Thermocouple Measurements

PRACTICAL
THERMOCOUPLE
THERMOMETRY
Second Edition
Thomas W. Kerlin
Mitchell Johnson
Notice
The information presented in this publication is for the general education of the reader. Because
neither the author nor the publisher has any control over the use of the information by the reader,
both the author and the publisher disclaim any and all liability of any kind arising out of such use.
The reader is expected to exercise sound professional judgment in using any of the information pre-
sented in a particular application.
Additionally, neither the author nor the publisher has investigated or considered the effect of
any patents on the ability of the reader to use any of the information in a particular application. The
reader is responsible for reviewing any possible patents that may affect any particular use of the
information presented.
Any references to commercial products in the work are cited as examples only. Neither the
author nor the publisher endorses any referenced commercial product. Any trademarks or trade-
names referenced belong to the respective owner of the mark or name. Neither the author nor the
publisher makes any representation regarding the availability of any referenced commercial prod-
uct at any time. The manufacturers instructions on use of any commercial product must be fol-
lowed at all times, even if in conflict with the information in this publication.
Copyright 2012 International Society of Automation (ISA)
All rights reserved.
Printed in the United States of America.

10 9 8 7 6 5 4 3 2
ISBN: 978-1-937560-27-0
No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or
by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior writ-
ten permission of the publisher.
ISA
67 Alexander Drive
P.O. Box 12277
Research Triangle Park, NC 27709
Library of Congress Cataloging-in-Publication Data in process.

Preface to the Second
Edition
Mitchell Johnson, President of JMS-Southeast, joined Dr. Kerlin in preparing
this second edition. He brings a wealth of knowledge about real-world
applications of thermocouples.
The descriptions of thermocouple principles, the tools needed to analyze

thermocouple performance, the causes of thermocouple errors, and the
characteristics of the commonly-used thermocouples in the 1999 edition of
this book are still as pertinent and correct as they were in 1999.
The second edition updates the book with increased coverage of topics related
to thermocouple applications. It provides new solved sample problems that
include illustrations of the use of the thermocouple loop analysis method. It
includes new or revised sections to discuss new developments and to expand
treatments of important technologies. It includes case studies of real-world
problems and their solutions.
Part of the motivation for preparing this second edition is the apparent lack of
widespread use of thermocouple loop analysis to characterize thermocouple
performance and problems. We contend that this method is an essential tool
for those who are responsible for measuring temperature with thermocouples.
One might argue that internet information now makes a book on

thermocouples unnecessary. Certainly, almost everything found in this book
can be found on the internet. However, the book eliminates the need to search
through, evaluate, and digest a huge information resource. The book is
intended as an easy-to-use reference that organizes and explains the subject in
a concise fashion and is convenient to access.
xi
Table of Contents
Preface to the Second Edition xi
Chapter 1Introduction 1
1.1 The Thermocouple 1
1.2 The Competition 2
1.3 Standards 5
1.4 Key References 6
References 6
Chapter 2Fundamentals 9
The Main Points 9
2.1 Temperature Scales 9
2.2 What Causes the Thermocouple Voltage? 11
2.3 The Seebeck Coefficient and Thermocouple Loop Analysis 12
2.4 Thermocouple Types 15
2.5 Lead Wire Effects 16
2.6 Junction Construction Effects on Thermoelectric Performance 19
2.7 The Differential Thermocouple 20
2.8 Multiple Thermocouple Circuits 23
2.9 Thermoelectric Heaters, Coolers and Generators 25
2.10 The Laws of Thermoelectric Circuits 26
References 29
Chapter 3Measuring Temperature with a

Thermocouple 31
The Main Points 31
3.1 Converting EMF to Temperature 31
3.2 Equations for EMF versus Temperature 34
3.3 Modern Thermocouple Instrumentation 36
3.4 Zone Boxes 44
3.5 Installation Effects 46
3.6 Measuring Transient Temperatures 50
References 64
Chapter 4Thermocouple Configurations 65

The Main Points 65
4.1 General Considerations 65
4.2 Make-Your-Own Thermocouples 65
4.3 Sheathed Thermocouples 66
4.4 Insulators for Sheathed Thermocouples 73
4.5 Surface Temperature Measurements 75
4.6 The Back End of a Sheathed Thermocouple 78
vii
viii Practical Thermocouple Thermometry
4.7 Thermowells and Protection Tubes 79

4.8 Special Applications of Thermocouples 85
References 88
Chapter 5Thermocouple Types 89

The Main Points 89
5.1 The ASTM Thermocouples 89
5.2 EMF versus Temperature for ASTM Thermocouples 91
5.3 Chemical Compatibilities 91
5.4 Tolerances on ASTM Thermocouples 93
5.5 Seebeck Coefficients for ASTM Thermocouples 95
5.6 Temperature Limits for ASTM Thermocouples 96
5.7 Color Coding of ASTM Thermocouples 97
5.8 Compensating Extension Wires for ASTM Thermocouples 97
5.9 High-Temperature Thermocouples 99
5.10 International Standards 100
References 104
Chapter 6Thermocouple Degradation 105

The Main Points 105
6.1 Basics 105
6.2 Decalibration Tendencies of ASTM Thermocouples 112
6.3 Insulation Resistance 114
6.4 Mechanical Problems with Thermocouples 115
6.5 Thermocouple Diagnostics 117
6.6 Self-Validating Sensors 122
References 125
Chapter 7Thermocouple Selection and Performance 127

The Main Points 127
7.1 The Factors in Selecting a Thermocouple 127
7.2 Costs 128
7.3 Thermocouple Accuracy 130
7.4 Thermocouple Durability 132
7.5 Options for Achieving Performance Requirements 133
7.6 The Future 135
References 136
Appendix AHypothetical Thermocouple Problems and

Solutions 137
Appendix BCircuit Analysis for Parallel

Thermocouples 145
Appendix CThermocouple Tables (Temperature in C) 147

Table of Contents ix
Appendix DPolynomial Equation Coefficients 155
Appendix EThermowell Selection 167
Appendix FCase Studies 171
Appendix GQuick Reference Thermocouple Guide 177
Index 179
1
Introduction
1.1 The Thermocouple

The thermocouple must surely be one of the simplest measuring devices ever
conceived. What could be simpler than two different wires joined at one end?
With this arrangement, a voltage is produced along the wires that increases in
magnitude as the temperature difference between the joined end and the open
end increases. All that is needed to determine the temperature at the junction
of the wires is to measure the voltage at the open end, make adjustments to
compensate for differences between the open-end temperature and the open-
end temperature used in calibration, and convert this compensated voltage
into temperature using the calibration for the wire types.
This approach is a proven technology for temperature measurement in

industry. Thermocouples account for more temperature measurements in U.S.
industry than any other sensor type. Thermocouples are rugged, inexpensive,
and easy to use. However, they have significant inherent inaccuracies and a
tendency to degrade with use. Users should understand these phenomena so
they can properly assess the accuracy of their measurements, select the proper
thermocouple for a given application, and install and operate the
thermocouple in the most advantageous way.
This short book focuses on the practical aspects of thermocouple

thermometry: how thermocouples work; how they go bad; how to assess
measurement accuracy; and how to select, install, and operate them. In this
book, a thermocouple will usually be shown schematically, as in Figure 1-1. In
practical applications, however, the arrangement is often as shown in Figure
1-2. In the case illustrated in Figure 1-2, the wires are contained in a metallic
sheath where the junction is formed. The wires come in three categories: base
metal (such as copper, nickel, and iron and are cheapest and most common),
refractory metal (such as tungsten and rhenium and used for very high
temperatures) and noble metals (such as platinum and rhodium and used for
high accuracy and high temperature). The open end is connected to a readout
that automatically measures the voltage, corrects for effects caused by the
temperature at the open end, and then computes and displays the
1
2 Practical Thermocouple Thermometry
Figure 1-1. Schematic View of a Thermocouple
Figure 1-2. Thermocouple in Practical Applications
temperature. This simplicity of implementation is both a blessing and a curse.

On the one hand, it is very easy to obtain a measurement: just turn the system
on and the result appears. On the other hand, this ease of use often
discourages users from expending enough time to understand what is
happening, and the unfortunate result may be undetected and unnecessary
measurement errors.
1.2 The Competition

Thermocouples are used routinely for temperature measurements ranging
from 270C to 2320C. Other sensor types are available for use over portions
of this range.1-3 Specifically, the sensors that are alternatives to thermocouples
(and their range of application) are as follows:
Introduction 3
Sensor Useful Temperature Range

