Nistspecialpublication960 12e2

i ii mci \_anui auui id
(2009 edition)
Jeff C Gust
Robert M. Graham
Michael A. Lombardi
Special
[T Publication
Ititut
ftitute of
ind Technology 960-12
m.t of Commerce
NIST Special Publication 960-12
Stopwatch and Timer

Calibrations
(2009 edition)
Jeff C. Gust
Richard J. Bagan, Inc.
Robert M. Graham
Sandia National Laboratories
Michael A. Lombardi
National Institute of Standards and Technology
January 2009
U.S. Department of Commerce

Carlos M. Gutierrez, Secretary

Patrick D. Gallagher, Deputy Director
Certain equipment, instruments or materials are identified in this paper in order to
adequately specify the experimental details. Such identification does not imply recom-
mendation by the National Institute of Standards and Technology nor does it imply the
materials are necessarily the best available for the purpose.
Official contribution of the National Institute of Standards and Technology; not subject
to copyright in the United States.

Special Publication 960-12 Natl. Inst. Stand. Technol.
Spec. Publ. 960-12
80 pages (January 2009)
CODEN: NSPUE2
U.S. GOVERNMENT PRINTING OFFICE
WASHINGTON: 2009
For sale by the Superintendent of Documents
U.S. Government Printing Office
Internet: bookstore.gpo.gov Phone:(202) 512-1800 Fax:(202) 512-2250
Mail: Stop SSOP, Washington, DC 20402-0001
Foreword >
FOREWORD
Stopwatch and timer calibrations are perhaps the most common calibrations performed
in the field of time and frequency metrology. Hundreds of United States laboratories
calibrate many thousands of timing devices annually to meet legal and organizational
metrology requirements. However, prior to the publication of the first edition of this
guide in May 2004, no definitive text had existed on the subject. This NIST Recommended
Practice Guide was created to a fill a gap in the metrology literature. It assists the
working metrologist or calibration technician by describing the types of stopwatches

and timers that require calibration, the specifications and tolerances of these devices, the
methods used to calibrate them, and the estimated measurement uncertainties for each
calibration method. It also discusses the process of establishing measurement traceability
back to national and international standards.
iii
Acknowledgements
ACKNOWLEDGEMENTS
The authors thank the following individuals for their extremely useful suggestions
regarding this new revision of the guide: Georgia Harris and Val Miller of the NIST
Weights and Measures Division, Warren Lewis and Dick Pettit of Sandia National
Laboratories, and Dilip Shah, the chair of the American Society for Quality (ASQ)
Measurement Quality Division.
v
Table of Contents
Contents
Section 1 introduction to Stopwatch and Timer Calibrations 1
1 .A. The Units of Time Interval and Frequency 1
1 .B. A Brief Overview of Calibrations 4

1. C. Traceability and Coordinated Universal Time (UTC) 5
Section 2: Description of Timing Devices that Require Calibration 9
2. A. Stopwatches 9
2.B. Timers 14
2. C. Commercial Timing Devices 15
Section 3: Specifications and Tolerances 17
3. A. Interpreting Manufacturer's Specifications 17

3.B. Tolerances Required for Legal Metrology 22
Section 4: Introduction to Calibration Methods 25
Section 5: The Direct Comparison Method 27
5. A. References for the Direct Comparison Method 27

5.B. Calibration Procedure for DirectComparison Method 31
5. C. Uncertainties of Direct Comparison Method 33
Section 6: The Totalize Method 41
6. A. Calibration Procedure for the Totalize Method 41

6.B. Uncertainties of Totalize Method 43
6. C. Photo Totalize Method 47
Section 7: The Time Base Method 51
7. A. References for the Time Base Method 51

7.B. Calibration Procedure for the Time Base Method 51
7.C. Uncertainties of Time Base Method 56
Section 8: How to Determine if the Calibration Method Meets the Required

Uncertainty 57
Stopwatch and Timer Calibrations
Section 9: Other Topics Related to Measurement Uncertainty 59
9. A. Uncertainty Analysis of Using a Calibrated Stopwatch to

Calibrate another Device 59
9.B. The Effects of Stability and Aging on Calibrations
of 32 768 Hz Crystals 60
9.C. Factors That Can Affect Stopwatch Performance 62
Appendix A: Calibration Certificates 65
Appendix B: References 67
6
1
List of Figures
List of Figures
Figure 1. The calibration and traceability hierarchy 6
Figure 2. Type I digital stopwatch 11
Figure 3. Type II mechanical stopwatch 1
Figure 4. Interior of digital (Type I) stopwatch 12
Figure 5. Inner workings of a mechanical (Type II) stopwatch or timer... 13
Figure 6. A collection of timers 14
Figure 7. Sample manufacturer's specifications for a digital stopwatch

(Example 1) 18
Figure 8. Sample specifications for a digital stopwatch (Example 2) 20
Figure 9. Typical performance of quartz wristwatches using 32 768

Hz time base oscillators 21
Figure 10. Portable shortwave radio receiver for reception of audio

time signals 30
Figure 11. Reaction time measurements (four operators, 10 runs each)

for the direct comparison method 36
Figure 12. Averaging measurement results for four different operators.... 3
Figure 13. Block diagram of the totalize method 41
Figure 14. Using the start-stop button of the stopwatch to start the
counter 42
Figure 15. Measured reaction times (four operators, 10 runs each) for
totalize method 44
Figure 16. Mean reaction times (four operators, 10 runs each) for the
totalize method 45
Figure 17. Photo totalize start reading. .47
ix
:ch and Timer Calibrations
List of Figures, cont'd
Figure 18. Photo totalize stop reading 48
Figure 19. Ambiguous photo totalize reading 49
Figure 20. Time base measurement system for stopwatches and timers. ..52
Figure 2 1 . A time base measurement system for stopwatches and timers

with an integrated graphics display 54
Figure 22. Graph of the frequency stability of two stopwatches 61
Figure 23. Factors that can change the quartz time base frequency 62
Figure 24. Stopwatch accuracy versus temperature 63
Figure Al. Sample calibration certificate, (page 1) 65
Figure Al. Sample calibration certificate, (page 2) 66

List of Tables
List of Tables
Table 1 - Metric prefixes (may be applied to all SI units) 2
Table 2 - Unit values, dimensionless values, and percentages 4
Table 3 - Type I and Type II stopwatch characteristics 9
Table 4 - Legal metrology requirements for field standard stopwatches and

commercial time devices 22
Table 5 - Comparison of calibration methods 25
Table 6 - Traceable audio time signals 27
Table 7 - The contribution of a 300 ms variation in reaction time to the
measurement uncertainty 32
Table 8 - Uncertainty analysis for direct comparison method (digital DUT)

using a land line 38

using a cell phone 39
Table 10 - Uncertainty analysis for direct comparison method (analog DUT)

using a land line 39
Table 1 1 - Uncertainty analysis for totalize method 46
Table 12 - The effect of the length of the measurement time on stability
(based on 25 readings) 53
Table 1 3 - Uncertainty analysis of using a calibrated stopwatch to calibrate

another device 60
XI
Intr lion
Section 1
Introduction to Stopwatch and Timer Calibrations
This document is a recommended practice guide for stopwatch and timer

calibrations. It discusses the types of stopwatches and timers that require
and tolerances, and the methods used to calibrate
calibration, their specifications
them. It measurement uncertainties and the process of establishing
also discusses
measurement traceability back to national and international standards.
This guide is intended to serve as a reference for the metrologist or calibration

technician. It provides a complete technical discussion of stopwatch and timer
calibrations by presenting practical, real world examples of how these calibrations
are performed.
There are nine sections in this guide. Section 1 provides an overview, and serves
as a good introduction if you are new to the field of metrology or to time and
frequency measurements. Section 2 describes the types of timing devices that
require calibration. Section 3 discusses specifications and tolerances. Sections 4
through 7 discuss calibration methods and their associated uncertainties. Section
8 provides additional information to help determine if the selected calibration
method can meet the required level of uncertainty, and Section 9 discusses other
topics related to measurement uncertainty. A sample calibration report and
references are provided in the appendices.
1.A. The Units of Time Interval and Frequency

Stopwatches and timers are instruments used to measure time interval, which is
defined as the elapsed time between two events. One common example of a time
interval is a person's age, which is simply the elapsed time since the person's
birth. Unlike a conventional clock that displays time-of-day as hours, minutes,
and seconds from an absolute epoch or starting point (such as the beginning of the
day or year), a stopwatch or timer simply measures and displays the time interval
from an arbitrary starting point that begins at the instant when the stopwatch is
started.
The standard unit oftime interval is the secon d (s) [1]. Seconds can be accumulated
toform longer time intervals, such as minutes, hours, and days; or they can be
sliced into fractions of a second such as milliseconds (10~ 3 s, abbreviated as
6
ms) or microseconds (10~ s, abbreviated as us). Table 1 lists these and other
i
prefixes that can be used with seconds, as well as the multiples/submultiples and
symbols used to represent them. The second is one of the seven base units in the
International System of Units (SI). Other units (most notably the meter and the
volt) have definitions that depend upon the definition of the second.
Table 1 - Metric prefixes

(may be applied to all SI units)
lVhiltinlps and Siibmiiltinlps Prpfix jy niuvji
1 000 000 000 000= 10 12

tera T
1 000 000 000= 10 9 giga G
1 000 000= 10 6
mega M
1 000= 10 3 kilo k
1 = 10°
0.001 = io-
3
milli m
0.000 001 = io-
6
micro u
0.000 000 001 = io-
9
nano n
0.000 000 000 001 = IO 12
pico P
0.000 000 000 000 001 = IO 15

femto f
The SI defines the second based on a property of the cesium atom, and for this
reason cesium oscillators are regarded as primary standards for both time interval
and frequency. A second is defined as the time interval required for 9 192 63 1 770
transitions between two energy states of the cesium atom to take place. The atomic
definition of the second, together with current technology, allows it to be measured
with much smaller uncertainties than any other SI unit. In fact, the National
Institute of Standards and Technology (NIST) can currently measure a second
15 9
with an uncertainty of less than 1 part in 10 , or more than 1 billion (10 ) times
smaller than the uncertainties required for the calibrations described in this guide!
The resolution of a stopwatch or timer represents the smallest time interval

that the device can display. Resolution is related to the number of digits on the
device's display for a digital stopwatch, or the smallest increment or graduation
on the face of an analog stopwatch. For example, if a stopwatch display shows
two digits to the right of the decimal point, it has a resolution of 0.01 s (10 ms, or
1/100 of a second). This means, for example, that it can display a value of 42.12
Introck : : m
s or 42.13 s. but that it lacks the resolution to display 42.123 s. Resolution of 10
ms is common for digital stopwatches, but some devices have 1 ms resolution
(0.001 s). or even smaller. For analog stopwatches, a common resolution is 1 5
of a second, or 0.2 s.
Although stopwatches and timers measure time interval, they do so by using

el frequency source. Frequency is the rate of a repetitive event, defined as the
number of events or cycles per second. The standard unit of frequency (f) is the
hertz (Hz), which is not a base unit of the SI. but one of the 21 derived SI units.
One hertz equals one event per second, one kilohertz (kHz) equals 103 events
6
per second, one megahertz (MHz) equals 10 events per second, and so on. The
period (I) is the reciprocal of the frequency. T= If. For example, a 1 MHz sine
6
wave would produce 10 cycles per second, or one cycle every microsecond.
A time base oscillator (sometimes called a clock or reference oscillator) produces

the frequency signals used by the stopwatch or timer to measure time intervals. In
today's devices, the time base oscillator is nearly always a quartz crystal oscillator.
However, older devices used either a mechanical oscillator, the alternating
current (AC) line frequency (60 Hz in the United States), or an oscillator based
on a tuned electronic circuit as their frequency source. The time base oscillator
serves as the reference for all of the time and frequency functions performed by
the device. The most common frequency used by quartz time base oscillators is
15
32 768 (= 2 ) Hz. In this case, when the stopwatch or timer has counted 32 768
oscillations from its time base oscillator, it then records that 1 s has elapsed. If
you want to think of this time base as a clock, it "ticks" 32 768 times per second,
or once every 30.518 us.
Throughout this guide, time interval is always presented in units of seconds (or
fractions of a second), and frequency is presented in units of hertz (or multiples
of a hertz). However, measurement uncertainties are presented as dimensionless
values that represent a fractional percentage error. Since these dimensionless
values are usually very small percentages, they are often expressed in scientific
notation. For example, if a stopwatch has a measurement uncertainly" of 1 s over a
time interval of 10 000 s. the uncertainty is expressed as either a percentage (0.01
%) or as a dimensionless value ( 1 x 10"). Table 2 provides more examples.
3
Table 2 - Unit values, dimensionless values, and percentages.
Time Length Dimensionless Dimensionless Percentage

un- of test uncertainty uncertainty uncertainty
certainty (literal) (scientific (%)
notation)
1 s 1 minute 1 part per 60 1.67 x 10- 1.67
1 s 1 hour 1 part per 3600 2.78 x 10-4 0.027 8
1 s 1 day 1 part per 86 400 1.16 x lO" 5 0.001 16
x lO" 2
1 s 100 s 1 part per hundred 1 1
x lO 3
1 s 1000 s 1 part per thousand 1 0.1
1 s 10 000 s 1 part per 1 thousand 1 x 10-4 0.01
x lO 5
1 s 100 000 s 1 part per 100 thousand 1 0.001
1 ms 100 s 1 part per 100 thousand 1 x io- 5 0.001
1 ms 1000 s 1 part per million (1 ppm) 1 x IO' 6 0.000 1
1 ms 10 000 s 1 part per 10 million 1 x 10" 0.000 01
1 ms 100 000 s 1 part per 100 million 1 x IO' 8 0.000 001
1.B. A Brief Overview of Calibrations

