1

United States Department of Commerce

j National Institute of Standards and Technology
NAT L. INST. OF STAND & TECH R.I.C.

NIST

I A111D3 MSDD3M
NIST Special
PUBLICATIONS

Publication 559 (Revised 1990)

Time and Frequency Users Manual

George Kamas
Michael A. Lombardi
NATIONAL INSTITUTE OF STANDARDS &
TECHNOLOGY
Research Information Center
Gakhersburg, MD 20699
NIST Special Publication 559 (Revised 1990)

Time and Frequency Users Manual

George Kamas
Michael A. Lombardi

Time and Frequency Division

Center for Atomic, Molecular, and Optical Physics
National Measurement Laboratory
National Institute of Standards and Technology
Boulder, Colorado 80303-3328

(Supersedes NBS Special Publication 559 dated November 1979)

September 1990

U.S. Department of Commerce

Robert A. Mosbacher, Secretary

National Institute of Standards and Technology

John W. Lyons, Director
National Institute of Standards U.S. Government Printing Office For sale by the Superintendent
and Technology Washington: 1990 of Documents
Special Publication 559 (Rev. 1990) U.S. Government Printing Office
Natl. Inst. Stand. Technol. Washington, DC 20402
Spec. Publ. 559 (Rev. 1990)
160 pages (Sept. 1990)
CODEN: NSPUE2
 ABSTRACT
This book is for the person who needs information about making time and
frequency measurements. It is written at a level that will satisfy those with a casual
interest as well as laboratory engineers and technicians who use time and frequency every
day. It includes a brief discussion of time scales, discusses the roles of the National

Institute of Standards and Technology (NIST) and other national laboratories, and
explains how time and frequency are internationally coordinated. It also describes the
available time and frequency services and how to use them. It discusses the accuracies
that can be achieved with the different services, and the pros and cons of using various
calibration methods.

Key words: frequency calibration; high frequency; Loran-C; low frequency; oscillators;
relative frequency; satellite broadcasts; standards; time calibration; time code; time
interval.

iii
 PREFACE

This book intended to assist users of time and frequency calibration services in
is

the United States and throughout the world. The book deliberately avoids using complex
derivations or mathematical analysis. Instead, simple explanations have been given in the
hope that more people will find the material useful.

Since the topics in this book are not covered in great depth, this book is intended
as an overview of time and frequency calibration methods, and not as a comprehensive
reference work. If you need more detailed information about a calibration method
described in this book, please contact NIST or a manufacturer of time and frequency
equipment.

v
 CONTENTS

ABSTRACT iii

PREFACE v

Chapter 1 - AN INTRODUCTION TO
TIME AND FREQUENCY 1

WHO NEEDS TIME AND FREQUENCY? 2

WHAT ARE TIME AND FREQUENCY? 5
WHAT IS A STANDARD? 6
THE NIST STANDARDS FOR TIME AND FREQUENCY 7
HOW TIME AND FREQUENCY STANDARDS ARE
DISTRIBUTED 8
FREQUENCY CALIBRATIONS 9
TERMS USED 11

Chapter 2 - TIME SCALES 15

SOLAR TIME 15
ATOMIC TIME 17
COORDINATED UNIVERSAL TIME 18
TIME ZONES 19
USING TIME SCALES FOR NAVIGATION AND
ASTROMONY 19
INTERNATIONAL COORDINATION OF FREQUENCY
AND TIME ACTIVITIES 21
The Role of the National Institute
of Standards and Technology (NIST) 21
The Role of the U.S. Naval
Observatory (USNO) 22

Chapter 3 - AN INTRODUCTION TO
FREQUENCY SOURCES 25

FREQUENCY SOURCES AND CLOCKS 25

THE PERFORMANCE OF FREQUENCY SOURCES 28
PRIMARY AND SECONDARY STANDARDS 30
QUARTZ OSCILLATORS 31
Temperature and Aging of Crystals 33

vii
 Quartz Oscillator Performance 35
ATOMIC OSCILLATORS 36
Rubidium Gas Cell Oscillators 36
Cesium Beam Oscillators 37
Hydrogen Masers 38
COMPARISON OF FREQUENCY SOURCES 40

Chapter 4 - MEASURING FREQUENCY 41

THE THEORY OF FREQUENCY MEASUREMENTS 41

Time Interval Measurements 42
Using a Measurement System 44
A COMPUTER-CONTROLLED MEASUREMENT SYSTEM 46
The Reference Frequency 47
The Frequency Divider 48
The Time Interval Counter 48
The Computer 49
Output of the Measurement System 50
Day to Day Operation of the System 51
Recordkeeping 53
TRACEABILITY FOR FREQUENCY MEASUREMENTS 53
WHAT A FREQUENCY CALIBRATION MEASURES 55
SUMMARY 60

Chapter 5 - CALIBRATIONS USING HF RADIO BROADCASTS 61

BROADCAST FORMATS 61
WVW/WWVH 61
CHU 63
HF STANDARD FREQUENCY AND TIME SIGNAL
BROADCASTS 64
RECEIVER SELECTION 65
ANTENNA SELECTION 66
Antennas for the 2.5- to 4-MHz Range 67
Antennas for the 4- to 5-MHz Range 68
Antennas for the 7- to 10-MHz Range 68
Antennas for the 14- to 15-MHz Range 68
Antennas MHz
for 20 69
USING HF BROADCASTS FOR TIME CALIBRATIONS 70
Measuring the Receiver Time Delay 71
Using an Adjustable Frequency Source to
Trigger the Oscilloscope 73
Delayed Triggering: An Alternate Method
That Doesn't Change the Clock Output 75
Using Oscilloscope Photography for Greater

viii
 Measurement Accuracy 78
USING THE WWV/WWVH TIME CODE 78
Time Code Format 78
USING HF BROADCASTS FOR FREQUENCY
CALIBRATIONS 80
Beat Frequency Method 81
Oscilloscope Pattern Drift Method 84
Frequency Calibrations by Time Comparisons 85
MEASURING RADIO PATH DELAY 85
SUMMARY 88

Chapter 6 - CALIBRATIONS USING LF AND VLF RADIO

BROADCASTS 89

GENERAL INFORMATION ABOUT LF AND VLF

STATIONS 90
LF and VLF Antennas 90
Signal Formats 90
Propagation Characteristics and Other
Phase Changes 91
Field Strengths of LF and VLF Stations 92
Monitoring Station Availability 93
FREQUENCY CALIBRATIONS USING WWVB 94
Computing Relative Frequency from a WWVB
Chart Recording 95
TIME CALIBRATIONS USING WWVB 96
Format of the WWVB Time Code 98
FREQUENCY CALIBRATIONS USING THE OMEGA
NAVIGATION SYSTEM 100
Operating Characteristics of Omega 101
Propagation Characteristics of Omega 103
OTHER LF AND VLF TIME AND FREQUENCY
STATIONS 105
TIME AND FREQUENCY CALIBRATIONS USING
LORAN-C 107
Format of Loran-C Broadcasts 109
Loran-C Reception 111
Getting Time From Loran-C 112
Frequency Calibrations Using Loran-C 113
SUMMARY 114

ix
Chapter 7 CALIBRATIONS USING
SATELLITE BROADCASTS 115

USING THE GOES SATELLITE BROADCASTS 116

GOES COVERAGE 117
GOES SIGNAL CHARACTERISTICS 118
GOES TIME CODE FORMAT 118
GOES TIME CODE PERFORMANCE 119
GOES EQUIPMENT 120
TIME CALIBRATIONS USING GOES 120
PRECAUTIONS FOR GOES USERS 122
THE TRANSIT NAVIGATION SYSTEM 123
GLOBAL POSITIONING SYSTEM (GPS) 124
Using GPS 127
GPS Performance 130
SUMMARY 130

Chapter 8 - CALIBRATIONS BY TELEPHONE 131

THE NIST (VOICE) TELEPHONE TIME-OF-DAY SERVICE 131

THE AUTOMATED COMPUTER TIME SERVICE (ACTS) 131
The ACTS Time Code Format 132
Performance of ACTS 133

APPENDIX - SUMMARY OF TIME AND FREQUENCY

RADIO CALIBRATION SOURCES 135

GLOSSARY 137

INDEX 143

x
 Chapter 1 - AN INTRODUCTION TO TIME AND FREQUENCY

This book is about time and frequency. It describes time interval, time of day, and
frequency calibrations. It explains the time and frequency broadcast services that are
available in the United States and other countries, and how you can use them. Your
requirements may be as modest as setting your watch, or as involved as calibrating precision
oscillators. In either case, you should find something of interest in this book.

Without realizing it, we use time and frequency every day. Knowing the correct time
allows us to function in an orderly manner. We need to know what time to meet a friend
for lunch, or to arrive at school or work. It's all right to get to church early, but it's

embarrassing to walk in during the sermon. And we'd all be disappointed if we missed our
airplane after months of planning a Hawaiian vacation.

In these examples, knowing the correct time to within a few minutes is usually
adequate. But even a few seconds can sometimes be quite important. For instance, every
day hundreds of people drop nickels, dimes, and quarters into parking meters, coin-operated
washers and dryers, and other machines that "keep" time. Businesses pay thousands of
dollars for the use of a computer's time. We all pay telephone bills based on the time we
spend using the telephone. These activities all require accurate time. Fifteen minutes on
a parking meter should really be 15 minutes and not 14. An error in the meter's timer
could mean a parking ticket. If we talk on the telephone for 7 minutes, we don't want to
be billed for 9 or 10.

Frequency is just as important as

time. Every timer has a frequency source
inside. In fact, time as we know it
(seconds, minutes, and so on) is simply the
accumulation of other more frequent time
pulses. Thus "time of day" is a label, a
name tag given to just one of many possible
pulses. But all of these pulses represent a
frequency. One pulse per second (like a
clock tick) has a frequency of 1 hertz.
From this we can build up a "day" by
counting 86 400-second ticks, or we can
start with a much higher frequency of a
million pulses per second and count these
faster pulses a million at a time to get a
pulse rate of once per second. These
examples show that "frequency" describes
the rate at which events occur, and that
"time" is the name or label that we give to
an interval. In fact, time interval is the

1
correct way to label a period of elapsed time. To measure frequency we will use time
interval. In this book there will be many references to such procedures (counting, timing,
and the calibration of devices) used to produce different pulse rates.

Accurate pulse rates or frequencies are used every day. TV stations must transmit
on the exact frequencies assigned by the Federal Communications Commission (FCC). Power
companies supply power to homes and offices at 60 Hz (cycles per second), so that electric
clocks keep the correct time. A stereo turntable needs to spin a record at 33-1/3 rpm
(revolutions per minute) instead of 32 or 34 rpm, so that the musical tones are reproduced
accurately.

Our voices range in frequency from about 87 Hz (bass) to

1175 Hz (soprano). Our ears can detect sounds ranging from
a telephone dial tone at about 400 Hz, a smoke alarm at 600
Hz, and a dentist's ultrasonic drill at 10 000 Hz. A piano has
notes from 27.5 to 4186 Hz. "A" above middle "C" in the
musical scale is 440 Hz, a frequency often used to tune musical
instruments.

Unlike the other basic physical standards (length, mass,

and temperature), frequency and time (or more properly, time
interval) are relatively easy to measure with great precision.
For example, a length measurement of 2.5 micrometers (0.0001
inch) is difficult to make. On the other hand, time
measurements of 1 nanosecond (0.000000001 second) are
commonplace in the world of science.

WHO NEEDS TIME AND FREQUENCY?

As we saw in the introduction, we all are involved with time and frequency to some
extent. For most of us, knowing the time to within 1 minute or so is usually enough.
However, some users need to know the exact second or even millionth of a second
(microsecond). Let's look at some of the more sophisticated users of time and frequency:

CELESTIAL NAVIGATORS need time of day to determine their exact location. An error
of 2 seconds could cause a ship to miss its destination by about 1 kilometer.

Other SHIPPERS and BOATERS need even more accurate time of day. When ships use
sophisticated electronic navigation systems, an error of only 3 microseconds could cause
unwanted errors.

2
POWER COMPANIES compare their generator frequency power flow at
to control electric
exactly 60 Hz. If they didn't, electric clocks wouldn't keep the correct time. They need
precise time to monitor the power grid to help prevent "brownouts" and massive power
failures. They also use time to help locate power outages and trouble on the lines. Time
is also important for keeping track of power flow for billing purposes.

RADIO & TV STATIONS must broadcast signals exactly on their assigned radio frequencies.
They also need accurate time of day so they can join their networks at the right instant.

The MEDICAL PROFESSION uses time and frequency sources to calibrate medical test
equipment, for date and time printouts in hospital care units, and for timing therapy and
observational procedures used in daily health care.

The OIL INDUSTRY needs accurate timing to help automate oil well drilling, especially
offshore.

JEWELERS & CLOCK/WATCH MANUFACTURERS need to set digital watches and clocks
to the correct time of day and rate before they leave the factory.

RAILROADS use time-of-day radio signals to set watches and clock systems, so that trains
arrive and depart on schedule.

The COMPUTER INDUSTRY needs

accurate time intervals for billing
purposes and for synchronizing
communication between systems many
miles apart.

POLICEMEN need accurate time to

check stopwatches and radar guns used
to clock speeders. They use frequency
sources to calibrate radar devices used i©
for traffic control.

SURVEYORS use time signals to

measure distance. For
example, 3
microseconds translate into 1 kilometer
in distance when using electronic
instrumentation.

3
MANUFACTURERS need time and frequency signals and sources to calibrate counters,
frequency meters, test equipment, timers used in electric appliances, and a variety of other
equipment.

The TRANSPORTATION INDUSTRY needs

accurate time sources to synchronize clocks
used in public transportation systems, and for
-(otfo) foV- vehicle location, dispatching, and control.

SPORTSMEN use elapsed time. Sports car rallies are timed to 0.01 second or less. So
are many track and field events.

The TELEPHONE INDUSTRY needs accurate clock time for billing and telephone time-
of-day services. They need accurate frequency signals so that long-distance phone calls
don't become garbled during transmission.

The COMMUNICATIONS INDUSTRY depends on accurate frequency control for its ability
to deliver messages to its users. Time is needed for labeling the time of occurrence of
important messages.

The MUSIC INDUSTRY uses frequency, obtained from tuning forks and crystal oscillators
and radio services (the 440 Hz tone, for example) to tune pianos, organs, and other musical
instruments.

The TELECOMMUNICATIONS INDUSTRY needs time intervals accurate to 1 microsecond

to synchronize satellite and communications terminals spread over wide geographical areas.

The AVIATION/AEROSPACE
INDUSTRY needs accurate time
interval sources for aircraft traffic
control systems and for
synchronization at satellite and
missile tracking stations. The
Federal Aviation Administration (FAA) records accurate time of day on its audio tapes along
with the air-to-ground communications from airplanes. Having an accurate record of when
particular events occurred can help determine the cause of a plane crash or equipment
failure.

4
ASTRONOMERS use time of day for
observing astronomical events, such as
lunar occultations and eclipses.

Accurate time setting is required for

MASTER CLOCK SYSTEMS in large insti-
tutions, such as airports, hospitals, large
factories, and office buildings, so that all

clocks in the system read the same time.

The MILITARY uses accurate time of day

to synchronize clocks on aircraft, ships,
submarines, and land vehicles. They also
use it to synchronize communications
between bases, and for radio navigation.

GEOPHYSICISTS/SEISMOLOGISTS studying lightning, earthquakes, weather, and other

geophysical disturbances need accurate timing to obtain data synchronously and
automatically over wide geographical areas. They use it for labeling geophysical events.
Other SCIENTISTS use time sources for controlling the duration of physical and chemical
processes.

WHAT ARE TIME AND FREQUENCY?

The introduction and examples show that time and frequency are closely related.
Frequency simply the rate at which things happen. For example, a human pulse beats
is

at a rate of60 to 100 times per minute. U.S. power companies vary the voltage at a rate
of 60 times per second. Each of these rates is a frequency. The basic unit of frequency is
the hertz (Hz). The hertz was named after Heinrich Hertz, an early radio scientist. If
something happens once per second, we say that it occurs at a frequency of 1 Hz.

As you might have guessed, we use a frequency source to get time interval and time
of day. For example, inside many wristwatches you will find a tiny quartz crystal a few
millimeters in length. This sliver of quartz vibrates at a frequency of about 32 000 times
a second (32 kilohertz, or 32 kHz). A mechanism inside the watch counts these vibrations.
When 32 000 vibrations (or time intervals) have occurred, 1 second has elapsed. At this
point, the watch will either update its digital display or move the "hands" of the clock.

In fact, the international definition of the second is based on counting a frequency.

A second is a time interval, and it is defined in terms of the cesium atom. This is explained

5
in some detail elsewhere in this book. For now, we can say that a second consists of
counting 9 192 631 770 periods of a property associated with the cesium atom. In other
words, the tiny time intervals of the oscillating cesium atom are totaled to get a new
(longer) time interval we call the "second."

Since we used the cesium "vibrations" to

getan interval of a second, we can, of course,
use time interval to describe frequency. If you
knew exactly when a second started and ended,
you could count events in that second and
define a frequency. For example, if you
electronically counted 1000 cycles of a certain
signal in 1 second, you would know that the
frequency of that signal was 1000 Hz, or 1
kilohertz (1 kHz). The definition of frequency
is based on properties associated with the
cesium atom. The atom causes vibrations or
oscillations. After some 9 billion of these have
occurred, we give that interval the name
"second," which is the time interval equal to
over 9 billion smaller time intervals. This is
then continued through minutes, hours, days,
years, and so on. Time interval is a very
important quantity. It defines time of day and
duration, and is the basis for using the term
"frequency."

WHAT IS A STANDARD?
When calibrations are made, a reference standard is needed. The calibration is then
referenced to the standard. According to Webster, the definition of a legal standard is
"something set up and established by authority, custom, or general consent as a model or
example." In the United States, the National Institute of Standards and Technology (NIST)
is legally responsible for maintaining and distributing all of the standards of physical
measurement.

The four basic standards are length, mass, time interval (often just called time), and
temperature. All other measurements can be derived from these four. As stated earlier,
time and frequency can be controlled and measured with greater precision than the other
basic standards. This is because the physical devices used (atomic and quartz crystal
oscillators) are very stable, and the measurements performed depend on electronic parts that
are capable of very fast changes in their on/off states.

6
 THE NIST STANDARDS FOR TIME AND FREQUENCY

Time of day is not a "standard" in the same sense as a meter stick or a standard set
of weights. The real standard is time interval, which is the length of time between two
events. By agreement, a day is simply the accumulation of 86 400 seconds. By referencing
to a standard time interval of 1 second, a number of organizations (including NIST) can
maintain accurate time-of-day clocks. These clocks will always agree with each other, and
the variations in time of day from country to country are extremely small.

Even though there is no physical "standard" for time of day, most countries keep a
carefully controlled frequency source with suitable electronics set "on time." They do this
by starting with the correct time interval (usually cesium derived) and then generating other
intervals such as the second or minute. As a result, the time of day throughout the world
is almost the same in every country.
Once a standard is set equal to other such standards,
depend on the time interval of the driving frequency
their differences (if any) in the future
source. In the same way that a watch depends on its balance wheel or quartz crystal, the
national standards of the world depend on their cesium "balance wheels."

The NIST standard for frequency and time interval is located in Boulder, Colorado.
It is the latest in a series of atomic oscillators built and maintained by NIST to provide the
official The current machine
reference for time interval or frequency in the United States.
is referred to as the "master" or "primary" frequency source. used to calibrate other
It is
oscillators ("secondary" standards), which are used to operate a time scale (a time scale is
a system of carefully operated oscillators), and then distributed to users through the NIST
services:

NIST Primary Standard

NIST Time Scale

NIST Services

Calibrations

?
 The NIST time scale isadjusted to agree with similar time scales in other countries.
The output of the time scale is distributed to users through the various frequency and time

broadcast services described in this book.

The NIST time scale works the same way as a clock or watch. has a frequency
It
source, a means of counting the frequency source and keeping time of day, and a way of
getting a useful output. In a clock or watch, the frequency source can be a 60-Hz power
line, a balance wheel, tuning fork, or quartz crystal. The counter for totaling the cycles of
frequency into seconds, minutes, and hours can be mechanical or electronic. The output
can be displayed digitally or by the position of the "hands" of the clock.

Anyone wanting to calibrate their frequency source or to compare their own time-
of-day pulses, uses the NIST time scale as a master clock. They then set their secondary
clocks to agree with the master clock. A similar situation has existed since man started
keeping time. For example, one of the first clocks used by man was the sundial. Sundials
worked by measuring the Sun's angle. Someone may have "set" an hourglass by carrying
it outdoors and reading the time on the sundial. Once they knew the correct time they
could turn the hourglass over and return indoors. By doing so, they could keep fairly
accurate time for the next hour. In this example, the sundial is the master clock, and the
hourglass is a secondary clock referenced to the master clock.

HOW TIME AND FREQUENCY STANDARDS ARE DISTRIBUTED

NIST distributes frequency and time signals to the general public. Most time and
frequency data are distributed by radio. Time of day, frequency, and time interval are
broadcast from radio stations WWV, WWVH, WWVB, and from the GOES satellites. These
services are controlled and operated by NIST. Precise time and frequency can also be
obtained by radio from Loran-C, Omega, and Global Positioning System (GPS) broadcasts.
These services are not operated by NIST, but they can provide an indirect reference to the
NIST time scale. Many of the broadcasts mentioned above are explained in detail elsewhere
in this book. NIST also distributes time and frequency data by telephone (both voice and
data) and through printed material.

Once obtained by radio, time and frequency signals are often distributed by cables.
For example, the standards laboratory of a manufacturing plant may provide signals for
users throughout the plant area. The type of distribution system used depends on the plant
layout and the level of accuracy needed. If you need to maintain the accuracy of an atomic
oscillator throughout a large area, you need the very best equipment and cables, and even
then you'll lose some accuracy. On the other hand, you can relax the specifications if all
you need is a time-of-day signal accurate to within 1 second.

8
 One main problems with distribution systems is that cables pick up noise along
of the
their path. This causes the signal at the end of the cable to be less useful than desired.
Not all cables work on all frequencies. It would be tempting to use commercial telephone
lines, but the highest practical frequency you can distribute in that way is 1 kHz. Any
attempt to send pulses over a telephone line intended for voice communication, but it
wouldn't work due to the limited bandwidth. Also, delays may be introduced because
telephone companies often combine several signals on the same line. Delays are also
introduced when calls are transmitted by satellite.

Designing a good distribution system may require a large investment in both time
and money. As a result, some users with high accuracy requirements find that it isn't
practical to maintain a local distribution system, and they simply re-establish time and/or
frequency at each destination point.

Of course, it's difficult to say how much accuracy can be obtained with any
distribution system. There is no simple formula for successful distribution of time and
frequency signals, although the more effort you expend, the better your results will be.

FREQUENCY CALIBRATIONS

If you have access to a frequency standard, you can use that standard to calibrate
other frequency sources. A calibration can take one of two forms:

o You can calibrate a frequency source by adjusting it to agree with the standard.

o If you are satisfied with the accuracy of the frequency source, you can simply
measure its frequency offset by comparing it to the standard and noting the error
when you make other calibrations using the source.

Calibration laboratories can measure the error of frequency sources with resolutions
ranging from a few parts per thousand to one part per trillion (1.00E-12). The resolution
needed depends upon the organization's end product. Precise calibrations require paying
more attention to detail but are not difficult to achieve.

Nearly all of time and frequency equipment are called frequency

calibrations
calibrations. This because the frequency of the device's internal oscillator is what is
is

actually being calibrated. For example, if your watch is 1 minute fast every day, you can
put it back on time by moving the minute hand. However, moving the minute hand doesn't
change the frequency at which the watch is running, and tomorrow it will be fast again.

9
To make the watch keep better time, you need to calibrate the internal oscillator by
adjusting its frequency and also reset the "hands" as required.

Since all calibrations are a standard, we can say that they are traceable
made using
to that standard. If you make a calibration
in the United States, it should be traceable to
NIST. This requires using a reference frequency distributed by NIST. Traceability to
NIST is always desirable and is the goal of many calibration laboratories.

To better illustrate traceability, let's look at an

example. Suppose you set your watch from a time
signal in your laboratory. The signal comes to you
from your company's distribution system. If you
trace the signal back through the system, you might
find that it comes from an oscillator on the
manufacturing plant grounds. If the oscillator is
calibrated using NIST radio signals, it can claim a
certain accuracy level compared to NIST. This
accuracy is transferred to your laboratory perhaps
only slightly reduced. Taking all these factors into
account, the signal in your laboratory (and therefore
your watch) is traceable to NIST.

Users who require legal evidence of traceability

should keep a record or log of their calibration
activities, even for something as simple as listening
to a radio or telephone and setting time of day.
Such a log book becomes evidence of the exact steps
followed and the time they were performed. If more
sophisticated means are used, then chart records or
photographs can be kept on file for a reasonable
time to demonstrate traceability.

What kind of accuracy is obtainable? As we

said before, frequency and time can both be
measured to very high accuracies with very great
measurement resolution. As this book explains,
there are many different ways to make calibrations.
Your accuracy depends on the technique you choose
and on the amount of error in your measurements.
frequency calibration accuracies range
Typically,
from parts per million by high frequency radio
signals to parts per trillion for satellite or Loran-C signals. A great deal depends on how
much effort you are willing to spend on getting a good, accurate measurement.

10
 Although some users take precise frequency measurement for granted, most time and
frequency devices need periodic calibration. For example, if a quartz oscillator is supposed
to produce a 5 MHz frequency, the frequency may still be in error by a significant amount.
Age affects the frequency of all quartz oscillators, and they can fail. Users need to know
their measurement requirements, and whether the method they choose meets those needs.

Once an oscillator is calibrated, it may not need to be calibrated again for a long
time. Today's oscillators are of excellent quality, and they may meet your accuracy
requirements for many weeks. Of course, nothing can be taken for granted. If you come
into your laboratory on Monday morning hoping that everything stayed put over the
weekend, you might be unpleasantly surprised. Digital dividers used to drive electronic
clocks do jump occasionally, especially if the power supplies are not designed to protect
against power glitches. It makes sense, therefore, to check both time and frequency
periodically to insure that the frequency is right and the clock is on time. If they are not,
you can then make an adjustment.

Many users find it convenient to maintain a continuous record of their oscillator's

frequency. This record will show when an oscillator needs calibration, and the results of
past calibrations. Normally, this record is a chart comparing the performance of the
oscillator to a radio signal which is traceable to NIST.

TERMS USED

This section covers some of the terminology used throughout the rest of this book.

Mega, Milli, Parts per and Percents

Throughout this book, we use terms such as kilohertz and megahertz, milliseconds
and microseconds. We also talk about accuracies of parts in 1.00E-09 cr 0.5%. The tables
on the following pages explain the meaning of these terms, and should serve as a convenient
reference for the reader.

11
 PREFIX CONVERSION CHART

Prefix Definition Example v

Pico 1 trillionth Picosecond - 1 trillionth of a second

Nano 1 billionth Nanosecond - 1 billionth of a second

Micro 1 millionth Microsecond - 1 millionth of a second

Milli 1 thousandth Millisecond - 1 thousandth of a second

Kilo 1 thousand Kilohertz - 1 thousand hertz

Mega 1 million Megahertz - 1 million hertz

Giga 1 billion Gigahertz - 1 billion hertz

Tera 1 trillion Terahertz - 1 trillion hertz

CONVERSIONS FROM PARTS PER ... TO PERCENT

PARTS PER PERCENT

1 part per hundred 1. OE-02 1. 0%

1 part per thousand 1. 0E-03 0. 1%

1 part per 100 thousand 1. OE-05 0. 001%

1 part per million 1.0E-06 0. 0001%

1 part per 100 million 1. OE-08 0. 000001%

1 part per billion 1. OE-09 0. 0000001%

1 part per 100 billion 1. 0E-11 0. 000000001%

1 part per trillion 1. OE-12 0. 0000000001%

1 part per 10 trillion 1. 0E-13 0. 00000000001%

12
 The unit used almost universally for frequency is the hertz, which means one cycle
or one pulse or one event per second. A
"cycle" can be generated in many different shapes.
It can be a sine wave or a square, triangular, or sawtooth wave. Keep in mind as you read
this book that not all of the waveforms being considered are sine waves.

Throughout this book, consideration is all ranges from the

given to frequencies of
1 Hz clock tick to many billions of hertz in the microwave region. The table below lists the
frequencies by bands. Most frequencies of interest are included in this table. The radio
frequency band contains the often heard references to high frequency, very high frequency,
low frequency, and so on.

RADIO FREQUENCY BANDS

RF Bands Frequency Range

VLF (Very Low Frequency) 3 - 30 kHz

LF (Low Frequency) 30 - 300 kHz

MF (Medium Frequency) 3 00 kHz - 3 MHz

HF (High Frequency) 3 - 30 MHz

VHF (Very High Frequency) 30 - 300 MHz

UHF (Ultra High Frequency) 300 MHz - 3 GHz

The difficulty of measuring frequency accurately is not directly related to the

frequency range. Making precise measurements at low frequencies is just as difficult as
making precise measurements at high frequencies.

we refer to the wavelength rather than the frequency

In some places in this book,
being used. Wavelength is especially convenient when calculating antenna lengths. In fact,
a glance at an antique radio dial shows that the band was actually marked in wavelengths.
For example, radio amateurs still refer to frequency allocations in terms of the 20-, 10-, or
2-meter bands.

13
 The conversion of wavelengths to frequency can be made for most purposes by using
the simple equation

300 000 000

Wavelength in meters =
f

where 300 000 000 meters per second is approximate speed of light, and f is the frequency
in hertz. From we can see that the frequency of the 10-meter band
this, is approximately
30 MHz, and 1000 kHz on the broadcast band is 300 meters.

Precise calculations of wavelength have to take the medium into account. For
example, signals travel slower inside a coaxial cable than they do through the atmosphere.

14
 Chapter 2 - TIME SCALES

There are a number of different time scales. The astronomical time scale is based
on the apparent motion of the Sun in the sky. The atomic time scale is based on the
"swings" of atomic oscillators. Time scales work the same way as clocks. Of course, the
pendulum in a time scale can be an atomic oscillator, and the "swings" are very rapid. For
example, a 1 MHz oscillator "swings" one million times per second (the "swings" are
electronic pulses). To keep track of these rapid swings we group the rapid pulses into
slower ones. Thus one million microseconds become 1 second, which is something a human
can deal with.

To see how the various time scales came about, let's take a brief look at the history
of time.