Typical resistance temperature 196C to 661C
detectors (RTDs)1
Thermistors 55C to 100C
Integrated circuit sensors 55C to 150C
Resistance temperature detectors and thermistors (the latter for a narrow range of
temperatures near ambient) are the only serious competitors for use as
immersion sensors in process environments that require a sheath or
protection tube to isolate the sensor from the process.
Integrated circuit sensors are used in benign environments such as for heating,
ventilating, and air conditioning systems or as components of electronic
instrumentation systems.
The competitors to thermocouples for process measurements have different

relative advantages, mainly with respect to measures of suitability for a given
application. These measures are allowable temperature range, accuracy, and
measurement system affordability (the measurement system consists of the
three components needed to make a measurement: the sensor, wiring and
instrumentation).
For a number of years, thermocouples have been losing market share to RTDs
in total temperature sensor sales. This trend is likely to continue. RTDs have
evolved from fragile, expensive laboratory sensors to quite rugged and
inexpensive industrial sensorslargely due to improvements in the quality of
thin film RTD elementsthough they are still not as rugged as
thermocouples. RTDs have lower decalibration tendencies and lower costs for
wiring between the sensor and its transmitter or readout. Greater achievable
accuracy is an advantage for RTDs over any type of thermocouple up to
around 460oC. Beyond this temperature, RTDs still have lower limits of error
than base-metal thermocouples, but larger limits of error than noble-metal
thermocouples.
Thermocouples remain the least expensive sensor for many applications, their
accuracy and decalibration tendency are improving as the subtleties of the
underlying principles of thermocouple thermometry are understood better
and improvements arise in composition control and sensor fabrication
procedures. They are suitable for use in unusual configurations, they are
rugged, and they are able to operate at high temperatures. These advantages
guarantee that thermocouples will continue to be very important sensors for
industry.
Table 1-1 summarizes the relative advantages and disadvantages of

thermocouples and RTDs.
Table 1-1. Comparison of Thermocouples and RTDs

Thermocouple RTD
Accuracy Limits of error wider than for Limits of error smaller than
RTDs (except for noble metal base-metal thermocouples at all
thermocouples above roughly temperatures and noble metal
460C) thermocouples below roughly
460C
Ruggedness Excellent Relatively sensitive to
temperature-induced strain,
thermal or mechanical shock
and pressure
Range 270C to 2320C 196C to 661C (typical)
(somewhat lower and higher
limits in special designs)
Size Can be as small as .01" and may Limited to 1/16", temperature
be tip sensitive sensitive for length of bulb
Drift Should be checked periodically Less drift than thermocouples
for drift (typically 0.01 to 0.1C / year)
Resolution Must resolve fractions of Must resolve fractions of ohms
millivolts per degree, lower per degree, higher signal-to-
signal-to-noise ratio noise ratio
Cold Junction Required Not Required
Lead Wire Must match lead wire calibration Can use copper wire for
to thermocouple calibration extension wire
Response Can be made small enough for Thermal mass restricts time to
millisecond response time seconds in most cases
Cost Low Higher than thermocouples
Noncontact temperature sensors are also available. They provide

measurement capability that includes situations where measurements with
thermocouples are not possible. Infrared temperature sensors and optical
pyrometers can measure temperatures that far exceed those possible by
means of any contact temperature sensors. These sensors work by measuring
the electromagnetic radiation emitted from an object. They are useful for
monitoring surface temperatures. Disadvantages of non-contact sensors
include high cost, error caused by emissivity uncertainties, the inability to
take an internal temperature and the fragility of the measuring device itself.
Introduction 5
1.3 Standards
Standards serve to define the acceptable performance levels of products such
as thermocouples. In the United States, consensus standards are prepared by
professional societies and are then approved and promulgated by the
American National Standards Institute (ANSI). The American Society for
Testing and Materials (ASTM) maintains Committee E.20 to address the needs
of thermometry standards in the United States. The Instrument Society of
America (now renamed the International Society of Automation) previously
developed and maintained a thermocouple standard (labeled MC 96.1), but
this standard was abandoned in 1982 in favor of the ASTM standard. The
ASTM standard has not received ANSI approval, but it is the pertinent and
universally used standard for thermocouples in the U.S.
International commerce involves the movement of products across national

boundaries, and its growth has created a need for international standards to
ensure compatibility and consistency of thermocouple performance. The
International Electrotechnical Commission (IEC) serves this function by
coordinating standards activities, publishing international standards, and
maintaining its Committee WG65B to deal with thermometry. However,
different standards still exist in different industrialized countries. These
differences, especially differences in color coding, often cause confusion in
selecting appropriate thermocouples for use in systems built in countries
where standards differ from local standards. Chapter 5 provides information
about U.S. and international standards.
Thermocouple standards define the nominal performance and tolerances for

the thermocouples used in most industrial applications. The tolerances are
chosen by defining products that are adequate in most applications but do not
require unrealistically costly manufacturing processes. In their purchase
specifications, purchasers of thermocouples often cite standards as minimum
performance requirements.
Standards serve a crucial role in industrial temperature measurement. They

greatly facilitate sensor replacement and interchangeability and the
assessment of thermocouple performance.
1.4 Key References

Many useful books are available that provide information on thermocouples,
their principle of operation, their construction, their degradation in use, and
their selection.4-25
References
1. Ball, K. E., Thermocouples and RTDs: The Controversy Continues,
InTech, Vol. 33, August 1986, pp. 4345.
2. Smith, J., Matching Temperature Sensors with Process Tasks,
Instrumentation and Control Systems, Vol. 67, April 1994, pp. 7782.
3. Waterbury, R. C., Hot Issue: RTDs vs. Thermocouples, InTech, Vol.
41, March 1994, pp. 4447.
4. The Theory and Properties of Thermocouple Elements, American Society
for Testing and Materials publication STP 492.
5. The Use of Thermocouples in Temperature Measurement, American Society
for Testing and Materials, ASTM 470B Fourth Edition, 1993.
6. Benedict, R. P., Fundamentals of Temperature, Pressure, and Flow
Measurements, John Wiley & Sons, New York, 1969.
7. Burns, G. W. and Scroger, M. G., The Calibration of Thermocouples and
Thermocouple Materials, NIST Special Publication 250-35, April 1989.
8. Burns, G. W., Temperature-Electromotive Force Reference Functions and
Tables for the Letter-Designated Thermocouple Types Based on the ITS-90,
National Institute of Standards and Technology publication NIST
Monograph 175, Superintendent of Documents, U. S. Government
Printing Office, Washington, DC, 1993.
9. Ripple, D.C. and Burns, G.W., Standard Reference Material 1749: Au/Pt
Thermocouple Thermometer, NIST Special Publication 260-134, March
2002.
10. Garrity, K., Ripple, D. C. et al., A Regional Comparison of Calibration
Results for Type K Wire from 100 C to 1100 C, TEMPMEKO, Vol. 29, Issue
5, pp.18281837, 5 June 2008.
11. Kinzie, P. A., Thermocouple Temperature Measurement, John Wiley &
Sons, New York, 1973.
12. Kerlin, T. W., and Shepard, R. L., Industrial Temperature Measurement,
ISA, Research Triangle Park, NC, 1982.
Introduction 7
13. Magison, E. C., Temperature Measurement in Industry, ISA, Research

Triangle Park, NC, 1990.
14. McGee, T. D., Principles and Methods of Temperature Measurement, John
Wiley & Sons, New York, 1988.
15. Michalski, L., Eckersdorf, K., and McGhee, J., Temperature
Measurement, John Wiley & Sons, New York, 1991.
16. Nicholas, J. V., and White, D. R., Traceable Temperatures, John Wiley &
17. Nicholas, J. V., and White, D. R., Traceable Temperatures, New Zealand
Department of Scientific and Industrial Research, DS.R Bulletin 234,
1982.
18. Pollock, D. D., Thermoelectricity: Theory, Thermometry, Tool, American
Society for Testing and Materials Special Technical Publication 852,
1985.
19. Pollock, D. D., Thermocouples: Theory and Properties, CRC Press, Boca
Raton, FL, 1991.
20. Quinn, T. J., Temperature, Academic Press, New York, 1983.
21. Schooley, James F., Thermometry, CRC Press, Boca Raton, FL, 1986.
22. Bentley, R. E., Handbook of Temperature Measurement, Vol. 3 Springer,
1998.
23. Liptak, B. G., Temperature Measurement, CRC Press, 1993.
24. McMillan, G. K. Advanced Temperature Measurement & Control, ISA, 2nd
Ed., 2010.
25. Kerlin, T. W., and Johnson, M. P., Thermocouples: What One Needs
To Know, InTech, Vol. 58, Sept/Oct. 2011, pp. 5253.
2
Fundamentals
The Main Points

Voltage is not produced at the junction of the thermocouple wires.
Voltage is produced along the portions of the thermocouple wires

that experience temperature differences.
Voltage for an ideal thermocouple is related to the temperature dif-

ference between the junction end and the open end.
Thermocouple loop analysis is simple and can explain all the

important phenomena in thermocouples related to temperature
measurement. Even casual users of thermocouples will benefit by
understanding and using this simple analysis method.
For temperature measurement, the quantity of interest is the open-

circuit voltage (OCV), that is, the voltage that occurs when there is
no current flowing.
It does not matter how thermocouple wires are joined (twisted,

welded, soldered, bolted, clamped, etc.) insofar as the thermocou-
ples temperature measuring capability is concerned.
2.1 Temperature Scales

It will be assumed that the reader knows what temperature is and why he or
she wants to measure it. Precise definitions of temperature may be based on
thermodynamics or on quantum physics.1-5 These have tremendous practical
importance to people working on defining the temperature scale or
performing high-accuracy sensor calibration, but they are usually not of much
importance in industrial temperature measurements. The user wants his or
9
her measurements to conform to a temperature scale that is universally

consistent.
The most common scale for scientific use is the Celsius scale, and for
industrial use both the Celsius and Fahrenheit scales are commonly used. The
Celsius and Fahrenheit scales are related to the Kelvin and Rankine absolute
scales, respectively.
The relationships between the scales are as follows:
F = 1.8 C + 32
C = (F - 32)/1.8
K = C + 273.15
R = F + 459.67
It is customary to refer to the temperatures as degrees C, degrees F,

degrees R, and kelvins. This special treatment of the terminology for the
Kelvin scale honors Lord Kelvins contributions to thermometry.
The temperature scales are revised periodically because scientists are

continually striving to improve the numbers used for the temperatures that
define reference thermal states. Here, a reference thermal state is defined as a
reproducible thermal condition such as a melting point for a pure material.
Scientists also strive to develop interpolations that define temperatures at
thermal states other than those that can be reproduced readily. This suggests
that there are correct temperatures, not just values that are arbitrarily
assigned (as in the creation of the Celsius and Fahrenheit scales). This
conclusion is certainly true. Temperature appears as a variable in many laws
of physics, and these variables cannot have arbitrary values. These values,
which are the object of the scientific efforts to define correct temperatures,
might be called physical temperatures but are commonly called
thermodynamic temperatures. One feature of a thermodynamic
temperature scale is that it has a zero value at some lowest possible
temperature. The Kelvin and Rankine scales have this feature.
Scientific experts meet regularly to evaluate new results in their effort to

establish correct temperatures and to prescribe procedures for conforming
to these values in industrial practice. During the twentieth century, this has
led to revised specifications about every twenty years. Through 1968, these
specifications were called the International Practical Temperature Scale and
were designated by the abbreviation IPTS followed by the last two digits of
the year of adoption. This led to IPTS-28, IPTS-48, and IPTS-68. The
terminology changed in 1990 when a new scale, called the International
Temperature Scale and designated ITS-90, was adopted.5 The difference
Fundamentals 11
between IPTS-68 and ITS-90 temperature scales is small (less than 0.4C for
temperatures below 1000C and about 0.05 percent of the Celsius temperature
above 1000C).
The obvious question is, How do these changes affect the industrial
practitioner? The answer is Very little. Thermocouples still provide the
same output when they experience the same thermal state. The small
differences in defining the scale result in small differences in the tables,
graphs, and equations used to provide thermocouple calibration information.
These differences are smaller than the uncertainties on industrial
measurements using thermocouples but are still a possible source of
confusion. In this book, the values used in all tables, graphs, and equations
will be based on ITS-90.
2.2 What Causes the Thermocouple