Like all calibrations, stopwatch and timer calibrations are simply comparisons
between the device under test (DUT) and a measurement reference, or standard.
When a stopwatch or timer is calibrated, either a time interval standard or a
frequency standard is used as the measurement reference. If a time interval
standard is used, it is compared to the DUT's display. If a frequency standard is
used, it is compared to the DUT's time base oscillator. Both types of calibrations
are described in detail later in this guide.
Most of the calibrations described in this guide are laboratory calibrations,

as opposed to field calibrations. To understand what this means, consider an
example where a stopwatch is calibrated in the laboratory against a standard,
and a calibration certificate and/or sticker are issued to the customer. That
same stopwatch or timer can then be thought of as a field standard, a working
standard, or a transfer standard, and used as the measurement reference during
a field calibration. In other words, it can be brought outside the laboratory and
used to calibrate another timing device, such as a parking meter. The same basic
4
Intr. m
principles that apply to laboratory calibrations apply to field calibrations,
although laboratory calibrations generally take longer and are made much
more carefully because the required measurement uncertainties are smaller.
Devices that are field calibrated are generally not used as a measurement
reference for performing other calibrations. Instead, they are working
instruments used for scientific, business, or legal purposes. Therefore, their
calibration can be thought of as a periodic test or inspection that ensures
that these devices are working properly and meeting their specifications.
Common sense dictates that the measurement reference for any calibration
(either the laboratory reference or the field standard) must always
outperform the devices it needs to test. A parking meter, for example,
might have an acceptable uncertainty of 1 % when timing a 5 minute
interval (±3 s). A field standard stopwatch used to test a parking meter
should be certified to an uncertainty small enough so that it contributes no
significant uncertainty to the parking meter calibration. In other words, we
need to be able to trust our reference so that we can trust our measurement
of the DUT.
When a laboratory calibration is completed, the metrologist has determined

the offset
1
of the DUT with respect to the reference. This offset can be

stated as a percentage or in units of time interval or frequency (or both)
on the calibration certificate, and should be quantified with a statement
of measurement uncertainty. Field calibrations are generally "pass/fail"
calibrations. This means that the device is tested to see whether it meets
its intended or legal metrology requirements, and it either passes or fails.
If it fails, it is removed from service until it can be adjusted, repaired, or
replaced.
1.C. Traceability and Coordinated Universal Time (UTC)

As previously discussed, when a device is calibrated by comparing it to a
measurement reference, the reference must be more accurate (have lower
compared to a more accurate standard. That more accurate standard
1 The term 'offset' is commonly used in the discipline of frequency

and time measurement and will be used throughout this text. In terms of
the ISO Guide to the Expression of Uncertainty in Measurement [2], the
offset would be considered the measurement result.
5
needs to be periodically compared to an even more accurate standard, and so on,
until eventually a comparison is made against a national or international standard
that represents the best physical realization of the SI unit that is being measured (in
this case, the SI second). This measurement traceability hierarchy is sometimes

illustrated with a pyramid as shown in Figure 1 . The series of comparisons back
to the SI unit is called the traceability chain. Metrological traceability is defined,
by international agreement, as:
The property ofa measurement result whereby the result

can be related to a reference through a documented
unbroken chain of calibrations, each contributing to
the measurement uncertainty. [3]
The definition of metrological traceability implies that unless the measured value
is accompanied by a stated measurement uncertainty, the traceability chain is
broken. It is the responsibility of the calibration laboratory to determine and

report the uncertainty of its measurements to its customers so that metrological
traceability is maintained [4].
Typical Calibration Hierarchy
Figure 1. The calibration and traceability hierarchy.
The International Bureau of Weights and Measures (BIPM) located near Paris,
France, is responsible for ensuring the worldwide uniformity of measurements
Introduction
and their traceability to the SI. The BIPM collects and averages time interval and
frequency data from more than 60 laboratories around the world and creates a
time scale called Coordinated Universal Time (UTC) that realizes the SI second
as closely as possible. Thus, UTC serves as the international standard for both
time interval and frequency. However, the BIPM does not produce a physical
representation of the second; it simply calculates a weighted average that is
published weeks after the actual measurements were made. This document, known
as the BIPM shows the time offset of each contributing laboratory
Circular T,
with associated uncertainties, and can be downloaded from the BIPM web site
(www.bipm.org). The laboratories that provide data to the BIPM maintain the
oscillators and clocks that produce the actual signals that are used as measurement
references. Most of these laboratories are national metrology institutes (NMIs)
that serve as the caretakers of the national measurement references for their
respective countries. Thus, to establish traceability to the SI for time interval and
frequency calibrations, the traceability chain for a measurement must link back
to signals that originate from an NMI or a national timekeeping laboratory. The
NMI that is chosen as a reference must contribute to the derivation of UTC by
sending its measurement data (with associated uncertainties) to the BIPM.
NIST is the ultimate reference point for most measurements made in the United
States, and as such, submits time and frequency data to the BIPM. NIST provides
it own real-time representation of UTC, designated UTC(NIST), that is distributed
to the public using a variety of radio, telephone, and Internet signals. These
signals are described in more detail in Section 5, and can serve as references for
measurements that are traceable back to the SI.
The traceability chain is easy to visualize if you think of it as a series of calibrations.

Every link in the chain is a calibration; comparing a device to a higher standard,
until eventually a comparison is made to the SI unit. Every calibration has some
measurement uncertainty. At the top of the chain, the measurement uncertainties

are so tiny they are insignificant to those of us who calibrate stopwatches and
timers. For example, the difference between the best possible estimate of
16
the SI second and UTC(NIST) is measured in parts in 10 . This represents a
time offset of less than 0.1 ns (10~ 10
s) over the course of a day. As we move
down the traceability chain to the actual calibration of a stopwatch or timer, the
|
uncertainties become larger and larger. For example, if the calibration laboratory
uses an audio signal from NIST radio station WWV (Section 5) to calibrate its
standard, the uncertainty of the received tones might be 1 ms. This uncertainty
7
is still small enough for the WWV tones to be used as a stopwatch calibration
reference, because the uncertainty introduced by an operator starting and stopping
the watch (human reaction time) is much larger, typically tens or even hundreds
of milliseconds. In summary, as long as each link of the chain and its uncertainty
are known, traceability to the SI can be established.
Stopwatch and timer calibrations are among the least demanding of all time
and frequency measurements. Relatively speaking, the instruments requiring
calibration are low cost, and the acceptable measurement uncertainties can be
quite large. Even so, for legal, technical, and practical reasons it is very important
to establish traceability to the SI for these calibrations. If a valid traceability
chain to the SI is established, it ensures that the working device was properly
calibrated, and, if correctly used, will produce valid results.
8
Description of Tin
Section 2
Description of Timing Devices that Require Calibration
This section describes the various types of stopwatches (Section 2. A. Stopwatches)

and timers (Section 2.B. Timers) that are calibrated in the laboratory. These
types of stopwatches and timers are often used as transfer standards to perform
field calibrations of the commercial timing devices described in Section 2.C.
2.A. Stopwatches
Stopwatches can be classified into two categories, Type I and Type II [5]. Type
I stopwatches have a digital design employing quartz oscillators and electronic
circuitry to measure time intervals (Figure 2). Type II stopwatches have an
analog design and use mechanical mechanisms to measure time intervals (Figure
3). The key elements of Type I and Type II stopwatches are summarized in
Table 3.
Table 3 - Type I and Type II stopwatch characteristics.
Type I Stopwatch Type II Stopwatch

Description
(digital) (analog)
Time measured by division Time measured by

Operating principle
of time base oscillator mechanical movement
• Mechanical
mainspring
Time base • Quartz oscillator • Synchronous
motor, electrically
driven
• Corrosion resistant metal

Case
• Impact resistant plastic
• Protects display
• Protects dial/hands
• Allows for proper
• Allows for proper
viewing
Crystal viewing
• May be tinted
• Must be clear and
• May employ
untinted
magnification
• 48 hours without
• 3 hours without
Minimum time interval replacement of
rewinding
battery
9
Table 3 - Continued
Type II Stopwatch Type II Stopwatch

Description
(analog) (analog)
• Single control to start/stop
Start and stop

• Audible signal of
start/stop
• Must reset stopwatc i to zero

Reset
• Must indicate
whether display
Split time (if equipped)
mode is regular or
split time.
• Must not exceed 1 .8 N (0.40 lbf)

Force to operate controls
• Face must be white

• Graduations must
Dial and hands be black or red
• Hands must be
black or red
• Unique, nondetachable serial number

• name or trademark
Manufacturer's
Required markings
• Model number (Type I only)
• Provide delimiting
character for
Digital display hours, minutes,
seconds (usually
colon)
Minimum increment • <0.2s
Minimum elapsed time

• 1 hour • 30 minutes
at rollover
• Stopwatch meets to erance regardless of

Physical orientation
physical orientation
10
Description ofTiming Devices
Figure 2. Type I digital stopwatch. Figure 3. Type II mechanical

stopwatch.
2.A.I. Basic Theory of Operation
Every stopwatch is composed of four elements: a power source, a time base,
a counter, and an indicator or display. The design and construction of each

component depends upon the type of stopwatch.
2.A.l.a. Digital (Type

Stopwatches
I)
The power source of a Type I stopwatch is usually a silver-oxide or alkaline

battery that powers the oscillator and the counting and display circuitry. The
time base is a quartz crystal oscillator that usually has a nominal frequency of
32 768 Hz, the same frequency used by nearly all quartz wristwatches. The
32 768 Hz frequency was originally chosen because it can be converted to a 1
pulse per second signal using a simple 15 stage divide-by-two circuit. It also
has the benefit of consuming less battery power than higher frequency crystals.
Figure 4 shows the inside of a typical digital device, with the printed circuit
board, quartz crystal oscillator, and battery visible. The counter circuit consists
of digital dividers that count the time base oscillations for the period that is
initiated by the start/stop buttons [6, 7]. The display typically has seven or eight
digits.
11
Figure 4. Interior of digital (Type I) stopwatch.
2.A.l.b. Mechanical (Type II) Stopwatches