SOLAR TIME

Consider a time system that uses the Sun and a sundial. The Earth rotates once
every 24 hours, and we call this period 1 day. The shadow on the sundial can indicate
fractions of cycles (time of day). As complete days elapse, calendars can be used to count
the days and divide them into weeks and years.

However, clocks based on the Earth's rotation

do not run at a constant rate. This is because the
Earth's orbit around the Sun is not circular, and the
Earth slows down and speeds up depending on its
distance from the Sun. The early astronomers and
mathematicians understood these laws of motion and
were able to correct the "apparent solar time" to obtain
a more uniform time called "mean solar time." This
correction is called the Equation of Time and is often
found engraved on sundials. Universal Time (UTO) is
equal to mean solar time if you make the correction at the Greenwich meridian in England.
This time scale was given the name UTO, the first in a series of time scale designations that
have evolved through the years.

If you use a star that is farther away than the Sun, the fact that the Earth's orbit
is not perfectly circular becomes less important. This system of timekeeping is named
"sidereal time" and is similar to mean solar time since both are based on the Earth's
spinning on its axis. The rate is different by 1 day per year, since the Earth circles the
Sun once per year.

15
 As better clocks were developed, astronomers began to notice a discrepancy in
Universal Time measured at different locations. This difference was eventually identified
as being caused by a wobble in the Earth's axis. The amount of wobble is about 15 meters
(nearly 50 feet) at the pole. By careful measurements made at various observatories
throughout the world, this wobble was corrected for, and a new time designation called UT1
was born. In its search for uniformity, the world community had now taken care of both
the non-circular orbit (UTO) and the axis wobble of the Earth (UT1).

When better clocks were developed,

it was found that UT1 displayed
fluctuations whose origin was unknown.
Due to the availability of stable
electronic clocks, these fluctuations
could beand were removed. This
resulted in an even more uniform time
scale called UT2.

To review, UT1 is a true

navigator's scale related to the Earth's
angular position. UT2 is a smooth
time and does not reflect the seasonal
variations in the Earth's position.
When the world's timekeepers went to
UT2, they bypassed the navigators' real
needs. A little later we'll describe the
present-day system which fixes the
problem.

Up until now we have talked about

the Universal Time family. Let us now
look at the other members of the time
family. The firstis "ephemeris
of these
time." An ephemeris
simply a table
is

that predicts the positions of the Sun,

Moon, and planets. Astronomers soon
found that these predicted positions on
the table did not agree with the
observed positions, because the
rotational rate of the Earth was not a
constant. In response to this problem,
the astronomers created what is called
"ephemeris time." It is determined by
the orbit of the Earth about the Sun,
and not by the rotation of the Earth on
its axis.

16
 )

ATOMIC TIME
Another kind of time that can be generated and used is Atomic Time. Atomic Time
is from Universal Time, but the concept is similar. Universal Time is obtained by
different
counting cycles of the Earth's rotation from some agreed-upon starting point. Atomic Time
is obtained from counting cycles of a signal from an atomic frequency source.

In 1971, the General Conference of

Weights and Measures officially recognized
the Atomic Time Scale and endorsed an
International Atomic Time Scale (TAI).
Since January 1972, TAI has been used by
most countries in the world.

Atomic time scales give us very

accurate time of day because they use very
Furthermore, they
stable atomic oscillators.
give us essentially constant intervals of
time. In other words, atomic time scales
are uniform timekeepers. Uniformity is
desired when we are trying to make two
events occur at the same time.

review for a moment the

Let's
several time scales we have discussed.
First, the universal time family is

dependent on the Earth's spin on its axis. Second, Ephemeris Time depends on the orbital
motion of the Earth about the Sun. And finally, Atomic Time, which is very uniform and
precise, depends on a fundamental property of atoms.

As shown in the table below, measurement uncertainties limit the stability of

Ephemeris Time to about 0.05 second in a 9-year period. Universal Time can be determined
to a few thousandths of a second or several milliseconds in 1 day. Atomic Time is stable
to a few billionths of a second in a minute or less. From these numbers, it is easy to see
why scientists have been leaning toward a time scale based on atomic clocks.

Universal Time (UTO UT1 UT2 , , 3 milliseconds in 1 day

Ephemeris Time (ET) 50 milliseconds in 9 years

Atomic Time <100 nanoseconds in 1 minute

17
 COORDINATED UNIVERSAL TIME
Coordinated Universal Time (UTC) was adopted in 1971 and became effective in 1972.
UTC isbased on an Atomic Time Scale. As a result, a UTC clock gradually gets out of step
with the Sun. This is the same situation that causes us to have leap years. Since the year
is not an exact multiple of the day, we add a day every 4 years to keep our calendar in

step with the seasons.

The same scheme was adopted to keep clocks in step with the Sun, and the "leap
second" was born. To make adjustments in the clock, a particular minute would contain
either 61 or 59 seconds instead of the conventional 60 seconds. You could, therefore, have
either a positive or a negative leap second. It was expected and proved true that leap
seconds would normally occur about once a year.

By international agreement, UTC is maintained within 0.9 second of the navigator's

time scale UT1. By adding positive or negative leap seconds, a good clock can keep
approximate step with the Sun. Since the rotation of the Earth is not uniform, we cannot
predict exactly when leap seconds will be added or deleted, but this is usually done on June
30 or December 31.

you tune to a frequency and time broadcast (radio station WWV, for example),
If
the announced time is often UTC. This time is uniform. It will never differ from UT1 by
more than 0.9 second. Most users, such as radio and television stations and telephone
time-of-day services, use UTC so they don't care how much it differs from UT1. Even most
navigators don't need to know UT1 to better than 0.9 second, so UTC also meets their
needs.

However, a small number of users need UT1 time to less than 0.9 second. To meets
the needs of these users, most standard frequency and time radio stations broadcast a
correction which can be applied to UTC to obtain UT1. On WWV, for instance, the
corrections, in units of 0.1 seconds, are encoded into the broadcasts by using double ticks
or pulses after the start of each minute. The amount of correction is determined by
counting the number of successive double ticks heard each minute, and the sign of the
correction is given by the location of the double ticks within the minute (most frequency and
time stations worldwide have some such scheme for UT1).

Keep in mind that UTC prevents you from simply subtracting the dates of the events
to get the time difference between them. You must take into account any leap seconds that
were added or deleted.

18
 TIME ZONES

Standard frequency and time stations

usually broadcast Coordinated Universal EASTERN CENTRAL
Time (UTC), which is referenced to the
Greenwich meridian. However, many users
want to display the local time used in their
city. If the time is being decoded from a
time code (as opposed to a voice time-of-
day announcement), the problem can be
solved by using clocks that can display time
for any of the world time zones, even
though they are receiving and decoding
UTC. Each time zone differs from UTC by
a specific number of hours. The map on MOUNTAIN PACIFIC
the next page shows time zones
the
currently in use in the continental United
States.

Contrary to popular opinion, NIST is

not involved with determining time zones in
the United States. This responsibility
belongs to the Department of
Transportation, because the need for time zones came about when railroads were first used
for interstate commerce. Information about time zones can be obtained from the
Department of Transportation, Washington, DC 20590.

USING TIME SCALES FOR NAVIGATION AND ASTRONOMY

Although most users like the uniformity of Atomic Time, there is one application that
needs the variability of Solar Time. This is in the area of celestial navigation where Earth
rotation (even though it fluctuates) is used in finding out where you are on the Earth.

Navigators who find their positions from the stars are among the largest user group
of standard time broadcasts. Since the Earth rotates once in 24 hours, a person can find
his position (in longitude) if he knows his local time and the time difference between the
Greenwich meridian and himself. As an example, a person trying to find his position uses
an instrument like a sextant to measure the local solar time wherever he might be on the
Earth. He then needs to know the time at the Greenwich meridian. This is the same
problem that generated all the interest several hundred years ago in developing a good
chronometer. Those clocks were used for many years until radio clocks were invented.
Even today, most vessels do not leave home without a good chronometer.

19
 Many countries operate high frequency radio stations to serve navigators and many
other users. The United States operates WWV
in Fort Collins, Colorado, and WWVH in
Hawaii. Similar stations are operated by the Canadians and the Japanese and many
countries in Europe and Asia. The U.S. Navy also broadcasts time signals from a number
of radio stations. \

Most measure and record astronomical

large countries also maintain observatories to
time. This done using telescopes. The official U.S. observatory is operated by the Navy
is

in Washington, DC. Astronomical time is difficult to measure accurately. Errors of a few

milliseconds are realized even after a whole evening's sightings have been taken and
averaged.

Standard Time Zones of the Continental United States

20
 INTERNATIONAL COORDINATION OF FREQUENCY AND TIME ACTIVITIES

Unlike other physical standards, time interval or frequency can be obtained from
many sources. Both Canada and the United States have primary frequency standards that
also act as flywheels for accurate timekeeping. In addition, the U.S. Naval Observatory
(USNO) makes astronomical observations for determining UT1 and keeps many frequency
sources running for use by the Department of Defense. Many laboratories throughout the
world maintain primary frequency standards for their countries.

The agency responsible for UT1 is the International Earth Rotation Service (IERS)
located in Paris, France. The IERS determines when leap seconds are needed.

The agency responsible for atomic time (UTC and TAI) is the International Bureau
of Weights and Measures (BIPM), also located in Paris, France. The BIPM determines the
frequency (rate) of UTC
and TAI by comparing the primary frequency standards located in
several countries (including the United States, Canada, Japan, and Germany). The time for
UTC is based on international time scale comparisons between hundreds of clocks located
in many laboratories around the world. The time scale comparisons are currently made
using the Global Positional System (GPS) or Loran-C. The BIPM evaluates the data from
each time scale and corrects the data from each contributor. By international agreement,
all UTC time scales must agree with the UTC time scale operated by the BIPM to within

± 1 millisecond. The result is a uniform world frequency (and time) system that differs
from country to country by only a small amount. So, whether you get your frequency and
time from CHU in Canada or the PTB in Germany, it will differ only slightly from the
U.S. standard.

The Role of the National Institute of Standards and Technology (NIST)

NIST is in the U.S. Department of Commerce and has been authorized by Congress
to undertake the following functions:

"The custody, maintenance, and development of the national

standards of measurements and the provisions of means and
methods for making measurements consistent with those
standards, including the comparison of standards used in scientific
investigations, engineering, manufacturing, commerce, and
educational institutions, with the standards adopted or recognized
by the Government"

The Time and Frequency Division of NIST, located in Boulder, Colorado, carries out
the above functions related to frequency or time interval and keeps time-of-day clocks

21
running on its time scale system. The Time and Frequency Division has many
responsibilities. The division operates the NIST standard of frequency and time interval,
as well as several time scales based upon this standard. In addition to maintaining the
standard, research efforts are carried out to improve it. This group also offers a direct
service for calibration of oscillators and clocks.

The Time and Frequency Division also distributes the standards and continually finds
new and improved methods of distribution. At this writing, the distribution services consist
of high frequency radio stations WWV
and WWVH, low frequency radio station WWVB,
telephone time-of-day services (both voice and data), and several services based on radio
signals from GOES and GPS satellites or Loran-C navigation signals. This division also
produces literature and conducts seminars on how to use these services, an„ provides
calibration and monitoring data to the general public. The services are used to provide
national and international frequency and time coordination.

To summarize, NIST maintains services based upon the primary frequency standards.
These services are constantly being improved and are designed to meet the needs of users
at all levels of accuracy for frequency and time calibration. Training seminars are offered,
and a number of publications are available to assist the public. For more information,
contact:

Time and Frequency Division

National Institute of Standards and Technology
325 Broadway
Boulder, CO 80303

The Role of the U.S. Naval Observatory (USNO)

The U.S. Naval Observatory (USNO) is in the Department of Defense (DOD). Under
a DOD directive, the USNO is responsible for:

" maintaining a reference standard (astronomical and atomic)

for use by all DOD components, other agencies of the Federal
Government, DOD contractors, and related scientific laboratories.
This responsibility includes that of programming the necessary
resources to maintain the reference standard and to disseminate
precise time to DOD users."

22
 The USNO contributes (along with a number of other laboratories throughout the
world) data to the International Earth Rotation Service (IERS) which combines the results
into the final values of UT1. The USNO is the only organization in the United States that
determines UT1 operationally.

The USNO serves as the main time and frequency reference for the DOD and its
contractors. The DOD provides frequency and time services throughout the world. These
include many different radio transmissions, satellite signals, and portable clocks.

The USNO accomplishes its global responsibilities with a system of cooperating

worldwide stations that make measurements and keep very good frequency and time
references. The USNO also publishes information about its services. For more information
about these publications, contact:

U.S. Naval Observatory

Time Services Department
34th & Massachusetts Avenue, NW
Washington, DC 20392-5100

The three military services of the DOD have established focal points for matters
dealing with frequency and time. These facilities coordinate DOD
efforts and provide
information to their organizations and contractors. The focal points are:

1. Aerospace Guidance and Metrology Center, Newark Air Force Station, OH 23055 (for
the Air Force).

2. Navy Electronic Systems Command, Code 510, Washington, DC 20360 (for the Navy).

3. U.S. Army
Material Development & Readiness Command, DRCQA-PC, Alexandria, VA
20315 (for the Army).

23
 Chapter 3 - AN INTRODUCTION TO FREQUENCY SOURCES

This chapter is about electronic frequency oscillators, how they work, and how they
can be used. It talks about some of the physical properties of frequency sources and
includes a table comparing the different kinds of devices that are available.

FREQUENCY SOURCES AND CLOCKS

Most and
clocks (especially the very accurate
precise ones) are based on high-quality frequency
sources. If we look at an event which occurs
regularly (the sunrise, for example), we can state
how many of these events occur in a given time
period. This number is the frequency of this
series of events. In our example, we could say
that the frequency of sunrises is 7 events per
week, or 365 events per year. "Events per week"
or "events per year" would be the unit which we
used for our frequency number. This number
changes if we use different units. In our example,
we assumed that we knew the definition of a week
and a year, and the numbers change accordingly.

Now, what is the time interval between the

events? The answers for our examples are simple:
One sunrise follows the other after t = 1/7 week,
or t = 1/365 year, where we used "week" and
"year" as two possible choices for our unit of time.

We have learned two things. First, for periodic events, the time between the events
is related to the frequency of their occurrence in the following simple way:

1
Frequency =
Time Interval

Second, periodic events can be counted to define time. The frequency source
becomes a clock when we add a counting mechanism for the events.

25
 In this example, the frequency used is that of the rotating Earth. The time between
recurring events is one day. The rotating Earth has served mankind for thousands of
years and until recently was the source of the definition of time interval. The counting
mechanism which made it a clock was the recording of days and years.

The needto get along for many days which might

be without sunlight, and to measure time
cloudy,
intervals much shorter than a day, brought about the
invention of clocks. The first "modern" clocks were
based on the pendulum. Pendulum clocks were
invented about 400 years ago and are still widely used
today. The pendulum may be a suspended weight
pendulum) like in "grandfather" clocks, or
(gravitational
the balance wheel (torsion pendulum) of modern
mechanical wristwatches. A pendulum clock has all of
the same features found in quartz crystal and atomic
clocks.

In a pendulum clock, the pendulum has to be kept

in motion by some type of energy source. In a
wristwatch, this energy source is typically the winding
spring or battery. The pendulum or spring produces
small bursts of energy that drive the gears inside the
clock. The moving pendulum is a frequency source. In
order to have a clock, a read-out mechanism is
necessary which counts and accumulates the ticks from
the frequency source (more accurately, the time interval
between the ticks) and displays the result. In our
example of a wristwatch, this is accomplished by using
a digital display or a set of gears that move the hands
on the clock face.

We have just discussed how adding a counting

mechanism to a frequency source creates a clock. This
task can also be performed by an electronic frequency
divider. For example, a signal from a 5-MHz oscillator
can be divided to 60 Hz (see figure on next page). The
60-Hz voltage can then be used to drive an electrical
clock motor similar to those driven by the 60-Hz power
line frequency that we use at home and at work. Or,
an additional electric pulse generator may be used that
generates one very sharp electrical pulse per second. The time interval of 1 second
between the pulses (corresponding to a frequency of 1 Hz) comes directly from our
frequency source. These pulses can be used directly in time comparisons with those of
other similar clocks, or they can drive gears or a digital display.

26
 Frequency
,
5 MHz
. ^ Frequency 60 Hz

Source Divider

Divider & ClocX

1 Hz
Pulse (Counter/

Generator Accumulator'

A Typical Clort System

Clock-Tick Output

The electric power line is a very good frequency source for many applications. A
clock driven from the power keep excellent time. This is because the power
line will
system is carefully controlled to maintain its frequency within definite limits. Each power
company is notified in advance to set its frequency to a particular value, so that the
millions of clocks on the system (in homes and offices across the country) will gain or lose
time as required to keep the clocks correct. The time corrections are usually done at
night.

The basic unit of time is the second (symbol s). The second has been defined by
using the property of a cesium atom. A second equals 9 192 631 770 periods of the
radiation corresponding to the transition between the two hyperfme levels of the ground
state of the cesium-133 atom. So, the frequency of a "cesium pendulum" is 9 192 631 770
events per second (the cesium atom is a very "fast" pendulum). The unit used for
describing frequency is the hertz (Hz), which means one event per second.

27
 THE PERFORMANCE OF FREQUENCY SOURCES
The performance of a frequency source is usually described in terms of its accuracy,
reproducibility, and stability. These are defined as:

ACCURACY
How well does it relate to the definition? In the case of frequency, accuracy is a
measure of how well the frequency source relates to the definition of a second (in
terms of cesium) mentioned previously. The very high frequency of the cesium
oscillator is divided to a lower easier-to-use value and compared.

REPRODUCIBILITY

If you built anumber of frequency sources and adjusted them, how well would they
agree in frequency? This term obviously applies to sources that are manufactured
and then tested to see how they differ.

STABILITY

Once a frequency device is set to a given frequency, how well does it generate that
value during some period of time? At any given moment, an oscillator generates an
output signal whose frequency depends on a number of factors: temperature, time
since turn-on, line voltage, vibration and shock, and so on. Some time later its
frequency will change, again due to a number of causes. The difference between its
frequency at one moment in time and another moment is called stability. The
specification for stability is usually given for a number of time periods: 1 second, 10
seconds, minutes, hours, or even days, months, or years. Stability is an important
factor in oscillator cost. More stable oscillators cost more.

The three characteristics are all important. Accuracy is sought by those who
evaluate and compare high-quality frequency sources. Reproducibility is important in
applications when several frequency sources are expected to agree with each other.
However, the stability of a frequency source is usually the most important characteristic
for the average user.

The frequency frequency source depends on many things that might

stability of a
cause frequency changes. Frequency stability can be measured by taking a reasonably large
number of successive readings on an electronic counter which counts the frequency of the
device to be evaluated. Each counter reading (in hertz) is obtained by counting the output
frequency for some specified time period. This sampling time can usually be chosen by
simply adjusting a knob on the counter; for example, a sampling time of 0.1 second or 1
second may be chosen. The result will change as the sampling time changes.

28
 Furthermore, variations in the readings of measured frequency might be expected
to average out if observed long enough. This is not always so. The stability of the
frequency source usually depends on the sampling time of the measurement and tends to
get smaller with longer sampling times. But, again, there are many exceptions to this.

The frequency fluctuations at some later time may be partially due to the previous
fluctuations. Then, the computed value depends on the particular way in which the many
counter readings are averaged and evaluated. Another influence on measured stability
depends on whether the counter starts counting again immediately after completion of the
preceding count or if some time elapses ("dead-time") before counting starts again.

Finally, the electronic

circuits used in
measuring frequency have a finite response
stability
time. This means that they cannot follow frequency
fluctuations faster than some given rate. For
example, our eyes cannot register light fluctuations
which occur faster than about every 1/10 of a second;
we say that the eye has a frequency response
(bandwidth) of 10 Hz, or that the eye cannot follow
frequencies higher than 10 Hz.

In order to measure frequency stabilities for

sampling times smaller than some value, our
measurement equipment has to provide for an
electronic frequency bandwidth which is large enough.

To summarize: A recommended way of

properly measuring and describing frequency stability
is the following: (a) make sure that the frequency
bandwidth of the total measuring setup is large
enough; (b) use a counter with a dead-time as small
as possible (the dead-time should be less than the
reciprocal bandwidth; if not, use the computation procedures that exist to allow for larger
dead-times); (c) take enough readings at a given sampling time. The result is that you'll
see how the frequency of a source will change if you watch it for 0.01 second, 0.1 second,
1 second, and so on. Such short observation times have applications in radar,
communications, and other fields.

Often, results are expressed as a relative (sometimes called fractional) value. The
value obtained for the frequency stability is divided by the nominal frequency to normalize
it. This lets us "rate" the oscillator as being 1 in a million or 5 in 10 million, and so on.

29
 There is a sound basis for using the relative frequency stability instead of using
frequency stability directly. Relative frequency is a numeric and is independent of the
actual operating frequency of the oscillator being discussed. This makes it possible to
compare the stability of a 10-MHz oscillator to a 10-kHz oscillator.

In fact, many high-quality oscillators have 1-, 5-, and 10-MHz outputs. However,
calibrating these sources might require the generation of a different frequency, for example,
a signal divided to 1 Hz. If the original source had a relative frequency stability of one
part per million (1.00E-06), then the 1-Hz signal would have the same relative frequency
stability. This assumes, of course, that the generation of new frequencies does not change
the stability. This assumption is true for most applications.

PRIMARY AND SECONDARY STANDARDS

At this point, we should briefly

discuss the terms "primary frequency
standard" and "secondary frequency
standard." These terms refer to how
oscillators are used. Any frequency
source, regardless of its accuracy or
stability, can be a primary frequency
standard if it is used as the sole
calibration reference for other
frequency sources. A secondary
frequency standard is a device that is
calibrated against a primary frequency
standard, and then used to calibrate
other frequency sources.

The machines used for primary

standards, like the one at NIST and
those in other countries, are in a class
by themselves. Because of the way
they have been built and operated,
they can be evaluated. This means
that experimental data are taken and
used to calculate the errors in their
output frequency due to all known
causes. It is because of this careful
evaluation that their accuracy can be
stated without having to check these
machines against others. This concept
will be discussed in more detail in the
section on cesium beam oscillators.

30
 QUARTZ OSCILLATORS
Quartz (sometimes called crystal oscillators) are widely used. They range
oscillators
from the tiny units found in wristwatches to elaborate instruments found in laboratories.
Quartz oscillators provide good performance at a reasonable price and dominate the field
of frequency sources.

The quartz a mechanical resonator.

crystal in the oscillator is The resonator's
oscillations have to be excited and sensed. This is done by taking advantage of the
piezoelectric effect in the quartz crystal; that is, mechanical compression of the crystal
generates a voltage across the crystal. Conversely, the application of an external voltage
across the crystal causes the crystal to expand or contract depending on the polarity of the
voltage.

Quartz

Crystal

++++++++++
4

Applied Voltage
Causes Contraction

Applied Pressure
Generates Voltage The Piezoelectric Effect

A crystal not a homogeneous medium but has a certain preferred direction; thus,
is

the piezoelectric effect has a directional dependence with respect to the orientation of the
crystal. In order to take advantage of the piezoelectric effect, the crystal block must be
cut in a well-defined way with respect to the crystallographic directions. The raw material
used today is natural or synthetic quartz. A block is cut out of the raw crystal material
in the desired orientation with the aid of optical techniques which allow the determination
of the crystallographic axes. The high-precision, final orientation of the cut and the tuning
to the desired frequency are then done by grinding and etching, controlled by x-ray
methods.

31
 The quartz crystal can be cut and electrically excited in a variety of ways (as shown
below). The most common types of vibrations (modes) are the longitudinal and thickness
modes, the flexure (bending) mode, the torsional mode, and the shear mode. In order to
use the piezoelectric effect, metal electrodes have to be attached to the crystal surfaces so
that the desired mode is excited.

The electrodes are typically created as

extremely thin metallic coating by vacuum
evaporation of metals. Electric leads are
attached to the electrodes by soldering. They
also serve as the mounting support, thus freely
suspending the quartz crystal. In order to least
upset the mechanical vibrations of the crystal,
the electrode-support leads are attached at
points where no vibrational motion occurs
(nodes). The crystal is then packaged, and the
case is either filled with a protective gas or else
evacuated.

Using the crystal, an oscillator can be built

by adding an electronic amplifier feedback and
a power supply. The oscillator's output
frequency is determined by the quartz crystal
Flexure Torsional
resonator. The resonating frequency of the
crystal is determined by the physical dimensions
Longitudinal
of the crystal and the type of crystal used. For
example, the resonance frequency for a
longitudinal mode of vibration is approximately

v0 = 2.7 x 10 3 x

where L is the length of the crystal. If L is in

meters, the resonance frequency v0 will be in
hertz. The equation allows us to estimate the
size of the crystals. For example, a 100-kHz
crystal is just a few centimeters long. Making
quartz crystals with resonance frequencies above
Shear 10 MHz is hardly possible. However, we can
Thickness excite resonators not only in their fundamental
mode (previously discussed), but also at
multiples (overtones) of this fundamental resonance frequency. For example, a violin string
can vibrate at frequencies which are multiples of its fundamental. Quartz crystals designed
for excitation at multiples of their fundamental resonance frequency are called overtone
crystals.

32
 Temperature and Aging of Crystals

Temperature and aging are two factors that influence quartz oscillator performance.
There is a temperature dependence of the quartz crystal that affects its resonance
frequency, and there is also a drift of the resonance frequency due to aging.

The temperature dependence is caused by a slight change in the elastic properties

of the crystal. Certain cuts (crystallographic orientations of the crystal) minimize this
effect over a rather wide range of temperatures, most notably the "AT" and "GT" cuts.
Temperature coefficients of less than one part in 100 million per kelvin (or degrees Celsius)
are possible. In other words, the fractional frequency change will be less than 1.00E-08
with 1 degree of temperature change. Crystal oscillators must be carefully designed if
very high frequency stabilities are desired. If large environmental temperature fluctuations
are to be tolerated, the crystals themselves are enclosed in electronically regulated ovens
which maintain a constant temperature. In some crystal oscillators this is done to better
than 1/1000 of a degree.

Feedback

Quartz
Crystal
Resonator
Output

Quartz Crystal
Oscillator

An alternate solution to the temperature problem is the so-called temperature-

compensated crystal oscillator or TCXO. An additional frequency-determining element in
the oscillator, which can be just a small capacitor, allows us to tune the oscillator over a
limited range. If a temperature sensor is added to cause a change in this capacitor, the
change in resonance frequency of the crystal resonator can be cancelled. Capacitors whose
values change with an applied voltage (varactors) are used. The applied voltage is derived
from a temperature-sensing circuit.

33
 The TCXO does not necessarily require further temperature control by an oven.
However, there a drawback to this approach. By adding another frequency-determining
is

element, the crystal resonator loses a corresponding part of its control on the output
frequency of the whole oscillator. The long-term stability (days) of TCXO's is therefore
below that of crystals with a good oven control. TCXO's are used in small, usually
portable units of relatively lower performance. They are used where frequency stabilities
from day to day and frequency changes (over tens of degrees of temperature) of not better
than 1.00E-09 are needed.

Aging is a common trait of all quartz

oscillators. It is a nearly linear (uniform)
change in resonance frequency with time.
Drift often is negative, meaning that the
resonance frequency decreases. A frequency
decrease could indicate an increase in the
crystal size. There are many possible causes:
contamination (depositing of foreign material)
on the surfaces; changes in the electrodes or
the metallic plating; reforming of loose (from
grinding and etching) surface material, or
changes in the internal crystal structure. All
of these are possibly caused or enhanced by
the vibrating motion of the oscillating crystal.
Recent crystal-holder design improvements,
combined with clean vacuum enclosures, have
led to a reduction of aging to about 1.00E-11
per day. For a 5-MHz crystal with a
thickness of a little less than a millimeter,
this aging corresponds to an absolute
thickness change of only 1.00E-11 of a
millimeter, or less than 0.1% of the diameter
of an atom.

When a quartz oscillator is first turned on, not usually oscillate at its original
it will

frequency. It will go through a "warm-up" period while the temperature of the crystal
resonator and its oven stabilizes. The warm-up period may last several days or more.
During this time, a large but diminishing drift occurs until the oscillator reaches its normal
operating temperature.

34
 Quartz Oscillator Performance

Quartz oscillators are very stable. The best available oscillators have stabilities of
a few parts in 1.00E-13 for sampling times from 1 second to 1 day. The limitations in
stability are mainly due to noise from electronic components in the oscillator circuits.
This noise can be reduced by selection of low-noise components and by special circuit
designs. Thus, there is a fair chance that crystal oscillator stability can reach values of
less than 1.00E-13 for sampling times of seconds to hours. For times shorter than one
second, stability is often determined by additive noise in the output amplifiers. This can
be reduced by a crystal filter in the output. The long-term stability beyond several hours
sampling time is determined by aging and by external factors like line voltage variations,
temperature fluctuations, and so on.

It is apparent, therefore, that crystal oscillators require calibration at least once a

year. They may need it more often depending on the application. Frequency adjustments
are made with a small added capacitor in much the same way as was discussed in
connection with the TCXO.

The most stable crystal oscillators, with the lowest aging rate, may cost several
thousand dollars or more. They have a volume of a few thousand cubic centimeters and
require input power of about 10 watts. They have an elaborate crystal oven for
temperature control, well-designed electronics, and usually several output frequencies which
are derived from the oscillator with frequency dividers and multipliers. These high-
performance devices may use 2.5- or 5-MHz crystal resonators and have a relative
frequency of 1.00E-11 per day.

Cheaper and smaller crystal oscillators are available in a variety of designs. As price
goes down, the performance also goes down. Costs can go down to below $100, sizes to
a few cubic centimeters, and power requirements to less than 0.1 watt. Relative frequency
can be as much as 1.00E-06.

For information on measuring the performance of quartz oscillators, please see

chapter 4.

35
 ATOMIC OSCILLATORS

This section discusses the design and performance of the three types of atomic
oscillators which are currently in use.

Rubidium Gas Cell Oscillators

Rubidium atomic resonance is at 6 834 682 608 Hz. Rubidium oscillators use a gas
cell containing rubidium gas. In order to reduce the effect of collisions among the
rubidium atoms, argon, an inert buffer gas, is introduced into the cell at about 1/1000 of
atmospheric pressure. This allows lifetimes of the rubidium atom oscillations of about
1/100 second. The oscillation lifetime is still limited by atom collisions. Atomic collisions,
as well as the simultaneous action of the light and the microwave signals on the same
atom, cause frequency shifts of the order of 1.00E-09. These frequency shifts depend
strongly on the composition, temperature, and pressure of the buffer gas and on the
intensity of the light. As a result, rubidium gas cells vary in their resonance frequency by
as much as 1.00E-09, depending on the particular setting of the frequency shifting
parameters when the oscillator is manufactured.