Voltage?
It is not necessary to undertake a detailed analysis of the physics behind the
thermoelectric voltage produced by a conductor in a temperature gradient. It
is, however, useful to have a qualitative feel for the underlying physics so the
behavior of thermocouples becomes understandable.6-7
Consider first a single conductor in a temperature gradient. The conductor

experiences an electrical potential that can be viewed as being caused by
variations in the density of free electrons in the conductor. The electrons in the
high-temperature region have a higher kinetic energy than those in the low-
temperature region. This electron diffusion causes production of a potential
difference along a wire that experiences a temperature gradient. The
magnitude of the effect depends on the composition of the conductor, its
metallurgical state, and the absolute temperature of the conductor.
One might be tempted to conclude that the existence of a potential difference

in a conductor that experiences a temperature gradient would permit the
temperature to be measured by measuring the voltage on a single wire. Not
so! The measurement of potential must be made by an instrument to which
the wire is connected. Therefore, the potential increases along one leg of the
loop and decreases by an equal amount in the other, giving a net potential of
zero at the measuring instrument.
If one wire will not work, then how about two? Consider a situation involving
two different conductors, as shown in Figure 2-1. Because of the different
tendencies of the two conductors to generate variations in free electron
densities (and therefore different tendencies to generate electrical potentials),
the two wires produce different electrical potentials. The net result is a
potential difference at the open end (where the measuring instrument is
connected). This is the basis for thermocouple thermometry. The open end is
also called the reference end of the thermocouple.
EMF
*2+
CONDUCTOR A
(43):(947&
MEASURED
2*&8:7*)
VOLTAGE
;41&9,* ^ CONDUCTOR B
(43):(947'
DISTANCE ALONG CONDUCTORS

).89&3(*&143,(43):(9478
Figure 2-1. Voltage Produced by Two Dissimilar Conductors
It should be noted that the voltage at the open end is the open-circuit voltage
(OCV). That is, it is the voltage produced in the absence of electrical current in
the thermocouple loop. If a current existed, it would reduce the differences in
free electron density that are responsible for the thermoelectric electromotive
force (emf). Consequently, the measurement of the thermoelectric emf must be
done in a way that ensures insignificantly small current flows. In a practical
sense, this means that the input impedance of the voltage-measuring
instrument must be large.
2.3 The Seebeck Coefficient and

Thermocouple Loop Analysis
A homogeneous section of a conductor that experiences a temperature T0 at
one end and a temperature T1 at the other end experiences a voltage
difference, V, between the two ends. The voltage is given by the following
equation:7-9
Fundamentals 13
V = S (T1 T0) (2-1)

where
S = the Seebeck coefficient (V/C)
The Seebeck coefficient (also called the thermoelectric power) is the

fundamental thermoelectric property related to thermocouple thermometry.
It is a physical property of a material, like its density, thermal conductivity, or
electrical resistivity. It is independent of the size and shape of the conductor
but does vary with temperature. Because of this temperature dependence, the
relation shown in Equation 2-1 is an approximation. This approximation is
adequate for the qualitative analysis of thermocouple circuits but is
inadequate for predicting the voltage that would be observed for a specific
thermocouple in a specific temperature gradient. However, for the uses to
which it is put in this bookunderstanding how various thermocouple
configurations workit is quite satisfactory.
The simple relation between voltage and temperature difference along the
conductor may be used to predict thermocouple performance, analyze
thermocouple configurations, and troubleshoot problems with thermocouple
thermometry. This procedure is called thermocouple loop analysis.7-9 The
procedure may be illustrated for the basic thermocouple shown in Figure 2-2.
The approach is simply to sum up the voltage contributions for each
homogeneous portion of the conductor. For example, if we choose to start the
summing process at the open end of conductor A, the voltage is as follows:
V = SA(T1 T0) + SB(T0 T1) (2-2)
contribution contribution
from from
conductor A conductor B
This is algebraically the same as
V = SA(T1 T0) SB(T1 T0) (2-3)
or
V = (SA SB)(T1 T0) (2-4)
Note that the difference in the Seebeck coefficients for the two conductors
appears in Equation 2-4. This always happens in thermocouple loop analysis,
and it is the property that is of practical interest in thermocouple
thermometry. It is called the relative Seebeck coefficient (between material A
and material B) and is written SAB. That is,
SAB = SA SB (2-5)
Consequently, Equation 2-4 may be written as follows:
V = SAB(T1 T0) (2-6)
This is the fundamental relation in thermocouple thermometry.
Thermocouple loop analysis provides the ability to characterize all

thermocouple configurations and the consequences of damage to any part of a
thermocouple circuit, which often accompanies typical applications.
Appendix A contains hypothetical problems and their solutions that illustrate

the use of thermocouple loop analysis for characterizing both normal and
abnormal thermocouple configurations. These examples illustrate the power
of loop analysis for understanding how thermocouples work, both as-
installed and after degradation experienced in use. Readers are encouraged to
study these examples in order to become proficient in using the loop analysis
method.
Figure 2-2. The Basic Thermocouple
An important use of thermocouple loop analysis is prediction of the voltage

contribution of segments along a thermocouple circuit. Consider again the
thermocouple circuit shown in Figure 2-2. The thermocouple consists of two
homogeneous wires operating with a temperature difference of T1 T0. At
some point along the wires, there is a location where wires experience some
other temperature, T2. Thermocouple loop analysis gives
V = SA(T2 T0) + SA(T1 T2) + SB(T2 T1) + SB(T0 T2) (2-7)

Fundamentals 15
or
V= (SA- SB )(T2 T0) + (SA- SB )(T1 T2) (2-8)
or
V = SAB(T2 T0) + SAB(T1 T2) (2-9)
That is, the voltage for a thermocouple operating between T0 and T1 is equal
to the sum of voltages from a thermocouple operating between T2 and T0 and
a thermocouple operating between T1 and T2. Stated differently
V(T3 T1) = V(T2 T0) + V(T1 T2) (2-10)
This result is often called the Law of Intermediate Temperatures in statements

of the Laws of Thermoelectricity (see Section 2.10.1).
2.4 Thermocouple Types

In principle, any two different conductors may be used to make a
thermocouple. In practice, however, only a few combinations of conductor
materials are used. Materials are chosen on the basis of the magnitude of their
relative Seebeck coefficient, chemical stability, metallurgical stability,
ductility, strength, and cost.
The data processing for converting measured voltage to temperature is

different for every different pair of conductors, so it is necessary to have a
reasonably small number of standard types to avoid complexity, cost, and
confusion.
There are nine standard types of thermocouples used in the United States. The
designations are based on the emf vs. temperature relation for the
thermocouples, not on their compositions. These types, which are given letter
designations by the American Society for Testing and Materials (ASTM), are
shown in Table 2-1 along with a specification of their main constituents.
Different thermocouple wire manufacturers use slightly different
concentrations of main constituents and may include trace materials to
achieve desired thermoelectric properties or to improve durability and
resistance to decalibration. The various manufacturers have their own trade
names for their products.
Table 2-1. ASTM Thermocouple Types

Type Principle Wire Constituents
J Iron vs. nickel-copper alloy
T Copper vs. nickel-copper alloy
K Nickel-chromium alloy vs. nickel-manganese-silicon-aluminum alloy
E Nickel-chromium alloy vs. nickel-copper alloy
N Nickel-chromium-silicon alloy vs. nickel-silicon-magnesium alloy
C Tungsten-rhenium alloy vs. tungsten-rhenium alloy
S Platinum-rhodium alloy vs. platinum
R Platinum-rhodium alloy vs. platinum
B Platinum-rhodium alloy vs. platinum-rhodium alloy
2.5 Lead Wire Effects