In a traditional mechanical stopwatch, thepower source is a helical coil spring,
which stores energy from the winding of the spring. The time base is usually
a balance wheel that functions as a torsion pendulum. The rate at which the
spring unwinds is governed by the balance wheel, which is designed to provide a
consistent period of oscillation relatively independent of factors such as friction,
operating temperature, and orientation. In most mechanical stopwatches, the
balance wheel is designed to oscillate at 2.5 periods per second, which produces
5 "ticks" or beats per second. The balance wheel is connected to an escapement
that meters the unwinding of the coil spring and provides impulses that keep the
balance wheel moving. It is this metered unwinding of the coil spring that drives
the counter indicator. In this type of device, the counter is composed of a gear
train that divides the speed of rotation of the escapement wheel to the appropriate
revolution speed for the second, minute, and hour hands. The time interval from
the counter is displayed either on a face across which the second and minutes
hands sweep, or on a series of numbered drums or discs that indicate the elapsed
time (Figure 5) [6].
12
Description ofTiming Devices
Figure 5. Inner workings of a mechanical (Type II) stopwatch or timer.
13
Another form of the Type II stopwatch uses a timer driven by a synchronous
motor that also drives the hands or numbered wheels. For this device, the power
source is the 60 Hz AC line voltage. The power source drives an electric motor
within the timing device. The time base is derived from the controlled regulation
of the 60 Hz frequency of AC electric power as supplied by the power utility
company. The frequency AC power in the United States are

limits for distributed
59.98 Hz to 60.02 Hz, or 60 Hz ± 0.033 % of the nominal value. However, the
actual frequency is controlled much more accurately than this, in order to advance
or retard the grid frequency and synchronize the power distribution system [8].
The counter and display circuitry are similar to those used in the mechanical
stopwatches previously discussed.
2.B. Timers
Timers, unlike stopwatches, count down from a preset time period instead of
counting up from zero. They can be small, battery-operated devices that are used
to signal when a certain time period has elapsed, or they can be larger devices
that plug into a wall outlet and control other items (Figure 6). A parking meter
is an example of a countdown timer. Inserting a coin starts the internal timer
counting down from an initial preset point. When the preset time has elapsed,
the EXPIRED flag is raised.
Figure 6. A collection of timers.
14
=
Description of Tir . ; : is
One type of timer used extensively in industry is the process control timer. As
their name implies, process control timers measure or control the duration of a
specific process. For example, when a product is made, it may need to be heat
treated for a specific length of time. In an automated manufacturing system, the
process control timer determines the length of time that the item is heated. In
some applications, such as integrated circuit manufacturing, the timing process
can be critical for proper operation.
Process control timers are also used in many different types of laboratory
environments. Calibration laboratories use timers to calibrate devices such as
radiation detectors, by regulating the amount of time that the device is exposed to
the radiation source. The uncertainty in the time of exposure directly influences
the overall measurement uncertainty assigned to the detector calibration.
Timers are also used in the medical field. For example, medical laboratories use
process control timers when specimen cultures are grown. Hospitals use timers to
regulate the amount of medication given to patients intravenously.
2.C. Commercial Timing Devices

Many types of timing devices are used every day in commercial applications.
Parking meters, automatic car wash facilities, taxicab meters, and commercial
parking lots are examples of entities that either charge a certain amount for a
specified period, or provide a certain period of service for a specified amount of
money.
The calibration requirements and allowable tolerances for these devices are
usually determined on a state-by-state basis by state law, or locally by city or
municipality ordinances. The allowable uncertainties are often 1 % or larger.
Generic guidance is provided in Section 3.
15
Specifications an ices
Section 3
Specifications and Tolerances
Whether we are developing a calibration procedure, or performing an uncertainty

analysis for a given calibration process, we need to be able to understand and
interpret the specifications and tolerances for both the DUT and the test equipment
associated with the calibration. This section reviews both manufacturer's
specifications and legal metrology requirements for stopwatches and timing
devices.
3.A. Interpreting Manufacturer's Specifications

When reviewing manufacturer's specification sheets, it quickly becomes obvious
that not all instrument manufacturers specify their products in the same way. This
section defines and describes the most common types of specifications quoted for
stopwatches and timing devices.
3.A.I. Absolute Accuracy Specifications

The absolute accuracy 2 of an instrument is the maximum allowable offset from
nominal. Absolute accuracy is defined in either the same units, or a fractional
unit quantity of themeasurement function for an instrument. For example, the

absolute accuracy of a ruler might be specified as ±1 mm for a scale from to 15
cm.
In the case of timing devices, it isn't useful to provide an absolute accuracy

specification by itself. This is because a device's time offset from nominal will
increase as a function of the time interval. If the timing device were able to
measure an infinite time interval, the offset (or difference in time from nominal)
of the device would also become infinitely large. Because of this, when timing
devices are specified with an absolute accuracy number, it is also accompanied by
a time interval for which this specification is valid. As an example, the absolute
accuracy specification for the stopwatch shown in Figure 7 is 5 s per day.
2 In this section, the term "accuracy" is used in order to allow the reader to correlate the concepts
of this chapter directly with published manufacturer specifications. In terms of the ISO Guide
to theExpression of Uncertainty in Measurement [2], the quantities associated with accuracy are
understood to be uncertainties.
17
Handsome stopwatch with large

display provides timing to
1 /1 00th of a second over a range
of 9 hours 59 minutes and 59.99
seconds.
Accurate to ± 5 s/day
Built-in memory recalls up to ten
laps.
Clock function (1 2 or 24 hrs)

features a programmable alarm
with an hourly chime plus built-
in calendar displays day, month
and date.
Countdown timer function

features input ranges from one
minute to 9 hours, 59 minutes.
Dimensions/Weight: 2.5 x 3.2 x.8:
(63x81 x 20 mm): 2.8 oz.
Water resistant housing is com

plete with lithium battery
Figure 7. Sample manufacturer s specifications for a digital stopwatch

(Example 1).
If the stopwatch in Figure 7 were used to measure a time interval longer than one
day, we could determine a new absolute accuracy figure by simply multiplying
the original specification by the longer time interval. For example 5 s per day
becomes 10 s per two days, 35 s per week, and so on.
While it is usually acceptable to multiply the absolute accuracy specification by

longer time intervals than the period listed in the specifications, caution must
18
Specifications and Tolerances
be used when dividing the absolute accuracy specification for shorter time
intervals than the period listed in the specifications. This is because for shorter
measurements periods, a new source of uncertainty, the resolution uncertainty of
the instrument, becomes an important consideration. For example, the absolute
accuracy of the example stopwatch (Figure 7) during a time interval of 30 s is
determined as
day
s
x 30 s = — x—
day
^-day = 0.0017s
2880
We can see from the specifications in Figure 7 that the stopwatch has a resolution
of 1/100 of a second, or 0.01 s. In this case, computing the absolute accuracy
specification for a 30 s interval results in a number (0.00 1 7 s) that is about six times
smaller than the smallest value the stopwatch can display. Most manufacturers of
timing devices do not consider the resolution of the product in their specifications,
but we will include resolution uncertainty in our examples.
3.A.2 Relative Accuracy Specifications

While absolute accuracy specifications are helpful, sometimes it is more
desirable to specify accuracy relative to the measured time interval. This makes
its significance easier to understand. For this purpose, we define a quantity called
relative accuracy:
„ . .
A Absolute Accuracy
Relative Accuracy
Measured Time Interval
Using the previous example from Figure 7, the stopwatch has an absolute accuracy
specification of 5 s per day, so the relative accuracy is:
Relative Accuracy =
lday
= ——— = 0.000 058 = 0.0058 % = 5.8 x 10" 5
.
86 400 s
Note that because the absolute accuracy specification and the measured time
interval are both expressed in seconds, the unit cancels out; leaving a dimensionless
number that can be expressed either as a percentage or in scientific notation.
Relative accuracy specifications can also be converted back to absolute time units
if necessary. For example, Figure 8 shows the manufacturer's specifications for
19
a stopwatch that is accurate to 0.0003 % (although not stated, it is assumed that

this percentage has been stated as a percent of reading or relative accuracy).
To compute the time accuracy for a 24 h measurement, we simply multiply the
relative accuracy by the measurement period:
0.0003 % x 24 h = 0.000 072 h = 0.2592 s
This computation shows that this stopwatch is capable of measuring a 24-hour

interval with an accuracy of about 0.26 s. However, it is again important to
note that when measuring small time intervals, the resolution uncertainty of the
stopwatch must be considered. For example, if the stopwatch in Figure 8 is used
to measure a time interval of 5 s, the computed accuracy is much smaller than the
resolution of the stopwatch:
0.0003 %x5s = 0.000 015 s
Timer: 9 hours, 59 minutes, 59 seconds,

99 hundredths.
Stopwatch: Single-action timing: time-

in/time-out; continuous timing; cumula-
tive split, interval split and eight memo-
ries. Triple display shows cumulative
splits, interval splits and running time
simultaneously.
Features: Captures and stores up to

eight separate times. After timed event is
complete, stopwatch displays informa-

tion in its memory. Counter box show
number of split times taken. Solid-state
design with an accuracy of 0.0003 %.
Durable, water-resistant construction
makes stopwatch suitable for field use
(operates in temperature from 1 ° to 59° C
[33° to 1 38° F]). With triple line LDC: top
two lines are each 1/8 in. high (3.2 mm);
third line is Vfc in. high (6 mm).
Figure 8. Sample specifications for a digital stopwatch (Example 2).
20
Specifications and Tolerances <
3.A.3. Typical Performance

During the NIST centennial celebration of 200 1 an exhibit at the NIST laboratories
,
in Boulder, Colorado, allowed visitors to measure the time base accuracy of their
quartz wristwatches. Over 300 wristwatches were tested. Nearly all quartz
wristwatches contain a 32 768 Hz time base oscillator, the same technology
employed by a Type I digital stopwatch. The results of these measurements,
showing the loss or gain in seconds per day for the watches, are summarized in
Figure 9, and give some idea of the typical performance of a quartz stopwatch
or timer. Roughly 70 % of the watches were able to keep time to within 1 s per
day or better, a relative accuracy of approximately 0.001 % (1 x 10~ 5 ). About 12
% had a relative accuracy larger than 5 s per day, or larger than 0.005 %. It is
interesting to note that nearly all of the watches in this study gained time, rather
than lost time; and were presumably designed that way to help prevent people
from being late. This characteristic will not necessarily apply to stopwatches and
timers.
Quartz Wristwatch Performance
40
35
3D
25-
O 20-"'
15
0* 10
d> 5
EL
^ L
«5 -1 *0,5 03 1 2 5 More
Seconds Per Day

Figure 9. Typical performance of quartz wristwatches using 32 768 Hz time
base oscillators.
21
3.B. Tolerances Required for Legal Metrology
General specifications for field standard stopwatches and commercial timing
devices are provided in Section 5 5 5 of NIST Handbook 44
.
[9] , and are summarized
in Table 4. NIST Handbook 44 is recognized by nearly all 50 states as the legal
basis for regulating commercial weighing and measuring devices. However,

some state governments and some regulatory agencies have other specifications
that need to be satisfied for a given calibration. Therefore, be sure to check and
understand the required tolerances and regulations for the types of calibration
that a calibration laboratory is asked to perform [10].
Table 4 - Legal metrology requirements for field standard stopwatches and

commercial time devices.
Commercial Overregistration Underregistration
Interval
timing
Measured
device Requirement Uncertainty Requirement Uncertainty
1 s per
30 minutes minute, but
None NA
or less not less than 2
minutes
Over 30
5 minutes
minutes
plus 4 s per
Parking up to and None NA 11.7% to
minute over
meter including 1 16.7%
30 minutes
hour
7 minutes
plus 2
Over 1
None NA minutes per
hour
hour over 1
hour
3 s per hour,
Time clocks 3 s per hour, not to exceed
and time not to exceed 1
0.07 % 1 minute per
0.07 % to
recorders minute per day
to 0.08 % day
0.08 %
9 s per minute
on the initial
interval, and 10% to

Taximeters 3 s per minute 5 % 6 s per minute 15%
on subsequent
intervals
5 s for any
Other timing 6 s per minute
interval of 1 NA 10%
devices
minute or more
22
Specifications an< .
trances <
The terms overregistration and underregistration are used when specifying the
accuracy of commercial timing devices. The terms are used to describe conditions
where the measurement device does not display the actual quantity. In timing
devices, underregistration is the greatest concern, because an underregistration
error occurs when the timing device indicates that the selected time interval has
elapsed before itactually has. An example of underregistration would be paying
for 10 minutes on a parking meter, and then having the meter indicate that time
had expired when only 9 minutes and 45 s had actually elapsed.
NIST Handbook 44 [9] specifies that instruments that are required to calibrate
timing devices must be accurate to within 15s per 24 h period (approximately
0.017 %). If stopwatches are used as the calibration standard, this becomes
theminimum allowable tolerance for the stopwatch. Another reference, NIST
Handbook 105-5 [5] states that the tolerance for instruments used to calibrate
timing devices must be three times smaller than the smallest tolerance of the device
being calibrated. Handbook 105-5 also provides a nearly identical specification
for stopwatches, stating that the tolerance for stopwatches is ±0.02 % of the time
interval tested (approximately 2 s in 3 hours), rounded to the nearest 0.1 s.
The uncertainties listed above were meant to be achievable with Type II
(mechanical) devices, but Type I devices are normally capable of much lower
uncertainties. As a result, organizations and jurisdictions that rely exclusively
on digital stopwatches (Type 1) might require that devices be calibrated to a
tolerance of 0.01 %, or even 0.005 %. For example, the State of Pennsylvania
code [11] uses the same specifications as NIST Handbook 44 for mechanical
stopwatches (15 s per 24 hours), but states that a quartz stopwatch shall comply
with the following more rigorous standards:
(i) The common crystal frequency shall be 32 768 Hz

with a measured frequency within plus or minus 3 Hz,
or approximately 0. 01 % of the standard frequency.
(ii) The stopwatch shall be accurate to the equivalent
ofplus or minus 9 seconds per 24-hour period.