Quartz Crystal
Oscillator &

Gas Cell

Rubid ium Gas Cell Oscillator

Since rubidium oscillators do change frequency, they need initial calibration and
also recalibration just like quartz oscillators. The stability performance of rubidium
oscillators is still quite spectacular. At 1-second sampling times, they display a stability

36
less than l.OOE-11 and perform near 1.00E-13 for sampling times up to 1 day. For longer
averaging times, the frequency stability is affected by the frequency drift, which is typically
l.OOE-11 per month. This is much better than a quartz oscillator. As with quartz
oscillators, the performance of a rubidium oscillator generally goes down as the price goes
down.

Rubidium oscillators are used when excellent medium-term few minutes

stability (a
to a day) is needed. They are smaller and less expensive than cesium beam oscillators and
work well in situations where a quartz oscillator (with its need for more frequent
calibration and its greater environmental sensitivity) would not suffice. Since rubidium
oscillators are more stable, they give better results with fewer adjustments than quartz
oscillators.

Cesium Beam Oscillators

For a cesium atomic resonance is at 9 192 631 770 Hz. An oven contains
oscillator,
the cesium metal. Ifheated to about 100 degrees Celsius, enough cesium gas will be
produced to form an atomic beam, which leaves the oven through one or many channels
into a vacuum chamber. The beam traverses first the state-selecting magnet, then the
microwave cavity. Typically, a cavity with separated interrogation regions is used. In the
cavity, an external microwave signal acts on the beam. The beam finally reaches the atom
detector after passing another state-selecting magnet. The atom detector is simply a
tungsten wire which is heated to about 900 degrees Celsius by passing an electric current
through it. This wire is biased with a few volts dc, and cesium atoms which hit it become
electrically charged or ionized and can be collected on an auxiliary electrode. The stream
of electrically charged atoms at this electrode represents an electric current which is
amplified, detected, and used in the feedback network.

Atoms move through the cavity at speeds of about 100 meters per second. In
commercially available cesiums (which have to be reasonably small), the cavity is only
about 0.2 meter long. The corresponding interaction time is two-thousandths of a second.
In laboratory devices (like the NIST primary standard), the cavity may be 4 meters long,
or even longer.

The fractional frequency stability of laboratory and commercial cesium standards can
reach 1.00E-14 at sampling times of less than 1 hour to days. The short-term frequency
stability is limited by fluctuations in the atomic beam intensity, "shot noise," which is basic
and unavoidable. These fluctuations affect the frequency stability less as more intense
atomic beams are used. This approach, which is becoming available in both commercial
and laboratory cesium standards, improves the stability. Laboratory cesium standards
allow a more complete and easier evaluation of all effects on the frequency than
commercial standards.

37
 I

Quartz Crystal
Oscillator k Feedback
Frequency
Synthesizer

V
State State
Selector Selector
Cavity
Cesium I I

J
Oven
u i — Atom
Detector
Cesium Beam Oscillator

Cesium used extensively where high reproducibility and long-term

oscillators are
stability are needed sampling times of more than a day. They are the workhorses in
for
most of today's accurate frequency and time distribution services. By definition, cesium
oscillators are a primary frequency source. This means that (when operated correctly) a
cesium oscillator will be very close to its correct frequency without any calibration.
However, since the expected level of performance is so high, checking the performance is
a very important part of operating a cesium oscillator.

Hydrogen Masers

Maser is an acronym meaning "microwave amplification by stimulated emission of

radiation." The atomic resonance frequency is at 1 420 405 752 Hz. While not in
widespread use, hydrogen masers represent the state of the art in commercially available
oscillators for averaging times of 1 second to 1 day.

Hydrogen masers work like this: All natural hydrogen gas is composed of hydrogen
molecules. Each hydrogen molecule is formed by chemical bonding of two hydrogen atoms.
The beam source a radio frequency gas discharge in molecular hydrogen. This discharge
is

produces atomic hydrogen with high efficiency. The atomic hydrogen beam leaves the
source through one or many channels into a vacuum chamber. The beam then traverses
a state-selecting magnet and enters a storage bulb in the microwave cavity.

38
 The storage bulb is made from quartz
glass which has low electric losses. Its inner
walls are lined with Teflon. This coating allows many collisions of the hydrogen atoms
with the walls without significantly disturbing the oscillations of the atoms. The
underlying physical mechanisms are not yet fully understood. The storage bulb is typically
0.15 meter in diameter and dimensioned in such a way as to hold hydrogen atoms for
about 1 second. After about 1 second, the atoms leave the bulb and thus also leave the
microwave cavity. If the intensity of the hydrogen beam, which consists only of upper
state atoms (emitting atoms), is sufficiently large and if the cavity losses are sufficiently
low, self-oscillation will start in the cavity. The maser itself will generate a microwave
signal. We then have a maser-oscillator with an output frequency directly derived from
the atomic resonance. A quartz oscillator can be locked to this frequency by frequency
comparison techniques.

The hydrogen maser is not quite as accurate as a cesium oscillator. This is because
of experimental difficulties in the evaluation of the frequency shift due to the collisions of
the hydrogen atoms with the Teflon surface of the storage bulb. This limits the long-
term than several days. Values of long-term stability are not
stability over periods longer
better than those of cesium oscillators. However, for periods of a few seconds to a day,
the hydrogen maser has the best stability of all existing oscillators. Its application is
limited to uses where these stabilities are critical and where a rather bulky device is not
a handicap. Unlike cesium and rubidium oscillators, hydrogen masers have not yet been
evaluated under adverse environmental conditions. The number of hydrogen masers in use
is very small compared to the numbers of cesium beams and rubidium devices, and the cost

of a maser is still very high.

The table on the next page summarizes the oscillators discussed in this chapter.

Hydrogen Feedback
Maser
Quartz Crystal
Oscillator &. Frequency
Frequency Comparator
Synthesizer

Microwave
Cavity With
Storage Bulb

39
 COMPARISON OF FREQUENCY SOURCES

Type of Oscillator
Feature
Rubidium Hydrogen
Quartz Gas Cell Cesium Maser
Output 10 kHz to 1, 5, or 1, 5, or 1, 5, or
Frequencies 100 MHz 10 MHz 10 MHz 10 MHz
(typical)

Relative 1. 00E-06 2 00E-11

. 5. 00E-11 5. 00E-13
Frequency to to to
(1 second) 1. 00E-12 5.00E-12 5. 00E-13

Relative 1. 00E-06 5. 00E-12 1. 00E-13 1. 00E-13

Frequency to to to to
(1 day) 1. 00E-12 3 . 00E-13 1. 00E-14 1. 00E-14

Causes of Aging of Aging of Aging of Cavity

Long-Term Crystal & gas cell, Electronic Pulling
Instability Electronic filter, & Components
Components light

Time for 1 second 1 day 1 week 1 week

clock to to to to to
be off 1 10 days 10 days 1 month 1 month
microsecond

Start-Up seconds 10 minutes 30 minutes several

after to to to hours
being off hours 1 hour 1 hour

Resonator many more than 3 years no data

Lif e-Span years 3 years

Weight in 0.1 to 10 1 to 20 16 to 400 over 100

kilograms (2 oz to (3 to 45 (35 to 7 0 kilograms
(pounds) 20 lbs) lbs) lbs)

Power 0.1 12 30 40
consumed to to to to
(watts) 15 35 200 100

Estimated $5 $5 000 $30 000 over

Cost to to to $200 000
$10 000 $20 000 $50 000

40
 Chapter 4 - MEASURING FREQUENCY

In the last chapter, we discussed the different types of frequency sources (oscillators)
that are available. This chapter discusses how to measure an oscillator's performance. It
covers the equipment needed to make time and frequency measurements, what the
equipment is used for, how it is used, and the kind of results that you can expect.

Determining the frequency of an electrical signal involves counting the number of

cycles or pulses that occur over a given time interval. For example, a digital clock may
change its display at a rate of 1 Hz (once per second). This rate has been given the name
of a famous German inventor, Heinrich Hertz. One Hz equals one event per second.

This chapter uses scientific notation when discussing the accuracy of an oscillator.
For example, a relative frequency of 1 part in 10 to the 11th is written as 1.00E-11.

THE THEORY OF FREQUENCY MEASUREMENTS

Frequency and time interval can be
measured with greater precision than any
other physical quantities. Frequency
measurements are made by comparing signals
from two frequency sources. This is true in
music, for example, where the orchestra tunes
its instruments to the oboe. In the same
way, oscillators are measured, adjusted, and
calibrated by comparing their output with the
output from a higher quality frequency
source. The higher quality frequency source
can be a radio signal or another oscillator.

A simple and direct way to measure

frequency is to use a frequency counter. The
problem with this method is the time it takes
to make measurements of very high
resolution. The resolution of an ordinary
counter is increased by increasing the gate time. For example, a 1-second count of a 1-
MHz signal might give a resolution of 1 Hz. The resolution improves to 0.1 Hz after 10
seconds. This resolution was obtained by waiting longer. It would take many hours to
reach the resolution of a precision oscillator. Even if the time were available, this method
is not recommended because it allows only a few readings per day, and the chances for

error become great. A system that must run continuously for hours to get one reading is
susceptible to errors due to power outages and other causes.

41
 1

A better approach is to measure time interval instead of frequency. This method

takes advantage of modern circuit chips and is faster. For example, if two oscillators with
1-Hz outputs are compared using a time interval counter, a high resolution reading can
be obtained every second. A typical counter might have a 100-MHz time base. This
means the measurement resolution obtained is the period of the 100-MHz signal (10
nanoseconds). Therefore, changing from frequency to time interval measurements results
in a dramatic increase in resolution.

Time Interval Measurements

The technique of measuring time interval to determine frequency is commonly used

for precise calibrations. Time interval measurements are made using an electronic device
called a time interval counter (TIC). To use a TIC, you connect signals from two different
frequency sources. One signal serves as a start pulse, and the other serves as a stop pulse.
The TIC measures the interval between the occurrence of the two pulses.

Time interval counters differ widely in specification and design details, but they all
share several basic parts. These are: the time base, the main gate, and the decade
counting assembly (DCA).

Time UJJJ 1 1 DCA and Readout

Base

J L

Start Signal J-
Gate Display
J. Control Control
Stop Signal

Amplifiers and
Signal Shapers

Diagram of Time Interval Counter

42
 The time base provides uniformly spaced pulses for counting and control of the TIC's
circuitry.The time base must be stable because time base errors will directly affect the
measurements. Most counters use an internal quartz oscillator for a time base.

The may
be followed by decade dividers to allow the user a choice of
oscillator
several different gate signals. Adding more dividers to the chain makes it possible to
obtain longer measurement intervals. Increasing the frequency of the time base makes it
possible to count shorter time increments (higher resolution).

The main and ends. The gate may

gate controls the time at which the count begins
be actuated automatically by pulses from the time base or manually by means of a switch
on the control panel. Pulses which pass through the gate are routed at the decade
counting assembly where they are displayed on the counter's digital readout. After a pre-
set display period, the counter resets itself and starts another count.

Input
A

Input
B

Main Gate
Enable Pulse

Time Base
ii
Pulses

to Counter

Measuring the time interval between

two input signals

43
 Other counter sections include amplifiers, pulse shapers, power supplies, and control
circuits.The input channels are usually provided with level controls and attenuators to set
the exact amplitude limits at which the counter responds to input signals. Proper setting
of the level controls helps reject noise or other unwanted signals.

A counter in"frequency measurement" mode will display the result as a frequency

its

(Hz, MHz, and On the other hand, readings produced by a time interval counter
so on).
are in time units. The readings measure the length of the interval (in time units) between
the signals being calibrated. If the two signals have exactly the same frequency, the time
interval will not change. If the two frequencies differ, the time interval will change,
although usually very slowly. It is exactly as if each signal were a clock, and the readings
tell whether one clock gained or lost time relative to the other clock.

It takes two readings to get one data point. A single reading of the time interval
between two clocks is not useful. It is the second reading, subtracted from the first, that
tells whether time is gained or lost. This gain or loss of time is a measure of the
frequency difference between the two frequency sources.

Using a Measurement System

A typical standards laboratory has a primary oscillator which it uses as a frequency

reference. In many cases, the standards lab has made a considerable investment in this
oscillator. Quartz oscillators (based on 1990 prices) range from $500 to $10 000; rubidium
oscillators from $5 000 to $20 000; and cesium oscillators from $30 000 to $50 000.

The primary oscillator is used to calibrate equipment that is now in the lab or that
is expected soon. It too must be periodically calibrated so that its accuracy is known.
Calibrating the primary oscillator must be done on-site. Since oscillators are sensitive to
shipment, it is not practical to send them to another laboratory for calibration. Being
turned on and off can cause their frequency to change.

The is to backup its primary oscillator with batteries,

ideal situation for a laboratory
keep it running all theand continuously measure its performance using a
time,
measurement system. A good measurement system allows even small laboratories to
perform calibrations at a relative frequency of 1.00E-12. To get an idea of what this
relative frequency (or frequency offset) means, if such an oscillator were used to drive a
time-of-day clock it would gain or lose only one ten-millionth of a second a day!

44
 Calibration labs can use several types of frequency measurement systems.
Manufacturers offer VLF, LF, and Loran-C receivers that make measurements and record
the data on chart paper or via computer data logging. This equipment often uses the
primary oscillator as a receiver input. The receiver then makes the comparison and draws
the chart. Front panel indicators assist in operating the system. Receiver options include
different types of antennas, some distribution capability, and even a way to discipline the
primary oscillator to make it agree in frequency with the radio signal. A block diagram
of this type of system is shown below.

Reference
Frequency
>-
Low Strip
Phase
Pass Chart
Comparator Recorder
Filter

Oscillator
Being Tested

A Typical Commercially— Available

Frequency Measurement System

All measurement systems work basically the same way: they make comparisons
between two One signal is a
signals. reference frequency, and the other signal is the
signal being measured. The reference is assumed to be more accurate than the signal

being measured and is most often a radio signal traceable to the NIST time scale

(traceability discussed later in this chapter). The different radio signals you can use as
is

reference frequencies are described in chapters 5-7. The signal being measured is from
an oscillator like those discussed in chapter 3.

The data from the system must be recorded in some way. The readings can be
manually recorded in a logbook, or continuously recorded by a chart recorder. However,
modern measurement systems often use a computer. A computer-controlled measurement
system is described in the next section.

45
 A COMPUTER-CONTROLLED MEASUREMENT SYSTEM
This section describes the design and operation of a typical computer-controlled
measurement system. Each part of the system is discussed in the following paragraphs.
A block diagram of the entire system is shown below.

ANTENNA

Radio Receiver STOP

(Reference)

Tine
Counter Tine Base
Interval
Counter

Oscillator Frequency START

Being Measured Divider

COMPUTER

Time-of-Day Disk
Printer Modem
Clock Drives

46
 The Reference Frequency

The designer of a computer-controlled measurement system needs to decide which

radio signal should be used as a reference frequency. The different radio signals you can
use are discussed in chapters 5-7. Help is also available from the manufacturers of time
and frequency equipment and also from the national laboratories that control the radio
signals. NIST assists users in this area with publications, phone consultation, and training
seminars.

The decision should be based on the accuracy required by the system. If the
required accuracy is very high, or if it can be expected to increase in coming years, then
consider only those radio signals which support that accuracy. For example, WWV
can be
used to calibrate an oscillator of modest performance, but a high-performance oscillator
may require using signals from WWVB
or Loran-C. Loran-C is a radio navigation system
maintained by the U.S. Coast Guard (Loran stands for LOng RAnge Navigation).

The choice of radio signal is also

limited by the location of the laboratory,
reception problems with certain radio
signals, available equipment, and past
experience. Another factor is whether the
radio signal lends itself well to automation.
It may be better to use a signal that is
easy to automate by computer, even if the
initial cost of the equipment is higher. If
low-cost equipment is purchased, it may
require a lot of attention. In these
instances, the labor costs could soon exceed
the amount of money that was originally
saved.

Since a radio signal is you also

used,
need to consider the type of antenna
required. The antenna needs to be on the
roof, and the receiver may need to be
grounded. Several antenna types are currently offered. VLF and Loran-C receivers
generally operate with short vertical whips or loops, but HF services like WWV need long
wires for good, noise-free reception. The antenna should be mounted in an area free from
obstructions and interference.

The system described here uses Loran-C radio signals as a reference frequency. It
includes a radio that receives a signal from a Loran-C station. Loran-C is described in
more detail in chapter 6.

47
 Loran-C signals make an excellent reference frequency for a number of reasons.
They are extremely accurate, with a relative frequency of about 1.00E-12 per day. This
is good enough to check the performance of cesium, rubidium, or quartz oscillators. The
continued accuracy of Loran-C is insured since the stations are controlled by cesium
standards monitored by the U.S. Coast Guard and the U.S. Naval Observatory, in addition
to NIST. Loran signals are also traceable to NIST. Traceability is described in more
detail later in this chapter.

A is that some Loran-C receivers are inexpensive and automatic.

second benefit
Finally, each Loran-C station has an effective range of over 1000 miles (1600 kilometers).
Since numerous Loran-C transmitters are in operation, this means that Loran can be
received in just about any location in the Northern Hemisphere (including all 50 States).

The Frequency Divider

The signals connected to a time interval counter for comparison must have the same
(or a related) frequency. For example, in order to compare a 5-MHz signal to a 1-MHz
signal,both signals must first be divided to a lower frequency (usually 1 Hz). The system
includes a frequency divider for this purpose.

accepts input frequencies of 1, 5, or 10 MHz and divides them to 1

The frequency
Hz comparison. It also produces a signal that matches the frequency (about 10 Hz)
for
of the output pulse from the Loran-C receiver.

The frequency divider accepts signals from one to four oscillators.

This allows the
primary oscillator to be compared to Loran-C (the reference frequency), and the other three
oscillators to be compared to the primary oscillator.

The Time Interval Counter

The system's time interval counter can measure up to four channels simultaneously.
It has no knobs or controls and therefore reduces measurement errors due to knob
settings. The counter's time base frequency is obtained from the primary oscillator. The
time base frequency can be 1, 5, or 10 MHz. A light indicates whether a time base signal
is present.

Channel 1 of the dedicated to the calibration of the primary oscillator in

counter is

the laboratory. The primary oscillator is compared to the Loran-C reference frequency.
Channels 2, 3, and 4 can be used to compare other oscillators to the primary oscillator (see
the figure on the next page).

48
 Loran-C (Reference Frequency)

CHANNEL
1
User's Primary Oscillator

User's Primary Oscillator

(Reference Frequency)
CHANNELS
2-4
Other Oscillators in the
Calibration Lab

Calibration Procedure Using a Typical Measurement System

The Computer

Just about any computer can be used to control a measurement system. This NIST
system is controlled by a low-cost microcomputer with two disk drives, a monitor, and a
printer.

The computer averages time interval

readings and stores the data and the time the data
were taken on disk. The time is obtained from a
clock inside the microcomputer. The computer
runs 24 hours a day and restarts automatically
after power outages.

Every 24 hours, the computer stops taking

data and plots the data recorded in the last 24
hours. It then prints a copy of this plot on the
printer. The information contained on the plots is
described in the next section. The plotting is done i i

h i i i i i i i i i i i

automatically, without user interaction.

r i,

55 BSffiga
i l l, i i i

g i i

49
 After the system plots, it starts taking readings for the next day. The process
continues in this way: recording, plotting, recording, plotting, and so on. The only
operator attention required is to occasionally add paper to the printer.

The computer software used by the system is "transparent." It requires little

attention andmakes users feel that they are using a completely integrated system.

Output of the Measurement System

As mentioned, the system stops taking data once every 24 hours and produces a plot
of the performance of each oscillator being measured. Since one to four oscillates can be
measured, from one to four plots are printed each day.

Since radio signals are affected by events along their path (like bad weather and
electrical interference),data from a radio signal must be averaged when compared to a
primary oscillator (which may have a relative frequency of less than 1.00E-11 per day).
A 24-hour measurement is sufficiently long and still provides the user with a daily report
of each oscillator's performance.

A sample phase plot is shown below. Each plot shows the number of microseconds
that were gained or lost in the 24-hour interval of the measurement. The unit for the
vertical axis is microseconds. The unit for the horizontal axis is time of day (hours).

50
 The plots alsoshow the computed relative frequency of the oscillator. The relative
frequency isprinted below the plot. Relative frequency is calculated by fitting a linear
least-squares line to the data and taking the slope of the line. The system can show
relative frequency offsets ranging from 1.00E-06 to 1.00E-12. This allows the system to
measure the full range of oscillators, from low-quality quartz oscillators to high-quality
atomic oscillators.

Day to Day Operation of the System

The computer-controlled system was designed to make daily operation easy. The
automatic plotting and relative frequency features make the lab technician's job easier,
make the calibrations less confusing, and reduce the amount of operator attention.
However, the frequency measurement system should be checked daily to make sure that
everything is working. The manuals included with the equipment explain which front
panel lights and meters are important for daily system checks. Once a lab technician
becomes familiar with the equipment, the daily check should take only a few minutes.

Part of the check requires looking at the system's computer screen. A picture of the
screen is shown below. The screen was designed so that the operator can check the status
of the system with just a quick glance.

National Institute of Standards and Technology

6 7 8 9 18 11 12

Oscillator Performance (last 24 hours)

87-89-89 Press <C0NIR0L-X> to exit Change in Counter

1748 UTC Press <C0MTR0L-P> to plot Tine (Ps) Reading (ps)

^§ Cesiun us Loran-C 9940 (Fallon, Nevada) -8. 81 45.56

1 Brand X Quartz us Cesium -8.82 53.07
pllj Brand Y Quartz us Cesiun * +8. 83 68.51
I Rubidium us Cesiun -8.88 68.08

51
 The measurement screen is divided into two parts. The bottom part of the screen
contains a box with information about the signals being measured. The left side of this
box shows the present time (UTC) and date. Below the date and time are the titles given
to each measurement channel. On our sample screen (on the previous page) the title for
Channel 1 is: Cesium vs Loran-C 9940 (Fallon)

This us that a cesium oscillator is connected to Channel 1 and is being

title tells
compared to a Loran-C signal.Loran-C is the reference frequency, and the cesium
oscillator is the oscillator being measured. The cesium oscillator is the reference frequency
on Channels 2, 3, and 4. By looking at the titles for those channels, you can see that
quartz oscillators are being measured on Channels 2 and 3, and that a rubidium oscillator
is being measured on Channel 4.

The right side of the box has two columns. The rightmost column is labeled
Counter Reading. This shows the last actual reading obtained from the counter. This
number is the time interval (in microseconds) between the signal from the reference
frequency and the signal being measured (see page 43). The leftmost column is labeled
Change in Time. This shows the difference (in microseconds) between the last two counter
readings. For example, if this number is 0.01, it means that the oscillator being measured
has moved 0.01 microsecond (10 nanoseconds) since the last counter reading was taken.
The system measured the performance of the oscillator by accumulating these time
differences.

The top part of the measurement screen contains a bar graph. This graph shows
the performance of each oscillator being measured over the last 24 hours. Each bar is a
different color. The color of the bar matches the color of the title for the channel.

The numbers at the top of the bar are a measure of oscillator performance. For
example, 6 represents a relative frequency of 1.00E-06, 9 represents a relative frequency
of 1.00E-09, and so on. The longer the bar, the better the performance of the oscillator.
If the bar is all the way to the right, the relative frequency of the oscillator is 1.00E-12
or better. This means that the oscillator is drifting less than 0.1 microsecond (100
nanoseconds) in 24 hours. If the bar is all the way to the left, the relative frequency of
the oscillator is 1.00E-06 or worse. This means the oscillator has drifted more than 0.1
seconds (100 milliseconds) in 24 hours.

The exact performance of the oscillator is shown on the daily phase plots (page 50).
The bar graph gives the operator a quick indicator of each oscillator's performance. By
looking at the bar graph, the operator can tell if there were any losses of the signals being
measured; for example, if the Loran-C receiver tracked properly for the last 24 hours. The
operator will also soon get a feel for how well each oscillator is expected to perform and
will quickly be able to notice if an oscillator is not performing properly.

52
 Keep in mind that the quantity being measured is a very small number. This means
that even a small error at any time during the daily measurement run will affect the data.
For this reason the equipment should be left on; sudden temperature changes should be
avoided, and electrical interference (from cleaning equipment, power devices, generators,
etc.) should be minimized.

Recordkeeping

Users of a measurement system should keep enough records to satisfy the needs of
the laboratory. For example, users of the system described above could keep the daily
plots in a notebook for future reference and diagnostic purposes. They can also keep the
data on computer disk and record the daily relative frequency values.

Users of an integral radio-oscillator-

recorder (page 45) can keep the chart
records provided to support claims of
traceable calibrations. If data from the
signal being received are listed in a NIST
or USNO publication, those documents
should also be kept.

All calibration labs that use a

measurement system should draw up a
system block diagram showing how signals
are generated, distributed, and calibrated.
When changes are made, they can be noted
on copies of the diagram. This block
diagram keeps the lab manager aware of
any changes and possible problem areas.

TRACEABILITY FOR FREQUENCY MEASUREMENTS

Traceability means that a measurement can be traced back to the national frequency
standard (in the United States, that standard is the NIST time scale). The NIST primary
frequency standard is physically located in Boulder, Colorado. Access to the NIST
frequency standard is by radio methods. Measurements made using the radio signals
discussed in chapters 5-7 are traceable to NIST.

53
 The rule for traceability is simple: The radio signal used must either be directly
controlled by NIST, or monitored by NIST. Signals controlled by NIST include WWV,
WWVB, WWVH, and GOES. The signals monitored (directly and indirectly) by NIST
include Loran-C, GPS, the Navy VLF signals, the Omega navigation signals, and Canadian
and European broadcast signals. All of these signals are useful because they are referenced
to atomic oscillators. Since they are carefully controlled, monitored, and intercompared
between national laboratories, traceability to NIST is assured. For example, users of the
signals from radio station CHU in Canada can be assured that these signals can be traced
to NIST, since the relationship between the Canadian and U.S. national standards is
known.

The accuracy of the traceability is by the number of steps in the

limited, however,
traceability path. Users should investigate the limitations so that no confusion exists over
what can or cannot be accomplished with a given signal.

We can illustrate traceability using Loran-C as an example. Loran-C is traceable to

NIST because its signals are monitored by NIST. The received signals from Loran-C are
continuously compared to the nation's frequency standard. Through this monitoring, NIST
is able to measure the accuracy of the Loran-C signals relative to the NIST time scale.

NIST (and the USNO) publish Loran-C data and make it available to Loran-C users. A
sample is shown below. By looking at this data, users can see that the day-to-day
variations between Loran-C and the NIST time scale are very small, a fraction of a
microsecond every 24 hours.

September Modified Loran-C Loran-C

1989 Julian Date (Dana 8970) (Fallon 9940)

15 47784 -0. 16 -0. 10

16 47785 -0.23 -0.25
17 47786 + 0.30 + 0.20
18 47787 + 0.21 -0. 10
19 47788 + 0. 06 + 0. 06
20 47789 + 0. 06 -0.21
21 47790 -0. 16 + 0.01
22 47791 -0. 16 -0.20
23 47792 -0. 16 + 0.01
24 47793 + 0. 12 -0.20
25 47794 -0. 11 + 0.01
26 47795 + 0.31 + 0.46
27 47796 -0. 04 -0. 11
28 47797 + 0. 17 + 0.26
29 47798 -0.34 -0. 58
30 47799 + 0. 05 + 0. 13

54
 The measurement system we just described used Loran-C as a reference frequency.
Users of the system measure the performance of their oscillator by comparing it to Loran-
C. The data in the figure were obtained by using the NIST time scale as a reference
frequency and using it to measure the performance of Loran-C.

The small variations in Loran-C do not concern most users. For example, the
relative frequency of Loran-C about 1.00E-12 over a 24-hour period when compared to
is

the NIST time scale. If the user is calibrating an oscillator to within 1.00E-09 of Loran-
C, the small errors in the Loran-C phase will not cause any problems. In this instance, the
performance of the reference frequency greatly exceeds the accuracy requirements of the
laboratory. This makes it fairly easy to establish traceability. From the user's viewpoint,
using Loran-C to make the calibration works just as well as using the NIST time scale.
Traceable radio signals like Loran-C make it possible for even the smallest lab to make
state of the art calibrations.

WHAT A FREQUENCY CALIBRATION MEASURES

Frequency calibrations measure the relative frequency of the oscillator. The time
interval overwhich measurements are performed is important. For the purposes of this
book, a time period of 1 day (24 hours) has been chosen.

From this starting point, two things

need to be considered: the mathematics of
calibrations, and the names and symbols to
be used. This section follows the
international practices of the frequency and
time industry and the major calibration
laboratories.

A measurement becomes a
calibration if it compares the oscillator's
long-term frequency to UTC(NIST), which
means Coordinated Universal Time at
NIST. Traceability to NIST means that
the oscillator was compared to UTC(NIST)
over a traceable path (using WWVB,
Loran-C, and so on). The oscillator then
becomes a traceable frequency source. The
normal length of a calibration is 24 hours
(1 day). There are cases, however, when
this time period can be shortened.

55
 The desired result of a calibration is to obtain the relative frequency of the oscillator
with respect to UTC(NIST). The international symbol for relative frequency is F. Relative
frequency is defined by the following equation:

f (actual) - f (nameplate)

F =
f (naneplate)

The Definition of
Relative Frequency

The nameplate frequency (usually 1, 5, or 10 MHz) is what the frequency of the

oscillatorshould be. If an oscillator operated exactly at its nameplate frequency, it would
be a perfect frequency source. In the real world, however, there is always some frequency
error, or a difference between the actual frequency and the nameplate frequency. This
frequency error is what a calibration measures. It is usually a very small number. In
much of the literature, the frequency error is referred to as delta-f (delta means a small
difference).

The equation finds the size of the frequency error. Dividing delta-f by the nameplate
value normalizes the equation. This lets the operator ignore the actual oscillator frequency
and concentrate on the frequency error.

To illustrate what we have covered so far, let's look at the numbers that calibration
technicians use to report their results. These are the same kinds of numbers used by
manufacturers when they quote specifications for their oscillator or frequency source. This
notation is also used internationally by many different countries.