The thermocouples used in industry are often located far from the readout
instrumentation to which they are connected. If wires made of the
thermocouple wire materials are used all the way from the junction to the
instrument, the system is as shown in Figure 2-2 and the electrical potential is
as given by Equation 2-6. But what if different kinds of wire are used? Why
would anyone do that, and what is the consequence?
Let us first consider the situation shown in Figure 2-3. Here, identical
conductors are connected to each side of the thermocouple. Loop analysis
gives the following:
V = SC(T1 T0) + SA(T2 T1) + SB(T1 T2) + SC(T0 T1) (2-11)
or
V = SAB(T2 T1) (2-12)
Note that the potential depends on the difference in temperature at the

junction and at the temperature where the thermocouple is connected to the lead
wire. The lead wire may be any conductor so long as it is the same in both
branches. It has no effect other than to move the reference temperature
location from the instrument to the connection point of the wires. Does this
have any real significance? Yes, but mainly historical (insofar as industrial
applications are concerned).
Consider the case in which the transition to identical lead wires in each branch
is submerged in an ice bath (see Figure 2-4). In this case, T1 is 0C, and the emf
Fundamentals 17
Figure 2-3. A Thermocouple with Identical Extension Wires
is the result of the temperature difference between the junction temperature

and 0C. This is the way thermocouples were actually used before modern
readout instrumentation was developed.
Figure 2-4. Thermocouple with Ice Bath Reference
Another setup for dealing with the reference junction temperature is shown in
Figure 2-5. Thermocouple loop analysis gives the following:
V = SA(T2 T0) + SB(T1 T2) + SA(T0 T1) (2-13)
or
Figure 2-5. An Alternate Ice Bath Reference
V = SA(T2 T1) + SB(T1 T2) (2-14)
or
V = SAB(T2 T1) (2-15)
This result shows that if the junction between A and B in the lower leg is
placed in ice water, then the result is again referenced to 0C. This analysis not
only shows an alternate way to use an ice bath to establish the reference
temperature; it also shows how the simple thermocouple loop analysis
procedure can be used to understand how a configuration will work.
Now, let us consider the configuration shown in Figure 2-6. In this case, wires
with Seebeck coefficients A and B are used to connect the thermocouple
wires to the readout. Thermocouple loop analysis gives the following:
V = SA(T1 T0) + SA(T2 T1) + SB(T1 T2) + SB(T0 T1) (2-16)
or
V = SAB(T1 T0) + SAB(T2 T1) (2-17)
Now, if the wire pair AB is chosen so as to have approximately the same

relative Seebeck coefficient as AB, we obtain the following:
Fundamentals 19
Figure 2-6. Thermocouple with Compensating Lead Wires
SAB ~ SAB (2-18)
and
V ~ SAB(T2 T0) (2-19)
This causes the reference junction to move to the point where the wires
connect to the readout instrument (just as if wires A and B are used
throughout).
Wires that have a relative Seebeck coefficient that is approximately the same
as the relative Seebeck coefficient of the wires to which they are attached are
called thermocouple extension grade wires. They are cheaper than
thermocouple wire, and they introduce little error.
2.6 Junction Construction Effects on

Thermoelectric Performance
All of the depictions of thermocouples in previous sections have showed the
two thermocouple wires joined at the junction, but there was no mention of
how they were joined. Were they twisted together, welded, soldered, bolted,
clampedor what? Thermoelectrically, it does not matter! Let us turn to
thermocouple loop analysis to understand why this is so. Consider the
configuration shown in Figure 2-7. Material C represents any material present
because of the joining operation. Loop analysis gives the following:
V = SA(T1 T0) + SC(T1 T1) + SB(T0 T1) (2-20)

or
V = SAB(T1 T0) (2-21)
That is, the material at the junction has no effect on temperature measurement
if the temperature is the same at both of the points where it meets the
thermocouple material. The main issue in junction construction is obtaining a
junction that is rugged and durable. The procedures for constructing junctions
are discussed in Section 4.2.
Figure 2-7. A Thermocouple with a Third Material at the Junction
2.7 The Differential Thermocouple

In some applications, it is more useful to know the temperature difference
between two points than to know the temperature at each of these points. For
example, in some material processing operations, uniformity of temperature
in a batch of material is important. Also, in performing energy balances,
temperature differences between different points in the process are important.
The differential thermocouple may be useful for applications such as these.
The basic differential thermocouple is shown schematically in Figure 2-8.

Note that two identical leads are bridged by another conductor (material B)
operating between temperatures T1 and T2. The loop analysis for this
arrangement gives the following:
Fundamentals 21
V = SA(T1 T0) + SB(T2 T1) + SA(T0 T2)

= SA(T1 T2) + SB(T1 T2)
= SAB(T1 T2) (2-22)
This shows that the voltage is a function of the temperature difference

between the two points where different conductors connect.
Figure 2-8. Basic Differential Thermocouple
Two thermocouples can be configured for differential measurements as

shown in Figure 2-9. A standard thermocouple readout cannot be used to
obtain the temperature difference directly. Also, since the relative Seebeck
coefficient is temperature dependent, one cannot simply use a tabulated value
of the Seebeck coefficient to obtain the temperature difference. One possible
approach is as follows:
1. Measure T1 and T2 individually.

2. Measure the emf, V, for the sensors connected in the differential
configuration.
3. Estimate an effective Seebeck coefficient using
S AB V ( T 1 T 2 ) (2-23)
4. Use this Seebeck coefficient in subsequent measurements of T1 T2

with the sensors connected in the differential configuration.
This procedure may seem to violate the basic premise of a differential

measurement: a direct measurement of temperature differences is preferable
to subtracting temperatures from two separate measurements. However, if we
use the procedure, small changes in the temperature difference are detectable
&
; '
&
Figure 2-9. Two Thermocouples Configured for Differential Measurements
with good accuracy so long as the temperatures remain close to their values
when SAB was evaluated.
The differential thermocouple can also be configured with the monitored

object as the bridging component of the thermocouple circuit (see Figure 2-
10). In this case, there may be little or no information on the relative Seebeck
coefficient between the wire and the monitored object. However, the
approach just outlined may still be used.
Figure 2-10. Thermocouple Bridged by Monitored Object

Fundamentals 23
2.8 Multiple Thermocouple Circuits

Two or more thermocouples may be incorporated in a thermocouple circuit.
They may be arranged in a series or in a parallel configuration. The
consequences of these arrangements are described in the following sections.
2.8.1 Series Thermocouple Circuits
Thermocouples may be wired in a series as shown in Figure 2-11. The usual

thermocouple loop analysis procedure may be used to determine the output
of this arrangement. For N thermocouples arranged in a series, the output is N
times the output that would be obtained with a single thermocouple
operating over the same temperature difference. This configuration, called a
thermopile, may be used to obtain a larger signal than would be obtained with
the normal single thermocouple arrangement.
Figure 2-11. Thermocouples Wired in Series
2.8.2 Parallel Thermocouple Circuits
Thermocouples may be wired in parallel as shown in Figure 2-12. Because

electrical currents can flow around the loops, the standard thermocouple loop
analysis is not applicable. It can be shown (see Appendix B) that the net
output is a weighted average of the individual thermocouple outputs. The
result for N parallel thermocouples is as follows:
Figure 2-12. Thermocouples Wired in Parallel
1 1
E T = --------------------------------------------------- E 1 + --------------------------------------------------
2
- E2 + (2-24)
+ + + +
1 2 n 1 2 n
where
ET = total emf from the circuit
i = 1/Ri = electrical conductance of thermocouple I
Ei = emf from thermcouple I
Parallel thermocouples can be used to measure the average of the

temperatures at each of the parallel junctions; however, the equation above
shows that the total output is the simple arithmetic average of all the
thermocouple outputs only if all of the thermocouples have equal
conductances. Consequently, the parallel arrangement is potentially useful for
measuring average temperatures, but caution must be exercised to ensure that
there are equal conductances in each loop.
The equation also shows the consequence of shorting a thermocouple at some

point between the junction and the open end. Shorting creates two parallel
thermocouples, but the conductance in the loop created by the short is much
higher than the conductance in the other loop. Therefore, a measurement
gives the temperature at the location of the short. This is potentially very
important. Consider a case in which the insulation on thermocouple wires is
lost and the wires touch at some point behind the junction. If the
thermocouple feeds a controller, the invalid measurement would lead to an
incorrect, possibly catastrophic, control section.
Fundamentals 25
2.9 Thermoelectric Heaters, Coolers and

Generators
The focus of this book is temperature measurement with thermocouples, but
there are other important applications of thermoelectricity. Temperature
measurements with thermocouples must be made with insignificant current
flowing in the circuit. Thermoelectric circuits can also achieve heating, cooling
and electricity generation and all of these applications involve electric current
in the circuit.
Consider a circuit of dissimilar conductors containing a source of direct

current electricity as shown in Figure 2-13. This arrangement causes one
junction to heat and the other to cool. Thermoelectric coolers find application
in devices such as beverage coolers.
WIRE A
DC POWER
SUPPLY
COLD HOT
WIRE B
Figure 2-13. A Thermoelectric Heater/Cooler
Now consider a circuit of dissimilar conductors that is heated at one junction

and cooled at the other junction as shown in Figure 2-14. This arrangement
causes current to flow in the circuit, thereby providing a source of electric
power. Thermoelectric generators find application in powering low-power
devices and are being considered for large-scale applications such as using
ocean temperature gradients to produce electricity.
WIRE A
COOLED ELECTRIC HEATED

JUNCTION CURRENT JUNCTION
WIRE B
Figure 2-14. A Thermoelectric Generator
2.10 The Laws of Thermoelectric Circuits
2.10.1 The Useless Laws of Thermoelectric