I
23
atch and Timer Calibrations
3.B.I. Time Clocks and Time Recorders

The specification for both overregistration and underregistration is 3 s per hour,
not to exceed 1 minute per day.
3.B.2. Parking Meters

The specifications for parking meters have no tolerance for overregistration.
Parking meters with a time capacity of 30 minutes or less are specified to have a
maximum underregistration error of 1 s per minute, but not to exceed 2 minutes
over the 30 minute period. For parking meters with a capacity of greater than 30
minutes, but less than 1 hour, the tolerance for underregistration is 5 minutes, plus
4 s per minute for every minute between 30 minutes and 60 minutes. Parking
meters that indicate over 1 hour have an underregistration tolerance of 7 minutes
plus 2 minutes per hour for time intervals greater than 1 hour.
3.B.3. Other Timing Devices

All other timing devices have an overregistration tolerance of 5 s for any time
interval of 1 minute or more and an underregistration tolerance of 6 s per indicated
minute. If the instrument is a digital indicating device, the tolerance is expanded
by one half of the least significant digit.
24
Met
Section 4
Introduction to Calibration Methods
There are three generally accepted methods for calibrating a stopwatch or timer:
the direct comparison method, the totalize method, and the time base method.
The first two methods consist of time interval measurements that compare the
time interval display of the DUT to a traceable time interval reference. In the
case of the direct comparison method, the time interval reference is normally a
signal broadcast by an NMI, usually in the form of audio tones. In the case of the
totalize method, the time interval reference is generated in the laboratory using
a synthesized signal generator, a universal counter, and a traceable frequency
standard. The third method, the time base method, is a frequency measurement. It
compares the frequency of the DUT's time base oscillator to a traceable frequency
standard [12]. The properties of the three methods are briefly summarized in
Table 5, and the following three sections are devoted to the three methods. Each
section explains how to perform a calibration using each method, and how to
estimate the measurement uncertainties.
Table 5 - Comparison of calibration methods.
^^^^ ^ftlethod
— Direct comparison
Time base
Properties — Totalize
measurement
Equipment Requirements Best Better Better
Speed Good Better Best
Uncertainty Good Good Best
Applicability Good Best Better
The methods used to estimate the uncertainty of measurement are described in

the ISO Guide to the Expression of Uncertainty in Measurement (GUM) [2].
This guide does not attempt to summarize the GUM, but does strive to produce
estimates of uncertainty that are consistent with the GUM. The resulting expanded
uncertainty of measurement is presented with a coverage factor that represents an
approximate 95 % level of confidence.
25
The Direct Comparison Method
Section 5
The direct comparison method is the most common method used to calibrate
stopwatches and timers. It requires a minimal amount of equipment, but has
larger measurement uncertainties than the other methods. This section describes
the references used for this type of calibration and the calibration procedure.
5.A. References for the Direct Comparison Method

The direct comparison method requires a traceable time-interval reference. This
reference is usually an audio time signal, but in some cases a traceable time display
can be used. The audio time signals are usually obtained with a shortwave radio
or a telephone. Since time interval (and not absolute time) is being measured,
the fixed signal delay from the source to the user is not important as long as
it remains relatively constant during the calibration process. A list of traceable
audio time sources is provided in Table 6.
Table 6 - Traceable audio time signals.
National Radio
Metrology T A. Telephone call Broadcast
t j • . , Location ,
Institute (NMI) numbers letters frequencies
National Institute _ _ 1t .
*-cu
of
a^
and
a
u
a
Standards
Technology 1
Fort Collins,
Colorado,
'
m^ /lftft7111
(303)499-7111
,
* WWV
_ in
2.5,5,10,
1C OAAyrTT
15, 20 MHz
TT . A ,
United States
(NIST)
National Institute
j^ aua j
of Standards
JrT, Hawaii,
.'.
/oao^k /nn*
(808) 335-4363 *
um VU
WWVH r 2.5,5,10,
c \l TT
and Technology
, ,
TT . , '
v 7 *
15 MHz
United States
(NIST)
United , .
Naval
States
^r^T*
Umted <202 ) 762 " 1401 *
*
Observatory (202) 762- 1 069
MatCS
(USNO)
27
United Colorado
States Naval Springs, *
(719)567-6742
Observatory Colorado.
(USNO) United States
• (613) 745-1576
National Ottawa,
(English language) 3.33.7.850.
Research Council Ontario,
(613) 745-9426(§, 14.67 MHz
(NRC) Canada
(French language)
(442) 215-39-02 *
(442)211-05-06 !
(442) 21 1-05-07 #
(442) 21 1-05-08 ##
Time
announcements
Centro Nacional XEQK
Queretaro. are in Spanish, a
de Metrologia (Mexico 1.35 MHz
Mexico country code must
(CENAM) City)
be dialed to access
these numbers
from the United
States, see www.
cenam.mx for more
information.
Taedok,
Korea Research
Science
Institute of
Town, HLA 5 MHz
Standards and
Republic of
Science (KRISS)
Korea
National Time Lintong,

2.5, 5, 10.
Service Center Shaanxi, BPM
15 MHz
(NTSC) China
* Coordinated Universal Time (UTC)

@ Eastern Time
! Central Time
# Mountain Time
## Pacific Time
28
The Direct Com 3 3 son
Please note that local "time and temperature" telephone services are not traceable
references and should not be used. For traceable calibrations, use only sources
that originate from a national metrology institute, such as those listed in Table
6. The following sections briefly describe the various radio and telephone time
signals, and provide information about the types of clock displays that can and
cannot be used.
5.A.I. Audio Time Signals Obtained by Radio

The radio signals listed in Table 6 include a voice announcement of UTC and
audio ticks that indicated individual seconds. WWV, the most widely used station,
features a voice announcement of UTC occurring about 7.5 s before the start of
each minute. The beginning of the minute is indicated by a 1500 Hz tone that
lasts for 800 ms. Each second is indicated by a 1 kHz tone that lasts for 5 ms.
The best way to use these broadcasts is to start and stop the stopwatch when the
beginning of the minute tone is heard.
Most of the stations listed in Table 6 are in the high-frequency (HF) radio
band (3 MHz to 30 MHz), and therefore require a shortwave radio receiver. A
typical general-purpose shortwave receiver provides continuous coverage of the
spectrum from about 150 kHz, which is below the commercial AM broadcast
band, to 30 MHz. These receivers allow the reception of the HF time stations
on all available frequencies. The best shortwave receivers are designed to work
with large outdoor antennas, with quarter-wavelength or half-wavelength dipole
antennas often providing the best results. However, in the United States, adequate
reception of at least one station can usually be obtained with a portable receiver
with a whip antenna, such as the one shown in Figure 10. This type of receiver
typically costs a few hundred dollars or less.
HF radio time stations normally broadcast on multiple frequencies because

some of the frequencies are not available at all times. In many cases, only one
frequency can be received, so the receiver might have to be tuned to several
different frequencies before finding a usable signal. In the case of WWV, 10
MHz and 15 MHz are probably the best choices for daytime reception, unless the
laboratory is within 1000 km of the Fort Collins, Colorado station, in which case
2.5 MHz might also suffice. Unless the receiver is near the station, the 5 MHz
signal will probably be the easiest to receive at night [13].
! 1 Timer Calibrations
Figure 10. Portable shortwave radio receiver for reception of audio time
signals.
5.A.2. Audio Time Signals Obtained by Telephone

The telephone time signals for NIST radio stations and are WWV WWVH
simulcasts of the radio broadcasts, and the time (UTC) is announced once per
minute. The length of the phone call is typically limited to 3 minutes. The
format of the other broadcasts varies. The USNO phone numbers broadcast UTC
at 5 s or 10 s intervals. The NRC phone number broadcasts Eastern Time at 10
s intervals, and CENAM offers separate phone numbers for UTC and the local
time zones of Mexico.
5.A.3. Time Displays

It might be tempting to use a time display from a radio controlled clock or from a
web site synchronized to UTC as a reference for stopwatch or timer calibrations.
As a general rule, however, these displays are not acceptable for establishing
traceability. Nearly all clock displays are synchronized only periodically, and
between synchronizations rely on a free running local oscillator whose frequency
uncertainty is usually unknown. And of course, an unknown uncertainty during
any comparison breaks the traceability chain. For example, a low cost radio
controlled clock that receives a 60 kHz signal from NIST radio station WWVB
is usually synchronized only once per day. In between synchronizations, each
"tick" of the clock originates from a local quartz oscillator whose uncertainty
is unknown, and that probably is of similar or lesser quality than the oscillator
inside the device under test. The NIST web clock at (time.gov) presents similar
problems. It synchronizes to UTC(NIST) every 10 minutes if the web browser is
left open. However, between synchronizations it keeps time using the computer's
clock, which is usually of poorer quality than a typical stopwatch, and whose
30
The Direct Comp< i ' n M
uncertainty is generally not known. In contrast, each "tick" of an audio broadcast
from WWV originates from NIST and is synchronized to UTC. Therefore, WWV
audio always keeps the traceability chain intact.
There are a few instances where a time display can be used to establish traceability.
One example would be a display updated each second by pulses from a Global
Positioning System (GPS) satellite receiver. In this case, if the traceable input
signal were not available, the display would stop updating. Therefore, if the
display is updating, then it is clear that each "tick" is originating from a traceable
source. However, nearly all GPS receivers have the capability to "coast" and
keep updating their display even when no satellite signals are being received. In
order for a GPS
display to be used as a reference, there must be an indicator on
the unit that shows whether the display is currently locked to the GPS signal, or
is in "coast" mode. If the receiver is in "coast" mode, it should not be used as a
calibration reference.
Another example of a traceable time display would be a digital time signal obtained
from a telephone line, such as signals from the NIST Automated Computer Time
Service (ACTS), which is available by dialing (303) 494-4774 [13]. With an
analog modem and simple terminal software (configured for 9600 baud, 8 data
bits, 1 stop bit, and no parity), you can view time codes on a computer screen,
and use these codes as a reference in the same way that you would use the audio
time announcements from WWV.
However, the length of a single telephone call
is limited to 48 s. In theory, Internet time codes could be used in the same way,
but the transmission delays through the network can vary by many milliseconds
from second to second. For this reason, the currently available Internet signals
should not be used as measurement references.
5.B. Calibration Procedure for Direct Comparison Method

Near the top of the hour, dial the phone number (or listen to the radio broadcast)
of a traceable source of precise time. Start the stopwatch at the signal denoting
the hour, and write down the exact time. After a suitable time period (depending
on the accuracy of the stopwatch), listen to the time signal again, and stop the
stopwatch at the sound of the tone, and write down the exact stopping time.
Subtract the start time from the stop time to get the time interval, and compare
this time interval to the time interval displayed by the stopwatch. The two time
intervals must agree to within the uncertainty specifications of the stopwatch
for a successful calibration. Otherwise, the stopwatch needs to be adjusted or
rejected.
31
5.B.I. Advantages of the Direct Comparison Method
This method is relatively easy to perform and, if a telephone is used, does not
require any test equipment or standards. It can be used to calibrate all types of
stopwatches and many types of timers, both electronic and mechanical.
5.B.2. Disadvantages of the Direct Comparison Method

The operator's start/stop reaction time is a significant part of the total uncertainty,
especially for short time intervals. Table 7 shows the contribution of a 300 ms
variation in human reaction time to the overall measurement uncertainty, for
measurement periods ranging from 1 s to 1 day.
Table 7 - The contribution of a 300 ms variation in reaction time to

the measurement uncertainty.
Hours Minutes Seconds Uncertainty (%)
10 3
1 60 0.5
10 600 0.05
30 1800 0.01666
1 60 3600 0.00833
2 120 7200 0.00416
6 360 21 600 0.00138
12 720 43 200 0.00069
24 1440 86 400 0.00035
As Table 7 illustrates, the longer the time interval measured, the less impact the
operator's start/stop uncertainty has on the total uncertainty of the measurement.
Therefore, it is better to measure for as long as practical to reduce the uncertainty
introduced by the operator, and to meet the overall measurement requirement.
To get a better understanding of the numbers in Table 7, consider a typical

stopwatch calibration where the acceptable measurement uncertainty is 0.02 %
(2 x 10
4
). If the variation in human reaction time is known to be 300 ms for the
direct comparison method, a time interval of at least 1500 s is needed to reduce
32
The Direct Comparison Method 4
the uncertainty contributed by human reaction time to 0.02 %. However, we
if
use a 1500 s interval, we could possibly be measuring the variation in human

reaction time, and nothing else. Therefore, we need to extend the time interval so
thathuman reaction time becomes an insignificant part of the measurement. For
example, NIST Handbook 105-5 [4] states that a stopwatch is considered to be
within tolerance if its time offset is 2 s or less in during a three hour calibration.
Three hours is a long enough time interval to exceed the 0.02 % requirement and
to ensure that the uncertainty contribution of human reaction time is insignificant.
There is no hard and fast rule, the length of the calibration can vary according to
each laboratory's procedures, but it must be long enough to meet the uncertainty
requirements for the device being tested. If your uncertainty requirement is
0.01 % or lower, the direct comparison method might not be practical.