56
 For example, consider a 1-MHz oscillator that is high in frequency by 1 Hz. The
numbers needed to compute the relative frequency of the oscillator are shown below. All
numbers have been converted to Hz so that we have a common unit to work with.

f(nameplate) = 1 000 000 Hz (1 MHz)

f( actual) = 1 000 001 Hz (1.000 001 MHz)

delta-f = 1 000 001 Hz - 1 000 000 Hz = 1 Hz

The quantity delta-f by itself is very useful. It shows us that we have a frequency
error of 1 Hz. But our use of mathematical notation will be more useful if we take one
more step. That step is called normalization and consists of dividing the frequency error
by the nominal frequency like this and labeling it as relative frequency (using the equation
we looked at earlier):

1 000 001 - 1 000 000

F = = 0.000 001 = 1.00E-06
1 000 000

This tells us that our source has a frequency error of 1 Hz (the numerator) but that
it only 1 Hz out of a million (the denominator). Using powers of ten for ease of
is

writing, we have 1.00E-06. The usefulness of this notation is obvious. We can describe
this oscillator as having a 1-Hz error in one million, a one part per million error, or we
can say that its relative frequency is 1.00E-06.

This form of notation is useful for the very small numbers that occur in the
measurement and calibration of high quality oscillators. Also, the notation gives us the
"feel" we need to make judgments about the devices we are using. The same 1-Hz error
for a 10-MHz oscillator is a smaller part of the nominal frequency and gives us a smaller
relative frequency. By dividing the delta frequency or frequency error by the nominal
frequency, we no longer have to worry about whether the nameplate frequency is 1 MHz,
10 MHz, or 3.5875 MHz. The relative frequency value (F) serves as a sort of report card
for oscillators. It lets us compare the performance of oscillators, regardless of their
nameplate frequency.

this notation is that it doesn't make any difference if we use the

Another benefit of
oscillatoroutput directly or if we use a divided version of the oscillator. For example, we
can divide the 1-MHz output of an oscillator to 1 Hz without changing the results. This
means that we can deal with sources of any frequency by using dividers to make the
measurement problems more manageable. It is much easier to deal with lower frequencies,
and the final results are the same.

57
 The sign of F changes from + to - if the oscillator output is lower than the
nameplate. This follows international convention and is the practice followed by many
manufacturers. F is a number and has no dimensions like hertz or percent. F does not
depend on the oscillator frequency. For example, if a 10-MHz oscillator had an actual
frequency output of 10 000 010 Hz, it would also have an F of 1.00E-06.

The graphs below show the frequency performance of a quartz, rubidium, and a
cesium oscillator. All three are graphs of actual data recorded at NIST and are typical
of the type of performance you can expect from each type of oscillator.

The first graph is of a low-cost quartz oscillator. This oscillator is typical of the
oscillatorsused as timebases in counters and other electronic test equipment. Its relative
frequency is 2.81E-08 (shown below the plot). The amount of oscillator drift over the 24-
hour period is nearly 3 000 microseconds.

Model V Quartz Oscillator us Cesium

1588 _ ,

1588 J

11/22/89 (MJD 47852) Channel 3 (MJD 47853) 11/23/89

16:88 [tic narks = 1 hour] 16:88
Data points= 25 Relative Frequency: +2.81E-88 r=+l.B8

The next graph (shown at the top of the next page) shows the performance of a
rubidium oscillator. Over the 24-hour period, the amount of oscillator drift is only about
4 microseconds. The relative frequency of the rubidium is 4.27E-11.

58
 Brand X Rubidium us Cesium

And below shows the performance of a cesium oscillator when compared

finally, the plot
directly to Loran-C. Because a radio signal is being used, this plot is not as smooth as the
other plots. However, notice that the two signals never differed by more than 0.5
microsecond (500 nanoseconds) over the 24-hour period.

11/22/89 (MJD 47852) Channel 1 MJD 47853) 11/23/89

16:88 [tic narks = 1 hour] 16:80
Data points= 25 Relative Frequency: -2.71E-12 r=-8.58

59
 SUMMARY

Time interval measurements are the method of choice when making accurate
frequency calibrations. You can buy or build many types of measurement systems based
on the time interval method, but all should run 24 hours a day and use a reference
frequency that is traceable to NIST. We encourage users to discuss their needs with
manufacturers of time and frequency equipment and to read the available literature.

is a way to talk about an oscillator's performance in a way

Relative frequency, or F,
that easy to understand; the oscillator is either higher or lower in frequency than it is
is

supposed to be. Measuring relative frequency requires dealing with very small quantities,
since the difference between the actual frequency and the nameplate frequency is usually
very small.

The next three chapters discuss radio signals that can be used as the reference
frequency for a measurement system. All of these signals are traceable to NIST. They
include high and low frequency broadcasts and signals broadcast from satellites. Obviously,
you will not want (or need) to use all of the services. Your accuracy requirements,
available manpower, and budget are all contributing factors in choosing a service that best
meets your needs.

60
 Chapter 5 - CALIBRATIONS USING HF RADIO BROADCASTS

High frequency (HF) shortwave radio broadcasts are a popular source of time and
frequency information. HF signals from stations such as WWV (Ft. Collins, Colorado),
WWVH (Kauai, Hawaii), and CHU (Ottawa, Canada) are readily available and provide
essentially worldwide coverage. A number of foreign countries operate services in these
frequency bands, and the signals can be received with relatively low-cost receivers.

In addition to the simpler, low-cost receivers, several manufacturers now provide

more elaborate receiving equipment. These receivers pick the best signal automatically by
re-tuning from one frequency band to another. Also, you can now buy receivers with a
built-in computer interface, making it possible to obtain a time code to reset a computer
clock. Of course, any shortwave radio receiver from the simplest to the most elaborate can
be used.

BROADCAST FORMATS
Many HF radio stations worldwide can be used for time and frequency measurements.
Many of these stations are listed in the table on pages 64-65. The following paragraphs
discuss the stations located in the United States and Canada.

WWV/WWVH
Standard time and frequency stations
WWV and WWVH are operated by the
National Institute of Standards and
Technology (NIST). WWV is in Ft. Collins,

Colorado, and WWVH is in Kauai, Hawaii.

Both stations broadcast continuous time and

frequency signals on 2.5, 5, 10, and 15 MHz.
WWV also broadcasts on 20 MHz. All
frequencies carry the same program, and at
least one frequency should be usable at all
times. As a general rule, frequencies above
10 MHz work best in the daytime, and the
lower frequencies work best at night.

The stations get their signals from a

cesium beam frequency source. They each use
three "atomic clocks" to provide the time of
day, audio tones, and carrier frequencies.

61
 The rates or frequencies of the cesium oscillators at the stations are controlled to be
within 1.00E-12 of the NIST frequency standard located in Boulder, Colorado. Time at the
stations is kept within a few microseconds of the NIST atomic time scale, UTC(NIST).

The seconds pulses or "ticks" transmitted by WWV and WWVH are obtained from the
same frequency source that controls the carrier frequencies. They are produced by a double
sideband, 100 percent modulated signal on each RF carrier. The first pulse of every hour
is an 800-millisecond pulse of 1500 Hz. The first pulse of every minute is an 800-
millisecond pulse of 1000 Hz at WWV and 1200 Hz at WWVH. The remaining seconds
pulses are brief audio bursts (5-millisecond pulses of 1000 Hz at WWV and 1200 Hz at
WWVH) that sound like the ticking of a clock. All pulses occur at the beginning of each
second. The 29th and 59th seconds pulses are omitted.

8.82b
—
Second

1880 Hertz WWV

1288 Hertz

WWVH

Each tick preceded by 10 milliseconds of silence and followed by 25 milliseconds

is

of silence to avoid interference from other time stations and to make it easier to hear the
tick. The total 40-millisecond protected zone around each seconds pulse is illustrated in the
figure. This means that the voice announcements are also interrupted for 40 milliseconds
each second. This causes only a small audio distortion. The ticks have priority and must
be received clearly.

The complete broadcast format for WWV

and WWVH, showing exactly what is

broadcast during each minute of the hour, is listed on pages 78-80.

62
 CHU
Canada has many frequency and time services similar to those in the United States.
The Canadian HF broadcast station CHU is located near Ottawa. Its signals can be heard
over much of the United States and are a valuable alternative to the WWV
signals. It has
the same propagation characteristics, and the same receiving techniques are used.

The CHU two ways. They are not

signals differ from those of the U.S. stations in
and they use a different format. The format is principally
in the standard frequency bands,
voice and ticks. At CHU, cesium frequency sources are used to generate the carriers and
the seconds pulses. Two systems are used to complement each other in cases of
maintenance or failure. The output from the cesium oscillator is fed to a frequency
synthesizer which produces the 3330-, 7335-, and 14670-kHz carrier signals for the
transmitters.

A 100-kHz signal is fed into one of two digital clocks, where it is divided into seconds
pulses of 100-Hz tone. The clock gates out the 51st to 59th second pulses of each minute
to permit the voice announcements to be inserted, and also gates out the 29th pulse to
identify the half-minute point.

The same 100-kHz signal is fed into two "talking clocks." One announces the station
identificationand time in English, and repeats the time in French. The other announces
the station identification and time in French, and repeats the time in English.

In 1975, CHU converted its operation to single sideband. It now broadcasts upper
sideband with full carrier. This is called the 3A3H mode. Users can still get the carrier
as a frequency standard, and an ordinary AM
radio will allow reception of the audio signals.

A time code appears in the 31st to the 39th seconds pulses of each minute. The
modulation is the commercial 300 baud FSK (frequency shift keying) at frequencies of 2025
and 2225 Hz. The code is a modified ASCII code in which each 11-bit character contains
two BCD digits, with one start bit and two stop bits. The first digit is the number 6 for
identification, which should be verified in the receiving clock; the remaining 9 digits give
the day, hour, minute, and second of UTC. The entire message is then repeated so that the
receiving clock can check for identical messages before updating. The code ends, and update
occurs at 0.5 second. This half-second must be added, along with the time zone hour, to
give the correct time. The same code is also available by telephone. Details can be obtained
by writing to the Time and Frequency Section, National Research Council, Ottawa, Ontario,
Canada.

The table on the following pages contains information about other HF Time and
Frequency Stations.

63
 ,

HF STANDARD FREQUENCY AND TIME SIGNAL BROADCASTS

Name Country Carrier Broadcast Days/ Hours/

Power (kW) Frequency (MHz) Week Day

ATA India 8 5, 10, 15 7 24

BPM China 10-20 2.5, 5, 10, 15 7 24

CHU Canada 3-10 3 .330, 7 . 335, 7 24

and 14.670

HLA Republic 2 5 5 7
of Korea

I AM Italy 1 5 6 2

IBF Italy 5 5 7 2 .75

JJY Japan 2 2.5, 5 , 8 7 24

10, 15

LOL Argentina 2 5, 10, 15 7 5

OMA Czecho- 1 2 . 5 7 24
slovakia

RCH USSR 1 2.5, 5, 10 7 21

RID USSR 1 5.004, 10.004, 7 24

15. 004

RIM USSR 1 5, 10 7 20.5

RTA USSR 5 10, 15 7 20.5

RWM USSR 5-8 4.996, 9.996, 7 24

14.996

VNG Australia 10 5, 10, 15 7 24

WWV United 2.5-10 2.5, 5, 10, 7 24

States 15, 20

WWVH United 5-10 2.5, 5, 10, 15 7 24

States

64
 5

HF STANDARD FREQUENCY AND TIME SIGNAL BROADCASTS (continued)

Name Country Carrier Broadcast Days/ Hours/

Power (kW) Frequency (MHz) Week Day

ZLFS New 0.3 2 . 1 3

Zealand

ZUO South 4 2.5, 5 7 24

Africa

RECEIVER SELECTION

This and the following sections discuss the use of HF broadcasts for time and
frequency calibrations. For convenience, we will refer mainly to WWV; however, the
discussions also apply to other HF broadcasts.

Almost any shortwave receiver can be used to receive HF signals for time and
frequency calibrations. However, for accurate time and frequency calibrations, it is
important to have a good antenna-ground system and a receiver with optimum sensitivity,
selectivity, image rejection, and frequency or phase stability. A block diagram of a typical
high-performance receiver is shown below.

V Antenna
OFF
Signal

ON
«—OH AGC —(\*) Strength Meter

RF 1st IF 2nd IF AF
Anapl Mixer Ampl Mixer Ampl Ampl

Speaker

Xtal VFO
Osc
o
Output
Preselector Main Tuning

Block Diagram of a High — Performanance

HF Receiver

65
 The first requirement of a good receiver is sensitivity. Therefore, a tuned RF
amplifier is desirable because it increases sensitivity. The next requirement is selectivity.
This is the ability of a receiver to reject interference. Often, either crystal or mechanical
filters are used.

The third feature of a good receiver

is the ability to reject interference from an

undesired signal at what is image frequency. Since the IF signal is the difference
called the
between the local oscillator frequency and the incoming signal, there are always two
different incoming signals that can produce the same IF signal. One is above and the other
below the oscillator frequency. Refer to radio texts for further explanations.

A local crystal oscillator can be used in these receivers to provide tuning stability.
Other information regarding well-designed receivers can be found in such reference material
as the Radio Amateur's Handbook.

Several receivers will automatically tune themselves to WWV. If timing is a special

problem in your home or business, of course check the receiver very carefully before you
purchase it. The single best option to look for in a receiver is the provision for an external
antenna. This may be just a short wire or an elaborate rooftop installation. Fancy knobs
and adjustments won't help much if you can't get a good signal.

ANTENNA SELECTION
Some HF stations transmit on more than one frequency. Because of changes in
ionospheric conditions, most receivers are not able to pick up the signals on all frequencies
at all times in all locations. However, unless there are severe magnetic disturbances (which
make radio transmissions almost impossible), users should be able to receive the signal on
at least one of the broadcast frequencies. As a general rule, frequencies above 10 MHz
provide the best daytime reception, while the lower frequencies are best for nighttime
reception.

All-band antennas capable of covering the entire HF band are commercially available.
These are used for around-the-clock communication, where the entire HF band must be used
for maximum reliability. Depending on location, an antenna and a receiver capable of
receiving two or more frequencies may be required. Any of the different WWV
frequencies
transmitted may be received depending upon distance, time of day, month, season, and
sunspot cycle. This section discusses the type of antenna best suited to each frequency. For
more information, refer to antenna manuals such as those published by the American Radio
Relay League.

66
 Antennas for the 2.5- to 4-MHz Range

In order to use these frequencies during the daytime, your receiver should be located
within 320 kilometers (200 miles) of the transmitter. This is because the ground wave from
the transmitter will travel only a short distance. These frequencies become more useful at
night, however, especially during the winter season in the higher latitudes, where longer
nights prevail. Reception is then possible over distances of several thousand miles.

There are two types of antennas which work in the 2.5- to 4-MHz range. The
vertical monopole quarter-wavelength antenna is very effective in receiving long distance
skywave signals normally arriving at angles of 20 degrees or less. It is also useful for
receiving weak ground wave signals.

For nighttime reception on paths up to several thousand miles, the skywave is

predominant, and a horizontal dipole half-wavelength antenna is recommended. The antenna
should be located a quarter-wavelength or higher above ground. Choose an area free from
obstructions and interference.

67
 Antennas for the 4- to 5-MHz Range

These frequencies can be received at greater distances than the 2.5-MHz frequency
throughout the day or night, especially during the minimum of the sunspot cycle. Reception
is possible up to 1600 kilometers (1000 miles) under ideal conditions, but the range is

usually less than that during the day.

At night, 5 MHz becomes a very useful frequency for long-range reception except
during maximum sunspot cycle. It is excellent during early dawn and early evening in the
winter months. Horizontal dipole antennas are a good choice for this band. You can also
consider using a multiband antenna.

Antennas for the 7- to 10-MHz Range

At these frequencies, reception over great

distances is possible during the day or night, but

they are still dependent upon the sunspot cycle.

During the minimum sunspot cycle, great
distances can be covered using 7-10 MHz in the
daytime when higher frequencies cannot be
received. And during the maximum sunspot cycle,
Copper these are probably the best frequencies to use at
Wire night when the lower frequencies cannot be
heard. However, during the maximum sunspot
cycle, reception is limited to short distances
during the day with limitations comparable to
those noted for 5 MHz. The 7-10 MHz range can
Antenna also provide daytime reception at fairly close
Height =
Support range, and can be used when 5-MHz reception is
Wavelength poor. The half-wavelength horizontal dipole
antenna should be selected for short distances. A
quarter-wavelength vertical monopole antenna is
suitable at greater distances.

Antennas for the 14- to 15-MHz Range

Cable to These frequencies are best suited for long-

Receiver range reception during the day. They are not
usable for short-range reception except during
periods of maximum sunspot activity. However,
for long-range reception, they are the most
favored frequencies during both sunspot cycle
Quarter-Wavelength Vertical conditions. Under average conditions, the
maximum wave angle is limited to 30 degrees or
Monopole Antenna less.

68
 During maximum sunspot cycle, reception may even be possible during the night in
some During minimum sunspot cycle, they are useful only during the daylight
locations.
hours and dawn and dusk periods. Both horizontal and vertical antennas work well in this
range.

ANTENNA DIMENSIONS FOR 2.5, 5, AND 10 MHz

Quarter-WaveLength Half-WaveLength Horizontal
Frequency Vertical Antenna Dipole Antenna

(MHz) Height Length Height

28.37 m 56 . 1 m 15 to 30 m
2 . 5
94 1
- 7" 187 1
50' to 100'

14 . 05 m 28.07 m 9 to 15 m
5.0
46' - 10" 93' - 7" 30' to 50'

7.05 m 14.05 m 7.5 to 15 m

10. 0
23' - 6" 46" - 10" 25' to 50'

Antennas for 20 MHz

Normally, 20 MHz is the best frequency to use for daytime reception and will be
optimum at either noon or a few hours past noon. Signals at this frequency arrive at very
low wave angles and are useful only for long distance reception. During the minimum of
the sunspot cycle, reception is poor but improves during the winter. During the maximum
of the sunspot cycle, the reception is excellent at night and during the day. The vertical
monopole that favors low wave angle radiation has been used at this frequency with
favorable results.

69
 USING HF BROADCASTS FOR TIME CALIBRATIONS

Time of day is available from many sources in the United States and Canada. Radio
and television stations mention the time frequently and, in fact, use the time of day to
schedule their own operations. Telephone companies offer time-of-day services in many
locations.

Most time-of-day services in the United States start at NIST. A NIST telephone
service is available by calling (303) 499-7111. In addition, WWV and WWVH broadcast voice

time-of-day once each minute. This is also the case for CHU, where the time is given
alternately in French and English. Using the WWV voice announcement and the tone
following the words, "At the tone, XXXX hours XXXX minutes Coordinated Universal Time,"
a person can check a wall clock or wristwatch to within a fraction of a second. The UTC
time that is announced can be converted to local time by using a time zone map.

Recovering time at higher

accuracies using electronic
requires
equipment. For example, if an
oscilloscope is used to see the seconds
ticks, the user can set clocks with much
greater accuracy than is possible by
only listening to the ticks.

The following sections describe

several methods of recovering time of
day with resolutions better than 1
millisecond (1000 microseconds) by
using an oscilloscope. Under very
favorable conditions, it is possible to
recover time to within 100
microseconds. In each case, path and
equipment time delay corrections are
necessary for accurate results.

For best results using any of these methods, the following guidelines are recommended:

o Make measurements at the same time every day.

o Avoid twilight hours, when the ionosphere is the least stable.

o Use the highest frequency that has good reception.

o Look at the received signals on the oscilloscope for a few minutes to judge
the stability of propagation conditions.

70
 Measuring the Receiver Time Delay

For accuracy in time-setting, the receiver delay must be known. Typical delays in
receiver circuits are usually less than a millisecond. However, they should be known to the
user who wants the very best timing accuracy. The actual delay in the receiver will vary
with tuning and knob settings.

To measure the receiver time delay, the following equipment is required (connected
as shown in the diagram):

o an oscilloscope with a calibrated, externally triggered time base

o an HF signal generator

o an audio signal generator

RECEIVER
Oscilloscope
Audio HF
Signal Signal
T
IN OUT VERT Generator Generator

1 kH;

Equipment Setup for Receiver

Time-Delay Measurements

The receiver should be tuned to receive WWV/WWVH signals. This is because the
receiver delay time varies with slightly different receiver dial positions. Therefore, the
receiver tuning should be set and marked where the maximum signal is received. The
frequency of the HF signal generator is then adjusted for peak receiver output.

71
 The audio is set to a 1-kHz output frequency.
signal generator An accurate 1-kHz
signal is not required. The HF
generator is externally modulated by the 1-kHz signal.
The oscilloscope sweep rate is set to 100 microseconds per division with positive external
triggering from the 1-kHz signal. The vertical amplifier gain is set high for a large vertical
deflection. The vertical position control is adjusted for zero baseline with no input signal.

Initially, the 1-kHz signal generator

is connected to the vertical input of the

oscilloscope. The adjusted so that the trace crosses or touches the horizontal
trigger level is

center line at the left. The horizontal position can be adjusted so that the signal crosses
the first division on the left as shown below. The crossover point of the undelayed signal
will serve as the zero delay reference point.

—1—

\ r/
4

Delay
Time
Delayed Zero Crossover

Undelayed Zero Crossover

Without touching any of the oscilloscopecontrols, disconnect the 1-kHz signal from
the vertical input and replace with the delayed 1-kHz signal from the receiver output.
it

Since a receiver delay is almost always less than 1 millisecond (1 cycle of a 1-kHz signal),
there is little chance of ambiguity on which cycle to measure. The delay is equal to the
sweep time from the reference undelayed crossover to the first delayed crossover point.

72
 If the delayed signal's phase is the opposite of the reference 1-kHz signal, the receiver
has an inverted output signal. However, the receiver delay time remains unchanged, and
the only difference will be that the output seconds pulse will be inverted with a negative
leading edge. For receiver delays of less than 500 microseconds, the sweep rate can be 50
microseconds per division.

This technique produces a local signal that approximates the timing signal. It uses
the same frequency on the dial and uses the same tone frequency. If you have a two-
channel scope, you can display the tone before it enters the RF signal generator and after
it comes out of the receiver.

Using an Adjustable Frequency Source to Trigger the Oscilloscope

This method (sometimes called the direct trigger method) requires an oscilloscope
with an external sweep trigger and accurately calibrated time base, and a receiver with
electronic audio output. The frequency source (clock) must be adjustable so its 1-pps output
can be advanced or retarded. Connect the equipment as shown:

Seconds Oscilloscope Hz
HF 1
Freq.
<
Revr Vertical Trigger Source
Tick

A 1-Hz pulse from the frequency source is used to trigger the oscilloscope sweep.
At some time interval later during the sweep, the seconds tick appears on the oscilloscope
display. The time interval from the start of the sweep to the point where the tick appears
is the total time difference between the local clock and the transmitting station. By
subtracting the path and receiver time delays from the measured value, the local clock time
error can be determined (path delays are discussed on page 85). The equation to determine
time error at a receiving location is shown on the next page.

73
TIME ERROR = TR - TT = TD - (PPD + RTD)

where: TR = time at receiving station

TT = time at transmitting station (WWV, WWVH)

TD = total time difference (measured using oscilloscope)

PPD = propagation path time delay (computed)

RTD = receiver time delay (measured)

Note: The units (usually milliseconds) should be the same for all terms.

Tune the receiver to the station and set the oscilloscope sweep rate
to 0.1 second per
division. You can listen to the broadcast tocheck the quality of reception. The tick will
then appear on the oscilloscope display. If the tick is one division or more from the left
side of the scope display, adjust the time of the local clock until the tick falls within the
first division from the left side. If the local time tick is late, the received tick will be heard
before the sweep starts. If this is the case, adjust the local clock until the tick appears.

After the local seconds pulse has been properly adjusted and appears within the first
division (0.1 second in time), increase the sweep rate. Then adjust the clock again until
the leading edge of the received pulse starts at a time equal to the propagation delay time
plus the receiver delay time after the trigger.

The sweep rate should be increased to the highest rate possible. However, the total
sweep time should not exceed the combined propagation and receiver deiay time minus 5
milliseconds (to allow for the length of the received seconds pulse).

With a sweep rate of 1 millisecond per division, for example, greater resolution can
be realized by measuring the second zero crossover point of the 5 millisecond received ticks.
Although the leading edge of the seconds pulse as broadcast from these stations is "on
time," it is difficult to measure due to the slow rise time at the beginning of the burst and
distortion due to propagation. For this reason, the second zero crossover should be used.
The second zero crossover of the WWV or CHU pulse is delayed exactly 1000 microseconds,
and the WWVH crossover is delayed 833 microseconds. The delay is called the cycle
correction.

74
 At a sweep rate of 1 millisecond per division, any changes in arrival time (jitter) are
readily apparent. After watching the pulses for a period of a minute or two, select a cycle
that is undistorted and relatively large in amplitude. In determining the time at a receiving
location, include the delay of the chosen zero crossover point, then add the cycle correction
to the propagation and equipment delay using the following equation:

TIME ERROR = TR - TT
= TD - (PPD + RTD + cycle correction)

where: TR = time at receiving station

TT = time at transmitting station (WWV, WWVH)

TD = total time difference (measured using oscilloscope)

PPD = propagation path time delay (computed)

RTD = receiver time delay (measured)

Cycle correction = 1000 microseconds per cycle (WWV or CHU)

= 833 microseconds per cycle (WWVH)

Delayed Triggering: An Alternate Method That Doesn't Change

the Clock Output

To improve the resolution of measurement, the oscilloscope sweep must be operated

as fast as possible. You can sweep the scope faster by generating a trigger pulse
independent of your clock and then positioning the pulse for maximum sweep speed.
However, you then must measure the difference between the clock and the trigger pulse.
This can be accomplished by using an oscilloscope with a delayed sweep circuit built in or
with an outboard trigger generator. The latter method is discussed here, but an oscilloscope
with a delayed sweep could also be used. Refer to the instrument manual for assistance.

On a typical digital delay generator, a delay dial indicates the delay between the
input local clock tick and the output trigger pulse. If the user already has a variable rate
divider to produce delayed pulses, a time interval counter can be used instead of the delay
generator. In either case, the trigger delay must be accounted for in measuring the time
delay (TD) of the received tick with respect to the local master clock.

75
 Measurements should be made at the same time every day (within 10 minutes) for
consistent results. Choose the time of day when it is approximately noon midway between
the transmitting station and the receiver's location. For night measurements, choose a time
when the midpoint is near midnight. Do not make measurements near twilight.

Connect the equipment as shown below. A commercially available frequency divider

and clock can be used in place of the controlled delay generator. A time interval counter
is then used to measure the output of the delayed clock to the master clock. The output
of the delayed clock is used to trigger the oscilloscope. The procedures described previously
also apply to this method.

100 kHz
Delayed 1 Hz

f Antenna
Delay Gen
Master
or Divider

& Clock Clock

1 Hz

With the oscilloscope sweep adjusted to 1 millisecond per division, the trigger pulse
should be delayed by an amount equal to the propagation delay in milliseconds. Don't worry
about fractional milliseconds in the delay. The sweep should be adjusted so that it begins
exactly at the left end of the horizontal graticule and is vertically centered.

The second zero crossover point of the tick (see figure on next page) should be
observed and carefully measured. With the sweep at 1 millisecond per division, the delay
of the second zero crossover on the oscilloscope is measured to the nearest one-tenth of a
millisecond and added to the trigger delay, resulting in an approximate total time delay.

76
If the local clock 1-pps time is exactly coincidentwith the UTC(NIST) seconds pulse, the
total measured time delay be approximately equal to the sum of the propagation delay
will
time, the receiver delay time (typically 200 to 500 microseconds), and the cycle correction
(1000 microseconds for WWV
or CHU and 833 microseconds for WWVH).

< > Second Zero Crossover

To further increase the resolution of delay measurement, increase the oscilloscope

sweep rate 100 microseconds per division and adjust the trigger pulse from the generator
to
to be approximately 500 microseconds less than the total delay time previously measured.
At these settings, the second zero crossover of the tick will be somewhere near the middle
of the oscilloscope face.

The vertical centering of the sweep should be rechecked and centered if necessary.
The tick is measured to the nearest 10 microseconds. The result should be within ± 100
microseconds of the result obtained at the 1 millisecond per division sweep rate. If the
result of this measurement falls outside this tolerance, then the procedure should be
repeated by measuring the time delay at a sweep rate of 1 millisecond per division. Use the
equations on pages 74-75 to obtain the time.

77
 Using Oscilloscope Photography for Greater Measurement Accuracy

By photographing five or more overlapping exposures of the WWV/WWVH tick, you

can estimate an average of the tick arrival time with more accuracy. The exposures are
made when consistently strong and undistorted ticks appear on the oscilloscope. To
determine the time, the average of the second zero crossover point of the tick is measured
using the procedure described previously.

To make measurements using this technique, you need an oscilloscope camera that
uses self-developing film. Place the camera shutter in the time exposure position so that
it can be opened and closed manually. The lens opening of the camera, the oscilloscope
trace intensity, and the scale illumination must be determined by experiment. Refer to your
camera manual.

One of the previously described procedures is followed to obtain the seconds tick.
At a sweep opened before the sweep starts
rate of 1 millisecond per division, the shutter is

and closed after the sweep ends. This is repeated each second until five overlapping
exposures are completed. The pictures should be taken when the ticks begin to arrive with
the least distortion and maximum amplitude.

This procedure can also be used at a faster sweep rate of 100 microseconds per
division with the second zero crossover point appearing in the middle of the trace (one
complete cycle of the tick should be visible). By taking overlapping exposures of the ticks,
you can obtain an average reading from the photograph.

USING THE WWVAVWVH TIME CODE

The broadcasts from WWV and WWVH include a time code. The time code signal is
100 Hz away from the main carrier and is called a subcarrier. The code pulses are sent
out once per second. With a good signal from a fairly high-quality receiver, you can hear
the time code as a low rumble in the audio. HF receivers that receive and decode this
signal can automatically display the time of day. The next section describes the time code
format.

Time Code Format

The WWV and WWVH time continuously broadcast in binary coded decimal
code is

(BCD) format on a 100-Hz subcarrier. The time code is a modified version of the IRIG-
H code. The code is transmitted serially on a 100-Hz subcarrier at a rate of one pulse per
second.