Circuits
Three laws of thermoelectric circuits were formulated long ago and they came
to be considered essential knowledge required for proper use of
thermocouples. They became well known largely because of an important
book published by Robert Benedict in 1969.2 The laws (as stated by Benedict)
along with comments are as follows:
1. Law of Homogeneous Metals: A thermoelectric current cannot be

sustained in a circuit of a single homogeneous material, however
varying in cross section, by application of heat alone.
Recall that measuring temperature with a thermocouple requires

measurement of the open circuit voltage, the voltage that exists when
no current is flowing. (Electrical current does flow in thermoelectric
heaters, coolers and generators.) Referring to a current when
discussing thermocouple behavior clouds the issue and could lead to
incorrect notions about how they work. Therefore, for temperature
measurement, the law might be restated as follows:
Revised Law of Homogeneous Metals: A thermoelectric emf cannot be

created in a circuit of a single homogeneous material, however
varying in cross section, by application of a temperature difference.
As shown in Section 2.3 the thermoelectric voltage produced by two

wires, A and B, with a junction at temperature T2, and the open end at
temperature T1, is given by (SA-SB) (T2-T1) where SA and SB are the
absolute Seebeck coefficients for wire A and B. If wires A and B are
Fundamentals 27
identical, the factor containing the Seebeck coefficients is zero and

there is no voltage produced, regardless of the temperatures.
This law is trivial since it is doubtful that anyone would think that
temperature could be measured by placing a loop of homogeneous
wire into an environment whose temperature is to be determined.
2. Law of Intermediate Materials: The algebraic sum of the

thermoelectromotive forces in a circuit comprised of any number of
dissimilar materials is zero if all of the circuit is at a constant
temperature.
This statement is correct, but a slightly less ponderous statement is as

follows:
Revised Law of Intermediate Materials: The algebraic sum of the emfs

in a circuit comprised of any number of dissimilar materials is zero, if
all of the circuit is at a constant temperature.
Loop analysis shows that the emf for each segment of a thermocouple
composed of two wires, A and B, produces an emf that is proportional
to the temperature difference across the segment. Consequently, the
total emf is zero if there is no temperature difference across any of the
segments
This law is also essentially trivial.
3. Law of Successive or Intermediate Temperatures: If two dissimilar

homogeneous metals produce a thermal emf of E1 when the
junctions are at temperatures T1 and T2 and a thermal emf of E2
when the junctions are at T2 and T3, the emf generated when the
junctions are at T1 and T3 will be E1 + E2.
This statement is correct, but a more general statement applies for the
configuration that is important in practical applications. That
configuration is a series of two parallel wire segments, each composed
of dissimilar metals with the final segment terminated at a junction.
The following statement applies for that configuration:
Revised Law of Intermediate Temperatures: If a segment of two

parallel dissimilar homogeneous metals produces a thermal emf of E1
when the temperatures at the ends of the segments are T1 and T2 and a
thermal emf of E2 when the temperatures at the ends of the segment
are at T2 and T3, the emf generated when the end of the segments are at
T1 and T3 will be E1 + E2.
This result follows from loop analysis as proved in Section 2.3.
Thermocouple loop analysis, as presented in this book, eliminates the need for
the three laws that served practitioners in the past. The traditional laws
provide little help to the practitioner who strives to make accurate
temperature measurements with thermocouples. The loop analysis method is
simpler, more comprehensive and easier to remember for occasional users.
Loop analysis, unlike the Laws, explains how thermocouples work when they
are used properly and it explains the consequences of using damaged or
improperly installed thermocouples. The traditional Laws of
Thermoelectricity, even as revised above, are essentially useless and should
be forgotten and replaced by widespread reliance on loop analysis.
2.10.2 The Useful Law of Thermocouple

Thermometry
Thermocouple loop analysis provides the following concise and useful law
that replaces the traditional three Laws of Thermoelectricity in applications of
thermocouples for temperature measurement.
The Law of Thermocouple Thermometry: The emf produced by a

segment of parallel dissimilar wires that experiences a temperature
difference across the segment is proportional to the temperature
difference. The total emf produced by the total circuit is the algebraic
sum of the emfs produced by each segment between the open end and
the junction of the wires.
The constant of proportionality is called the relative Seebeck coefficient. The

relative Seebeck coefficient has the following properties:
It depends on the composition of the two wires in the segment, but

is independent of the dimension or shape of the conductors.
It may be positive or negative.
It depends on the temperature in the segment.

Fundamentals 29
References
1. The Theory and Properties of Thermocouple Elements, American Society
2. Benedict, R. P., Fundamentals of Temperature, Pressure, and Flow
Measurements, John Wiley & Sons, New York, 1969.
3. Schooley, James F., Thermometry, CRC Press, Boca Raton, FL, 1986.
4. Burns, G. W., Temperature-Electromotive Force Reference Functions and
Tables for the Letter-Designated Thermocouple Types Based on the ITS-90,
National Institute of Standards and Technology publication NIST
Monograph 175, Superintendent of Documents, U. S. Government
Printing Office, Washington, DC, 1993.
for Testing and Materials, ASTM 470B Fourth Edition 1993.
6. Reed, R. P., Thermoelectric Thermometry: A Functional Model,
Temperature: Its Measurement and Control in Science and Industry, Vol. 5,
Part 2, James F. Schooley, editor, American Institute of Physics, New
York, 1982.
9. Moffat, R. J., The Gradient Approach to Thermocouple Circuitry,
Experimental Technique, April 1984, pp. 23-25.
3
Measuring Temperature with
a Thermocouple
The Main Points

Thermocouples measure temperature differences. To obtain the
temperature at the closed end, we must know the temperature at
the open end and account for it.
Ice was used to establish the open-end temperature in early tem-

perature measurements using thermocouples.
Temperature versus thermocouple emf tables or formulas must be

based on some fixed open-end temperature. The ice point (0C) is
by far the most common.
Modern readout devices handle the open-end temperature com-

pensation automatically.
Installation effects can influence the accuracy of temperature mea-

surements.
Temperature measurements always lag behind changing process

temperatures. The speed of response of a temperature sensor
depends strongly on the conditions (especially flow rate) in the
monitored medium.
3.1 Converting EMF to Temperature

We have seen previously that the open-circuit voltage (OCV) of a
thermocouple depends on the temperature difference between the measuring-
junction end and the open end. To find the temperature at the measuring
junction, one must know the temperature at the open end and account for it.
31
The most convenient and reproducible reference temperature available is 0C.

A mixture of ice chips and water is all that is needed to hold the open end at
0C.
In early temperature measurements with thermocouples, an ice bath was used

for the reference end. (This approach is still used in calibration laboratories.) It
became standard practice to develop thermocouple calibration data for a
reference temperature of 0C. Figure 3-1 shows the emf versus measuring-
junction temperature for the standard thermocouple types for the reference
temperature of 0C. If the reference end were placed in an ice bath, this
calibration could be used to obtain the temperature of the measuring junction.
Condensed tables for all common U.S. thermocouple types are given in
Appendix C. The internet provides easy access to tables with greater
resolution. A web search for thermocouple tables provides numerous
options for obtaining high-resolution tables.
ds

:
ds

<
E

Z
^

d
Figure 3-1. Thermoelectric EMFs for Standard Thermocouples
Now, let us consider the situation in which the reference-end temperature is

not 0C but is known. If the known temperature is T1, then we can write
Measuring Temperature with a Thermocouple 33
V ( 0C T 2 ) = V ( 0C T 1 ) + V ( T 1 T 2 ) (3-1)
where
V(0CT2)= voltage produced by the thermocouple with the refer-
ence end at 0C and the measuring junction at tempera-
ture T2
V(0CT1)= voltage produced by the thermocouple with the refer-
ence end at 0C and the measuring junction at tempera-
ture T1
V(T1T2) = voltage produced by the thermocouple with the refer-
ence end at temperature T1 and the measuring junction
at temperature T2
The emf V(T1T2) is what is measured. The emf V(0CT1) is what must be
added to the measured emf to obtain the emf that would have been measured
if the reference end had been at 0C. After this addition is done, standard
calibrations based on a 0C reference temperature can be used.
Let us use an example to clarify this procedure.
EXAMPLE
A Type N thermocouple produces an emf of 10.610 mV when the open-end

temperature is 20C. What is the measuring-junction temperature?
SOLUTION
According to Appendix C, V(0C20C) is 0.525 mV.
Therefore,
V(0CT2) = 0.525 + 10.610 = 11.135 mV
This is the emf that would have been measured if the reference temperature
had been 0C. Again, using Appendix C, we find that T2 = 350C.
Another example further illustrates the use of Equation 3-1.
EXAMPLE
A Type J thermocouple is connected to copper wires that connect to a readout

instrument. What voltage is produced if the junction is at 400oC and the
connection to copper is at 100oC?
SOLUTION
The copper section contributes no voltage because both conductors are

identical. The Type J segment contributes the following voltage:
V = V(400oC 100oC)
Using Equation 2-10 gives
V(400oC 100oC) = V(400oC 0oC) V(100oC 0oC)
That is, we can use the thermocouple tables (referenced to 0oC). Using the
table in Appendix C for Type J thermocouple gives
V(400oC 100oC) = 21.848 5.269 = 16.579 mv
3.2 Equations for EMF versus

Temperature
There are three ways to present the nominal calibration data for standard
thermocouple types: tabular, graphical, and analytical. Neither the graphical
nor tabular approach is well suited for use in instruments that measure
thermocouple emf and convert to temperature. For this application, it is
necessary to have an equation (or a set of equations for different temperature
ranges) to represent the relationship between emf and temperature.
In practical thermocouple measurements, it is useful to have equations for

temperature as a function of voltage and for voltage as a function of
temperature. Consider the first example in the previous section. The first step
is an evaluation of the voltage that would have occurred if the open end were
at 0C and the measuring junction were at 20C. This step requires a relation
for voltage as a function of temperature. (We used a table in the example, but
an equation would be needed for automatic readout systems.) The next step is
to evaluate the temperature that corresponds to the voltage obtained by
adding the measured voltage and the voltage from the previous step. This
requires a relation that gives temperature as a function of voltage.
If the relationship were linear (the curve representing voltage versus

temperature is assumed to be a straight line), the equations would be as
follows:
T = a 0 + a 1V (3-2)
or
V = b0 + b1T (3-3)
where
T = temperature
V = thermocouple voltage (corrected for a 0C reference
temperature)
a0, a1, b0, b1 = constants
Unfortunately, the emf versus temperature relationships for thermocouples

are not linear. The linear approximation is useful only for making rough
estimates or for portions of the whole range of the thermocouple over which
the relationship is nearly linear.
If the nonlinearity is to be handled explicitly by an equation, the usual form is

as follows:
T = a0 + a1V + a2V2 + + anVn (3-4)
or
V + b0 + b1T + b2T2 + + bnTn (3-5)
The terms raised to the second and higher powers account for the curvature of
the relations. The highest power, n, is called the order of the equation. It has
been found that the equation order must be high (n = 5 to 14, depending on
thermocouple type) to accomplish adequately the conversion from emf to
temperature or temperature to emf in standard thermocouples for a wide
range of temperatures.1-3 Polynomials for the standard U.S. thermocouples
are shown in Appendix D.
Lower-order (even linear) polynomials are adequate over a narrow range.