5.C. Uncertainties of Direct Comparison Method
The Direct Comparison Method has three potentially significant sources of
uncertainty that must be considered: the uncertainty of the reference, the reaction
time of the calibration technician, and the resolution of the DUT.
5.C.I. Uncertainty of the Traceable Time Interval Reference

If the reference signal is one of the telephone services listed in Table 6, two
phone calls are usually made. The first call is made to obtain the signal to
start the stopwatch, and the second call is made to obtain the signal to stop the
stopwatch. If both calls are made to the same service and routed through the
same phone circuit, the delay through the circuit should be nearly the same for
both calls. Of course, the delays will not be exactly the same, and the difference
between the two delays represents the uncertainty of the time interval reference.
In most cases, this uncertainty will be insignificant for our purposes, a few
milliseconds or less. For example, callers in the continental United States using
ordinary land lines can expect signal delays of less than 30 ms when dialing
NIST at (303) 499-71 1 1, and these delays should be very repeatable from phone
call to phone call. Even in a theoretical case where the initial call had no delay,
and the final call had a 30 ms delay, the magnitude of the uncertainty would be
limited to 30 ms.
However, if an ordinary land line is not used, the uncertainties associated with
telephone time signals must be assumed to be larger. For example, wireless
phone networks or voice over Internet protocol (VOIP) networks sometimes
introduce larger delays that are subject to more variation from phone call to
phone call than ordinary land lines. If wireless or VOIP networks are used,
33
Stopwatch arid Timer Calibrations
however, it is reasonable to assume that the transmission delay does not exceed
150 ms, since the International Telecommunication Union (ITU) recommends
that delays be kept below this level to avoid the distortion of voice transmissions
[14]. Calls made from outside the continental United States might occasionally
be routed through a communications satellite, introducing delays of about 250
ms. Although the use of satellites is now rare, if the first call went through a
satellite, and the second call didn't (or vice versa), a significant uncertainty would
be introduced. Therefore, common sense tells us that a laboratory in Illinois (for
example) shouldn't start a calibration by calling the NIST service in Colorado,
and then stop the calibration by calling the NIST service in Hawaii. Based on
this discussion, it is recommended that an uncertainty of 150 ms be assigned
if wireless or VOIP networks are used, or 250 ms if calls are routed through a
satellite.
During a single phone call, the uncertainty of the time interval reference is
essentially equal to the stability of a telephone line (the variations in the delay)
during the call. Phone lines are surprisingly stable. The ITU recommendations
for timing stability within the telephone system call for stabilities of much less
than 0. 1 ms for the North American Tl system [15]. A NIST study involving the
Automated Computer Time Service (ACTS), a service that sends a digital time
code over telephone lines, showed phone line stability at an averaging time of 1
s to be better than 0.1 ms over both a local phone network and a long distance
network between Boulder, Colorado and WWVH

in Kauai, Hawaii [16]. While it
is not possible to guarantee this stability during all phone calls, it is reasonable to
assume that the stability should be much less than 1 ms during typical calls, which
are limited to about 3 minutes in length. Thus, the uncertainties contributed by
phone line instabilities are so small that they can be ignored for our purposes.
If the radio signals listed in Table 6 are used as a reference instead of a telephone
signal, the arrival time of the signal will vary slightly from second to second as the
length of the radio signal path changes, but not enough to influence the results of a
stopwatch or timer calibration. Shortwave signals that travel over a long distance
relyon skywave propagation, which means that they bounce off the ionosphere
and back to Earth. A trip from Earth to the ionosphere is often called a hop, and
a hop might add a few tenths of a millisecond, or in an extreme case, even a full
millisecond to the path delay. Normally, propagation conditions will remain the
same during the course of a calibration, and the variation in the radio signal will
be negligible, often less than 0.01 ms. Even if an extra hop is added into the radio
path during the calibration; for example if a one hop path becomes a two hop
34
path, the received uncertainty of the radio signal will still not exceed 1 ms [13].
If a traceable time display is used instead of a radio signal or telephone signal,
it can generally be assumed that the uncertainty of the display is less than 1
ms. This is because instruments that continuously synchronize their displays to

tracable signals will normally have repeatable and stable delays. However, in
order for this uncertainty estimate to be valid, be sure to use only time displays
that meet the traceability requirements discussed in Section 5. A. 3.
5.C.2. Uncertainty Due to Human Reaction Time

To understand the effect of human reaction time on stopwatch and timer calibration
uncertainties, a small study was conducted at Sandia National Laboratories. Four
individuals were selected and asked to calibrate a standard stopwatch using the
direct comparison method. Two separate experiments were conducted. In the
first experiment, the operators were asked to use a traceable audio time signal,
and second experiment, the operators were asked to use a traceable time
in the
display. The time base of the stopwatch was measured before and after each test
(using the Time Base Method), and its offset from nominal was found to be small
enough that it would not influence the test. Therefore, differences in readings
between the stopwatch being tested and the standard would be due only to the
operator's reaction time. Each operator was asked to repeat the measurement
process 10 times, and the resulting 10 differences between the standard and the
stopwatch were recorded and plotted (Figure 11).
As shown was usually less than ±100

in Figure 11, the average reaction time
ms, with a worst-case reaction time exceeding 700 ms. The mean and standard
deviation for each operator was computed and graphed in Figure 12. This graph
indicates that the average (mean) reaction time of the operator can be either
negative (anticipating the audible tone) or positive (reacting after the audible
tone). Figure 12 also shows that in addition to the average reaction time having
a bias, the data is somewhat dispersed, so both elements of uncertainty will need
to be considered in a complete uncertainty budget. For this experiment, the
worst case mean reaction time was 120 ms and the worst case standard deviation
was 230 ms. It should be noted that in the measurements recorded in Figure 12,
Operators 1 and 2 had no previous experience calibrating stopwatches. Based on
these results, it is recommended that each calibration laboratory perform tests to
determine the uncertainty of their operator's reaction time.
35
Human Reaction Time, Direct Comparison, Audio Method

Measured Reaction Time Deviation over 10 Runs
80
BL7Q
60
50
40
30
vT 0-20
C 10
o
5g 00
« *
TO
> -0.10
i » t
Q -0.20
-0.30
-0 40
-0.50
»0.60
-0 70
-0.80
Run Number
Figure 11. Reaction time measurements (four operators, 10 runs each) for the
direct comparison method.
Human Reaction Time, Direct Comparison, Audio Method

Mean Bias and Standard Deviation
>
<
f
Figure 12.
12
Averaging measurement
Operator Number
3 4
results for four different operators.
36
The Di »arison Method
When was used, the uncertainty due to human reaction

a traceable time display
time was found to be approximately the same as the human reaction time for an
audible tone. Keep in mind that these results are presented to illustrate the nature
of uncertainty due to human reaction time, and to provide a very rough estimate
of its magnitude. We strongly encourage each person calibrating stopwatches
and timers to perform repeatability and reproducibility experiments to help better
determine the uncertainty of human reaction time.
5.C.3. Device Under Test (DUT) Resolution Uncertainty

Since the direct comparison method requires observing data from the DUT
display, the resolution of the DUT must also be considered. For digital indicating
devices, resolution uncertainty is understood to be half of the least significant
digit, with an assumed rectangular probability distribution. For an analog watch,
the same method of determining resolution uncertainty may be used because the
watch moves in discrete steps from one fraction of a second to the next.
5.C.4. Uncertainty Analysis

This section provides an example of how data collected using the direct comparison
method can be used to perform an uncertainty analysis. For this estimate of
uncertainty we will include the mean bias as an estimate of uncertainty, rather
than correcting for it, because the mean bias can be either negative or positive,
and may vary from time to time for the same user [17]. In this calibration
process, the mean bias can be considered a measurement of reproducibility, and
the standard deviation a measure of repeatability.
5.C.4.a. Uncertainty Distributions

Because of the lack of knowledge regarding distributions of the mean bias and
delay deviation between telephone calls, both components of uncertainty are
treated as rectangular distributions. Since the resolutions of digital and analog
stopwatches have known, discrete quantities, their distribution is also rectangular
[12]. All other data are considered to be normally distributed.
5.C.4.D. Method of Evaluation

Even though the data provided in previous sections were treated statistically, they
were collected from previous measurements, and not during the actual stopwatch
calibration. Because the metrologist does not have statistical data based on a
series of observations to support these uncertainties, they are identified as Type
B.
37
5.C.4.C. Combination of Uncertainties
In the following examples, the uncertainty budgets were developed for a
calibration using traceable land lines (Table 8 and Table 9), and for a calibration
using a cellular phone or satellite signal (Table 1 0) based upon the data previously
provided. The human
reaction time was based on the worst case data presented
in Section 5.C.2. The uncertainties are rounded to the nearest millisecond. The
uncertainty components are considered to be uncorrected, so they are combined
using the root sum of squares method.

using a land line.
Source of Magnitude, Method of Distribution Standard

uncertainty ms evaluation uncertainty, ms
Human reaction Octangular
120 TypeB 69
time bias
Human reaction Normal
time standard 230 TypeB (k= 1) 230
deviation
Telephone delay Rectangular

30 TypeB 17
deviation
Rectangular
X
A DUT resolution TypeB
Combined uncertainty 241
Expanded uncertainty (k = 2, representing approximately a 95 % 482

level of confidence)
38
i he Direct Comparison I e t 3d

using a cell phone.
Source of Magnitude, Method of Distribution Standard

Human reaction
120 TypeB Rectangular 69
time bias
Human reaction
Normal
time standard 230 TypeB 230
deviation
Telephone delay
deviation
Vi DUT resolution 5 TypeB Rectangular 3

Table 10 - Uncertainty analysis for direct comparison method (analog

DUT) using a land line.
Source of Magnitude,
& Method of .
A. Standard
Distribution A .
A
Human reaction
time bias
Human reaction
Normal
(*=i)
deviation
Telephone delay
deviation
Vz DUT resolution 100 Type B Rectangular 58

39
Section 6
The Totalize Method
The totalize method partially eliminates the measurement uncertainty from human
reaction time, but requires two test instruments: a calibrated signal generator,
and a universal counter.
6.A. Calibration Procedure for the Totalize Method

The counter is set to TOTALIZE, with a manual gate. A signal from a calibrated
synthesized signal generator is connected to the counter's input, and the
laboratory's primary frequency standard is used as the external time base for the
synthesizer and the counter (Figure 13). The frequency should have a period at
least one order of magnitude smaller than the resolution of the stopwatch. For
example, if the stopwatch has a resolution of 0.0 1 s ( 1 ms); use a 1 kHz frequency
(1 ms period). This provides the counter with one more digit of resolution than
the stopwatch.
Laboratory Frequency
Reference Standard
© Connect to
Synthesizer
and Counter
Frequency Synthesizer External
Time Base
Inputs
0—1
Universal Counter
Figure 13. Block diagram of the totalize method.
41
To begin the measurement, start the stopwatch and manually open the gate of
the counter at the same time. One way to do this is by rapidly pressing the start-
stop button of the stopwatch against the start button on the counter (Figure 14).
Another method is to press the start/stop button of the stopwatch with one hand
and simultaneously pressing the start/stop button of the counter. After a suitable
period of time (determined by the calibration requirements of the stopwatch or
timer being calibrated), use the same method to simultaneously stop the stopwatch
and close the gate of the counter.
Figure 14. Using the start-stop button of the stopwatch to start the counter.
Once the counter and stopwatch are stopped, compare the two readings. Use the
equation At/T to get the results, where At is the difference between the counter
and stopwatch displays, and T is the length of the measurement run. For example,
if At = 100 ms and T = 1 hour, the time uncertainty is 0.1 s / 3600 s or roughly
5
2.8 x 10- (0.0028 %).
The Totalize Method >
6.A.I. Advantages of the Totalize Method
When using the stopwatch's start/stop button to open and close the counter's
method partially eliminates human reaction time, and therefore has a
gate, this
lower measurement uncertainty than the direct comparison method.
6.A.2. Disadvantages of the Totalize Method

This method requires more equipment than the direct comparison method.
6.B. Uncertainties of Totalize Method

The factors that contribute to the measurement uncertainty of this method are
discussed below.
6.B.I. Uncertainty of the Frequency Input