78
 The time code is in binary coded decimal (BCD) format. Groups of binary digits
used to represent decimal numbers. The binary-to-decimal weighting scheme is
(bits) are
1-2-4-8. The least significant bit is always sent first. The table below shows the BCD
groups and the equivalent decimal number:

BINARY GROUP DECIMAL

Weight: 1 2 4 8 EQUIVALENT

0 0 0 0 o
1 0 0 0 1
0 1 0 0 2
1 1 0 0 3
0 0 1 0 4
1 0 1 0 5
0 1 1 0 6
1 1 1 0 7
0 0 0 1 8
1 0 0 1 9

The decimal number is obtained by multiplying each bit in the binary group by the
weight of its respective column and then adding the four products together. For example,
the table shows that the binary group 1010 is equal to 5. This is derived by:

(1 x 1) + (0 x 2) + (1 x 4) + (0 x 8) = 1 + 0 + 4 + 0 = 5

In the standard IRIG-H code, a 0 bit consists of exactly 20 cycles of 100-Hz amplitude
modulation (200 milliseconds duration), and a 1 bit consists of 50 cycles of 100 Hz (500
milliseconds duration). The WWV/WWVH code differs from IRIG-H because all tones are
suppressed briefly while the seconds pulses are transmitted.

Tone suppression also deletes the first 30 milliseconds of each binary pulse in the
time code. This makes the WWV/WWVH
bits 30 milliseconds shorter than the IRIG-H bits.
Therefore, 170 millisecond pulses are recognized as 0 bits, and 470 millisecond pulses are
recognized as 1 bits. The leading edge of each pulse coincides with the positive-going
crossing of the 100-Hz subcarrier; but due to the tone suppression, it occurs 30 milliseconds
after the start of the second.

Within minute, enough bits are sent to express the minute, hour, and day of year,
1
the UT1 and a Daylight Saving Time (DST) indicator.
correction, The coded time
information refers to the time at the start of the 1-minute frame. Seconds are determined
by counting pulses within the frame. Two BCD groups are needed to express the hour (00

79
to 23) and minute (00 to 59); and three groups are needed to express the day of year (001
to 366). Some bits in the BCD groups are unused, but may provide additional information
in the future. To represent units, tens, or hundreds, the basic 1-2-4-8 weights are multiplied
by 1, 10, or 100, as appropriate.

Each frame begins with a unique spacing of pulses that mark the start of a new
minute. During the first second of the minute, no pulse is transmitted. This creates a 1-
second (1000-millisecond) hole. Since the pulses are already delayed 30 milliseconds by the
tone suppression, the UTC minute actually begins 1030 milliseconds (1.03 seconds) earlier
than the first pulse in the frame. For synchronization purposes, a position identifier pulse
is transmitted every 10 seconds. The position identifier pulse lasts for 770 milliseconds (77

cycles of 100 Hz).

UT1 corrections are sent during the final 10 seconds of each frame. These
corrections are to the nearest 0.1 second. The UT1 correction is expressed with bits called
control functions. Control function #1 occurs at 50 seconds and tells whether the UT1
correction negative or positive. If a 0 bit is sent, the correction is negative; if a 1 bit is
is

sent, the correction is positive. Control functions #7, #8, and #9 tell the amount of the
UT1 correction. They occur at 56, 57, and 58 seconds, respectively. Since the UT1
corrections are in tenths of seconds, the binary-to-decimal weights are multiplied by 0.1.

DST sent by control function #6, at 55 seconds. If DST is in effect,

information is

a 1 bit is sent. If Standard Time is in effect, a 0 is sent. The setting of this bit is changed
a few hours prior to 0000 UTC on the date of change. This schedule notifies users in the
continental United States of the time change several hours before it occurs locally (usually
at 2:00 a.m.). Receivers that display local time can read control function #6 and make the
1-hour adjustment automatically when time changes occur.

Additional changes to the WWV/WWVH time code formats are planned for
implementation during early 1991. These involve addition of two digits of the current year,
a leap second warning indicator, and improved indicators of Daylight Saving Time. Contact
NIST for current status information regarding these changes.

USING HF BROADCASTS FOR FREQUENCY CALIBRATIONS

In addition to the widely used time service of WWV

and WWVH, standard frequencies
are also available to the broadcast listener. Both stations provide a calibrating frequency
that is readily available for comparison and measurement.

There are two ways to use WWV

(or WWVH) as a standard frequency. You can
measure your own frequency source against WWV
and write the difference down for
reference, or you can adjust your frequency source to make it agree with WWV.

80
 In this section, several calibration methods are discussed.These methods use WWV-
type signals to calibrate lower accuracy quartz oscillators. Under good conditions, relative
frequencies of 1.00E-07 are possible. Direct frequency comparison or measurement with
WWV can usually be accomplished to about 1.00E-06. This resolution can be improved by
long-term (many weeks) time comparison of clocks operated from a frequency source rather
than direct frequency comparison. Three methods of calibrating frequency sources using the
broadcasts of WVW/WWVH are discussed: (1) beat frequency method; (2) oscilloscope pattern
drift method; and (3) frequency calibrations by time comparisons.

Beat Frequency Method

The beat frequency or heterodyne method is a simple technique commonly used by

radio operators to calibrate transmitters and tune receivers. A frequency offset of about
1.00E-06 can be accurately determined. This means that this method can be used to
1-MHz signal to within 1 Hz.
calibrate a

The below shows an arrangement for calibrating a 100-kHz oscillator. A 100-

figure
kHz signal containing harmonics is coupled to the receiver input along with the WWV
signal from the antenna.

WWV /WWVH
Broadcast

Speaker

22 pf cap
100 kHz
Square Square 100 kHz
Wave Wave Quartz
Generator Oscillator

Equipment Setup for

Beat Frequency Method

81
 The beat frequency method works by heterodyning or mixing a known and accurate
frequency (like a WWV RF signal) with the output of an oscillator. The mixing is
accomplished by the converter circuit in any superheterodyne receiver. The difference
frequency of the two RF signals can be amplified and detected. The result is an audio
output signal called the beat frequency or beat note.

The frequency of the beat note is the difference of the two input frequencies. When
the two frequencies are made equal, their difference decreases to zero and is called "zero
beat." When zero beat is reached, the oscillator is equal in frequency to the WWV
frequency.

To calibrate a frequency source or quartz oscillator with an output frequency lower

than that broadcast by WWV, the correct harmonic equal to the WWV
signal is required.
For example, if a 100-kHz signal is to be calibrated with the WWV
5-MHz carrier frequency,
then it must also contain a harmonic 50 times itself. This means that the signal to be
calibrated has to be a submultiple of the WWV
carrier frequency.

Theoretically, a wave does not contain any harmonics. In practice, though, all
sine
sine wavesignals contain enough harmonics to produce a beat note. A square wave signal,
on the other hand, is very rich in harmonic content and is ideal for generating harmonics
to calibrate receivers and transmitters in the HF and VHF band. A simple method of
generating a square wave from a sine wave is by clipping the signal with a diode clipping
circuit. To obtain a strong harmonic signal for beat notes requires a large amplitude signal
to produce heavy clipping. A better method is to digitally condition the 100-kHz signal to
produce square waves.

input impedance is low (50 to 100 ohms), a 10- to 20-picofarad (pF)

If the receiver
capacitor can be used to couple the high frequency harmonic to the receiver input and to
reduce the level of the lower fundamental frequency. If the receiver has a high input
impedance with unshielded lead-in wire from the antenna, the harmonic signal can be loosely
coupled to the receiver input by wrapping a few turns of an insulated wire around the
antenna lead-in and connecting it directly to the output of the oscillator. For receivers with
built-in or whip antennas, you must experiment to find a way to inject the oscillator signal.

The relationship between the oscillator error and the beat note that is measured
during the calibration looks like this:

f (actual) = f(nameplate) + delta-f

The actual oscillator output is is made up of two

designated as f(actual). This parts,
the correct frequency f(nameplate), plus an error that we can designate as delta-f.

82
 When we make our measurement, a harmonic of f(actual) is beat against the carrier
frequency, f(carrier). The resulting beat note f(beat) is the difference between the two and
is written as:

f(beat) = f (carrier) - harmonic of f (actual)

A negative sign in the answer is ignored, so the answer is always a positive number.

To find the frequency error, delta-f, we divide f(beat) by the number of the harmonic (N):

f (beat)
delta-f
N

For example, if a beat frequency of 100 Hz is measured between the 5-MHz WWV
carrier frequency and the 50th harmonic of a 100-kHz oscillator signal, the frequency error
of the 100-kHz signal is:

100 Hz
delta-f = 2 Hz
50

The oscillator frequency is in error by 2 Hz. The relative frequency of the oscillator
is 2 parts in 100 000 or 2.00E-05. To determine whether the oscillator is high or low in
frequency, the oscillator frequency must be changed to note which way the beat frequency
decreases. If increasing the oscillator frequency decreases the beat note, it indicates that
the oscillator frequency is lower than the WWV/WWVH
frequency.

If the above 50 Hz, headphones, a speaker, or a counter can be used.

beat note is

Below that frequency, a dcoscilloscope can be connected to the receiver detector. A signal
strength meter can be used and the beats counted visually. The automatic gain control
(AGO should be disabled, if possible, for the meter fluctuations to be more noticeable. The
manual RF gain can be adjusted to compensate for loss of the AGC.

83
 To correct the oscillator frequency, turn the adjustment knob in the direction that
lowers the frequency of the beat note. Between 50 Hz and about 1 Hz, the beat note cannot
be heard, and the signal strength meter will begin to respond to the beat note as it
approaches 1 Hz. As zero beat is approached, a very slow rise and fall of the background
noise or the WWV
audio tone can also be heard on the speaker. The meter effect is much
easier to follow. As it approaches zero beat, the very slow rise and fall of the signal
strength may sometimes become difficult to distinguish from signal fading due to propagation
effects.

Oscilloscope Pattern Drift Method

The oscilloscope pattern drift method is good for comparing two frequencies if you
have an oscilloscope with external triggering. The method requires an oscilloscope with
an accurately calibrated sweep time base. External triggering is obtained from the signal
to be calibrated. This signal can be any integer submultiple of the tone being received from
WWV/WWVH. The receiver (tuned to WWV
or WWVH) has its audio output connected to
the vertical input of the oscilloscope. With the sweep rate set at 1 millisecond per division,
the trigger level is adjusted so that a zero crossover of the corresponding 600- or 500-Hz
signal is near midscale on the scope.

By measuring the phase drift during a given time interval, the frequency error is
determined. If the zero crossover from the oscillator being measured moves to the right,
the oscillator is higher in frequency than the WWV
signal, and if it moves to the left, the
oscillator is lower in frequency (see figure below). Under ideal conditions, a relative
frequency of 1.00E-06 can be determined by increasing the observation time.

k2i

Initial Final

84
 Frequency Calibrations by Time Comparisons

you are already making daily time calibrations using WWV or WWVH, you can use
If
the data you obtain to make frequency calibrations of the oscillator driving your clock. You
can do this by recording the daily time differences and then calculating the frequency rate.
The operation is similar to telling your jeweler that your watch gains or loses so many
seconds a day. He then adjusts the rate to compensate.

The operation proceeds something like this. Each day, note the amount the clock
output differs from WWV. Then keep an accurate record of the history of the oscillator in
terms of the time gained or lost with respect to WWV
ticks. This change in time can then
be converted to the relative frequency of your oscillator. See chapter 4 for more information
on the mathematics involved.

In summary, frequency calibrations can be made by recording only the time error
produced by an oscillator. This technique is recommended for lower quality oscillators and
for situations when you are already making time calibrations.

MEASURING RADIO PATH DELAY

Measuring radio path delay involves two things. First, the great circle (curved)
distance must be found between the transmitter and receiver. A simple computer program
to accomplish this (written in BASIC) is listed below. This program calculates the great
circle distance between any two points on the earth.

'
Great Circle Distance Program (written in BASIC)

'
function to convert radians to degrees
def fnRadToDegrees(x) = x/57.2957795131
'
function to calculate the arccosine of a number
def fnArcCos(x) = ((-atn(x/sqr(-x*x+l)) + 1.5707633)*57.2957795131)

els '
clear the screen
dim pointa(6) '
array for point A coordinates
dim pointb(6) '
array for point B coordinates

print "This program calculates the distance and the path delay between"
print "points A and B. You must enter the coordinates for both points."
print
print "Enter Coordinates for Point A"
print
input "Is the Latitude in Degrees North (Y or N, Default is Y): ", ql$
input "Is the Longitude in Degrees West (Y or N, Default is N): ", q2$
if ql$o"n" and ql$o"N" then NorthA=-l else NorthA=0

85
print
input "Latitude Degrees: ", pointa(l)
input "Latitude Minutes: ", pointa(2)
input "Latitude Seconds: ", pointa(3)
input "Longitude Degrees: ", pointa(4)
input "Longitude Minutes: ", pointa(5)
input "Longitude Seconds: ", pointa(6)

'
reduce latitude and longitude of Point A to two floating point numbers
yl=pointa(l) + (pointa(2)*60+pointa(3))/3600
y2=pointa(4) + (pointa(5)*60+pointa(6))/3600
yl =fnRadtoDegrees(yl) Point A latitude '

y2=fnRadtoDegrees(y2) '
Point A longitude
if q2$="N" or q2$="n" then y2=fnRadtoDegrees(360)-y2 ' adjust longitude if in eastern hemisphere

print
print "Enter Coordinates for Point B"
print
input "Is the Latitude in Degrees North (Y or N, Default is Y): ", ql$
input "Is the Longitude in Degrees West (Y or N, Default is N): ", q2$
if ql$o"n" and ql$o"N" then NorthB=-l else NorthB=0
print
input "Latitude Degrees: ", pointb(l)
input "Latitude Minutes: ", pointb(2)
input "Latitude Seconds: ", pointb(3)
input "Longitude Degrees: ", pointb(4)
input "Longitude Minutes: ", pointb(5)
input "Longitude Seconds: ", pointb(6)

'
reduce latitude and longitude of Point B to two floating point numbers
zl=pointb(l) + (pointb(2)*60+pointb(3))/3600
z2=pointb(4) + (pointb(5)*60+pointb(6))/3600
zl =fnRadtoDegrees(zl) Point B latitude '

z2=fnRadtoDegrees(z2) Point B longitude '

if q2S="N" or q2S="n" then z2=fnRadtoDegrees(360)-z2 adjust longitude if ' in eastern hemisphere

p=cos(abs(y2-z2)) '
cosine of difference in longitude between points A and B
y3=sin(yl) '
sine of Point A latitude
y4=cos(yl) '
cosine of Point A latitude
z3=sin(zl) '
sine of Point B latitude
z4=cos(zl) '
cosine of Point B latitude

if NorthAoNorthB then '

points are on opposite sides of equator
C=y4*z4*p-y3*z3
else
C=y4*z4*p+y3*z3 '
points are on same side of equator
end if

c=fnArcCos(c) '
distance in nautical miles
miles=c*60*1.151 '
distance in miles
kilos=c*60*1.8522 '
distance in kilometers
pathdelay=miles*5.376344086 '
path delay in microseconds
print
print
print "Distance in Miles: ";miles
print "Distance in Kilometers: ";kilos
print "Path Delay in Microseconds: "jpathdelay
end

86
 The second thing we need to know is the number of times the signal bounces from
the ionosphere to the Earth (hops). This factor part of the propagation delay that the
is

signal experiences along its path. The computer program we just looked at estimates the
path delay, but does not take propagation delay into account. For groundwave signals
transmitted at low frequencies (like those discussed in chapter 6), this estimate is probably
very accurate. However, for HF signals it probably isn't. The reason for this is that at any
particular time, the path delay may not be at its average value. Since the path is constantly
changing by small amounts, it cannot be estimated with great accuracy.

Knowing the exact number of hops between the Earth and ionosphere is important;
otherwise, estimates of path delay can be off by 500 to 1000 microseconds (0.5 to 1
millisecond). If you are trying to make time calibrations to within 1 millisecond, you need
more accurate estimates of the path delay.

The ionosphere is made up of a number of layers whose actual height varies both
daily and seasonally. This changes the path delay a lot if the receiver and transmitter are
close toeach other. Since the Earth is round, there is a maximum ground distance a single
hop can span. This is about 4000 kilometers or 2200 miles. For greater distances, the radio
waves must obviously be reflected a number of times. Any error in the wave path
computation for one hop must be multiplied by the total numbers of hops.

To reduce the error in estimating the path delay, the frequency used for reception
should be selected for the least number of hops. This should also result in a stronger
signal. Some frequencies will penetrate the ionosphere and not reflect. Other frequencies
will skip over your receiving site, so a lower frequency must be selected. Use of the
maximum frequency that is receivable assures the least number of hops. This maximum
usable frequency is called the MUF. A frequency about 10 percent below the MUF provides
the best reception.

For distances under 1600 kilometers (1000 miles), the single-hop mode of
transmission dominates. For short distances, the height must be estimated carefully. For
distances of less than 1600 kilometers, where only a single-hop mode of transmission occurs,
a much wider range of error in the height can be tolerated.

Errors can be reduced by estimating height according to time of day, season, latitude,
and sunspot cycle. An average height of 250 kilometers is reasonable for the winter months
when propagation conditions are good at 10 MHz and above. In summer, the average height
can be increased to 350 kilometers. In the fall, the estimated height can be decreased again.
Using this method can reduce the error.

87
 For distances of 1600 to 3200 kilometers (1000 to 2000 miles) where multiple-hop
transmission occurs, the height must be determined with greater accurac}'. Heights for a
particular latitude, longitude, and time of day are available from the World Data Center,
NOAA, Department of Commerce, Boulder, CO 80303.

SUMMARY
High Frequency (HF) radio broadcasts are commonly used for time and frequency
calibrations. Accuracies ranging from 1 millisecond for time calibrations to 1.00E-07 for
frequency calibrations can be obtained. If your accuracy requirements are higher, you should
use one of the radio broadcasts discussed in the next two chapters.

88
 Chapter 6 - CALIBRATIONS USING LF AND VLF
RADIO BROADCASTS

Along with the widely used high frequency broadcasts discussed in chapter 5, NIST
also operates a frequency and time service on WWVB, a low frequency (LF) radio station.
WWVB is on the WWV site in Ft. Collins, Colorado. WWVB continuously broadcasts time
and frequency signals at 60 kHz for the continental United States.

WWVB operates at 60 kHz to

take advantage of the stable radio paths
in that frequency range. Both frequency
and time signals are provided, but no
voice transmissions are made due to the
narrow bandwidth of
transmitter/antenna combination.
the
Many
WWVB
countries have services in this band,
which ranges from 30 to 300 kHz, as
well as in the VLF (very low frequency) Fort Collins, CO
band from 3 to 30 kHz. In addition to
discussing the use of WWVB, this
chapter also covers other LF and VLF
stations, including the 100-kHz Loran-C
broadcasts.

It may seem unusual

to send
signals a frequency band that is
in
almost in the audio range, and these
signals pose some special problems for
the transmitter and receiver design
engineers. However, low frequencies
such as the 60-kHz signal of WWVB are
used because of their remarkable
stability. Radio waves at low frequencies
use the Earth and the ionosphere as a
waveguide and follow the Earth's
curvature for long distances. Accuracies
of 1.00E-11 or better for frequency and
500 microseconds for time can be
achieved by using LF or VLF broadcasts.
Users can do even better with Loran-C.
60 kH

89
 GENERAL INFORMATION ABOUT LF AND VLF STATIONS
The following sections contain some general information about LF and VLF time and
frequency broadcasts.

LF and VLF Antennas

Using a quarter-wavelength antenna to receive LF and VLF signals is obviously

impossible; at 60 kHz, this antenna would have to be about 2 miles long! As a compromise,
antennas that are electrically "short" are used with tuning boxes and special couplers. On
vertical antennas, top loading is often used. This consists of radial wires extending from
the top of the antenna to the ground. Of course, as with any antenna system, a good
ground is essential.

The physical location of LF and VLF antennas is important. The location affects
signal strength and noise. Keep the antenna away from metal Long-wire antennas
objects.
should be at least 15 to 20 feet above ground. Several commercial antennas use
preamplifiers so the connecting coaxial cable supplies power to the amplifier. In those cases,
care must be taken to avoid shorting the cable.

Manufacturers of VLF-LF radio receivers offer a variety of antenna types. Long-

wire antennas up to several hundred feet are available. Whip antennas 8 to 10 feet long
are used where space is a problem. On the other hand, air loop antennas are able to reject
interference but do not have as much gain as whips. Ferrite loop antennas are often
smaller than the air loops.

antennas benefit from having couplers and/or amplifiers incorporated into

All of these
the antenna structure to allow amatch to be made with the shielded antenna cable. For
more information about LF and VLF antennas, consult an engineering handbook or an
antenna manual.

Signal Formats

WWVB transmits a carrier for frequency information and changes the level (-10
decibels (dB)) of that carrier to transmit a binary time code. The Omega Navigation System
sends only the carrier. The Loran-C navigation system transmits pulses. Receivers for all
of these signals usually "lock" onto the carrier and thus recover frequency information. In
addition, some means are provided to hear the station. This is usually just a tone output.
For example, if you were to listen to WWVB, you would hear the 1-second code segments
as a tone that changes from loud to soft. There are no voice signals on any of the LF or
VLF stations. The bandwidth used cannot transmit voice signals.

90
 ,

Propagation Characteristics and Other Phase Changes

Phase records made of VLF and LF stations show phase shifts caused by the daily
and seasonal changes along the radio path. These phase shifts occur when sunrise or sunset
occur on the path from the transmitter to receiver. For instance, as the path is changing
from all darkness to all daylight, the ionosphere lowers. This shortens the path from
transmitter to receiver. This shortened path causes the received phase to advance. This
phase advance continues until the entire path is in sunlight. The phase then stabilizes until
either the transmitter or receiver enters darkness. When this happens, the ionosphere
begins to rise, causing a phase retardation.

A strip chart recording of the phase of WWVB is shown below. The chart shows the
signal from WWVB being compared to a very stable oscillator, so the phase shifts can be
attributed to WWVB and not the oscillator. The phase shifts that take place at sunrise and
sunset can be easily seen. The magnitude of the change is a function of the path length,
and the rate of the change is a function of the path direction.

Sunset
-Sunset (rcvr) - Si inrise (rcvr) (WWVB)
Sunrise —
(WWVB)
— Sunset — Sunrise „
Sunset
,

Sunrise —
(WWVB) (WWVB) (rcvr) (Rcvr)
—J

2 ° - o 2 h
2 m N m a 2 a
2
o 2 o $ 5 2
° §
5 o o S

Time (UTC)

Phase of WWVB as received in the eastern United States (chart is 50 microseconds wide)

A phase recording from a stable VLF or LF radio station contains a great deal of
information. The user's job is understand what
to sort out this information so that he can
is happening to the frequency source that is being calibrated. Most stations operate with
a nearly perfect record, but mistakes can happen. This happens just often enough so that
the user needs additional information. This information consists of monthly or weekly
notices of the actual phase of the signal (measured in microseconds). NIST publishes data
for WWVB, and the U.S. Naval Observatory (USNO) publishes data for a number of LF and
VLF stations. This information is free upon request.

91
 Do not rush through LF or VLF calibrations. Certain practices are highly
recommended. Be careful when you adjust the receiver. Once it is working properly, do not
change knob settings or cabling. Always operate the equipment continuously, if possible.
This gives you a continuous chart record so that you know when to expect sunrise/sunset
phase shifts. You can also detect local interference and noise conditions. If your receiver
has a mechanical phase counter, jot down the readings on the chart. This is a great help
in trying to reconstruct events that happen on weekends or at night.

A typical occurrence for a tracking receiver is the cycle phase shift. Since the
receiver is faithfully following a zero crossing of the received phase, it doesn't know which
of the thousands of crossings it is locked to. If the receiver loses lock, it will simply go to
the next crossing that comes along. On a phase chart this will show up as a phase shift
equal to one cycle of the carrier.

For the 60-kHz WWVB

signal, one cycle of phase equals 16.167 microseconds. If the
chart recorder/receiver combination is producing a chart that is 50 microseconds wide (see
the chart on the previous page), the pen will move about one-third of the way across the
chart for each cycle change. You need to identify and ignore these phase shifts when
making calibrations. Keep in mind that phase charts are ambiguous by an amount equal
to the period of one cycle of the carrier frequency.

How can you whether the recorded phase difference indicates a change in your
tell

oscillator, the path, or the transmitted signal? The answer is that you must know from
experience. If you plan to use VLF or LF signals for frequency calibrations, you must
become very familiar with the characteristics of the signals you are using. Battery backup
for receivers is highly desirable, in fact, almost necessary if you plan to calibrate precise
oscillators over a period of many days. All of this may sound difficult, but it really isn't.
The results are worth the effort when you consider that you can perform calibrations up to
1.00E-11 in 24 hours.

Field Strengths of LF and VLF Stations

The field strength of the WWVB signal has been measured along nine radial paths
from the station. These measurements are summarized on the field contour map shown on
the next page. The map does not show the nonlinear field gradient between the 500- and
100-microvolt per meter contour.

Destructive interference occurs between the first hop skywave and the groundwave
at approximately 1200 kilometers (750 miles) from the station. On some radial paths, this
dip in field intensity is quite severe and will cause signal loss at certain times. The distance
between the station and from day to
this null varies night. It is also seasonal. The
sharpness of the null is much less pronounced in the winter. In general, the signal should
be stronger during the winter months.

92
 Measured Field Strength of

Monitoring Station Availability

No matter which station is chosen as a basis for calibration, WWVB, Omega, or VLF,
you should be on a mailing list to receive notices of changes in operation schedule. WWVB
currently operates 24 hours a day, every day. No changes in operating format are made
without advance notification. The Omega stations do schedule maintenance outages and
other changes in operation. Announcements regarding the status of the various Omega
stations in the Omega Navigation System are given hourly on both WWV and WWVH.

Currently, NIST and the USNO are monitoring and reporting the phase differences
of some of the U.S. Loran-C stations. In addition, NIST monitors and reports phase
differences for WWVB, and the USNO offers data on the VLF and Omega stations. These
data are available to interested users by contacting the Time and Frequency Division of
NIST or the USNO.

93
 FREQUENCY CALIBRATIONS USING WWVB
Commercially available WWVB receivers are often used to compare a local oscillator
with the received WWVB signals. These receivers are often complete frequency measurement
systems like the one shown on page 45. WWVB
serves as the reference frequency for these
systems. They may include a built-in chart recorder that shows the phase change of the
oscillator being measured relative to the WWVB signal. This type of system is discussed in
this section. If it meets your needs, WWVBcan also be used as the reference frequency in
a computer-controlled measurement system like the one shown on page 46.

The phase change depends on the relative difference between your oscillator and the
cesium oscillator controlling the WWVB signal at the transmitter. If a precision oscillator
(like a rubidium or cesium) is being used, most of the phase changes are caused by WWVB.
For example, in the chart shown on page 91, the phase shifts are due to the diurnal effect,
or the rise and fall of the Sun along the radio path between the transmitter and the
receiver. During the daytime and nighttime, WWVB serves as a very stable reference
frequency with a relative frequency of about 1.00E-11.

If you use a WWVB compare a lower quality oscillator, the diurnal phase
receiver to
shifts will be less noticeable. For example, if your oscillator is performing at a relative
frequency of 1.00E-08, it is drifting about 1000 microseconds (1 millisecond) per day. With
this much drift, the diurnal phase shifts may not be noticeable at all. However, the chart
recording may be difficult to interpret. If you are using a chart record that is 50
microseconds wide, the chart will "overflow" nearly once an hour. This means that the pen
on the chart recorder will drop from the top of the chart back to the bottom and start over.
The result will be a jagged record like the one shown below:

On low-quality oscillators you may want to increase the chart width as much as
possible (to 100 microseconds, for example). On higher quality oscillators you may want to
decrease the chart width to show more resolution. You can also vary the speed of the chart
recorder. You need to study the manuals for the receiver and chart recorder so that you
can produce charts that are easy to interpret and that suit your calibration needs.

94
 When you use WWVB
for frequency calibrations, remember that the signal started
out at the transmitter as a nearly perfect frequency source. By the time it arrives as your
receiver, it has been altered by diurnal phase shifts, and noise along the radio path. If you
know what to look for, you can interpret the chart recordings obtained from your receiver
and measure the performance of your oscillator at accuracies as high as 1.00E-11.

Computing Relative Frequency from a WWVB Chart Recording

There are two ways to compute relative frequency from a VLF or LF plot. You can
take the slope of the daytime portion of the plot (the straight portion), or you can take
the time difference between two discrete points 1 day apart. This second method is easier
to use. For example, after comparing your oscillator to your WWVB
receiver for over 24
hours, you can select one point on the chart that was recorded during the daytime (to avoid
diurnal phase shifts). You can then take the difference between this point and a point
recorded 24 hours later. For the purposes of our example, let's say that this difference is
25 microseconds, or one-half the width of a typical chart.

Since WWVB is the reference frequency, the 25-microsecond phase shift is attributed
to the oscillator being tested. The number of microseconds in a 24-hour period is shown
below:

86 400 000 000

Since the oscillator moved 25 microseconds, we can say that its relative frequency is

equal to 25 parts in 86 400 000 000 parts, or:

25

86 400 000 000

This number represents the relative frequency of the oscillator under test. Since the
number is so small, we use scientific notation:

2.89E-10

you use the slope of the daytime phase to compute relative frequency, you need to
If
account for cycle jumps and transmitter phase shifts. On days when the radio propagation
is poor or the signal is noisy, it may not be possible to get an accurate relative frequency

reading for that day. For more information on relative frequency and what it means, see
chapter 4.

95
 TIME CALIBRATIONS USING WWVB
Like the high frequency stations WWV
and WWVH, WWVBalso broadcasts a time
code. There are three ways to use the time code. (and most expensive) way is
The first
to use a WWVB
receiver that automatically decodes and displays the time. There are a
number of manufacturers that produce this type of equipment. A second way is to use a
receiver with internal logic circuits that provide a level shift code output. This code may
be applied directly to a strip chart or to an oscilloscope for manual use. And finally, the
simplest way is to observe the amplified signal using a tuned radio frequency receiver and
an oscilloscope.

The illustration below shows the signal envelope as seen on an oscilloscope at the
transmitter with a scale of 5 milliseconds per centimeter. The on-time point is easily seen.
With the horizontal scale of the oscilloscope expanded to 200 microseconds per centimeter,
the on-time point can be determined to within about ±2 cycles of the 60-kHz carrier. Of
course, the farther the receiving site is from the transmitter, the more the signal-to-noise
ratio degrades until some averaging techniques must be employed. Since the WWVB
data
rate only one bit per second, visual integration is very difficult, especially when the
is

horizontal scale is expanded to allow greater time resolution. Averaging or signal

integration using a signal averager or an oscilloscope camera works quite well at remote
receiving locations.