Since open-end compensation usually involves ambient temperatures of 0C
to 40C, linear equations for emf as a function of temperature are often used,
and this causes little error for this application.
The form of Equation 3-4 results in some very small coefficients being
multiplied by factors (powers of V or T) that are very large numbers.
Therefore, it is necessary to process some very large numbers and some very
small numbers. This is handled adequately with the precision available in
modern computers, but numerical errors are possible in calculations with
lower precision. A way to improve the situation is to use the nested form of
the general equation. Equation 3-4 can be rewritten to accomplish this. Taking
the fifth order case as an example, we obtain
a1 a2 a3 a4 a5
T = a 0 1 + ----- V 1 + ----- V 1 + ----- V 1 + ----- V 1 + ----- V (3-6)
a0 a1 a2 a3 a4
3.3 Modern Thermocouple

Instrumentation
Modern thermocouple instrumentation operates as follows:2, 4-8
Measure the thermocouple emf, V(T1T2).
Measure the temperature of the reference end, T1. (This must be

done with an auxiliary temperature sensor.)
Calculate the emf, V(0CT1), that is, the emf that would be pro-
duced by the thermocouple if the measuring junction were at T1
and the reference end were at 0C. An emf-versus-temperature
equation may be used for this.
Add V(0CT1) and V(T1T2). This gives the emf, V(0CT2),

which would have been measured if the measuring junction was at
T2 and the open end was at 0C.
Calculate the temperatures corresponding to V(0CT2). A tem-

perature-versus-emf equation may be used for this.
The reader may ask, Why use a thermocouple at all if it is necessary to use a
totally different temperature sensor in the instrumentation? The answer is
that the thermocouple and the reference temperature sensor have different
requirements. The thermocouple must operate over a wide temperature range
(possibly at quite a high temperature) and be rugged enough to tolerate harsh
industrial environments. The reference temperature sensor must operate only
over a narrow range near ambient, and it operates in a much more benign
environment. The sensors used for reference temperature measurements are
resistance thermometers, thermistors, and integrated circuit sensors.
The open-end compensation may be done electronically or computationally.

In the electronic approach, the auxiliary sensor is configured so as to add a
voltage to the thermoelectric emf of the thermocouple. The circuit is designed
so that the added voltage is the same as would have been produced by a
thermocouple operating between 0C and the actual temperature at the point

where the thermocouple emf is measured. Figure 3-2 shows an arrangement
that involves a resistance thermometer or a thermistor in a Wheatstone
bridge. The fixed resistors in the bridge are chosen to give the appropriate
voltage-versus-temperature relation for the thermocouple type that is to be
connected to the readout. Figure 3-3 shows an arrangement for computational
compensation for the open-end temperature. It uses a resistance thermometer,
thermistor, or integrated circuit sensor to provide a signal that is sampled by
an analog-to-digital converter. The thermocouple emf is likewise sampled by
an analog-to-digital converter. In the logic processor, the reference
temperature is determined, the emf (V(0CT1)) is calculated and added to
the thermocouple emf, and the temperature corresponding to this emf is
calculated and output to a display or other device.
Figure 3-2. Electronic Reference Junction Compensation
It has been argued that the open-circuit voltage (OCV) is the output of interest
for a thermocouple. That is, there should be no current flow in a thermocouple
circuit. However, voltage measurements in thermocouple instrumentation
involve measuring the voltage drop across a fixed resistor in the instrument.
This means that a nonzero current must flow through the resistor. To
approximate open-circuit conditions adequately, the input resistance must be
large, which results in a very small current.
In industrial applications, the cold junction compensation and the associated

signal processing is handled by indicators (usually with digital displays),
transmitters, loggers, controllers, recorders or Universal Serial Bus devices
that send the temperature measurement result to a computer.
Figure 3-3. Computational Reference Junction Compensation
A typical indicator is shown in Figure 3-4. They may be designed for bench-
top use or for rack mounting. Some are designed for one specific
thermocouple type, but most modern instruments will accommodate all
ASTM approved thermocouple types. Many instruments will accommodate
either thermocouples or resistance thermometers. Typical achievable
accuracies (in measuring the OCV of the thermocouple, performing the
reference junction compensation, and converting to temperature) are a
fraction of a degree at low temperature to several degrees at high
temperature. Users should consult manufacturers specifications if the
instrumentation accuracy is needed.
Typical transmitter designs are shown in Figure 3-5. A transmitter measures

the thermocouple voltage, applies the open-end compensation, and provides
an analog or digital output. An analog output is usually a 4-to-20-mA or 10-to-
50-mA current signal that is proportional to the temperature. A digital output
is usually a binary-coded decimal (BCD) output that corresponds to the
temperature. Transmitters are available for a single thermocouple type or for
multiple types (with software for selecting the appropriate type).
Communication protocols also exist to assist with device management as add-
ons to the 420mA signal (e.g., HART) or as digital substitutes for the 420
mA signal (e.g., FOUNDATION Fieldbus, Profibus).
Isolated transmitters are recommended for thermocouple applications as they

provide a safeguard against common mode noise. Common mode noise
results from ground-to-ground potential differences between a ground in the
thermocouple circuit and a ground at some other point or points in the plant.
If a grounded thermocouple is not properly isolated and is installed into a
Figure 3-4. Typical Indicator
system which has a different ground potential, then voltage which is

unrelated to the thermocouples emf can be introduced into the thermocouple
loop, thereby overwhelming the thermocouples signal. In many cases,
common mode noise can also be avoided in the thermocouple circuit by using
shielded thermocouple wire with the shield grounded at the thermocouple so
that the shield maintains the same ground potential as the thermocouple, thus
avoiding common mode noise.
In recent years, wireless means of transmitting data have emerged, permitting

operators to monitor process variables remotely without the need for
extension wire or conduit. A typical wireless transmitter is shown in
Figure 3-5. Typical Transmitters
Figure 3-6. These systems either operate as a point to point communication

between the receiver and the sensor-transmitter or in a mesh fashion with a
sensors transmitter acting as both a transmitter and repeater. Both styles of
wireless transmitter are extremely well suited for monitoring in applications
where line-of-sight clearances permit effective communication. Because of the
possibility of signal interruption, they are less well suited for applications in
which a given measurement will be used to control a process.
A typical temperature logger is shown in Figure 3-7. Loggers monitor the

thermocouple, determine the temperature and save the result for subsequent
transfer to another device.
A typical recorder is shown in Figure 3-8. Recorders provide graphical

records of temperature data.
A typical temperature controller is shown in Figure 3-9. Controllers operate

like an indicator, but they have the added capability of providing a control
signal output that is a function of the difference between the measured
temperature and a desired temperature (a set point) that is dialed or punched
Figure 3-6. Typical Wireless Transmitter
into the controller. Controllers may have proportional, integral, and/or

derivative control action.
A typical USB device is shown in Figure 3-10. It measures the temperature

and transmits the result to a computer. It may have the capability to handle
multiple thermocouples. It operates like a transmitter, with digital signals
transmitted to the computer. Depending on the computer software that is
employed, the computer may operate as an indicator, logger, or controller.
Figure 3-11 shows a typical hand-held calibrator and a typical sensor

calibrator with heat block.
Figure 3-7. Typical Logger
It is important to check the accuracy of thermocouple instrumentation and, if

an adjustment capability is included in their design, to recalibrate them when
necessary. Special calibrators are manufactured for this purpose. The hand
held calibrator provides the voltage that would be provided by a
thermocouple (of a specific type) that operated between a specified
temperature and the temperature of the calibrator.
A sensor calibrator allows the user to compare the signal provided by the
probe under test to a standard probe of known accuracy (often NIST
traceable). However, thermocouple calibration should only be performed on
new thermocouples. Attempts to recalibrate thermocouples that have been
used in a process are inadvisable. Decalibration in use usually is caused by
development of inhomogeneities residing in a temperature gradient. The
process conditions that cause measurement errors are not duplicated in a
calibration facility.
Figure 3-8. Typical Paper Recorder
Figure 3-9. Typical Controller

Figure 3-10. Typical USB Device
Figure 3-11. Calibration Equipment
3.4 Zone Boxes

It is sometimes desirable to minimize the wiring from a group of
thermocouples installed in a process. A zone box may be used for this purpose
(though they are not common in U.S. industrial practice).1 The configuration
is shown in Figure 3-12. All of the thermocouples are terminated in an
insulated box. The transition is to copper wire. Switches are used in the
copper portion within the zone box to select specific thermocouples for
measurement. The insulation ensures that the temperature in the zone box is
uniform and slowly varying, but the temperature of the zone box is not meant
to remain isothermal. Instead, the temperature of the zone box is monitored
with a separate thermocouple. As seen in Figure 3-10, four wires must be used
between the zone box and the reference temperature region. Two wires are
copper, and two are thermocouple material. Of course, additional wiring will
be required to provide the signals that are needed to actuate the switches.
T1
C A
C B T2
T0
C A
C C B T2
C A
C C B T2
C A
V T2
C C B
ZONE BOX
Figure 3-12. Zone Box
Thermocouple loop analysis may be used to determine the OCV measured in

this configuration:
V = Sc ( T1 T0 ) + SA ( T2 T1 ) + SB ( T1 T2 ) + Sc ( T0 T1 )
+ SA ( T1 T0 ) + SB ( T0 T1 )
(3-7)
= SA ( T2 T0 ) + SB ( T0 T2 )
= S AB ( T 2 T 0 )
That is, the voltage is the same as a thermocouple (consisting of wires A and
B) operating between T0 and T2.
3.5 Installation Effects

In considering the accuracy of temperature measurement, the usual focus is
on the accuracy of the transduction from the measurable output to
temperature. However, the measurement can never be more accurate than the
difference between the temperature of the monitored process or object and the
temperature of the sensing element itself. Such differences arise if heat
transfer effects cause the sensor temperature to differ from the temperature
being monitored. The five main effects that merit consideration in
thermocouple thermometry are discussed in the following five sections.5
3.5.1 Stem Losses
The casing in a metal-sheathed sensor is a good conductor of heat.