A synthesized signal generator that was recently calibrated typically has a
frequency uncertainty ranging from 1 x 10~ 6 to 1 x 10~ 9 . If the signal generator
time base is externally locked to a laboratory frequency standard such as a cesium
oscillator or a GPS disciplined oscillator, the frequency uncertainty can be much
12
smaller, typically parts in 10 or less.
6.B.2. Uncertainty Due to Human Reaction Time

This source of uncertainty is due to any difference between the starting of both
the stopwatch and counter, and the stopping of both the stopwatch and counter.
In order to estimate this source of uncertainty, a study was conducted at Sandia
National Laboratories. Four individuals were selected and asked to calibrate a
standard stopwatch using the totalize method, using one hand to start and stop
the stopwatch, and one hand to start and stop the frequency counter. The time
base of the stopwatch was measured before and after each test (using the time
base method), and its offset from nominal was found to be small enough so that
it would not influence the test. Therefore differences in readings between the
stopwatch being tested and the standard would be due only to the operator's
reaction time.
Each operator was asked to repeat the measurement process 10 times, and the
resulting difference between the standard and the stopwatch were recorded.
43
Reaction Time Measurements for Totalize Method (Two Hands)
0.10
0.08
0.06
0.04
0.02
0.00
-0.02
-0.04
-0.06
-0.08
-0.10
4 6 10
Run Number
Figure 15. Measured reaction times (four operators, 10 runs each) for totalize
method.
As shown in Figure 15, the average reaction time was usually less than 40 ms.
The was about 90 ms. The mean and standard deviations for
largest deviation
each operator were computed and graphed in Figure 16. The data demonstrates
that the worst case mean reaction time was 27 ms, and the worst case standard
deviation was 10 ms, which is a significant improvement over the direct
comparison method. Subsequent measurements of human reaction time were
made by placing the start/stop button of the stopwatch directly against the start/
stop button of the frequency counter (as shown in Figure 12) so that both buttons
were pushed at the same time. The experimental data showed that this method
provided no significant advantage over using one hand to start the stopwatch and
the other hand to start the frequency counter.
44
The Totalize Method
Mean Reaction Time for Totalize Method (Two Hands)
2 3
Operator Number
Figure 16. Mean reaction times (four operators, 10 runs each) for the totalize
method.
6.B.3. Uncertainty Due to the Counter

For this method, the uncertainty of the counter is related to the counter resolution.
The internal time base of the counter is not used, because the gate time is controlled
by the manual start/stop function. This method does not use the counter gate or
trigger, so the uncertainties associated with these functions are also eliminated.
The counter is simply used as an event counting device, and the uncertainty is
equivalent to ±1 least significant digit on the display.
6.B.4 Device Under Test (DUT) Resolution Uncertainty

Since the totalize method involves observing data from the DUT, the resolution
of the DUT must also be considered. For digital indicating devices, resolution
uncertainty is understood to be half of the least significant digit, with a rectangular
45
distribution. A rectangular distribution is also used for analog timers, since these
devices move in discrete steps from one fraction of a second to the next.
6.B.5 Method of Evaluation

Since the sources of uncertainty in this calibration are different types, such
as the absolute accuracy specification for reaction time, the relative accuracy
specification for the frequency source, and an absolute specification of ±1 count
for the counter, it is necessary to define the calibration conditions and convert all
specifications to a common format.
The uncertainty example in Table 1 1 assumes that the synthesized signal generator
can produce a 1 kHz signal with an uncertainty of 1 x 10" 8 or less. The stopwatch
is compared to the counter for 3 hours (10 800 s). The stopwatch is started and
stopped with one hand, and the counter is started and stopped with the other. The
stopwatch is digital and the resolution is 0.01 s. As in the previous uncertainty
example, all sources of uncertainty are combined by root sum of squares, and
uncertainties are rounded to the nearly millisecond.
Table 11 - Uncertainty analysis for totalize method.
Source of Magnitude, Method of Standard

Distribution
Human reaction
time bias
Human reaction
Normal
(k=l)
deviation
Synthesizer
TypeB Rectangular
accuracy
Totalize counter
resolution
A DUT
X
resolution

46
I he Totalize Method
6.C. Photo Totalize Method

An alternativemeasurement technique to the totalize method previously discussed
involves using a high speed camera (digital or film type). The equipment is set up
in thesame manner as in Section 6. A. In this method, both the stopwatch being
calibrated and the universal counter are started in no particular order. When
both instruments are counting, a photo is taken of the two instruments. Under
appropriate shutter and lighting conditions, the resulting photo will clearly show
the time displayed on both the stopwatch and counter. Both readings are recorded
as the initial measurements in the calibration. After an appropriate amount of time
has elapsed, a second photo is taken and the displays are recorded as the closing
measurements. The difference in elapsed time for the stopwatch is compared to
the difference in elapsed time for the counter to determine the measured offset
of the stopwatch.
Figure 17 shows that the initial counter reading is 53.980 s and the stopwatch
is at 57.52 s. Figure 18 shows that the counter reading is 82.049 s and the
stopwatch is at 1 minute 26.18 s (86.18 s). The elapsed time on the counter is
82.049 - 53.980 = 28.069 s, and likewise the elapsed time on the stopwatch is
86.18 - 57.52 = 28.66 s. This would indicate that over the 28 s interval, the
stopwatch was in error by 0.59 s or approximately 1820 s per day.
Figure 1 7. Photo totalize start reading.
47
HEWLETT
«U*ABO MPM „„,...„-
COUNTER
8? 049
QQ CZ
Figure 18. Photo totalize stop reading.
WARNING NOTE - In order for this method to provide accurate results, the
photo system must be set up in a manner that captures the stopwatch and counter
display correctly. If the shutter speed is too slow, the images will not produce
discernable information, as Figure 1 9 shows for the stopwatch. We could interpret
this stopwatch reading with the 1/10 s digit as 9, 7, 3, or 4, and the 1/100 s digit
as 3, 5, or 6.
Additionally, often when the photo yields a very clear display as in Figures 17
and 18, the resultingmeasurement is not valid. This stopwatch was previously
calibrated using other methods described in this book and was found to have an
offset of less than 1 second per day, which is significantly different from the photo
totalize method results indicated above. For this example and photo equipment
setup, themethod produced false results. Anyone using the photo totalize
method must validate the measurement results of their measurement process
by comparing results obtained for a particular instrument to other measurement
methods discussed in this book.
48
Figure 19. Ambiguous photo totalize reading.
Section 6.C1. Advantages of the Photo Totalize Method

Using measurement method, all uncertainty due to human reaction time is
this
reduced to zero. This makes the estimate of measurement uncertainty smaller,

and the amount of time needed to complete a calibration shorter than the direct
comparison or totalize method. This method also makes it possible to calibrate
multiple stopwatches at the same time, and then maintain the photos as archival
data.
Section 6.C.2. Disadvantages of the Photo Totalize Method

This method can VERY EASILY produce erroneous results. As previously
stated, themeasurement results for the particular cameras used by the laboratory
should always be compared to measurement results of the same unit under test
using other methods described in this book.
49
Th€ : Base Method <«
Section 7
The Time Base Method
The time base measurement method is method for stopwatch and
the preferred
timer calibrations, since it introduces the least amount of measurement uncertainty.
Because the DUT's time base is measured directly, the calibrating technician's
response time is not a factor.
The exact method of measuring the stopwatch's time base depends upon the type
of stopwatch or timer being calibrated. If the unit has a quartz crystal time base,
an inductive or acoustic pickup is used to monitor the stopwatch's 32 768 Hz
time base frequency on a calibrated frequency counter (the pickup is fed into an
amplifier to boost the signal strength). If the unit is an older LED-type stopwatch,
the frequency is usually 4.19 MHz. An inductive pickup can even be used to
sense the stepping motor frequency of analog mechanical stopwatches, or the
"blink rate" of a digital stopwatch display. Or, an acoustic pickup can be used to
measure the "tick" of a mechanical stopwatch.
7.A. References for the Time Base Method

The reference for a time base calibration is the time base oscillator of the measuring
instrument. For example, if a frequency counter is used, the measurement
reference is the time base oscillator of the frequency counter. In order to establish
traceability, the frequency counter time base must have been recently calibrated
and certified. However, a better solution is to have the laboratory maintain a
traceable 5 MHz or 1 MHz signal that can be used as an external time base for
the frequency counter and all other test equipment. If an external time base is
used and its measurement uncertainty is known, it is unnecessary to calibrate the

internal time base oscillator.
7.B. Calibration Procedure for the Time Base Method

Two methods of calibrating a stopwatch time base are described below. One
uses a commercially available measurement system; the other uses a frequency
counter with an acoustic pickup. Note that neither calibration method requires
opening the case of the stopwatch or timer. Keep in mind that you should never
disassemble a stopwatch or timer and attempt to measure the time base frequency
by making a direct electrical connection. The crystal oscillators in these units are
very small, low-power devices. Their frequency can dramatically change if they
are disturbed or loadeddown by the impedance of a frequency counter, and in
some cases they can even be destroyed by incorrect electrical connections.
51
7.B.I.Using a Commercial Time Base Measurement System

Time base measurement systems are available from several vendors. One
example of a commercially available time base measurement system (Figure 20)
is described here for the purposes of illustration. This unit measures the frequency
of the time base oscillator, and converts this information to a display of seconds
per day, or seconds per month. This same function could be performed with a
sensor (acoustic or inductive pickup), a frequency counter, and the conversion
formula described in the next section.
Figure 20. Time base measurement system for stopwatches and timers.
The example unit uses a 4.32 MHz quartz time base oscillator as a measurement
reference. In a 2 measurement period (the shortest period used by the instrument),
s
the oscillator produces 8 640 000 cycles, a number equal to the number of 0.01
s intervals in one day. Therefore, the instrument resolution is 0.01 s per day.
Cycles from the DUT time base are counted by a sensor that is connected to the
measurement system. The sensor uses the cycle count from the DUT to start
and stop the cycle count from the reference time base. For a 2 s measurement
interval, a 32 768 Hz DUT would stop the cycle count from the reference time
base after counting to 65 536 (32 768 x 2). If the reference time base counted to
8 639 900 during this same interval, then the system would estimate that the DUT
would gain about 1 s per day (100 x 0.01 s).
The time base measurement system shown in Figure 20 allows the measurement
interval to be set from 2 s to 960 s. Selecting an interval longer than 2 s is
52
The Time Base Method
effectively the same thing on a frequency counter. It
as increasing the gate time
is important to measure for a long enough period to get an accurate and stable
reading. When testing a 32 768 Hz quartz stopwatch, a 10 s to 12 s measurement

is normally sufficient to obtain a reading stable to ±1 count. When testing an
older mechanical (Type II) stopwatch, a longer measurement of 120 s or more
may be required. Table 12 shows the effect that the length of the measurement
time has on the stability of the stopwatch calibrator's readings.
Table 12 - The effect of the length of the measurement time on stability

(based on 25 readings).
Measurement time 2s 10 s 12 s 20 s
Mean -0.03 -0.06 -0.06 -0.06
Standard deviation of the mean 0.0050 0.0012 0.0011 0.0006
Maximum 0.00 -0.05 -0.05 -0.06
Minimum -0.09 -0.07 -0.07 -0.07
Range 0.09 0.02 0.02 0.01
To support 0.01 s resolution, the instrument's 4.32 MHz time base oscillator
7
must be calibrated to within 1.16 x 10~ . If the instrument is calibrated to within
specifications, the display uncertainty is ±0.05 s per day (maximum time base
frequency offset of about 6 x 10~ 7 ). In all cases, the uncertainty of the time base
oscillator relative to UTC must be known in order to establish traceability. The
system can be calibrated by applying either a traceable 1 pulse per second (pps)
signal to a pin connector on the back of the unit or a traceable 32 768 Hz reference
signal to the sensor, or by extracting and measuring the 4.32 MHz signal directly
with a traceable frequency measurement system.
The DUT can be a Type 1 Hz and 4.19 MHz devices can

stopwatch (both 32 768
be measured), or a Type 2 mechanical stopwatch. The 32 768 Hz signal is picked
up with an acoustic sensor, and then compared to the time base oscillator. A 1 Hz
offset in the 32 768 Hz signal translates to a time offset of about 2.6 s per day. A
capacitive sensor is used to detect the 4.19 MHz frequency of quartz time base
oscillators, an acoustic or inductive pickup is used to sense the stepping motor
frequency of analog mechanical stopwatches, and an inductive pickup is used to
sense the "blink rate" of digital stopwatches.
Front panel switches allow the operator to select the type of device being tested,
53
the measurement interval, and whether the time offset should be displayed as
seconds per day or seconds per month. Once these parameters have been chosen,
the device is measured by simply positioning it on top of the sensor until a usable
signal is obtained, waiting for the measurement interval to be completed, and
then recording the number from the display. It is always a good idea to allow
the stopwatch calibrator to complete at least two complete measurement cycles
before recording a reading.
Figure 21 shows a more elaborate time base measurement system with some
additional features. This device has a small LCD display with graphics capability,
built-in temperature measurement capability, an RS-232 interface that allows
measurement results to be transferred to a computer, and an optional Global
Positioning System (GPS) satellite receiver that corrects and calibrates the quartz
time base oscillator. Like the system shown in Figure 20, this system can measure
the 32 768 Hz quartz crystal frequency of Type 1 stopwatches, the stepping motor
frequency of analog stopwatches, and the "blink rate" of digital stopwatches. As
an additional benefit, this system can measure the quartz crystal frequency and
the stepping motor frequency at the same time. This can be useful if the device
under test compensates for the frequency offset of the quartz crystal by applying
corrections to the stepping motor, thus making the analog display more accurate
than the time base (a common practice in analog quartz wristwatches).
Figure 21. A time base measurement system for stopwatches and timers with
an integrated graphics display.
54
TheTis r i
;
is ;
7.B.2. Using a Frequency Counter and an Acoustic Pickup