0 1 2 3 4 5
milliseconds / centimeter

96
 The actual process of time recovery consists of viewing the signal envelope on an
oscilloscope that being triggered by a local clock at the receiving site. Having previously
is

determined the path delay, its value can be subtracted from the total observed delay on
the oscilloscope to obtain the local clock error.

In very noisy locations, where the on-time point is difficult to identify, a photograph
of the waveform can be helpful. Sometimes it is advantageous to allow a number of
oscilloscope traces to be exposed on the same piece of film. This has an averaging effect
that simplifies the location of the on-time point. However, a single exposure can be used.
Take the photo and draw one horizontal line through the average of the waveform amplitude
and another line along the average slope of the dropout. The intersection of these lines is
the on-time point.

Receiving long distances from the transmitter have an added problem

sites that are
when high resolution timing desired.
is If the delay to the receiver site is, say, 15
milliseconds and the sweep speed is set to 1 millisecond per centimeter, the on-time point
of the envelope will be off the oscilloscope face.

This problem can be solved if a second clock is available. The second clock can be
adjusted late so that when used for the oscilloscope trigger source, the on-time point will
be near the beginning of the trace. The sweepspeed can now be increased to obtain the
desired resolution. The time intervals to be accounted for are the time between the
beginning of the trace and the on-time point, and the interval between the local clock and
the second clock. The local clock error is:

Local Clock Error = D - delta(tc) - delta(to)

where: D is the propagation delay

delta(tc) is the interval between the two clocks

delta(to) is the trace delay on the oscilloscope

If the clock error is positive, then the local clock pulse occurs after the
local
transmitted time pulse. If the local clock error is negative, then the local clock pulse occurs
before the transmitted time pulse.

97
 Format of the WWVB Time Code

WWVB transmits a special time code that provides time information. This service
is used by seismologists, standards labs, commercial power companies, and others interested
in synchronization with accuracies of the order of 500 microseconds. The code provides
year, day, hour, and minute information. Seconds are resolved by counting bits. A
correction, applied to the transmitted time to obtain Earth time, UT1, is also provided for
use by some astronomers and navigators. Other code bits provide information about
Daylight Saving Time, leap years, and leap seconds.

Unlike a code that is designed primarily for machine decoding, the WWVB code was
originally designed to be manually decoded from strip chart recordings or tapes. To simplify
this process for a human decoder, the most significant bit occurs first in each code word.
The actual bit, whether a binary zero, binary one, or position identifier, is determined by
the duration of the power reduction of the transmitter carrier.

The WWVB time code is synchronized with the 60-kHz carrier and is broadcast
continuously at a rate of one pulse per second. The time code is sent in binary coded
decimal (BCD) format. Bits are sent by shifting the power of the 60-kHz carrier. The
carrier power is reduced 10 dB at the start of each second. If full power is restored 200
milliseconds later, it represents a 0 bit. If full power is restored 500 milliseconds later, it
represents a 1 bit. Reference markers and position identifiers are sent by restoring full
power 800 milliseconds later.

The time code is in binary coded decimal (BCD) format. Groups of binary digits
used to represent decimal numbers. The binary-to-decimal weighting scheme is
(bits) are
8-4-2-1. The most significant bit is always sent first. This is the reverse of the
WWV/WWVH time code. The table below shows the BCD groups and their decimal
equivalent:

BINARY GROUP DECIMAL

Weight: 8 4 2 1 EQUIVALENT

0 0 0 0 0
0 0 0 1 1
0 0 1 0 2
0 0 1 1 3
0 1 0 0 4
0 1 0 1 5
0 1 1 0 6
0 1 1 1 7
1 0 0 0 8
1 0 0 1 9

98
 The decimal number is obtained by multiplying each bit in the binary group by the
weight of its respective column and then adding the four products together. For example,
the table shows that the binary group 0101 is equal to 5. This is derived by:

(0 x 8) + (1 x 4) + (0 x 2) + (1 x 1) = 0 + 4 + 0 + 1 = 5

Every minute, the WWVB time code sends the current minute, hour, day of year, and
year; UT1 correction, and Daylight Saving Time (DST),
a leap second, and leap year
indicators. Two BCD groups are needed to express the hour (00 to 23) and minute (00 to
59); and three groups are needed
to express the day of year (001 to 366). To represent
hundreds, the basic 8-4-2-1 weights are simply multiplied by 1, 10, or 100 as
units, tens, or
appropriate. The coded information refers to the time at the start of the one-minute frame.
Seconds are determined by counting pulses within the frame.

Each minute begins with a frame reference pulse lasting for 0.8 second. A position
identifier pulse lasting for 0.8 second is transmitted every 10 seconds.

UT1 corrections are broadcast at seconds 36 through 44 of each frame. These

corrections are to the nearest 0.1 second. The
transmitted at 36, 37, and 38 seconds
bits
show whether UT1 is positive or negative with respect to UTC. If 1 bits are sent at
seconds 36 and 38, the UT1 correction is positive. If a 1 bit is sent at second 37, the UT1
correction is negative. The amount of the UT1 correction is sent in a 4-bit BCD group at
40, 41, 42, and 43 seconds. The binary-to-decimal weights are multiplied by 0.1, because the
UT1 corrections are expressed in tenths of seconds. Information about tens of year is sent
in seconds 45 through 48, and units of year information are sent in seconds 50 through 53.

The WWVB time code also contains information about leap years, leap seconds, and
DST. The leap year bit is transmitted at 55 seconds. If it is set to 1, then the current year
is a leap year. The bit is set to 1 during each leap year sometime after January 1 but
before February 29. It is set back to 0 shortly after January 1 of the year following the
leap year. Receivers that read this bit can automatically adjust themselves during leap
years. Receivers can also automatically adjust themselves for leap seconds, by reading the
leap second warning bit transmitted at 56 seconds into the frame.

Two DST bits are sent at 57 and 58 seconds. Using two bits allows a receiver that
is turned on during a time change day to set its time correctly. Bit 57 changes from 0 to
1 at 0000 UTC on the day of the change from standard time to DST. Exactly 24 hours
later, bit 58 also changes from 0 to 1. On the day of change back to standard time, bit
57 will go from 1 to 0 at 0000 UTC, followed exactly 24 hours later by bit 58. During all

other days both bits will be 0 or 1, depending on whether standard time or DST is in effect.

99
 FREQUENCY CALIBRATIONS USING THE OMEGA NAVIGATION SYSTEM
The Omega Navigation System consists of eight VLF radio stations operating in the
10- to 14-kHz range.The eight stations are listed below:

Omega Navigation System Stations

Call Location Power Carrier Days/ Hours/

Sign (kW) (kHz) Week Day

Omega Alda, 10 11.05 - F

A Norway 10.20 - A 7 24
11.33 - C
13.60 - B

Omega Monrovia, 10 11.05 - G

B Liberia 10.20 - B 7 24
11.33 - D
13.60 - C

Omega Haiku, Oahu, 10 11.05 - H

C Hawaii 10.20 - C 7 24
11.33 - E
13.60 - D

Omega La Moure, 10 11.05 - A

D North Dakota 10.20 - D 7 24
11.33 - F
13.60 - E

Omega La Reunion 10 11.05 - B

E 10.20 - E 7 24
11.33 - G
13.60 - F

Omega Golfo Nuevo, 10 11.05 - C

F Argentina 10.20 - F 7 24
11.33 - H
13.60 - G

Omega Woodside, 10 11.05 - D

G Victoria, 10.20 - G 7 24
Australia 11.33 - A
13.60 - H

Omega Tsushima 10 11.05 - E

H Islands, 10.20 - H 7 24
Japan 11.33 - B
13.60 - A

100
 If you decide to use an Omega signal for frequency calibrations, a phase-tracking
receiver recommended. If one of the navigation frequencies is to be used, then an Omega
is

commutator must also be used. This is a device that turns the phase-tracking receiver on
and off at the proper times to receive only the desired Omega station.

The frequencies and the format segments of the Omega stations are referenced to
cesium beam oscillators, so the transmissions are very stable. The USNO monitors and
reports the daily phase values for each Omega station. NIST and
radio stations WWV
WWVH each broadcast a 40-second message describing the current status
(see chapter 5)
of the Omega Navigation System. This message is broadcast at 16 minutes after the hour
on WWV and at 47 minutes after the hour on WWVH.

Operating Characteristics of Omega

The eight Omega stations transmit in the internationally allocated VLF navigational
band between 10 and 14 kHz. Since the Omega stations transmit at a very low frequency
and at a high power (10 kilowatts), it is possible to receive at least one Omega station just
about anywhere in the world. In some remote locations, Omega might be the only signal
available for frequency calibrations. The map below shows the location of the eight Omega
stations.

101
 Each Omega station transmits on four common navigational frequencies (10.2, 11.05,
11.33, and 13.6 kHz) and one unique frequency (from 11.8 to 13.1 kHz) using an
omnidirectional antenna. In order to prevent interference, transmissions from each station
are time-sequenced in an eight-segment frame as shown in the figure below- Each station
uses one channel of the frame for each of the four common frequencies, and four segments
for its unique frequency. The unique frequencies for each station are marked with an
asterisk. The unique frequency is normally used for frequency calibrations.

Segments with Frequencies in kHz

Station
1 2 3 4 5 6 7 8

Norway 10. 02 13 60 11. 33

. 12 . 1* 12 1* 11.05 12 1* 12 1*
. . .

Liberia 12 0* 10.20 13.60 11. 33 12 0* 12.0* 11. 05 12 . 0*

. .

Hawaii 11.8* 11.8* 10.20 13 60 11.33 11.8* 11.8* 11. 05

.

North Dakota 11. 05 13 1* 13. . 1* 10.20 13 60 11. 33 13. 1* 13 1*

. .

La Reunion 12 .3* 11. 05 12 .3* 12 .3* 10.20 13 . 60 11. 33 12 .3*

Argentina 12 .9* 12.9* 11.05 12.9* 12.9* 10.20 13.60 11. 33

Australia 11. 33 13 0* 13
. . 0* 11. 05 13 0* 13 0* 10.20 13 60
. . .

Japan 13 . 60 11. 33 12.8* 12.8* 11.05 12.8* 12.8* 10.20

* unique frequency

The Omega arranged so that only four stations transmit on the

signal pattern is

navigational frequencies during each segment. Since no stations broadcast on the same
frequency at the same time, each individual station can be identified. The length of each
segment varies from 0.9 to 1.2 seconds, and there is a 0.2-second silent interval between
segments. The entire signal pattern repeats itself every 10 seconds.

All Omega
transmitting stations are synchronized by means of very stable cesium
beam frequency standards. The synchronization of all transmissions is tightly controlled,
and the phase relationships between all signals are maintained to within a fraction of a
cycle. Although the signals are highly stable as transmitted, you need to be aware of the
propagation characteristics of Omega before you use the signals for frequency calibrations.
These characteristics are discussed in the next section.

102
 Propagation Characteristics of Omega

The same characteristics that allow Omega signals to be received at great distances
also limit their accuracy. Since Omega between the Earth and the
signals are propagated
ionosphere, the propagation parameters change as a result of changes in the Earth or
ionosphere. If you use Omega for frequency calibrations, you need to be aware of several
types of phase shifts that might show up in your data.

The most common type of Omega phase shift is due to the diurnal effect. Like
WWVB signals (page 91), Omega signals are subject to diurnal phase shifts due to the rise
and fall of the Sun along the radio path. However, since these diurnal phase shifts are
highly repeatable, it is possible to predict when they will occur and to correct for them
during calibrations.

Unpredictable short-term variations may also occur. Ninety-five percent of the time,
these are small variations related to random propagational variations which will not degrade
the signal's accuracy. Occasionally, however, large disturbances can occur as a result of
The emission of rays from the Sun occasionally
solar emission of ray or particle bursts.
causes a short-term disruption of Omega signals; this is called a Sudden Phase Anomaly
(SPA). The duration of a SPA is generally not greater than 1 hour, but it can cause a
large phase shift. A SPA can occur as often as 7-10 times per month. It usually affects
signals from only a few stations at a time since rays from the Sun usually enter only part
of the illuminated portion of the Earth's surface.

The Omega phase can also be shifted by a Polar Cap Disturbance (PCD). A PCD
occurs when a large quantity of protons are released from the Sun. Although an infrequent
occurrence, a PCD might shift the phase of an Omega signal for a period of several days.
However, a PCD only affects those transmissions involving arctic propagation paths.
Because of their possible long duration, PCD notices are broadcast as navigational warning
messages on WWV and WWVH
(see page 101).

Modal interference another form of signal interference. It occurs when the Omega
is

signal propagates in different ways over the radio path, and the various propagation modes
interfere with each other (the ground wave interferes with the skywave, for example). Under
ideal conditions, one mode would be dominant at all times, and the received phase from
Omega would be regular. However, in practice, the competing modes do not completely
disappear. The worst thing that can happen with modal interference is a situation where
the modal dominance changes. An example of this is a situation where one mode is
dominant during the day and a second mode is dominant at night. During the transitional
period between day and night, the two modes are equal, and abnormal transitions may occur
in which cycles are "slipped" or lost.

103
 The plots below show the phase of Omega as received at NIST in Boulder, Colorado,
and compared to the national frequency standard over a 1-week period (7 days). The first
plot shows the 11.8-kHz signal from Hawaii, and the second plot shows the 13.1-kHz signal
from North Dakota. The plots show that the Omega signals are stable and the diurnal
phase shifts are predictable. The plot of the Hawaii signal demonstrates that Omega signals
are usable over a very long radio path (several thousand miles).

Onega Hawaii (11.8 kHz) us UTC(NIST)

46

11/18/85 (MJD 46387) Channel 1 (MJD 46394) 11/25/85

16:18 [tic narks = 4 hours] 14:52
Data points= 263

Onega North Dakota (13.1 kHz) ws UTC(NIST)

48

11/18/85 (MJD 46387) Channel 2 (MJD 46394) 11/25/85

16:18 [tic narks = 4 hours] 14:52
Data points= 263

104
 , , , ,

OTHER LF AND VLF TIME AND FREQUENCY STATIONS

There are a number of other LF and VLF time and frequency stations located around
the world. Information about many of these stations is listed in the table below and on the
next page.

LF and VLF Time and Frequency Stations

Call Location Power Carrier Days/ Hours/

Sign (kW) (kHz) Week Day

DCF77 Mainf lingen, 20 77 50

. 7 24
Germany

GBR Rugby, United 60 15. 95 7 22

Kingdom 16. 00

HBG Prangins 20 75. 00 7 24

Switzerland

JJF-2 Sanwa, 10 40 00
. 7 24
JG2AS Sashima,
Ibaraki
Japan

MSF Rugby, United 25 60. 00 7 24

Kingdom
NAA Cutler, Maine 1000 24 00
. 7 24
United States

NCA Aguada 100 28.50 7 24

Puerto Rico

NTD Yoshima 50 17.40 7 24

Japan
NLK Jim Creek, 125 24.80 7 24
Washington,
United States

NPM Lualualei 600 23.40 7 24

Hawaii
United States

NSS Annapolis, 400 21.40 7 24

Maryland,
United States

105
 LF and VLF Time and Frequency Stations 1

Call Location Power Carrier Days/ Hours/

Sign (kW) (kHz) Week Day

NWC NW Cape, 1000 22.30 7 24

Australia

OMA Liblice, 5 50. 00 7 24

Czechoslovakia

RBU Moskva, 10 66. 67 7 24

USSR
RTZ Irkutsk, 10 50. 00 7 23
USSR

RW-166 Irkutsk, 40 200. 00 7 23

USSR

RW-7 6 Novosibirsk, 150 272 00

. 7 22
USSR
UNW3 Molodechno, 25.50, 25.10 7 2
USSR 25.00, 23.00
20.50

UPD8 Arkhangelsk, 25.50, 25.10 7 2

USSR 25.00, 23.00
20. 50

UQC3 Khabarovsk, 300 25.50, 25.10 7 2

USSR 25.00, 23.00
20.50

USB2 Frunze, 25.50, 25.10 7 3

USSR 25.00, 23.00
20. 50

UTR3 Gorky, 300 25.50, 25.10 7 2

USSR 25.00, 23.00
20. 50

106
 TIME AND FREQUENCY CALIBRATIONS USING LORAN-C

As we discussed in chapter radio signals from Loran-C can make an excellent

4,
reference frequency for calibrations. The success of Loran-C signals is due to a number of
things. Each station broadcasts 24 hours a day, 7 days a week. Each station is controlled
by cesium standards, and the signals are carefully monitored and controlled by the U.S.
Coast Guard, NIST, and the U.S. Naval Observatory (USNO). Loran-C groundwave signals
provide state-of-the-art frequency and time calibrations. Even the skywave signals, though
slightly less accurate, can become the basis of excellent high-quality calibrations. And since
many Loran-C transmitters exist, the signals are usable just about anywhere in the northern
hemisphere.

Like Omega, Loran-C is a radio navigation system (Loran is an acronym for LOng
RAnge Navigation). All Loran-C stations broadcast on a frequency of 100 kHz, using a
bandwidth from 90 to 110 kHz. At this low frequency, the radio waves follow the Earth's
curvature and are not subjected to the large diurnal phase shifts that occur with signals
from WWVB or Omega.

The Loran-C navigation system consists of many synchronized "chains" or networks

of stations. These chains provide groundwave coverage of most of the United States,
Canada, Europe, the North Atlantic, the islands of the Central and West Pacific, the
Philippines, and Japan. Each chain has one master station (designated as M), and from two
to four slave stations (designated as W, X, Y, and Z). Information about each Loran-C chain
is listed below and on the next two pages. This information lists all of the chains being
monitored by the USNO at the time of this writing, but new stations are continually being
added. For the exact status of the Loran-C system, please contact the U.S. Coast Guard.

Loran-C Time and Frequency Stations

Chain Stations Power (kW)

4990 Johnston Island, Hawaii, USA M 275

Upolu Point, Hawaii, USA X 275
Kure Island, Hawaii, USA Y 275

5930 Caribou, Maine, USA M 350

Nantucket Island, Mass., USA X 275
Cape Race, Newfoundland, Canada Y 1500
Fox Harbour, Labrador, Canada Z 800

5990 Williams Lake, BC, Canada M 400

Shoal Cove, Alaska, USA X 540
George, Washington, USA Y 1600
Port Hardy, BC, Canada Z 400

107
 Loran-C Time and Frequency Stations

Chain Stations Power (kW)

7930 Fox Harbour, Labrador, Canada M 800
Cape Race, Newfoundland, Canada W 1500
Angissoq, Greenland X *7 C f\

f
7960
~i f~\ r\
Tok, Alaska, USA M 540
Narrow Cape, Alaska, USA X 400
Shoal Cove, Alaska, USA Y 540

7970 r\ *7 r\
Ejde, Faeroe Islands, Denmark M 325
Sylt, Germany W 325
Boe, Norway X 165
Sandur, Iceland Y 1500
Jan Mayen, Norway Z 165
T OAA
7980 Malone, Florida, USA M r\ r\ r\
800
Grangeville, Louisiana, USA o
W 8 00
r\ r\

Raymondville, Texas, USA X 400

Jupiter, Florida, USA Y 275
Carolina Beach, N. Carolina, USA Z 550

7990 Sellia Marina, Italy M 165

Lampedusa, Italy X 325
Kargabarun, Turkey Y 165
Estartit, Spain Z 165

8970 Dana, Indiana, USA M 400

Malone, Florida, USA W 800
Seneca, New York, USA X 800
o
Baudette, Minnesota, USA Y 800 r\ r\

9940 Fallon, Nevada, USA M 4 00

George, Washington, USA W 1600
Middleton, California, USA X 400
Searchlight, Nevada, USA Y 540

9960 Seneca, New York, USA M 800

Caribou, Maine, USA W 350
Nantucket Island, Mass., USA X 275
Carolina Beach, N. Carolina, USA Y 550
Dana, Indiana, USA Z 400

108
 Loran-C Time and Frequency Stations

Chain Stations Power (kW)

9970 Iwo Jima, Japan M 1800

Marcus Island, Japan W 1800
Hokkaido, Japan X 1000
Gesashi, Japan Y 1000
Yap, Caroline Island z 1000

9980 Sandur, Iceland M 1500

Angissoq, Greenland W 760
Ejde, Faeroe Island, Denmark X 325

9990 St. Paul, Alaska, USA M 275

Attu, Alaska, USA X 275
Point Clarence, Alaska, USA Y 1000
Narrow Cape, Alaska, USA Z 400

The master station in a Loran-C chain transmits groups of pulses that are received
by the slave stations. The slave stations receive the master pulse groups, and then transmit
similar groups of synchronized pulses.

When Loran-C is used to navigate, the constant time differences between the
reception of the master pulses and the slave pulses are used to determine a line of position
(LOP). Determining an LOP requires receiving signals from three separate Loran-C
transmitters (the master and at least two slaves). However, for frequency and time
applications, you only need to receive one Loran-C station. This station can be either a
master or a slave, although some receivers require the master station to be received before
the slaves can be identified.

Format of Loran-C Broadcasts

All Loran-C stations broadcast on the same carrier frequency (100 kHz). Because of
this, a Loran-C receiver has to distinguish between signals broadcast from a number of
different transmitters.

Each Loran-C chain is identified by a unique Group Repetition Interval (GRI). The
length of the GRI is fixed, and each chain is named according to its GRI (divided by 10).
For example, the 7980 chain has a GRI of 79 800 microseconds. This means that every
79 800 microseconds (or about 12 times per second) every station in the chain transmits a
group of pulses. The GRI must be long enough for each station in the chain to transmit
its pulses and to accommodate for spacing between the pulses. Enough time must be

109
included between the pulses so that signals from two or more stations cannot overlap in
time anywhere in the coverage area. Therefore, the minimum GRI is determined by the
number of stations in the chain, and by the distance between the stations. Possible GRI
values range from 40 000 microseconds to 99 990 microseconds. \

Once a Loran-C chain has been identified, the stations within the chain can also be
identified by looking at the pulses. The master station sends its pulses first. The master
transmits 8 pulses separated by a 1000-microsecond delay. Then 2000 microseconds after
the 8th pulse, a 9th pulse is sent. The 9th pulse is used to identify the master station.
The slave stations then send their pulses in turn. For example, if a chain has three slave
stations (X, Y, and Z), they send their pulses in order. X goes first, then Y, then Z. Each
slave station transmits 8 pulses separated by a 1000-microsecond delay. The figure below
illustrates the way Loran-C pulses are transmitted.

GRI
—

Master X Y Z Master
Pulses Pulses Pulses
Pulses Pulses

Loran station identification is also aided by separately phase-coding the master and
slave pulses.Each group of pulses is coded by a phase reversal process which enables the
receiver eliminate certain types of interference.
to All these subtle differences in
transmission pulse rates make Loran-C receivers fairly complex. If you have a manual
Loran-C receiver, you need to know many of the details of the broadcast format before you
can acquire and track a Loran-C signal. However, some Loran-C receivers are now
completely automatic. If you key in the GRI, these receivers will acquire and track the
desired stations in just a few seconds.

110
 Lor an- C Reception

Radio energy from each Loran-C transmitter radiates in all directions. A portion of
the energy travels out from each transmitting station parallel to the surface of the Earth.
This is the ground wave.

Useful Loran-C groundwave coverage extends from approximately 2400 to 3200

kilometers (1500 to 2000 miles). During periods of good reception, this range may be
greater; and during periods of high noise and interference, it may be less. However, with
typical noise and interference levels, 2400 kilometers (1500 miles) is a good estimate of the
reliable groundwave range of Loran-C signals from a station transmitting 300 kilowatts.

Part of the Loran radio signal radiates upward from the transmitting antenna and
is from the electrified layer of the atmosphere known as the ionosphere. This
reflected
signal is called the skywave. If the groundwave signal has traveled a long distance, it will
be reduced in amplitude and weakened. The receiver will pick up both the groundwave and
the skywave signals. The receiver cannot easily lock to the groundwave because it is weak
and noisy, so it will lock to the skywave signal.

Receiving the skywave is less desirable than receiving the groundwave. This is
because the skywave signal "moves" around, since it is reflected off the ionosphere. This
movement is caused by the motion of the ionosphere due to the rise and fall of the Sun.
If you are receiving skywave signals from Loran-C, you may see diurnal phase shifts in your
data, similar to those (although smaller in magnitude) from WWVB
(page 91) or Omega
(page 104).

Receiving the skywave signals isn't

all bad, however. Because the skywave signals
are stronger than groundwave signals at great distances, skywave lock-on is possible at
distances beyond where the groundwave signals can be received. And although the skywave
moves around, it does so in a predictable way. With good recordkeeping, the Loran-C
skywaves can yield very good time and frequency data.

Part of the reason that Loran-C transmits pulses is so that the receiver can
distinguish between groundwave and skywave signals. The accuracy of the system is based
on the fact that the early part of the pulse (the part that leaves the transmitter first) will
travel along the ground and arrive before the less stable skywave pulse that bounces off the
ionosphere. By determining what cycle of the pulse is being tracked, you can tell if you are
tracking a skywave or a groundwave. Many receivers track the third cycle of the pulse.
The third cycle arrives early enough to be groundwave but has enough amplitude for the
signal to be strong. A picture of a Loran-C pulse with the third cycle identified is shown
on the next page. The shape of the pulse makes it possible for a receiver to identify one
particular cycle of a 100-kHz carrier, and to stay locked to that cycle.

Ill
 Getting Time From Loran-C

Loran-C pulses are very stable and can be used as the frequency source for a very
accurate clock. Loran-C does not have a time code, and the clock must originally be set to
the correct time (UTC), using a time signal from WWV
(chapter 5) or some other source.
However, once the clock is set, the Loran-C pulses should keep the clock accurate to within
10 microseconds or better.

you use Loran-C pulses to drive a clock, you must remember that the interval
If
between pulses (the GRI) is not an even multiple of 1 second. For example, the stations
in the Loran 7980 chain transmit their pulses every 79 800 microseconds*, or a little more
than 12 times per second. In order to get time from Loran-C, you need to know the time
of coincidence (TOC) when the Loran pulse coincides with a 1 second pulse. Depending
on the GRI, a TOC may occur as infrequently as every 16 minutes. You can obtain TOC
tables for all Loran-C chains by writing to the U.S. Naval Observatory, Time Services
Department, Washington, DC, 20392-5100.

Some Loran-C receivers have a TOC button. If the is locked to the incoming
receiver
signal, you can press this button whena TOC occurs to
synchronize a 1 pulse-per-second
(1 pps) timing signal to the Loran pulses. If you do not lose power, and the receiver does
not lose lock, the timing pulses will stay on time. It is important to keep a battery supply
connected to the receiver to prevent loss of synchronization in case of power outages.

112
 As mentioned previously, a clock driven by Loran-C pulses can keep time to within
10 microseconds. Of keeping time at this accuracy requires taking path and
course,
equipment delays into account. In addition, you need to know the sample point where the
receiver is locked on the signal. Since Loran-C operates at 100 kHz, the zero crossings of
the pulses are separated by the 10-microsecond period. Whenever your receiver moves from
one cycle to the next, the time jumps by 10 microseconds. Some Loran-C timing receivers
have oscilloscope and chart recorder outputs so you can see the exact position of the sample
point.

Frequency Calibrations Using Loran-C

Whenusing Loran-C for frequency calibrations, time interval measurements are the
method of choice. For details about making frequency calibrations using the time interval
method, please read chapter 4. That chapter discusses a computer-controlled frequency
measurement system based on the time interval method that uses Loran-C as a reference
frequency (page 46). This system uses a time interval counter to measure and record the
time interval between the Loran-C signal and the signal from the oscillator being calibrated.
Using a system like the one described in chapter 4 allows even the smallest laboratory to
make state-of-the-art frequency calibrations.

Loran-C is almost unequaled as a frequency calibration source. Loran-C signals are

good enough to calibrate quartz, rubidium, or cesium oscillators. One reason for this is that
the Loran-C groundwave (unlike other VLF or LF broadcasts) is not subjected to large
diurnal phase shifts. Of course, Loran-C signals change phase due to propagation noise
along the radio path. However, these phase shifts are quite small and average out over a
24-hour period. For example, the relative frequency of Loran-C is about 1.00E-12 over a
24-hour period when compared to the NIST time scale. This means that the amount of
Loran-C phase shift over a 24-hour period is only about 0.1 microsecond (100 nanoseconds).

The smallerrors in Loran-C phase generally do not cause problems when making
frequency calibrations. This is because the Loran-C signals are far more accurate than the
frequency output of most oscillators. For example, assume that Loran-C is being used to
calibrate an with a calibration requirement of 1.00E-10. Since Loran signals are
oscillator
accurate to about 1.00E-12 over a 24-hour period, this means they are 100 times more
accurate than the calibration requirement for the oscillator. If the performance of the
oscillator relative to Loran-C is plotted, the small errors in the Loran-C phase are not even
visible. A plot which illustrates this is on the next page. For this type of calibration, using
Loran works just as well as using the NIST time scale.

The second plot on the next page shows a cesium oscillator compared to Loran-C over
a period of nearly 4 weeks. This data is plotted on a much smaller scale than the previous
plot, and the data is not as smooth because the small errors in the Loran-C phase are now
visible. However, the plot still shows us that the cesium oscillator is high in frequency
(although only by 1.87E-12), relative to Loran-C.

113
 Rubidium Oscillator vs Loran-C 9948 (Fallon, Nevada)

20

11/01/89 (MJD 47831) Channel 1 (MJD 47832) 11/02/89

16:39 [tic narks = 1 hour] 14:54
Data points= 21 Relative Frequency: +2.42E-10 r=*1.00

Cesiun Oscillator vs Loran-C 9940 (Fallon, Nevada)

E 2.5

12/02/89 (MJD 47862) Channel 1 (MJD 47888) 12/28/89

17:00 [tic narks - 24 hours! 17:00
Data points^ 623 Relative Frequency: +1.87E-12 r=*0.98

SUMMARY
Signals broadcast from LF and VLF
radio stations are an excellent choice for time
and frequency calibrations. These signals provide more accuracy than the HF signals
discussed in chapter 5. However, using LF and VLF signals might require more time,
money, and effort than using HF signals.

The next chapter of this book discusses yet another alternative: time and frequency
signals broadcast by satellite.

114
 Chapter 7 - CALIBRATIONS USING SATELLITE BROADCASTS

Several satellite signals are available for time and frequency calibrations. An earth-
orbiting satellite is a nearly ideal vehicle for broadcasting time and frequency radio signals.
This is because the path between the satellite and the receiver is clear, and not influenced
by earth-bound noise sources.