Furthermore, the thermocouple wires can carry heat axially along the sensor,
and, if metallic thermowells or protection tubes are used, they also provide a
good path for axial heat conduction. This can affect temperature measurement
if the back end of the sensor is hotter or colder than the tip (the usual case).
For example, if the monitored temperature in a container is high and the back
end of the sensor is located in a cool region outside the container, heat will
flow from the tip to the back end. This will cool the tip and make the
temperature lower than the monitored temperature. Heat transfer will always
occur to some extent in this common situation where the tip and the back end
are at different temperatures.
The question is, does heat transfer along the sensor alter the temperature
measurement significantly? The answer can be found by comparing the radial
heat transfer between the sensor and the process with the axial heat transfer
between the sensor tip and its back end. The radial heat transfer is the
mechanism by which the sensor and the monitored medium achieve the same
temperature, and the axial heat transfer is the mechanism by which the sensor
assumes a different temperature than the monitored temperature.
Consequently, this problem, called the stem loss effect, can be reduced by the
following actions:
Reduce axial heat transfer

1. Use a longer sensor (longer heat transfer path)
2. Insulate the back end
Increase radial heat transfer
1. Use a sensor with a smaller diameter (shorter heat transfer path)
It should also be noted that process conditions have an effect on stem loss.
Processes with poor sensor-to-fluid heat transfer will suffer larger stem loss
errors than processes with good sensor-to-fluid heat transfer.
The axial heat conduction equation can provide a theoretical estimate of the
error due to stem losses. A relationship may be derived by approximating the
sensor as a hollow cylinder immersed in a fluid. This is a crude representation
of a real sensor, but it does provide a means to assess approximate stem loss
errors. The resulting formula is as follows:5
2F
E = ---------------------------------------------------------------- (3-8)
L L
( ( 1 + F )e ( 1 F )e )
where
E = error (percent of difference between tip temperature and back-
end temperature)
e = base of the natural logarithm (equal to 2.718)
= (2roh/(k(ro2 ri2)))1/2
L = sensor insertion depth (cm)
h = surface heat transfer coefficient (watts.cm2 oC)
k = thermal conductivity of sheath material (watts.cm oC)
ro = sheath outer radius
ri = sheath inner radius
F = k/h
Consider the following example:
Problem: A sheathed thermocouple is installed in a process where the tip

temperature is 300 oC higher than the back end. The sheath inner radius is
0.3175 cm (0.125 inches) and the outer radius is 0.635 cm (0.25 inches). The
thermal conductivity of the sheath material is 0.5 watts/cm oC
(28.9 BTU/hr ft oF). Estimate the stem loss error for surface heat transfer
coefficients of 0.5 watts/cm2 oC (typical of moving water) and 0.005 (typical of
moving air) and for thermocouple lengths of 1 to 20 cm. Note that the heat
transfer coefficient is 100 times smaller for the case representing immersion in
air.
Solution: The solution is shown in Figure 3-13. Equation 3-8 is only an

approximation, but its use shows the strong dependence of stem loss error on
the type and condition of the fluid into which the sensor is immersed. For
example, the predicted error for this example essentially disappears for
sensors longer than around 4 centimeters in the water case. In the air case,
very large errors occur for shorter immersion depths, and errors of at least
10 degrees even persist at 10 centimeters.
350
Error (degrees C)
300
250
200 h=0.5
150 h=0.005
100
50
0
0 10 20 30
Immersion Depth (cm.)
Figure 3-13. Stem Loss Errors for Example Problem
Another type of measurement error caused by heat transfer to or from the

sensing element occurs in surface-mounted thermocouples. This problem,
called thermal shunting, is discussed in Section 4.5.6.
Sometimes, it is desirable to use testing to determine whether a stem loss

problem exists. If an installation permits the sensor to be inserted into the
monitored process to different depths, it is possible to evaluate the depth of
insertion necessary to avoid stem loss errors. The sensor is inserted at
progressively greater depths in the process (which must stay at the same
temperature during the testing). At each insertion depth, the sensor is allowed
to remain stationary long enough to reach a constant temperature, and this
temperature is recorded. Stem losses are negligible for insertion depths
beyond those at which the indicated temperatures are the same. The sensor
should be permanently installed at a somewhat greater depth than the
minimum required to avoid stem losses. Once this depth is determined for
one sensor, it can also be used for other similar sensors installed at other
places in the same process so long as the process temperature and flow
conditions are similar.
3.5.2 Radiative Exchanges
Sensors can transfer heat to or from bodies by radiation through transparent

media. If the sensor can see an object that is significantly hotter or colder
than the medium whose temperature is being monitored, then measurement
errors can occur. In this case, the issue is heat transfer between the sensor and
the monitored medium versus heat transfer between the sensor and the hot or
cold region by radiation.
The radiative exchange problem can be reduced in the following ways:
Reduce or eliminate radiative exchange

1. Locate the sensor where it cannot see the hotter or colder region
2. Place a barrier or shield between the sensor and the hotter or
colder region
Increase radial heat transfer
1. Use a smaller-diameter sensor (to enhance the process-to-sensor
heat transfer relative to the radiative heat transfer)
Also, process conditions influence the radiative exchange problem. Increasing

the flow rate of the monitored fluid will improve the heat transfer between the
sensor and the monitored fluid and will reduce the radiative exchange
problem.
3.5.3 Kinetic Energy Dissipation
When a moving fluid strikes a stationary object such as a sensor, a portion of

the fluid is slowed down, causing the kinetic energy to transform into heat
energy. This heating is a significant problem only at very high velocities (on
the order of Mach 1 and higher). Since analysis procedures for evaluating this
problem are very uncertain, it is usually preferable to design to avoid the
problem. This involves using shields or wells to protect the sensor from direct
exposure to the high-velocity fluid. The shield or well will experience the
direct heating rather than the sensor. Of course, the presence of this hotter
body near the sensor may transfer some of this acquired heat to the sensor,
but the measurement error will be smaller than that for an unshielded sensor.
3.5.4 Internal Heating
Thermocouples used in strong nuclear radiation fields will experience heating

as a result of the absorption of the radiation by the sensor constituents. The
calculation of radiation heating is an involved procedure that generally
requires the use of radiation transport computer codes. The temperature rise
caused by this energy deposition requires a heat transfer calculation. It has
been found that it takes 30 to 100 millwatts (mW) of heat deposition to cause a
1C temperature rise for a typical sheathed thermocouple installed in flowing
water. Much larger temperature rises will occur in poorer heat transfer media
(such as stagnant air).
3.5.5 Surface-Mounted Sensors

Thermocouples are sometimes mounted on the surfaces of containers
containing a fluid. The purpose of the measurement may be to measure the
surface temperature or to estimate the temperature of the contained fluid. The
rules of thumb are as follows:
To measure a representative surface temperature, install the sensor

with several inches of wire near the junction on the surface and
insulate in the same way as the surface is insulated for normal use.
This is to ensure that the surface temperature is the same as would
have occurred without the thermocouple present.
To measure a temperature that more closely approximates the
contained fluid's temperature, again install the sensor with several
inches of wire near the junction on the surface but add extra insulation
on the sensor. This forces more of the temperature drop to occur in the
insulation, thereby minimizing the temperature drop through the
container wall.
3.6 Measuring Transient Temperatures

It is sometimes necessary to monitor changing process temperatures. Some
understanding of dynamic heat transfer is needed to assess the accuracy of
these transient temperature measurements.
3.6.1 Basics
When a monitored temperature is changing, the temperature indicated by a

sensor will always lag behind. This is strictly a heat transfer effect. The
dynamic response is controlled by the resistance to heat transfer between the
process or object being monitored and the temperature-sensitive portion of
the sensor and by the heat capacity of the sensor. For example, a sensor will
have a faster response in flowing water than in stagnant water (the surface
heat transfer resistance decreases as flow increases). Also, a smaller-diameter
sensor will have a faster response than a larger-diameter sensor in the same
environment (the larger sensor has more heat capacity and, because of the
longer heat transfer path, more heat transfer resistance).
3.6.2 Bare Sensors