If an acoustic pickup and amplifier are available, you can measure the frequency
of a stopwatch time base directly with a frequency counter. The reading on

the counter display can be used to calculate the frequency offset using this
equation:
f - f
/(offset) = measured nominal
^nominal
where Jfmeasured , is the reading

c displayed
r j by
j
the frequency
i j
counter, and f
j nominal is
. .
the frequency labeled on the oscillator (the nominal frequency it is supposed to

produce).
If Jf
nominal
,
is 32 768 Hz and Jfmeasured ^ is 32 767.5 Hz, then the frequency
~ J offset is
-0.5 / 32 768 or -1.5 x 10~ 5 or -0.0015 %. To get time offset in seconds per day,
multiply the number of seconds per day (86400) and the frequency offset:
86 400 x (-1.5 x 10 5
)
= -1.3 s per day ,
which means the stopwatch can be expected to lose 1.3 s per day. You might find
it easier to note that a 1 Hz error in a 32 768 Hz device equates to a time offset of
about 2.64 s, since 86 400 / 32 768 = 2.64. Therefore, a 2 Hz offset is about 5.3
s / day, a 3 Hz offset is about 7.9 s / day, and so on. If the acceptable tolerance is
10 s/ day, then you'll know that 3 Hz is well within tolerance.
These results show that even a low cost eight-digit frequency counter will
provide more measurement resolution than necessary when measuring 32 768
Hz devices. The last digit on an eight-digit counter represents 0.001 Hz (1 mHz),
and a 1 mHz frequency offset represents a time offset of just 2.6 ms per day. Very
few stopwatches or timers can perform at this level.
7.B.3. Advantages of the Time Base Method

The time base method completely eliminates the uncertainty introduced by
human reaction time. The measurement uncertainty can be reduced by at least
two orders of magnitude when compared to the direct comparison method, often
to 1 x 10 6 or less. This method is also much faster. The measurement can often
be performed in a few seconds, as opposed to the several hours often required for
the direct comparison method.
55
7.B.4. Disadvantages of the Time Base Method

This method requires more equipment than the direct comparison method, and
does not easily work on some electrical, mechanical, or electro-mechanical units.
It also does not test the functionality of the stopwatch or timer, only the time base.
Function tests need to be performed separately by starting the unit, letting it run
for a while (a few minutes to a few hours, depending on how the unit is used), and
stopping the unit. If the unit appears to be counting correctly, the displayed time
interval will be accurate.
7.C. Uncertainties of Time Base Method

This method utilizes either a time base measurement system or a frequency counter
with an acoustic or inductive pickup to measure the frequency of the device's
internal time base oscillator. If we use the time base measurement system shown
in Figure 20 as an example, and take into account its specified accuracy of ±0.05
s/day and its resolution of 0.01 s, then the measurement uncertainty equals 0.05 s/
day (50 ms/day). There is no uncertainty contributed by human reaction time, and
the resolution uncertainty of the stopwatch calibrator is insignificant compared to
its accuracy specification. Resolution uncertainty does not need to be considered,
since data are not observed from the DUT's display. Since there is only one
uncertainty component, we did not include an uncertainty analysis table.
56
Required Uncertainty
Section 8
How to Determine the Calibration Method Meets the
if
Required Uncertainty
The preceding sections have provided estimates of expanded measurement

uncertainty associated with various types of stopwatch calibrations. It is important
to ensure that the achievable measurement uncertainty is relatively small when
compared to the stopwatch tolerance we are attempting to verify. For example,
if we use the direct comparison method to calibrate a stopwatch with an accuracy
of 5 s per day (0.01 s resolution), and we decide to use a calibration period of 5
minutes, the accuracy of the stopwatch over the 5 minute interval would be about
17 ms. The first problem arises in that the resolution of the stopwatch is 0.01
s, and cannot display 0.017 making the specification impossible to verify for
s,
this time period. Furthermore, the measurement uncertainly for the calibration
process is 480 ms, approximately 28 times larger than the accuracy that we are
trying to verify! Clearly a 5 minute calibration period for this stopwatch and
calibration method is not adequate. Even a 1 hour calibration is inadequate,
because the quantity required to be measured, now about 208 ms, is still less than
half of the measurement uncertainty.
As stated in ISO/IEC 17025 [4], when a statement of compliance to a

metrological specification is made (for example, is the DUT "in tolerance" or
"out of tolerance" based on the manufacturer's specifications), the laboratory
must take the measurement uncertainty into account. It is the responsibility of the
calibration laboratory to decide whether the measurement uncertainty associated
with the calibration method is small enough to comply with the metrological
requirement. Calibration laboratories generally use some sort of decision rule
for determining whether the uncertainly associated with the calibration method is
adequate. Examples of such rules are the N: 1 decision rule (where the tolerance
of the instrument to be tested is N times larger than the calibration uncertainty
and is mostly commonly stated as 4:1 or 3:1), or some form of guardbanding
(where the calibration uncertainty or some fraction of it is subtracted from the
DUT accuracy specification and the calibration results are acceptable only if
the measured was within the guardband limit). Whatever the case, it is
offset
important to compare the calibration measurement uncertainty to the tolerance of
the DUT, in order to ensure that the calibration process is valid.
57
To illustrate this, assume that a laboratory employs a 4:1 rule for acceptable
uncertainty. This means that the expanded measurement uncertainty of the
calibration must be four times smaller than the acceptable tolerance of the DUT
in order to declare the unit in or out of tolerance. The laboratory is calibrating a
stopwatch to a tolerance of 0.02 %, using the direct comparison calibration
digital
method via a telephone landline (Table 8). The estimated expanded measurement
uncertainty for this method was 482 ms. In order for 482 ms to be one quarter
of the tolerance of the DUT, the tolerance must be 1928 ms. The amount of time
required to elapse on the stopwatch during the calibration would be at least:
1,928 S
=Xs = 9640 s = 2 h 41 minutes
0.02 %
For the same scenario, if we were trying to verify the stopwatchshown in Figure
8 to the manufacturer's specification of 0.0003 %, the stopwatch would have to
run for more than 1 week (178 hours, 31 minutes, and 7 s). This is impractical,
and the laboratory that is testing to 0.0003 % would clearly want to consider the
time base method as a much faster, more appropriate method.
58
OtherTopics Related to Measurement Uncertainty
Section 9
Other Topics Related to Measurement Uncertainty
This section discusses other important topics related to measurement uncertainty.
The first involves estimating the uncertainty of a field calibration, where a
reference standard stopwatch is brought into the field and used to calibrate another
device. Next, the stability and aging of 32 768 Hz quartz crystal oscillators used
in stopwatches and timers is by discussions of how the
discussed, followed
performance of a stopwatch can change when a new battery is installed, and how
extreme temperature conditions can affect stopwatch performance.
9.A. Uncertainty Analysis of Using a Calibrated Stopwatch to

Calibrate another Device
A calibrated stopwatch is often used to perform field calibrations of other timing
devices, such as an industrial timer or parking meter. In this example (Table
13), we are calibrating an industrial timer with 1 s resolution. The reference
for the calibration is a stopwatch that has a specified accuracy of 5 s per day
(about 208 ms per hour), 0.01 s resolution, and a calibration uncertainty of 0.2
s per day (roughly 8 ms per hour). As previously discussed in Section 5.C.2,
the human reaction time bias is 120 ms, and the human reaction time standard
deviation 230 ms. The measurement time for the calibration is 1 hour (3600
is
s). Using these values and the analysis provided in Table 13, the uncertainty of
the industrial timer when measuring a 1 hour interval is 789 ms, or 0.02 %. All
values in Table 1 3 are rounded to the nearest millisecond.
When performing this analysis, it is important to include the specified accuracy

of the reference stopwatch, which takes into account long-term sources of error
during the calibration interval that may not have been noticeable during the time
of calibration. It is also important to consider the uncertainty of the stopwatch
calibration as part of the budget when using a stopwatch as a reference, because
it may be relatively large when compared to the other sources of uncertainty in
the measurement process.
59
Table 13 - Uncertainty analysis of using a calibrated stopwatch to calibrate
another device.
Standard
Source of Magnitude, Method of
Distribution uncertainty,
uncertainty ms evaluation
ms
Unit under test
resolution 500 TypeB Rectangular 289

A digit
l
Human
reaction time 120 TypeB Rectangular 69
bias
Human
reaction time Normal
230 TypeB 230
standard (k=\)
deviation
Stopwatch
accuracy
Stopwatch
resolution TypeB Rectangular
A digit
l
Stopwatch
Normal
calibration TypeB
{k=2)
uncertainty
Expanded uncertainty (k = 2, representing approximately a

789
95 % level of confidence)
9.B. The Effects of Stability and Aging on Calibrations of 32 768 Hz
Crystals
Aging is the systematic change in frequency over time due to internal changes
in an oscillator. All quartz oscillators are subject to aging, and the rate at which
they age is often a key factor in determining their calibration interval. The aging
rate of a quartz crystal oscillator often depends upon its surface area tovolume
ratio, and small, low frequency crystals generally have low aging rates [18].
Thus, the 32 768 Hz crystals found in stopwatches and timers usually tend to age
slowly and have very good long term stability, and their frequency changes by a
surprisingly small amount over time.
60
OtherTopics Related to Measurement Uncertainty
Frequency Stability of Two Low-Cost Stopwatches

1.0E-07
„4 L
' -
ft 1
I
«A
jj r
* 8.0E-08 /—
C
« q
/
f
7.0E-08
t i i
\ >t i i i
£ 6.0E-08
> i i
\
r t j
i
tn V j
> \
j Stopwatch A
5.0E-08
jjj
0) \ s Stopwatch B
3 V j
CT
i
I! 4.0E-08
i I r
! ! !
3.QE-Q8 ,.. .i. —l~ _»™___L . ,.
Days
Figure 22. Graph of the frequency stability of two stopwatches.
To illustrate this, Figure 22 is a graph of the frequency stability of two stopwatches

that were continuously measured over a period of more than one month using
a previously described (Section 7.B.1) time base measurement system. Both
stopwatches were low cost devices; stopwatch A had a suggested list price of
about $25 USD, and stopwatch B sells for about $55 USD. The laboratory
temperature during the measurement was about 23 °C ± 1 °C. The graph shows
the Allan deviation (ADEV) of each stopwatch for averaging times ranging from
0.5 days to 7 days. ADEV is a commonly used statistic for estimating frequency
from the conventional standard deviation because it does
stability [19]. It differs
not use the average frequency as a point of reference. Instead, it compares the
frequency offset of the DUT during each measurement period with its frequency
offset during the previous measurement period. By doing so, it reveals how the
frequency of an oscillator changes over time due to effects such as aging.
Stopwatch A had an average frequency offset (accuracy) during the test of about
6
8 x 10" . As indicated in Figure 22, the stability (ADEV) was slightly better than
1 x 10" 7 after one week, or about 80 times better than the accuracy. Stopwatch
B had an average frequency offset (accuracy) during the test of about 5 x 10~
7
.
61
Its stability as estimated with ADEV was near 7 x 10~ 8 after one week, or about
seven times better than its accuracy.
Because of their slow aging rate, stopwatches tend to produce very repeatable
results over long periods of time. When compared to the requirements for
stopwatch calibrations discussed in Section 3, the small changes in frequency
due to aging are usually insignificant. This makes it possible for a laboratory to
allow long intervals (perhaps exceeding one year) between calibrations.
9.C. Factors That Can Affect Stopwatch Performance