NIST provides a continuous

time code through the GOES
(Geostationary Operational
Environmental Satellite) satellites.
These satellites are operated by the
National Oceanic and Atmospheric
Administration (NOAA). The time
code is referenced to UTC(NIST) and
is broadcast continuously to the entire

Western Hemisphere from two

satellites (GOES/East and
GOES/West). This cooperative
arrangement between NIST and
NOAA was made by formal
agreement. The current agreement
extends until 1997.

The U.S. Navy operates the

TRANSIT of
series navigation
satellites, with major participation by
the Applied Physics Laboratory of the
Johns Hopkins University and the
U.S. Naval Observatory. The
operation of this system is also
discussed in this chapter.

In addition to GOES and

TRANSIT, the Global Positioning
Satellite System (GPS) is also covered
in this chapter. GPS offers users
several advantages. Its frequency
band, signal strength, and the use of
atomic oscillators on board each
satellite allow high-accuracy
calibrations. Also, its popularity
should make equipment available that
is versatile and easy to use.

115
 USING THE GOES SATELLITE BROADCASTS

A no different than any other radio signal as far as the user is

satellite signal is
concerned, You should not be discouraged by the difference in frequency or the operating
techniques needed for frequency and time calibrations using GOES. In fact, tests have
shown that GOES is one of the easiest services to use and can be used in areas where
other signals are nearly impossible to receive. The equipment provided by manufacturers
is usually quite sophisticated and nearly automatic in its operation.

The GOES 36 000 kilometers above the equator

satellites are in geostationary orbit
and travel at a speed of about 11000 kilometers per hour. Geostationary means that the
satellites stay above the same spot on the Earth's surface. Because they are geostationary,
the path delay for the time code remains relatively constant at all times.

Geostationary satellites for time broadcasting are almost always in view and provide
a source of continuous synchronization. Non-geostationary satellites, on the other hand,
offer exposure to the user for only short periods of time at intervals ranging from about
1 hour to many hours or even days. Thus, non-geostationary satellite systems usually
have a number of satellites in orbit.

GOES can be used for frequency calibrations. You could use GOES signals as the
reference frequency for a frequency measurement system like the one shown on page 46.
However, GOES has a time-code output (perhaps its major application is for its time code),
and most commercially available GOES receivers were designed for time recovery.
Frequency calibrations obtained using GOES time data fall into the accuracy range from
1.00E-06 to 1.00E-09 (for a 1-day measurement) with respect to NIST, depending on the
sophistication of the equipment used.

The path from the receiver to a geostationary satellite is free from obstructions and
allows using high carrier frequencies that are largely unaffected by the ionosphere and
troposphere. This eliminates the fading and path length variations which are characteristic
of terrestrial HF signals. The clear path also means that the path delay can be computed
with greater accuracy than with HF signals.

The NIST master clock islocated on the ground rather than in the satellite itself.
The satellite is only a transponder used to relay signals. This allows for easy control and
maintenance of the system, thus guaranteeing better performance and reliability.

Because of these advantages, the GOES satellite can provide a continuous time
message to receivers in its view. It can control the frequency rate of the slaved ground
clock to eliminate the need for high-quality oscillators and can also provide position data
to correct for propagation delays.

116
 GOES COVERAGE
There are usually three GOES satellites in orbit, two in operation and a third
serving as an in-orbit spare (this situation can change due to failure of one or more
satellites). The two operational satellites are normally located at 135 degrees and 75
degrees West Longitude, and the spare is at 105 degrees West Longitude, however, this
may change at times depending upon the current operational status of the available
satellites. The western GOES operates on 468.825 MHz; the eastern on 468.8375 MHz.
The approximate coverage of the two operational satellites is shown on the map below.
As you can see from the map, much of the Western Hemisphere is covered by at least one
satellite. The continental United States is covered by both satellites.

It is relatively simple to point an antenna to either satellite. If the path to the

satellite is pointing in the general direction of the satellite is usually sufficient,
clear,
particularly for relatively low gain antennas (<10 dB). Satellite positions may change with
time. Users should check the monthly NIST Time and Frequency Bulletin for the latest
information.

1 - GOES/West
2 - Both Satellites
3 - GOES/East

117
 GOES SIGNAL CHARACTERISTICS
As mentioned earlier, there are three GOES two in operational
satellites in orbit,
status with a third serving as an in-orbit spare. The signal characteristics of the two
operational satellites are summarized in the table below:

uU£iO / H Cl 0 L.

Frequency 468.8250 MHz 468.8375 MHz

rOlailZaLlun

riuuuiauiun V-jrorv ^ZDU J

"«* r 0 Ja (IDU J

LJCX La J\d L-t: 1uU O IT aJJ 1 on Rpc;

Satellite Location 135 West 75 East

Signal Strength
(output from -131 dBm -131 dBm
isotropic antenna)

Coding Manchester Manchester

Bandwidth 400 Hz 400 Hz

* right hand circularly polarized

** coherent phase shift keying

GOES TIME CODE FORMAT

The GOES time code is generated and sent to the satellite from NOAA's facility at
Wallops Island, Virginia. The frequency standards at Wallops Island provide a stable and
accurate reference for the time code. They are compared to UTC(NIST) with Loran-C and
GPS. The transmitted accuracy from Wallops Island is kept within at least 10
microseconds of UTC(NIST).

The GOEStime code includes the current year, complete time-of-year information,
the UT1 correction, satellite position information, accuracy indicators, Daylight Saving Time
and leap second indicators, and system status information.

The time code is interlaced with interrogation messages that do not contain time and
frequency information. The interrogation messages are used by NOAA to communicate

118
with systems gathering weather data. The interrogation messages are broadcast at a rate
of 100 bits per second. They are one-half second, or 50 bits in length. The first 4 bits
form a BCD time code word. The remaining 46 bits do not contain timing information.

A time code frame consists of 60 BCD time code words. It takes 60 interrogation
messages, or 30 seconds, to complete a time code frame. The completed time code frame
contains a synchronization word, time code accuracy indicators, Daylight Saving Time
(DST) and leap second indicators, system status information, a time-of-year word (UTC day
of year, hour, minute, and second), the last two digits of the current year, the UT1
correction, and the satellite's position. The position information is updated every minute.
It includes the satellite's latitude, longitude, and height above the Earth's surface.

GOES TIME CODE PERFORMANCE

The GOES time code can be used at three performance levels: uncorrected for path
delay, corrected for mean path delay only, and fully corrected.

Uncorrected: The path delay from Wallops Island, Virginia (where the time code
total
originates), to the satellite and then back to Earth is about 260 milliseconds. To
compensate for this path delay, the signals are advanced by 260 milliseconds before
transmission from Wallops Island. The uncorrected signal arrives back on Earth nearly
on time (within 16 milliseconds), depending upon the receiver's location.

Corrected for Mean Path Delay. If the appropriate mean path delay correction is made,
the signal arrival time usually accurate to within ±0.5 milliseconds.
is (Occasionally,
however, NOAA uses satellites that produce delay variations of several milliseconds during
a 24-hour period.) For example, the mean path delay from San Francisco through
GOES/East is 255 milliseconds. The mean path delay consists of 130.5 milliseconds from
satellite to Earth, and 124.5 milliseconds from Earth to satellite. Since the time is
advanced by 260 milliseconds before leaving Wallops Island, it arrives at San Francisco 5
milliseconds early. You can correct for mean path delay in San Francisco by subtracting
5 milliseconds from the time signal.

Fully Corrected: The satellite's orbit is not perfectly circular and not in the plane of the
equator. Therefore, the path delay at a fixed location typically varies by a few hundred
microseconds throughout the day. Also, the satellites are sometimes moved to keep their
orbital position within assigned limits. This movement causes irregular changes in the
path delay. Obtaining a fully corrected time signal requires correcting for these position
changes. Since satellite position data are included in the time code, users (and automatic
receivers) can make these corrections. A fully corrected time signal is usually accurate
to within ±100 microseconds. The ultimate accuracy depends on equipment delays and
noise levels in addition to the path delay.

119
 GOES EQUIPMENT
GOES time code receivers are commercially available. Some of them are accurate
to within 100 microseconds over averaging periods of hours, months, or years. Other
versions are accurate to within 1-2 milliseconds over the same periods. For a list of
manufacturers, contact the Time and Frequency Services Group, 576.00, NIST, 325
Broadway, Boulder, CO 80303.

The manufacturers of GOES receivers also sell several different types of antennas.
The best performance has been obtained using a right hand, circularly polarized helical
antenna with about 10 dB gain. Excellent results have been obtained with the microstrip
antenna, also shown. Dipoles and loops have also worked at lower levels of performance.
All of these antennas had gains in the range of 3 to 10 dB. As a general rule, GOES
antennas are small and easy to use and install. They often work well indoors. This
makes it easier to receive GOES than many other time and frequency signals.

TIME CALIBRATIONS USING GOES

To use the GOES signals to calibrate or set another clock, you need to determine
the relationship of the satellite-controlled clock's 1 pps output relative to UTC(NIST) and
to the clock being calibrated. This relationship is usually determined by use of a time
interval counter. For example, the clock to be calibrated may start the counter; then the
satellite controlled clock stops it. The UTC(NIST) and the clock
relationship between
being calibrated is then known by combining measured result with the computed
this
difference between the satellite clock and UTC(NIST).

An example of the clock calibration procedure is given below. The clock to be

calibrated is located in Boulder, Colorado, where the GOES satellite signals are being
received. The calibration procedure here for informational purposes. Keep in
is listed
mind that few, if any, users would actually need to work through this example. Most
GOES receivers perform these calculations automatically.

Signals received in Boulder (105.26 degrees West Longitude and 40.00 degrees North
Latitude) showed the satellite location to be:

134.92 degrees West Longitude

0.38 degrees South Latitude
-1-46 microseconds distance

120
 The total path delay from NOAA's Wallops Island to the GOES receiver in Boulder,
Colorado, is 266 351 microseconds. This delay can be broken into three parts:

WALLOPS ISLAND TO SATELLITE 133 606 microseconds

SATELLITE TO BOULDER 127 533 microseconds
MEASURED EQUIPMENT DELAY 5 192 microseconds

TOTAL DELAY 266 351 microseconds

Since the GOES signal is late with respect to the clock at Wallops Island, we can
write our equation as follows:

GOES SIGNAL WALLOPS ISLAND CLOCK =

- -266 351 microseconds

To compensate path delay between Wallops Island and the GOES receiver,
for the
the Wallops Island clock is set 260 000 microseconds early with respect to UTC(NIST).
This means that a GOES signal received anywhere inside the coverage area will always be
within 16 000 microseconds of UTC(NIST). In the case of our receiver in Boulder, the
time difference is only -6 351 microseconds (260 000 - 266 351):

GOES SIGNAL - UTC(NIST) = -6 351 microseconds

The measured time interval between the clock being calibrated and the GOES signal
is stated as:

CLOCK BEING CALIBRATED - GOES SIGNAL = +6,548 microseconds

Then, by adding the last two equations:

CLOCK BEING CALIBRATED - UTC(NIST) = +197 microseconds

means that the clock being calibrated is within 197 microseconds of

This
UTC(NIST). From this example, you can see that GOES serves as an excellent reference
for time calibrations.

121
 PRECAUTIONS FOR GOES USERS
The following paragraphs describe some conditions you should be aware of while
receiving GOES signals.

Interference From Land-Mobile Radio Services

The land-mobile radio services can interfere with the GOES time code, since they
use the same frequencies (468.8250 and 468.8375 MHz). This interference occurs mainly
in urban areas with high land-mobile activity. Since the land-mobile frequency allocations
have priority over the GOES allocations, complaints to the FCC do not help.

Because of the specific frequencies involved, the land-mobile services affect

GOES/West more than GOES/East. Therefore, users in urban areas should use GOES/East
if land-mobile interference is a problem,

Larger-than-Normal Time Code Deviations

The GOES time code is as accurate as

stated more than 99 percent of the time.
However, some situations can cause larger-
than-normal time code deviations. One
such situation occurs when NOAA uses a
GOES/East and/or GOES/West satellite
with larger-than-normal orbit inclinations.
In this instance, users in "uncorrected"
mode may receive time with variations of
several milliseconds.

A more common problem (though of

much shorter duration) occurs when
stationkeeping maneuvers are performed on
the satellites. These maneuvers are
performed every 1 or 2 months to keep the
assigned orbit locations.
satellites in their
Receivers that use the position data can
compensate for these maneuvers; however,
the position data are not updated until
0000 UTC on the day after the maneuver.
This means that for up to 24 hours
following a maneuver, the time code may
Wallops Island, be off by more than the normal 100
microseconds. NIST changes the accuracy-
Virginia indicator bit in the time code to indicate
when this condition occurs.

122
 Continuity

NIST cannot guarantee the long-term continuance of the GOES time code, since
the satellites belong to NOAA. However, NIST and NOAA have agreed to include the time
code in the GOES satellite transmissions as long as possible. The current agreement
extends to 1997.

THE TRANSIT NAVIGATION SYSTEM

These Navy navigation have been operational since 1964. Known as the
satellites
TRANSIT power, and coverage provide an excellent source of time
satellites, their design,
and frequency signals for calibration purposes. The signal format used on the system
lends itself well to time recovery from which frequency calibrations can be derived using
the techniques previously described in this book.

TRANSIT, being an orbiting system, experiences a much larger Doppler shift of its
radio signal than GOES. This is no real problem for the user since the overall system
design takes this into account and provides enough information to the users so that
accurate time recovery can be achieved. The biggest difference between GOES and
TRANSIT is that the TRANSIT system can be used only during several brief periods each
day when a satellite is in view of the receiver.

This system is a good candidate for users who have a worldwide requirement.
TRANSIT will work anywhere in the world. GOES is limited to the Western Hemisphere
and its oceans.

The satellites broadcast at about 400 MHz. Their message contains information
that allows position fixing and time recovery. As with the GOES system, TRANSIT allows
a user to recover time signals that can be used as a clock or to steer an oscillator's
frequency. Unlike GOES, however, the TRANSIT time signal format does not include day-
of-the-year information. Accuracy of the overall system is slightly better than GOES. As
with the GOES system, the satellite message is designed to allow the receiver to
synchronize itself, locate the required bits in the data stream and produce a time mark.
At this writing, the TRANSIT system permits time recovery to within 10-50 microseconds.

AlthoughTRANSIT is primarily a navigation satellite system, the U.S. Navy provides

support for frequency and time function. Users planning to obtain equipment to use
its

TRANSIT, especially at remote locations, should contact the U.S. Naval Observatory (page
23) for the latest information. The manufacturers of TRANSIT equipment are another
good source of information.

123
 At this writing, the TRANSIT system is supported by These
five orbiting satellites.
are in a circular polar orbit at an altitude of about 1100 kilometers. Thus, a user on the
Earth will see a satellite every 90 minutes or so. In contrast to a GOES-type of
transponder, TRANSIT has an accurate clock onboard each of
its satellites. These are
monitored from earth-control stations, and corrections are sent to the satellites periodically
to keep them on time. In addition to position information, the satellites send a time
mark every 2 minutes. By monitoring the signals from four ground-monitor stations, the
Navy is able to steer the on-board clocks by special commands and to carefully control
the accuracy of the TRANSIT signals.

Equipment costs are higher for TRANSIT than for GOES, but still not unreasonable
considering the few alternatives. Antennas are small, vertical, omnidirectional, and simple
to install. Some TRANSIT receivers are computer-controlled and allow nearly hands-off
operation. However, TRANSIT receivers typically require that you obtain the time of day
(from another source) to within 15 minutes in order to resolve the time ambiguity of
TRANSIT. For some applications, it may also be necessary to key your longitude and
latitude into the receiver.,

GLOBAL POSITIONING SYSTEM (GPS)

The Global Positioning System (GPS) is a worldwide satellite-based radio navigation

system developed by the U.S. Department of Defense (DOD). The system allows users to
obtain highly accurate time. GPS also transmits navigation information that permits
positioning accuracy at two levels. The higher accuracy is obtained from the precision
positioning service (PPS). Reduced accuracy is available from the standard positioning
service (SPS).

The PPS signal not available for use by the general public. However, the SPS
is

signal is available to any properly equipped user. There are no fees associated with the
use of SPS.

GPS should be fully implemented by the mid-1990's. The finished system will
include 24 satellites (21 primary satellites and 3 in-orbit spares). The spares will usually
be active and available to users. Each satellite will carry two rubidium and two cesium
atomic standards. The 24 satellites will be located in 6 equally spaced orbital planes (4
satellites per plane) inclined at 55 degrees with respect to the equator. The altitude of
each satellite is about 10 900 nautical miles.

GPS is being implemented in phases. The first GPS satellites (called Block I) were
launched beginning in February 1978. A total of 11 Block I satellites were launched. As
of this writing, seven are still operating and six are still usable for precise time transfer.

124
However, the Block I satellites were preliminary test units and will not be part of the 24

satellites included in the finished system. The satellites that will be included in the
finished system are called BlockII satellites. The first Block II satellite was launched in
1989, and the launches will continue at a rate of about 5 per year until all 24 satellites
are in orbit. After that time, about 3 additional launches per year may be needed to
provide GPS with the necessary in-orbit spares.

Once all 24 satellites are in orbit, the probability is greater than 99.6% that a user
(at a random time and place on the Earth) will have coverage within the given threshold
for PDOP (position dilution of precision). A value of PDOP < 10 and an elevation angle
greater than 5 degrees are considered necessary to meet this condition.

Each GPS satellite broadcasts two carrier frequencies, called LI and L2 (where LI
= 1575.42 MHz and L2 = 1227.6 MHz). Specified minimum signal strengths are -163
dBW and -160 dBW, respectively. Two spread spectrum waveforms (PRN) are in use:

(1) a coarse acquisition code (C/A), with a 1.023-MHz chip rate and
a period of 1 millisecond

(2) a precision code (P), with a 10-MHz chip rate

The C/A code is associated with the Standard Positioning Service and the P code
with the Precision Positioning Service. The C/A code is intentionally degraded in accuracy.
The LI frequency from each satellite is modulated with both the C/A and P codes, while
the L2 frequency is only modulated with the restricted P code. Although all GPS
satellites transmit on the LI and L2 frequencies, the interference between satellites is
minimal. This is because each satellite is given an individual code assignment to prevent
interference.

Both the LI and L2 carrier frequencies are continuously modulated with a 50 bit
per second navigation message. This message contains the following information:

(1) The current status of each satellite, allowing you to reject

satellites that may be having problems.

(2) A "Handover Word (HOW)" needed in the acquisition of the P

code from the C/A code.

(3) Satellite clock corrections, providing differences between the

particular satellite clock and GPS system time and between

125
 GPS system time and UTC(USNO). GPS system time is
normally kept within 1 microsecond of UTC(USNO), except that
leap seconds are ignored by GPS system time. Users who apply
all time corrections contained in the GPS data message can
relate their clocks to UTC(USNO) with a precision of 100
nanoseconds and an accuracy of better than 1 microsecond.

(4) Satellite position information.

(5) Information about ionospheric delay effects.

(6) Information about all GPS satellites. This allows you to acquire
all satellites once the first one has been acquired.

The GPS space segment supported by a network. This network includes a master
is

control station in Colorado Springs, Colorado, and multiple monitoring and upload stations
located around the world. Each monitoring station transmits the data they receive to the
master control station. This information is then sent to the GPS upload stations and
uploaded into the satellites, so that the navigation messages can be updated.

GPS system time is produced by the reference clock at one of the monitoring station
sites. This role of system reference clock can be switched from clock to clock and from
station to station if necessary, without making significant changes to the GPS system time.

As mentioned Standard Positioning Service (SPS) are restricted

earlier, users of the
to signals with intentionally degraded accuracy. Since SPS is available to anyone, the
DOD adds errors to the data in the interest of national security. Currently, SPS users
can expect timing errors of about 300 nanoseconds, but DOD reserves the right to reduce
the accuracy even more.

Even with the reduced accuracy of SPS, GPS is still an excellent time and frequency
source. Most users should find it more than adequate. However, if you do need more
precise time, you have several options:

(1) You can receive signals from the Block I satellites. The current
Block I satellites provide unrestricted, full-accuracy timing
signals. Some of these satellites are expected to remain usable
for several more years.

126
 (2) You can use GPS in the "common-view" mode (described in the
next section). The common-view mode does not totally eliminate
the reduced accuracy of SPS, but it does minimize the effects.

(3) Since the data errors are random, you can improve the data
by averaging.

(4) You can apply corrections to the received data.

For more information about GPS, contact the Time Service Department of the U.S.
Naval Observatory, 34th and Massachusetts Ave., NW, Washington, DC 20392. Their
Automated Data Service contains a number of files accessible to civilian timing users of
GPS that provide both current and past GPS data, as well as current information on the
GPS system.

Using GPS

A variety of GPS receivers are available. These include sophisticated multi-

frequency versions, small handheld units, and specialized timing receivers used for civilian
timing applications. A single-frequency, SPS timing receiver can be purchased from a
number of commercial suppliers at prices ranging from $15 000-$25 000 (1990 prices).
These receivers are often completely automatic and require only a small omnidirectional
antenna.

A typical GPS timing receiver can automatically acquire the signal from each desired
GPS satellite, extract the timing information from the received signal, and apply
corrections to the received time to determine either GPS system time or UTC(USNO).
The receiver's coordinates must be known and entered into the receiver. This can be done
either manually (if the precise location of the site is known), or automatically by the
receiver, through performing a navigation solution from the GPS signals from multiple
satellites. Ideally, the position should be known within 1 meter or better.

To derive UTC at your location, several corrections need to be determined and

applied to the received satellite signal. First, the path delay from satellite to receiver
must be computed by the receiver. This is possible because the precise positions of the
satellite and receiver are known. The satellite position information is included in the
navigation message, and the receiver's position was entered during receiver initialization.
Next, the resulting path delay is further corrected for ionospheric and tropospheric delay
effects,again making use of data supplied in the navigation message. At this point the
time delivered to your site can be related to the time kept by the satellite's onboard clock.

127
 In order to refer the received time to GPS system time or to UTC, further
corrections are necessary. To relate to GPS
system time, a correction must be applied for
the amount of time offset between the satellite clock and GPS system time. This
information is available to the receiver from the navigation message. Finally, you may
apply a correction for the current difference between GPS system time and UTC(USNO).
This information is available from continuous monitoring of the received GPS signals by
the U.S. Naval Observatory. You can obtain the USNO data in various forms from the
USNO's Automated Data Service by using a computer and a modem. Contact the USNO
for a list of available GPS files and for instructions for accessing the computer database.

There are two principal methods for using GPS time. The first is the "time
distribution" mode. In this mode the user receives the direct GPS timing signals and uses
them as the local equivalents of GPS or UTC time (depending on which correction was
applied). Some receivers provide standard time and frequency outputs that are phase
locked to UTC(GPS) or UTC(USNO). The time distribution mode is capable of providing
time to within 100 nanoseconds, but will be less accurate when using SPS.

The second method is the "common-view" mode. This mode is used to synchronize
or compare time standards or time scales at two or more locations. For example, common-
view is often used to make time scale comparisons between international laboratories. A
diagram of the common-view mode is shown below:

128
 Users at the two sites, A and B, make simultaneous measurements of the same GPS
timing signal when the satellite is visible to each site with reasonable elevation angles.
Each site measures the difference between its local clock and the received signal.
Typically, repeated 6-second measurements are made during a satellite pass lasting for 10-
15 minutes. The individual measurements at each site are then averaged to produce
estimates of (Clock A - GPS satellite clock) and (Clock B - GPS satellite clock). If these
results are then subtracted, the satellite clock drops out and an estimate of Clock A -
Clock B remains.

One advantage of the common-view technique is that the errors associated with the
satellite clock are common to both and are thus eliminated in the comparison. This
sites
means that satellite clock errors introduced by the SPS service should not affect the
accuracy of common-view time comparisons. The common-view method also reduces the
effect of satellite position errors.Since these errors are not totally common to the two
paths they will still have some impact on the time difference results. If a comparison is
being made between two widely separated locations (like the United States and Japan), this
effect can produce uncertainties as large as several tens of nanoseconds.

In order to use the common-view technique, the sites being compared must be close
enough to each other to receive signals from the same satellite. However, the acceptable
distance is larger than might be expected. For example, it is possible for common-view
time comparisons to be made between the western United States and India.

The common-view method provides the basis for the NIST Global Time Service.
This service allows users to procure a turn-key GPS receiving system from NIST and
allows traceability to UTC(NIST) at accuracy levels of less than 100 nanoseconds. Also,
BIPM publishes tracking schedules for common-view links between various regions of the
world. Users wishing to compare time with other laboratories can consult these schedules
to determine which satellites and time periods are available for common-view
measurements. These comparisons can provide accuracies over intercontinental distances
to within 10 nanoseconds.

Tofurther improve the usefulness of GPS, other more specialized techniques are also
being developed. One example is a codeless, ionospheric calibration receiver being
developed by several laboratories. Since errors in predicting the ionospheric delay are a
major source of time comparison uncertainties, this new receiver could significantly improve
the results. The ionospheric calibration receivers make use of the fact that the two carrier
frequencies (LI and L2) are affected differently as they traverse the ionospheric region.
By measuring this different behavior, you can predict the actual path delay more
accurately. These new receivers can perform this calibration using the dual GPS
frequencies without having access to the restricted P code transmissions. Another possible
development is a receiver that can receive signals from both the GPS satellites and signals
from the GLONASS system under development by the USSR.

129
 GPS Performance

Although not all GPS satellites have been launched, the system's time transfer
of the
capabilities are available now, since only one satellite is needed to transfer time. The plot
below shows GPS performance over a 3-day period when compared to UTC(NIST):

GPS Timing Receiver vs UTC(NIST)

E .25

86/28/98 (MJD 48862) CHI Adjusted (MJD 48865) 86/23/98

21:47 ttic marks = 2 hours] 22:12
Data points= 43 Relative Frequency: +1.17E-13 r=+8.1B

The shows data obtained with a commercially available GPS timing receiver.
plot
These receivers are readily available, and their cost has gone down over the past few years.
The receiver was using the (C/A) code with intentionally degraded accuracy. Even so, the
range of the data is less than 0.5 microsecond (500 nanoseconds) over the 3-day period.

The plot GPS is an excellent source for precise time transfer. However,
shows that
users outside the DOD
should realize that the precision of GPS timing has been reduced
and may be reduced further in the future. If you need time at the highest possible
accuracy you should monitor future GPS developments very closely.

SUMMARY
Satellite signals are an excellent reference for time calibrations and may also be used
for frequency calibrations. Commercially available satellite receivers are sophisticated, easy
to use and install, and allow you to obtain time to within 100 microseconds or better.

The next chapter of this book discusses two NIST services that allow you to recover
time by telephone.

130
 Chapter 8 - CALIBRATIONS BY TELEPHONE

If your accuracy requirements are low, you may be able to get a perfectly acceptable
time signal by telephone. This chapter describes two telephone services offered by NIST.

THE NIST (VOICE) TELEPHONE TTME-OF-DAY SERVICE

The audio portions of the WWV
and WWVH
broadcasts (chapter 5) can also be heard
by telephone. The actual time you hear on the telephone will differ by a few milliseconds
from that heard by radio. This is because of the delay in the telephone lines. In addition,
if you are from any great distance, the way that the telephone company handles
calling
your call may few hertz. This means that if you are
limit the accuracy of the tones to a
making a calibration at 1000 Hz, you could have an error as great as 3 Hz. This accuracy
is not as good as the accuracy obtained using radio signals, but may be perfectly acceptable.

To hear these broadcasts, dial (303) 499-7111 for WWV, and (808) 335-4363 for WWVH.
Callers are disconnected after 3 minutes. These are not toll-free numbers, and callers
outside the local calling area are charged for the call.

THE AUTOMATED COMPUTER TIME

SERVICE (ACTS)

In 1988, NIST began the Automated

Computer Time Service (ACTS). This service
provides computers with telephone access to
NIST time at accuracies approaching one
millisecond. Since the time code uses the
standard ASCII character set, it works with
nearly all computer systems and modems.
Simple hardware can also be built to set non-
computer clock systems.

The phone number for ACTS is (303)

494-4774. ACTS operates at 300 or 1200 baud,
with 8 data bits, 1 stop bit, and no parity. In
addition to the UTC hours, minutes, and
seconds, the time code includes the date, the
Modified Julian Date (MJD), Daylight Saving
Time and leap second indicators, a leap second
flag, a UT1 correction, and other information.
Users at 1200 baud receive the full time code
each second, while 300- baud users require 2
seconds to receive the complete time code.

131
 The ACTS Time Code Format

The time code for ACTS is sent in ASCII at either 300 or 1200 baud. Both baud rates
require 8 data bits, 1 stop bit, and no parity. The 1200-baud format is shown below:

? = HELP
National Institute of Standards and Technology
Telephone Time Service

(1 second pause here)

D L D
MJD YR MO DA H M S ST S UT1 msADV OTM
47222 88-03-02 21:39:15 83 0 + .3 045. 0 UTC(NIST) *
47222 88-03-02 21:39:16 83 0 + .3 045. 0 UTC(NIST) *
47222 88-03-02 21:39:17 83 0 + .3 045. 0 UTC(NIST) *
47222 88-03-02 21:39:18 83 0 + .3 045. 0 UTC(NIST) *
47222 88-03-02 21:39:19 83 0 + .3 037 6 UTC(NIST) #
.

47222 88-03-02 21:39:20 83 0 + .3 037 6 UTC(NIST) #

.

© t- C • • • Gt-C^ • * * • • • •

The part of the time code contains the Modified Julian Date (MJD), the date
first
(year, month, day), and the time (UTC hours, minutes, seconds). This information is
followed by the Daylight Saving Time (DST) information. The DST code is always a two-
digit number (00 to 99). This code is normally 00 when Standard Time (ST) is in effect,
or 50 when DST is in effect. About 48 days prior to a time change, however, this code
starts counting the days until the change. When ST is in effect, the DST code counts down
from 99 to 51 in the 48 days prior to the time change. When DST is in effect, the DST
code counts down from 49 to 01 in the 48 days prior to the time change. In both cases, the
code is updated at 0000 UTC.

The leap second (LS) If no leap second is scheduled

flag is always "0," "1," or "2."

at the end of the current month, it is "0." It is a leap second is scheduled to be "1" if

added on the last day of the current month. The LS flag remains on for the entire month
before the leap second is added. Once the leap second is added, the LS flag is reset to "0".
The leap second flag is a "2" only if a leap second is to be deleted on the last day of the
current month. So far, a negative leap second has never been needed.

