB (the path delay was about 0.000 26 s), it was probably
accurate to within 0.01 s of UTC(NIST) at the time of each
The measurement results are shown in Figure 6. The watch
synchronizes five times daily, at midnight, 1, 2, 3, and 4 a.m.
The midnight synchronization occurs after the watch has run
for about 20 hours without adjustment and the amount of the
correction is about 0.45 s. The other four synchronizations
occur after the watch has run unadjusted for about one hour,
and the correction is slightly larger than 0.02 s. This particular
Source of inaccuracy Seconds
Radio station clock 0.000 000 050
Transmission system delay 0.000 017
Path delay (worst case) 0.020
OTM selection 0.001
Synchronization errors 0.010
RCC inaccuracy at time of
Table 2. Sources of RCC synchronization inaccuracy.
5. Measuring an analog radio controlled wristwatch between
synchronizations.Horological Journal March 2010 111
watch narrowly meets the requirement of always remaining
within ±0.5 s of UTC(NIST). Its performance can be
considered typical of many RCC products.
This brings us back to our original question: How accurate is
a radio controlled clock? The short answer is they should
always be accurate to within one second of UTC, assuming
that they synchronize at least every other day and that their
quartz oscillator is of reasonable quality. The key, of course, is
successful synchronization to a time signal station, because
without that advantage an RCC is just a run of the mill quartz
clock. Instead of the ‘20,000,000 years’ mentioned by the ad
writer, the watch measured in Figure 6 would lose a second in
about two days if it were unable to receive a radio signal and
On the other hand, if a RCC could miraculously be made to
run for ‘20,000,000 years’ and was somehow able to
synchronize once every 24 hours, it would never lose a full
second. Such are the benefits of atomic timekeeping.
The author is an employee of a US Institute making this
article a contribution of the United States government, and not
subject to copyright. The illustrations and descriptions of
commercial products are provided only as examples of the
technology discussed, and this neither constitutes nor implies
endorsement by the NATIONAL INSTITUTE OF STANDARDS
AND TECHNOLOGY (NIST). The author thanks Matt Deutch
and Glenn Nelson of NIST Radio Station WWVB and Etsuro
Nakajima of Casio Computer Co. Ltd. for many helpful
1. M. A. Lombardi, Radio Controlled Wristwatches, HJ,
148(5), May 2006, pp. 187-190.
2. M. A. Lombardi, T. P. Heavner, and S. R. Jefferts, NIST
Primary Frequency Standards and the Realization of the SI
Second, Measure: The Journal of Measurement Science, 2(4),
December 2007, pp. 74-89.
3. BIPM Circular T (numbers 249 through 260), available at:
4. J. Levine, Realizing UTC(NIST) at a Remote Location,
Metrologia, 45, pp. S23-S33, December 2008.
5. G. K. Nelson, M. A. Lombardi, and D. T. Okayama, NIST
Time and Frequency Radio Stations: WWV, WWVH, and
WWVB, NIST Special Publication 250-67, 161 pages, January
6. WWVB Radio Controlled Clocks: Recommended
Practices for Manufacturers and Consumers, NIST Special
Publication 960-14, 68 pages, August 2009.
7. M. A. Lombardi, The Accuracy and Stability of Quartz
Watches, HJ, 150(2), February 2008, pp. 57-59.
6. The performance of a typical radio controlled wristwatch.
Michael Lombardi has worked in the Time and Frequency
Division of the National Institute of Standards and
Technology (NIST) since 1981. NIST is a national metrology
institute that is responsible for maintaining the national
standards of frequency and time interval for the United
Michael is currently in charge of the remote time and
frequency calibration programs at NIST. He also works on
numerous other time and frequency projects, including a
time network that continuously compares the time standards
of North, Central, and South America. He has published
over 80 papers related to time and frequency. A member of
IEEE, and an associate editor of Measure: The Journal of
Measurement Science, Michael lives in Colorado with his
wife and two daughters.