Many factors can cause a quartz crystal oscillator to change frequency. Some of
these are shown in Figure 23. Sharp temperature changes, vibration, and shock
are some of the more common phenomena that can affect the time base on a day-
to-day basis. Also, if power is removed from an oscillator and later restored, there
can be a shift in the frequency. This is true of quartz crystal stopwatches when
the battery is changed. This frequency change is usually small when compared
to the stopwatch's time base specifications, usually less than 10 to 20 parts in
10 6 (-10 to 20 ms/day) and usually short-term, with the time base returning to
its original frequency after a period ranging from a few hours to a few days.
Oscillator
Turn Off
Vibration Shock &
Figure 23. Factors that can change the quartz time base frequency.
62
Other Topics Related tc >yrament Uncertainty
Temperature is perhaps the most important factor that affects quartz crystal time
base performance. The quartz crystals used in wrist watch time bases are cut in
such a way as to make them most stable near body temperature. The time base
acts as a thermally stable oscillator while it is on a person's wrist. Stopwatches,
on the other hand, use quartz crystals that have been cut to be stable at or near
room temperature. If they are used in very hot or very cold environments, the
6
time base can vary by over 20 parts in 1 . Figure 24 shows the accuracy of a
typical stopwatch over a temperature range of 0° C to 50° C (32° F to 122° F).
As can be seen for the figure, the flat portion of the performance curve is near
room temperature (23° C or 73° F), and drops off as the temperature increases
or decreases.
Stopwatch Frequency Offset Versus Temperature
<
-5
-10
\
\
-15 -V—
-20
R oom Temperature (23° C)

-25
10 20 30 40 50
Temperature (° C)
Figure 24. Stopwatch accuracy versus temperature.
63
Appendix A
Appendix A
Calibration Certificates
If a calibration certificate is issued for a stopwatch or timer, it should contain the

information listed in Section 5. 10 of ISO/IEC 1 7025 [4]. Some of the necessary
elements are the name and address of the laboratory, the name and address of the
client, the identification of the method used, and a description of the item being
calibrated. A sample calibration certificate that is compliant with ISO/IEC 1 7025
is shown in Figure Al.
XXX CALIBRATION LABORATORY CERTIFICATE

XXX dlitmuioo UtvniUTv, Cif>. Slats. Z»p
CtfJMUTlON
Fsk: 4782!
TIMER-CLOCK
Manufacturer. Fisher Scientific

Model Number I4.64X-J
Serial Number 1X1456*.
Submitted In Timing i'esi Labs.

12 3 TcaUbLmx
Amlimn. I sA
\> ! mind Condition: m Tolerance

As left Condition' Idl At Found
Certified N*< vcrotv? IS. 2002
This unit was unled using the I ahoratnn V Sli<fmatch Tinier C»ltbni<i»<« Sun ton. MMS f
^4i#»
#LAB 991 2 The calibration data imbcaies a devialum from ifw reftusnee
utilizing procedure
seconds per da> (• -). tolerance oflhe Device t "ntler Test, and Fvpsnded ? 'memttty
Kiandiird in
of Measurement tK 2, tef resenting apprimmiikfv 95»« confidence).
( ttUbntti<m n».tu
Measured V ahie Tolerance l-.vpjnded I nccrtainty

TL29*d S.64 vd 0S64sd
Calibration fiocedare; »apwat£b.Timcr Version L03-' 1 1, IW

Laboratory Tetapeiature 23.0 c tie c
Laboratory Ilinindily 40,(1 **- 5 ,0«n
Certificate Number, 146316
Itteftrtogtss: Calibration M. Tcefwiciaii Approved Iw J M.
K&mtpk Cal Cert Page I of 2
Figure Al. Sample calibration certificate, (page 1)
65
\\\ Calibration 1 1 burn lor?

C tty. Stale
lile: -17X21
Dale: \ov ember IS 2009
CgggCgl If atvabilHv Ma untrue V jiucs and I be associated uncertainties supplied b> the* XXX
Calihraaoti I jlMii,iiti> me iiiivcahlc to one 01 m«»c Offfcc loU-ncma
1, Tine val*iv» of the units ((j Ihcr lv*sc <>t ckn\«.dt i*n.nnt J-ncd and divw-rntnatvd b> the N.itMirul
lif WlCt <»t nikd SUiv» <4 Aidokj) »n. m spcvial easts jnd
ManiLiidv and icthnologY (I vsheiv
appti-pN.iic. In the National Standards I ablators ,>| amrthcT n.ifi-m
2 The accented valucis) of fimdamcnlal physical phenomena (infringe nandar*):
3. K3!KK*i<<roih^no»>-maint3incd*undard»crtaNishcdb> cither a Kll-t»Uhrauoo »nd <t a direct
ealilnalion UchnKiuv:
i. Standards inainlaiacd and disseminated by the \.V\ Calibration laboratory in special cases and
where wan anted,
5 V alues and uncertainties wising from participation in a National Measurement Svttcnt
Because «»l inherent eomptvAilx in 11k <.aiinralit>n prtn.es* and the isKefUmrx wowtributM* b> rath
aundaidi and calibrabng instruments, traccabilay always toquues ev aluation or a "ira-jcabilii) bee *
V
*fmc£ah(litv tree" *nah su can he assembled lor a specific calibration and \alid (or a particular and
IgMSlfifi poinl m bine, The "iravcabtlily tree" isill include copies o( relevant ecruticates and reports,
vvcvrplcd an .ippn>|*t.»li. Im brevity. However, the >.-•*{ nt preparation nl tin "baccabilits ttce" vsill bj
.iui v.il to the requester.
?Solf. 1 This certificate or report shall not be reproduced except in lull without the advance written
approval ol the \\\ Calibration I aKrraton
Not^.2
. : Ibe as rcsci*. cd <.t-ndiUnn of the standard, set Of standards, or measurement equipment described
herein was a* expected, unless olbenvtsc noted in the tuvdy ot the ccttificaic or report
Example Cal Cert
Figure Al. Sample calibration certificate, (page 2)

Appendix B
Appendix B
References
[I] B. N. Taylor and A. Thompson, (Eds.), "The International System of Units (SI),"
Natl. Inst. Stand. Technol. Special Publication 330, 2008 Edition, March 2008.
[2] BIPM/IEC/IFCC/ISO/IUPAC/IUPAP/OIML, Guide to the Expression of

Uncertainty in Measurement (GUM), 1995.
[3] International Organization for Standardization (ISO), International Vocabulary of

Metrology - Basic and General Concepts and Associated Terms (VIM), 2008.
[4] International Organization for Standardization (ISO), General Requirements for

the Competence of Testing and Calibration Laboratories, 2005.
[5] R. J. Anderson and G. L. Harris, "Specifications and Tolerances for Field Standard
Stopwatches," Natl. Inst. Stand. Technol. Handbook 105-5, October 1997.
[6] Robbins, Michael, "Electronic Clocks and Watches," Howard W. Sams Co.,
Indianapolis, IN, 1975.
[7] J. R. Vig, "Introduction to Quartz Frequency Standards," Army Research and

Development Technical Report, SLCET-TR-92-1, October 1992
[8] North American Electric Reliability Council (NERC), "Time Error Correction,"
Standard BAL-004-0, April 2005.
[9] T. Butcher, L. Crown, L. Sebring, R. Suiter, J. Williams, editors, "Specifications,

Tolerances, and other Technical Requirements for Weighing and Measuring Devices,"
Natl. Inst. Stand. Technol. Handbook 44, 2007 edition (updated annually).
[10] M. A. Lombardi, "Legal and Technical Measurement Requirements for Time and
Frequency," Measure: The Journal of Measurement Science, 1(3): 60-69, September
2006.
[II] State of Pennsylvania code, 67 § 105.71(2), 2005.
[12] R. M. Graham, "Calibration Techniques for Stopwatches and Timers," Proc. 2003
National Conference of Standards Laboratories (NCSLI), 6 p., August 2003.
67
[13] M. A. Lombardi, "NIST Time and Frequency Services," Natl. Inst. Stand.
Technol. Special Publication 432, 71 p., January 2002.
[14] International Telecommunications Union (ITU), "One-way Transmission Time,"

ITU Recommendation G. 1 14, May 2003.
[15] International Telecommunications Union (ITU), "The Control of Jitter and

Wander within Digital Networks Which are Based on the 1544 kbit/s Hierarchy," ITU
Recommendation G.824, March 2000.
[16] D. W. Allan, D. D. Davis, J. Levine, M. A. Weiss, N. Hironaka, and D. Okayama,

"New Inexpensive Frequency Calibration Service from NIST," Proc. 1990 IEEE Freq.
Cont. Symp., pp. 107-116, May 1990.
[17] S. D. Phillips, K. R. Eberhardt, "Guidelines for Expressing the Uncertainty of

Measurement Results Containing Uncorrected Bias," Natl. Inst. Stand. Technol. J.
Res., 102(5): 577-585, September-October 1997.
[18] J. R. Vig and T. R. Meeker, "The Aging of Bulk Acoustic Wave Resonators,
Filters, and Oscillators," Proc. 1991 IEEE Freq. Cont. Symp., pp. 77-101, May 1991.
[19] IEEE, "IEEE Standard Definitions of Physical Quantities for Fundamental

Frequency and Time Metrology - Random Instabilities," IEEE Standard 1139-1999,
March 1999.
68

Nistspecialpublication960 12e2

Uploaded by

Copyright:

Available Formats

Nistspecialpublication960 12e2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nistspecialpublication960 12e2

Uploaded by

Copyright:

Available Formats

i ii mci \_anui auui id

Stopwatch and Timer

U.S. Department of Commerce

National Institute of Standards and Technology

National Institute of Standards and Technology

working metrologist or calibration technician by describing the types of stopwatches

Section 1 introduction to Stopwatch and Timer Calibrations 1

1 .A. The Units of Time Interval and Frequency 1

1 .B. A Brief Overview of Calibrations 4

Section 2: Description of Timing Devices that Require Calibration 9

Section 3: Specifications and Tolerances 17

3. A. Interpreting Manufacturer's Specifications 17

Section 4: Introduction to Calibration Methods 25

Section 5: The Direct Comparison Method 27

5. A. References for the Direct Comparison Method 27

Section 6: The Totalize Method 41

6. A. Calibration Procedure for the Totalize Method 41

Section 7: The Time Base Method 51

7. A. References for the Time Base Method 51

Section 8: How to Determine if the Calibration Method Meets the Required

Section 9: Other Topics Related to Measurement Uncertainty 59

9. A. Uncertainty Analysis of Using a Calibrated Stopwatch to

Appendix A: Calibration Certificates 65

Figure 1. The calibration and traceability hierarchy 6

Figure 2. Type I digital stopwatch 11

Figure 3. Type II mechanical stopwatch 1

Figure 4. Interior of digital (Type I) stopwatch 12

Figure 5. Inner workings of a mechanical (Type II) stopwatch or timer... 13

Figure 6. A collection of timers 14

Figure 7. Sample manufacturer's specifications for a digital stopwatch

Figure 8. Sample specifications for a digital stopwatch (Example 2) 20

Figure 9. Typical performance of quartz wristwatches using 32 768

Figure 10. Portable shortwave radio receiver for reception of audio

Figure 11. Reaction time measurements (four operators, 10 runs each)

Figure 12. Averaging measurement results for four different operators.... 3

Figure 13. Block diagram of the totalize method 41

Figure 17. Photo totalize start reading. .47

Figure 18. Photo totalize stop reading 48

Figure 19. Ambiguous photo totalize reading 49

Figure 2 1 . A time base measurement system for stopwatches and timers

Figure 22. Graph of the frequency stability of two stopwatches 61

Figure 24. Stopwatch accuracy versus temperature 63

Figure Al. Sample calibration certificate, (page 1) 65

Figure Al. Sample calibration certificate, (page 2) 66

Table 1 - Metric prefixes (may be applied to all SI units) 2

Table 2 - Unit values, dimensionless values, and percentages 4

Table 3 - Type I and Type II stopwatch characteristics 9

Table 4 - Legal metrology requirements for field standard stopwatches and

Table 5 - Comparison of calibration methods 25

Table 6 - Traceable audio time signals 27

Table 7 - The contribution of a 300 ms variation in reaction time to the

Table 8 - Uncertainty analysis for direct comparison method (digital DUT)

Table 9 - Uncertainty analysis for direct comparison method (digital DUT)

Table 10 - Uncertainty analysis for direct comparison method (analog DUT)

Table 1 1 - Uncertainty analysis for totalize method 46

Table 12 - The effect of the length of the measurement time on stability

Table 1 3 - Uncertainty analysis of using a calibrated stopwatch to calibrate

This document is a recommended practice guide for stopwatch and timer

This guide is intended to serve as a reference for the metrologist or calibration