The UT1 correction is shown as either a positive or negative number in steps of 0.1
second. The remainder of the time code shows the millisecond advance (msADV) and the
on-time marker (OTM). The displayed time is valid at the OTM. The OTM is either an

132
"*" or a "#" character. When you first connect, the "*" OTM
is displayed. This marker is
transmitted 45 milliseconds early with respect to UTC(NIST). The 45 milliseconds accounts
for the 8 milliseconds required to send a character at 1200 baud, an additional 7
milliseconds to compensate for delay from NIST to the user, plus a 30 millisecond
"scrambler" delay. The "scrambler" delay approximately compensates for the internal delay
found in 1200-baud modems.

If you echo all characters, NIST measures the round trip delay and advances the
OTM so that the midpoint of the stop bit arrives at your computer on time. When this
happens, the msADV shows the actual required advance in milliseconds and the OTM
becomes "#." Four consecutive stable measurements are needed before the switches OTM
from "*" to "#." your 1200-baud modem has the same internal delay used by NIST (30
If
milliseconds), then the "#" OTM should arrive at your computer within ±2 ms of the correct
time. Different brands of 1200-baud modems have different internal delays, and the actual
offset of the "#" OTM may be as large as ±10 ms. This should still be more than adequate,
however, since many computer clocks can only be set within 20-50 milliseconds.

The 300-baud time code is potentially more accurate than the 1200-baud code. Due
to a simpler modulation scheme used at 300 baud, the OTM should arrive at your computer
within 1 millisecond of the correct time.

Users are allowed 55 seconds on ACTS unless all lines are busy. In that case, the
first call that reaches 15 seconds is terminated.

Performance of ACTS

With appropriate software, ACTS can set or check computer time-of-day clocks in one
of two different modes:

Fixed Delay Mode In this mode, the user receives the time code and an on-time marker
-

character. The marker character has been advanced in time by a fixed amount to
compensate for typical modem and telephone line delays. Unless the connection is routed
through a satellite, the accuracy in this mode should be better than 0.1 second.

Measured Delay Mode - In this mode, the user's computer echoes all characters back to
NIST and the round trip line delay is measured. The on-time marker character is then
advanced to compensate for the line delay. The accuracy in this mode should be better than
10 milliseconds using a 1200-baud modem, or about 1 millisecond using a 300-baud modem.
Accuracy at 1200 baud is limited by the internal delays in 1200-baud modems. Repeatability
at both 300 and 1200 baud is about 1 millisecond.

133
 The accuracies cited above assume that the telephone connection is reciprocal, that
is, that the connections follow the same path, so the path delay is the same in both
directions. Most phone connections are of this type.

NIST offers software which lets users use ACTS on several popular personal
computers. The documentation for this software includes more information about the
features of the service and a simple circuit diagram that can be used to obtain an on-time
pulse. This material costs $35.00 (subject to change without notice). To order, contact:
NIST Office of Standard Reference Materials, Room 205, Building 202, NIST, Gaithersburg,
MD 20899, telephone (301) 975-6776. Ask for the Automated Computer Time Service
Software, #RM8101. If you have questions or comments about ACTS, write to: NIST-
ACTS, Time and Frequency Division, 576.00, 325 Broadway, Boulder, CO 80303.

134
 )

APPENDIX - SUMMARY OF TIME AND FREQUENCY

RADIO CALIBRATION SOURCES

The table below serves as a quick reference to the time and frequency radio
calibration sources described in this book. When using the table, please keep in mind that
the accuracies stated are relative to UTC(NIST) and are possible only within the coverage
area of the signal. Some signals may not be usable in your area. For more specific
information about each calibration source, please refer to the chapter listed in the table.

Name of Source Accuracy Topic

and carrier Covered
frequency Frequency Time In:
(24 nours) (microseconds

nr olgrldlS
(2.5 to 25 MHz) 1. 00E-07 1000 Chapter 5

GOES Satellite
(468 MHz) 1. 00E-09 100 Chapter 7

Omega and VLF

(10 to 15 kHz) 1. 00E-10 1000 Chapter 6

WWVB and LF
(30 to 300 kHz) 1. 00E-11 1000 Chapter 6

Loran-C
(100 kHz) 1.00E-12 < 10 Chapter 6

GPS
(1575 MHz) 1.00E-13 < 1 Chapter 7

135
 GLOSSARY

The following definitions are offered as an aid to the reader. The definitions relate
to the way the terms appear in this book and may differ from the definitions listed in other
references.

ACCURACY - the degree of conformity of a measured value to its definition.

AGING the systematic change in frequency with time due to internal changes in the
-

oscillator.For example, a 100-kHz quartz oscillator may age until its frequency becomes
100.01 kHz.

AMBIGUITY having more than one possible value. For example, if a 24-hour clock
-

displays a time of 3 hours and 5 minutes, it is ambiguous as to the day. If we add a

number to the clock display to show that it is day 17, the time is still ambiguous as to the
month. If we add another number showing it is month 3, the time is still ambiguous as to
the year.

ATOMIC TIME (TA) - the time obtained by counting cycles of an atomic frequency source
as opposed to time based on the Earth's rotation. Atomic time is extremely uniform.

AUTOMATIC GAIN CONTROL (AGO a - circuit used in some radio receivers (including
HF receivers) so that both weak and strong signals can be heard at similar volume.

CALIBRATION a measurement that determines the frequency or time offset of

- an
UTC(NIST). Calibrations are usually made
oscillator relative to the national standard, by
comparing the output of an oscillator to a reference signal known to be traceable to
UTC(NIST). The reference signal is usually obtained by radio. Calibrations can also be
made by making a direct comparison between two oscillators. In this case, one oscillator
is assumed to be the better of the two and is used as the reference.

CESIUM OSCILLATOR (Cs) - any oscillator that uses cesium to obtain atomic resonance.
Cesium oscillators are often used as a laboratory's primary frequency standard.

DECADE COUNTING ASSEMBLY (DCA) - a circuit used in counters that counts from
0 to 9 before it overflows. This enables the counter to display units of tens, hundreds, and
so on.

DIURNAL PHASE SHIFT - the phase shift (diurnal means daily) associated with sunrise
and sunset on low frequency radio paths.

137
DOPPLER EFFECT - an apparent change in frequency caused by movement of either the
transmitter or receiver. In the case of radio waves, the Doppler effect occurs when the
ionosphere changes its position.

DRIFT (FREQUENCY DRIFT) - the systematic change in frequency with time of an

oscillator. Drift is due to aging plus changes in the environment and other factors external
to the oscillator (see Aging).

DUT1 - UT1 and UTC, expressed to the nearest

the approximate time difference between
0.1 second. The value of DUT1 sometimes included in time codes (WWV, for example)
is

so that listeners can correct the time as heard to make it agree with UT1 to within 0.1
second.

EPHEMERIS TIME (ET) - an astronomical time scale based on the orbital motion of the
Earth.

FREQUENCY - if T is the period of a repetitive event, then the frequency (f) is equal
to 1/T. Frequency is normally expressed in hertz, with 1 Hz meaning that an event occurs
once per second.

FREQUENCY-SHIFT KEYING (FSK) - a means of modulating a radio carrier by changing

its frequency by a small amount.

GREENWICH MEAN TIME (GMT) - a 24-hour time system based on the time at
Greenwich, England. Greenwich Mean Time can be considered equivalent to Coordinated
Universal Time (UTC), which is broadcast from all standard time and frequency radio
stations. However, several international organizations have recommended that UTC be used
rather than GMT in all applications.

GROUP REPETITION INTERVAL (GRI) - the rate of recurrence of specified groups of

pulses. In the case of Loran-C, the GRI is altered between different groups of stations to
avoid interference.

IONOSPHERE - the outer part of the Earth's atmosphere. The ionosphere consists of a
series of constantly changing layers of ionized molecules. Many radio waves reflect back to
Earth from the ionosphere.

138
JITTER - usually used to describe the small changes in a signal over time. Phase jitter
causes a counter to be triggered either early or late.

JULIAN DAY (JD) - the Julian Day is obtained by counting days with a starting point of
noon on January 1, 4713 B.C. (Julian Day zero). This is one way of telling what day it is
with the least possible ambiguity.

LEADING EDGE - the first occurring change in a pulse. The leading edge can be either
negative- or positive-going.

LEAP SECOND an intentional increment of 1 second used to adjust UTC to ensure

-

approximate agreement with UT1. An inserted second is called a positive leap second and
an omitted second is called a negative leap second. A leap second is needed about once per
year.

LONG-TERM STABILITY - describes the frequency change of an oscillator that occurs

over periods greater than 1 second.

MAXIMUM USABLE FREQUENCY (MUF) - the highest frequency that can be used for
radio transmissions without having the signal escape through the ionosphere. Signals below
the maximum usable frequency will reflect off the ionosphere.

MEAN SOLAR TIME - time based on a mean solar day, which is the average length of
all solar days in a solar year. The mean solar second is 1/86 400 of a mean solar day.

MIXER - an electrical circuit that mixes two signals and produces a new signal that is

usually either the sum or the difference of the two inputs.

MODIFIED JULIAN DAY (MJD) - this is equal to the Julian day, shifted so its origin
occurs at midnight on November 17, 1858. The MJD differs from the Julian Day by exactly
2 400 000.5 days.

NIST - abbreviation for The National Institute of Standards and Technology (formerly
the National Bureau of Standards).

OMEGA NAVIGATION SYSTEM - a low frequency (10 to 15 kHz) navigation system

also used for time and frequency calibrations.

139
PATH DELAY - the amount of time it takes for a radio signal to travel from the
transmitter to the receiver.

PHASE JUMP - a sudden phase change in a signal.

PHASE LOCK - a servo mechanism technique for causing one signal to follow another.
The signals can be (but do not have to be) at the same frequency. Phase locking in terms
of frequency sources is analogous to a mechanical servo, where one shaft or wheel follows
another.

PHASE SIGNATURE a deliberate phase offset to identify a signal.

- as broadcast WWVB
is deliberately phase shifted at 10 minutes after the hour. This tells you that are tracking
WWVB and not some other signal.

POWER LINE FREQUENCY - in the United States, the power lines are held to a
frequency of 60 Hz. This frequency is often used to drive electric clocks.

PRECISION - the degree of mutual agreement among a series of individual measurements.

Precision is often (but not necessarily) expressed by the standard deviation of the
measurements.

RECEPvTER DELAY - the delay experienced by the signal going from the receiving antenna
to the detector or output device, speaker, or oscilloscope terminal. A delay of several
milliseconds is typical in high frequency radio receivers.

REPRODUCIBILITY - the ability of a single device (or a set of devices) to produce the
same value when they are put into operation repeatedly without adjustment.

RESOLUTION - the degree to which a measurement can be determined is called the

resolution of the measurement. For example, a measurement made with a time interval
counter might have a resolution of 10 nanoseconds.

RUBIDIUM OSCILLATOR (Rb) - a precision atomic oscillator based on a resonance of

rubidium gas.

SELECTIVITY - the characteristic of a radio receiver that allows it to separate one signal
from another.

140
SENSITIVITY - the characteristic of a radio receiver that allows it to detect weak signals.

SHORT-TERM STABILITY - a description of the frequency fluctuations caused by random

noise in an oscillator. To properly state the short-term stability, the number of samples,
averaging time, repetition time, and system bandwidth must be specified.

SIDEREAL TIME - time based on observations of stars rather than the Sun. Sidereal
time is used by astronomers and a sidereal day is equal to about 23 hours, 56 minutes, and
4 seconds of solar time. Because it is based on observations of stars, it is more accurately
determined than solar time.

SIGNAL-TO-NOISE RATIO (SNR) -the ratio of the strength of a radio signal to that of
the noise. This is a more useful term than signal strength.

STANDARD -a universally accepted reference. The NIST frequency standard is a cesium

oscillator locatedat the National Institute of Standards and Technology in Boulder,
Colorado. Many frequencies generated and used in the United States are referenced to this
standard.

STANDARD FREQUENCY BAND - radio bands allocated expressly for the purpose of
distributing standard frequency and time signals. For example, all of the transmissions of
WWV occur in standard frequency bands centered on 2.5, 5, 10, 15, and 20 MHz.

SYNCHRONIZATION - in the context of timing, synchronization means to bring two

clocks or data streams into phase so they agree. They need not be on time.

TCXO (TEMPERATURE COMPENSATED CRYSTAL OSCILLATOR) - a crystal oscillator

that contains special components to minimize the effect of temperature on the crystal
frequency.

TIME BASE in a frequency counter the oscillator that provides the timing signals for
-

measurement and control. The accuracy of the counter is directly related to the time base.

TIME INTERVAL COUNTER - a counter designed to measure the time interval between
two input signals. Time interval counters are often used to measure the performance of an
oscillator.

141
TIME OF COINCIDENCE (LORAN) the time when the GRI pulse from a Loran-C
-

station coincident with an exact second of Coordinated Universal Time. These times of
is

coincidence occur at regular intervals of 16 minutes or less, depending on the GRI.

TIME SCALE - a system of unambiguous ordering of events.

TRANSFER STANDARDS the name given to a signal that is used to perform a

-

calibration when it can be referenced to a standard. For example, since WWVB

is referenced

to the NIST frequency standard, it can be used as a standard to calibrate other oscillators
at remote locations.

TRIGGER ERROR - the error associated with false triggering caused by phase noise or
jitter on a signal.

UNIVERSAL TIME (UT) FAMILY

Universal Time (UT) is the name given to time scales
-

based on the rotation of the Earth. UT is given in several ways. Apparent solar time is
first corrected by the equation of time to mean solar time, UTO. It is then again corrected
for migration of the Earth's poles to obtain UT1. This is further corrected by removing
fluctuations of unknown origin to obtain UT2.

ZERO BEAT - the condition between two signals when no beat is heard or seen between
the signals (when the frequencies are equal).

142
 INDEX

ACCURACY
Definition 28

AGING (OF QUARTZ CRYSTALS) 33-34

ANTENNAS
GOES Satellite 120
GPS Satellite 127
HF (High Frequency) 66-69
TRANSIT Satellite 124
VLF and LF (Very Low and Low Frequency) 90

APPARENT SOLAR TIME 15

ASTRONOMICAL TIME 20

ATOMIC
International Atomic Time (TAI) 17
Oscillators
Rubidium 36-37
Cesium 37-38
Hydrogen 38-39
Time Scales 17-18

AUTOMATED COMPUTER TIME SERVICE (ACTS)

Computer Software 134
General Information 131
Performance 133-134
Time Code Format 132-133

BASIC STANDARDS 6

BEAT FREQUENCY METHOD FOR FREQUENCY CALIBRATIONS 81-84

CALIBRATIONS
Frequency Calibrations
Automated Calibration Systems 44-55
Definition 9-11
Direct Measurements 41
Theory 41-44
Time Interval Measurements 42-44,60,113
What They Measure 55-60
Using HF Broadcasts 80-85
Using Loran-C 113-114

143
 Using Omega 100-104
Using WWVB 94-95
Time Calibrations
Using GOES , 120-121
Using the GOES Time Code 118-119
Using GPS 127-129
Using HF Broadcasts 70-78
Using Loran-C 112-113
Using WWVB 96-97
Using the WWVB Time Code 98-99
Using the WWV/WWVH Time Code 78-80

CESIUM BEAM OSCILLATORS 37-38

CHU (CANADA) 63

CLOCKS 25-27

COMPUTERS
Time Recovery for Computer Systems 61,131-134
Used in Measurement Systems 49-50

CONVERSIONS
Parts Per ... To Percents 12
Prefix 12
Wavelength to Frequency 14

COORDINATED UNIVERSAL TIME (UTC) 18-19

COORDINATION OF FREQUENCY AND TIME ACTIVITIES 21-23

COUNTERS
Frequency 28,41,44
Time Interval 42-44,48

DELAYED TRIGGER METHOD OF TIME CALIBRATION 75-77

DEPARTMENT OF DEFENSE, U.S. 22-23

DEPARTMENT OF TRANSPORTATION, U.S. 19

DIRECT TRIGGER METHOD OF TIME CALIBRATION 73-75

DISTRIBUTION SYSTEMS 8-9

DIURNAL PHASE SHIFTS 91,95,103-104

107,111

144
DRIFT, FREQUENCY 34

ELECTRIC POWER LINE FREQUENCY (60 Hz) 3,26-27

EPHEMERIS TIME (ET) 16-17

FEDERAL COMMUNICATIONS COMMISSION (FCC) 2

FREQUENCY CALIBRATIONS (See CALIBRATIONS)

FREQUENCY DIVIDER 48

FREQUENCY MEASUREMENT SYSTEMS

Computer-Controlled 46-51
Day to Day Operation 51-53
Output 50-51
Recordkeeping 53
Typical 45

GLOBAL POSITIONING SYSTEM (GPS)

Block I satellites 124-125
Block II satellites 125
Carrier Frequencies 125
Common-View Mode 128-129
Coverage 129
Performance 130
Precision Positioning Service (PPS) 124-26
Receivers 127
Standard Positioning Service (SPS) 124-126
Time Calibrations 127-129

GOES SATELLITE
Accuracy 116, 119-121
Antennas 120
Continuity 123
Coverage 117
Frequency Calibrations 116
Interference 122
Outages 123
Performance 119
Receivers 120
Signal Characteristics 118
Time Calibrations 120-121
Time Code Deviations 122
Time Code Format 118-119

GREAT CIRCLE DISTANCE CALCULATIONS 85-86

145
GREENWICH MERIDIAN 19

GROUNDWAVE, LORAN-C 107,111

GROUP REPETITION INTERVAL (GRI), LORAN-C 109-110

HF (HIGH FREQUENCY) BROADCASTS

Antennas 66-69
Broadcast Formats
CHU (Canada) 63
WWV/WWVH 61-62
Calibrations
Frequency 80-85
Time 70-78
Time Code 78-80
Ionospheric Effects on Reception 87
List of Stations 64-65
Receivers
Measuring Receiver Time Delay 71-73
Selection 65-66

HYDROGEN MASER 38-39

INTERNATIONAL ATOMIC TIME SCALE (TAI) 17

INTERNATIONAL BUREAU OF WEIGHTS AND MEASURES (BIPM) 21

LEAP SECOND 18,139

LF AND VLF (LOW AND VERY LOW FREQUENCY) BROADCASTS

Antennas 90
Calibrations
Frequency
Using Loran-C 113-114
Using Omega 100-104
Using WWVB 94-95
Time
Using Loran-C 112-113
Using WWVB 96-97
Using the WWVB Time Code 98-99
Field Strengths 92-93
Monitoring Station Availability 93
Propagation Characteristics 91-92
Signal Formats
General 90
Loran-C 109-110
Omega 101-102

146
 Stations
List of LF and VLF Stations 105-106
List of Loran-C Stations 107-109
List of Omega Stations 100

LORAN-C NAVIGATION SYSTEM (100 kHz)

Calibrations
Frequency 47-53,113-114
Time 112-113
Chains 107-110
Control (by USNO and U.S. Coast Guard) 107
Group Repetition Interval (GRI) 109-110
List of Stations 107-109
Performance 113-114
Published Data 54
Pulse Shape 111-112
Reception 111

MAXIMUM USABLE FREQUENCY (MUF) 87

MEAN SOLAR TIME 15

MODAL INTERFERENCE 103

NAMEPLATE FREQUENCY 56-58

NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY (NIST)

Publications 22,54,93,117
Responsibility 6,21-22
Services
Radio Broadcasts
GOES 115-123
WWVB 89-99
WWV/WWVH 61-88
Telephone Service
Automated Computer Time Service (ACTS) 131-134
Telephone Time-of-Day Service 131
Time and Frequency Standards 8-9
Traceability to NIST 10-11,53-56

NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION (NOAA) 115, 123

OMEGA NAVIGATION SYSTEM

Coverage Area 101
Frequency Calibrations 100-101
List of Stations 100
Operating Characteristics 101-102
Propagation Characteristics 103-104

147
OSCILLATORS
Comparison Table 40
Performance 28-30
Types
*
Cesium 37-38
Hydrogen Maser 38-39
Quartz 31-35
Rubidium 36-37

OSCILLOSCOPE
Photography 78,96
Used for Frequency Calibrations 84
Used for Time Calibrations 70-78,96-97

OSCILLOSCOPE PATTERN DRIFT METHOD 84

PATH DELAY
GOES Path Delay Corrections 119
GPS Path Delay Corrections 127
Measuring Radio Path Delay 85-88

PATTERN DRIFT CALIBRATION METHOD 84

PHASE PLOTS
Cesium Performance 50,59,114
Omega Performance 104
Quartz Oscillator Performance 58
Rubidium Performance 59,114
WWVB Performance 91,94

PIEZOELECTRIC EFFECT 31

POLAR CAP DISTURBANCE (PCD) 103

PRIMARY FREQUENCY STANDARD 30,38

PRIMARY OSCILLATOR 44-45,48-49

PROPAGATION CHARACTERISTICS
GOES 116
GPS 125-126
HF Broadcasts 87-88
LF and VLF Broadcasts 91-92
Loran-C 111
Omega 103-104
TRANSIT Satellites 123

148
QUARTZ OSCILLATORS
General Information 31-32
Performance 35,58
Quartz Crystals 31-32
Temperature and Aging Effects 33-34
Warm-Up Period 34

RADIO FREQUENCY BANDS 13

RECEIVER TIME DELAY, MEASURING 71-73

REFERENCE FREQUENCY 45,47-48,55,60

RELATIVE FREQUENCY
Defined 55-58
General 29-30,60
Measuring with WWV 82-84
Measuring with WWVB 95

REPRODUCIBILITY 28

RESONANCE FREQUENCY 32

RUBIDIUM GAS CELL OSCILLATORS 36-37

SATELLITE SERVICES
GOES 115-123
GPS 124-130
TRANSIT 123-124

SECONDARY FREQUENCY STANDARD 30

SECOND (DEFINED) 5-6,27

SIDEREAL TIME 15

SKYWAVE, LORAN-C 107,111

SOLAR TIME 15-16

STABILITY 28-30

STANDARDS
Definition 6
NIST Standards for Time and Frequency 7-10
Primary and Secondary Standards 30

SUNDIAL 8,15

149
TCXO (TEMPERATURE-COMPENSATED CRYSTAL OSCILLATOR) 33-35

TELEPHONE TIME SERVICES 131-134

TIME AND FREQUENCY

General Information 1-2,5-6,25-27
Terminology 11-14
Users of Time and Frequency 2-5

TIME AND FREQUENCY DIVISION OF NIST 21-22

TIME BASE 42-43,48

TIME CALIBRATIONS (See CALIBRATIONS)

TIME CODE
CHU (Canada) 63
GOES 118-119
WWVB 98-99
WWV/WWVH 78-80

TIME COMPARISON METHOD OF FREQUENCY CALIBRATIONS 85

TIME OF COINCIDENCE (TOO 112

TIME SCALES
Atomic Time Scales 17-18
Solar Time Scales 15-16
Using for Navigation and Astronomy 19-20

TIME ZONES 19-20

TRACEABILITY 10-11,53-56

TRANSIT SATELLITE NAVIGATION SYSTEM 123-124

U.S. NAVAL OBSERVATORY (USNO)

Data Monitoring and Publications 93,107,112,115
GPS Data 127
Responsibility 20-23

UTC (COORDINATED UNIVERSAL TIME) 18-19

UTO 15

UTl
Corrections
On ACTS 126
On GOES 119

150
 On WWVB 98-99
On WWV/WWVH 80
Defined 16
Determination of 21,23
Relationship to UTC 18

UT2 16

WAVELENGTH 13-14

WRISTWATCH 5,9-10

WWVB
Antennas 90
Cycle Phase Shifts 92
Field Strength 92-93
Frequency Calibrations 94-95
Performance 91-92
Propagation Characteristics 91-92
Signal Envelope 96
Signal Formats 90
Time Calibrations 96-97
Time Code Format 98-99

WWV/WWVH
Antennas 66-69
Broadcast Formats 61-62
Frequency Calibrations 80-85
Measuring Radio Path Delay 85-88
Receiver Selection 65-66
Time Calibrations 70-78
Time Code Format 78-80

ZERO BEAT METHOD OF FREQUENCY CALIBRATIONS 81-84

151

-fr U.S. GOVERNMENT PRINTING OFFICE. 1990 -261 -913 20704

 PUBLICATION OR REPORT NUMBER
BL-114A U.S.DEPARTMENT OF COMMERCE 1.

(5-90) NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY NIST/SP-559 (Revised 1990)

PERFORMING ORGANIZATION REPORT HUMBER

BIBLIOGRAPHIC DATA SHEET 3. PUBLICATION DATE

September 1990
4. TITLE AND SUBTITLE

Time and Frequency Users Manual

5. AUTHOR (S)

George Kamas and Michael A. Lombardi

6. PERFORMING ORGANIZATION (IF JOINT OR OTHER THAN NIST. SEE INSTRUCTIONS) 7. CONTRACT/GRANT NUMBER
DEPARTMENT OF COMMERCE
U.S.
NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY
BOULDER, COLORADO 80303-3328 8. TYPE OF REPORT AND PERIOD COVERED

9. SPONSORING ORGANIZATION NAME AND COMPLETE ADDRESS (STREET, CITY, STATE, ZIP)

10. SUPPLEMENTARY NOTES

Supersedes NBS Special Publication 559 dated November 1979

11. ABSTRACT (A 200-WORD OR LESS FACTUAL SUMMARY OF MOST SIGNIFICANT INFORMATION. IF DOCUMENT INCLUDES A SIGNIFICANT BIBLIOGRAPHY OR
LITERATURE SURVEY, MENTION IT HERE.)

This book is for the person who needs information about making time and frequency
measurements. It is written at a level that will satisfy those with a casual interest
as well as laboratory engineers and technicians who use time and frequency every day.
It includes a brief discussion of time scales, discusses the roles of the National
Institute of Standards and Technology (NIST) and other national laboratories, and
explains how time and frequency are internationally coordinated. It also describes
the available time and frequency services and how to use them. It discusses the
accuracies that can be achieved with the different services, and the pros and cons of
using various calibration methods.

12. KEY WORDS (6 TO 12 ENTRIES; ALPHABETICAL ORDER; CAPITAUZE ONLY PROPER NAMES; AND SEPARATE KEY WORDS BY SEMICOLONS)

frequency calibration; high frequency; Loran-C; low frequency; oscillators; relative

frequency; satellite broadcasts; standards; time calibration; time code; time interval

13. AVAILABILITY 14. NUMBER OF PRINTED PAGES

X UNLIMITED 160
FOR OFFICIAL DISTRIBUTION. DO NOT RELEASE TO NATIONAL TECHNICAL INFORMATION SERVICE (NTIS).
15. PRICE
ORDER FROM SUPERINTENDENT OF DOCUMENTS, U.S. GOVERNMENT PRINTING OFFICE,
WASHINGTON, DC 20402.
ORDER FROM NATIONAL TECHNICAL INFORMATION SERVICE (NTIS), SPRINGFIELD, VA 22161.

ELECTRONIC FORM
1 YiO J. Technical Publications
Periodical

—
Journal of Research of the National Institute of Standards and Technology Reports NIST research
and development in those disciplines of the physical and engineering sciences in which the Institute
is active. These include physics, chemistry, engineering, mathematics, and computer sciences.

Papers cover a broad range of subjects, with major emphasis on measurement methodology and
the basic technology underlying standardization. Also included from time to time are survey articles
on topics closely related to the Institute's technical and scientific programs. Issued six times a year.

Nonperiodicals

Monographs — Major contributions to the technical literature on various subjects related to the
and technical
Institute's scientific activities.
Handbooks — Recommended codes of engineering and industrial practice (including safety codes) de-
veloped incooperation with interested industries, professional organizations, and regulatory bodies.
Special Publications — Include proceedings of conferences sponsored by NIST, NIST annual reports,
and other special publications appropriate to this grouping such as wall charts, pocket cards, and
bibliographies.
Applied Mathematics Series — Mathematical tables, manuals, and studies of special interest to physi-
cists, engineers, chemists, biologists, mathematicians, computer programmers, and others engaged in
scientific and technical work.
—
National Standard Reference Data Series Provides quantitative data on the physical and chemical
properties of materials, compiled from the world's literature and critically evaluated. Developed un-
der a worldwide program coordinated by NIST under the authority of the National Standard Data
Act (Public Law 90-396). NOTE: The Journal of Physical and Chemical Reference Data (JPCRD)
is published quarterly for NIST by the American Chemical Society (ACS) and the American Insti-

tute of Physics (AIP). Subscriptions, reprints, and supplements are available from ACS, 1155 Six-
teenth St., NW„Washington, DC
20056.
Building Science Series — Disseminates technical information developed at the Institute on building
materials, components, systems, and whole structures. The series presents research results, test
methods, and performance criteria related to the structural and environmental functions and the
durability and safety characteristics of building elements and systems.
—
Technical Notes Studies or reports which are complete in themselves but restrictive in their treat-
ment of a subject. Analogous to monographs but not so comprehensive in scope or definitive in
treatment of the subject area. Often serve as a vehicle for final reports of work performed at NIST
under the sponsorship of other government agencies.
—
Voluntary Product Standards Developed under procedures published by the Department of Com-
merce of the Code of Federal Regulations. The standards establish nationally
in Part 10, Title 15,
recognized requirements for products, and provide all concerned interests with a basis for common
understanding of the characteristics of the products. NIST administers this program as a supplement
to the activities of the private sector standardizing organizations.
Consumer Information Series —
Practical information, based on NIST research and experience, cov-
ering areas of interest to the consumer. Easily understandable language and illustrations provide use-
ful background knowledge for shopping in today's technological marketplace.
Order the above NIST publications from: Superintendent of Documents, Government Printing Office,
Washington, DC 20402.
—
Order the following NIST publications FIPS and NISTIRs—from the National Technical Information
Service, Springfield, VA 22161.
—
Federal Information Processing Standards Publications (FIPS PUB) Publications in this series col-
lectively constitute the Federal Information Processing Standards Register. The Register serves as
the official source of information in the Federal Government regarding standards issued by NIST
pursuant to the Federal Property and Administrative Services Act of 1949 as amended, Public Law
89-306 (79 Stat. 1127), and as implemented by Executive Order 11717 (38 FR 12315, dated May 11.
1973) and Part 6 of Title 15 CFR (Code of Federal Regulations).
NIST Interagency Reports (NISTIR) —A
special series of interim or final reports on work performed
by NIST for outside sponsors (both government and non-government). In general, initial distribu-
tion is handled by the sponsor; public distribution is by the National Technical Information Service,
Springfield, VA 22161, in paper copy or microfiche form.
U.S. Department of Commerce
National Institute of Standards and Technology
(formerly National Bureau of Standards)
Gaithersburg, MD 20899

Official Business
Penalty for Private Use $300