For bare-wire thermocouples, the thermocouple wires are in direct contact
with the monitored process or object. These sensors have fast responses
because the heat transfer resistance is small and the heat capacity is small. For
ultrafast bare-wire thermocouples, very small wire diameters are used, but
care must be taken to ensure that the small wires are adequately durable and
resistant to decalibration.
3.6.3 Sheathed Sensors
Sheathed sensors are used for temperature measurements in which the sensor
must be protected from mechanical damage or chemical attack. The sheath
and associated insulation affect the speed of response of the temperature
measurement.
3.6.3.1 The Time Constant
Time response is a concern primarily for sheathed sensors installed in a

process fluid (usually called immersion sensors). This installation may also
include a thermowell or a protection tube. In this case, the components of the
sensor (and the thermowell or protection tube, if present) contribute to the
overall heat transfer resistance and heat capacity. Also, any gas-filled gaps (as
might occur between a sensor sheath and a thermowell or protection tube)
will increase the heat transfer resistance tremendously because air and other
gases are good thermal insulators.
The most common way to specify the dynamic response characteristics of a

temperature sensor is by means of its time constant, which is defined as
follows:5
Time constant: The time required for the response to complete 63.2% of
its total change following a step change in the monitored temperature.
This definition follows from a simplified approximation of the actual response

of a sensor. This approximation, called the first order approximation, gives the
following result for the response of a sensor to a step change in the monitored
temperature:
t
T ( t ) = T ( ) [ 1 e ] (3-9)
where
T(t) = measured temperature change at time t
T() = final measured temperature change
= the time constant
The behavior of a sensor following a step change in monitored temperature

according to the first order approximation is shown in Figure 3-14.
0.9
0.8
0.7
INDICATED TEMPERATURE CHANGE
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
TIME
Figure 3-14. Step Response for a First Order System
The step response is often used to characterize a sensor's dynamic

characteristics, but step changes are rarely encountered in the real world. A
more realistic transient, the ramp, is depicted in Figure 3-15. In this case, there
is an initial transient after which the measurement tracks the true temperature
with a constant delay. For the first order approximation, the delay is
numerically equal to the time constant, .
0.9
0.8
0.7
TEMPERATURE CHANGE
0.6 ACTUAL
TEMPERATURE
INDICATED
0.5 TEMPERATURE
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
TIME
Figure 3-15. Ramp Response for a First Order System
3.6.3.2 Real Sensor Response
At this point, the reader should be asking, How good is the first order
approximation? The answer is, Pretty good, but it helps to understand
actual behavior versus first order approximation behavior. The actual
response of a typical real sensor to a step change in monitored temperature is
as shown in Figure 3-16. This shows that the actual response is somewhat S-
shaped. That is, the response starts off tangent to the time axis, rises, and then
rolls over to approach its final value. In the first order approximation, the
response jumps immediately (tangent to the response axis). Clearly, the
behavior of the first order approximation is wrong. The temperature-
measuring region in a sheathed sensor is buried in a structure, and it takes

time for heat to diffuse through this structure and change the temperature in
the temperature-measuring region. (This is what causes the real sensor
response to have a slower initial response than the first order approximation.)
0.9
0.8
0.7
INDICATED TEMPERATURE CHANGE
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
TIME
Figure 3-16. Step Response for a Real System
The correct model for representing the actual response of a sensor to a step
change in monitored temperature is as follows:5
t 1 t 2
T ( t ) = T ( ) [ 1 + a 1 e + a2 e + ] (3-10)
That is, it takes an infinite sum of exponentials to characterize the response

exactly (though the terms have diminishing influence as the index on the term
increases).
Even though the first order model is inexact, the concept of an overall time
constant is still useful. The overall time constant (the 63.2% response) is
related to the i in Equation 3-10 as follows:5

overall = 1 1 ln 1 ----2- ln 1 ----3- (3-11)
1 1
The higher terms (those involving 2, 3, ) can contribute up to about 20%

of the total value of overall .
3.6.3.3 Influence of Sensor Characteristics and Process

Conditions on Response
Thermocouples are frequently used to monitor the temperature of a fluid with

time-varying temperature. It is useful to consider how sensor design features
and the conditions in the monitored fluid affect sensor response
characteristics. In general, sensor dynamic response has an internal
component and a surface component. As a consequence, the time constant
may be represented as follows:5,10
overall = internal + external (3-12)
The internal component depends on sensor diameter and on the temperature-

dependent physical properties of the sensors constituents. The external
component depends on sensor diameter, fluid flow rate, the properties of the
sensors constituents and the properties of the specific fluid whose
temperature is to be monitored.
Quantitative assessments of the effect of the fluid flow rate and sensor
diameter and a qualitative assessment of the effect of temperature on sensor
response appear below. The formulas shown below were derived for a solid
cylindrical sensor. This only approximates a real sensor, but use of the
formulas to correlate test data shows that the formulas are effective.
Effect of Flow Rate of Fluid Surrounding the Thermocouple
The internal component of the time constant does not depend on fluid
properties or flow rate. It can be shown that the surface component of the time
constant is inversely proportional to the surface heat transfer coefficient, h.
Therefore the flow dependence (for a specific sensor diameter and

temperature) is given by the following equation:
C
= C 1 + -----2- (3-13)
h
where
C1 = a constant that contains information about sensor
dimensions and physical properties
C2 = a constant that contains information about sensor
dimensions and physical properties
h = surface heat transfer coefficient (It contains all of the
information about fluid flow rate and physical proper-
ties)
Correlations provide the surface heat transfer coefficient. The correlations

give the Nusselt number as a function of the Reynolds number and the
Prandtl number as follows:
Nu = f (Re, Pr) (3-14)
where
Nu = the Nusselt number = hD/k (dimensionless)
Re = the Reynolds number = Du/ (dimensionless)
Pr = the Prandtl number = C/k (dimensionless)
h = convective heat transfer coefficient
D = sensor diameter
u = fluid velocity
= fluid density
= fluid viscosity
C = fluid specific heat capacity
k = fluid thermal conductivity
There are numerous correlations that cover different ranges of the parameters
in the correlations. Interested readers can find additional correlations in the
heat transfer literature if needed for a specific application.
A popular correlation for flow past a cylinder is as follows:
Nu = (0.35 + 0.47 Re0.52)Pr0.3

for Reynolds numbers between 0.1 and 1000 (3-15)
Nu = 0.26 Re0.6 Pr0.3

for Reynolds numbers between 1,000 and 50,000 (3-16)
Therefore,
= C1 + C3u-0.6 (for Reynolds numbers between 1,000 and 50,000) (3-17)
where
C3 = 0.26(D/)0.6Pr0.3 (for Reynolds numbers between 1,000 and 50,000) (3-18)
The flow dependence of the time constant for a typical sensor appears in
Figure 3-17. Experiments performed in a laboratory provide a check of the
validity of Equation 3-15. Data show agreement with the behavior indicated
by Equation 3-15. Figure 3-18 shows typical results.
Effect of Sensor Diameter
The formulas for C1 and C2 for a solid cylinder are as follows:
C1 = 0.24CR2/k (3-19)
C2 = CR/2 (3-20)
where
R = sensor radius
The sensor radius (or diameter) appears in both of the constants C1 and C2. C1
accounts for internal component of the time constant and C2 along with the
heat transfer coefficient account for the surface component of the time
constant. Therefore, the dependence of C1 on sensor diameter applies for
sensors in which the internal heat transfer resistance dominates (generally
large sensors and/or sensors in stagnant or low-flow environments). The
dependence of C2 on sensor diameter applies for sensors in which the surface
heat transfer resistance dominates (generally small sensors and/or sensors in
high-flow environments).
1.1
0.9
TIME CONSTANT
0.8
0.7
0.6
0.5
0.4
0 1 2 3 4 5 6 7 8 9
FLOW RATE
Figure 3-17. Flow Dependence of the Time Constant
SENSOR NO. 1 - 25C, 1 BAR

T = 0.063 + 0.321 u -0.6
0.5
TIME CONSTANT (S)
0.4
SENSOR NO. 1 - 300C, 147 BAR
T = 0.70 + 0.318 u -0.6
0.3
0.2 SENSOR NO. 2 - 25C, 1 BAR

T = 0.037 + .0233 u -0.6
0.1
0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
[ FLOW RATE, U (M/S) ] -0.8
Figure 3-18. Correlation of the Time Constant with Flow Rate

The diameter (or radius) of the sensor appears in the definitions of C1, C2 and
h. C1 varies as the diameter squared. C2 varies as the diameter and h varies as
the diameter to the diameter raised to the 0.6 power. Therefore,
internal = C4D2 (3-21)
surface = C5D1.4 (3-22)
where C4 and C5 are constants (assuming that sensor diameter is the only
variable under consideration).
Therefore, for sensors in which internal heat transfer resistance dominates

(generally large diameter sensors and/or high fluid flow rates), the time
constant varies as the sensor diameter squared and for sensors in which the
surface heat transfer resistance dominates (generally small diameter sensors
and/or low fluid flow rates), the time constant varies as the sensor diameter
raised to the 1.4 power.
In many industrial applications, the fluid flow rate is high, thereby causing
the sensors internal heat transfer resistance to dominate. In these cases, the
time constant varies as the sensors diameter squared. For example, switching
to a smaller sensor with half the diameter would decrease the time constant by
a factor of four.
Effect of Temperature
The effect of temperature on sensor response is more difficult to assess. The

heat transfer depends on expansion and contraction of sensor internals as well
as the temperature-dependent physical properties of the sensor internals and
of the surrounding fluid. Small cracks in the insulator can grow or shrink and
the fit between a sheathed sensor and its thermowell can become tighter or
looser, thereby causing increases or decreases in heat transfer resistance.
These effects can dominate the temperature dependence of a sensors time
response.
The overall conclusion is that the temperature dependence of the response

characteristics of a sensor or a sensor-thermowell assembly is essentially
impossible to predict.
3.6.3.4 Achieving Fast Response
One approach for achieving faster response is to use smaller sensors, but this
is not always practical. In general, any design that reduces the internal heat
transfer resistance provides faster response. There are several design options
that accomplish this.
It should be noted that designs that reduce a sensors time constant also
reduce measurement errors due to stem loss. Stem losses occur when the axial
heat transfer along a sheathed sensor is significant compared to the heat
transfer between the junction and the monitored medium (see Section 3.5.1).
Of course, exposed junction thermocouples have the fastest response because

the junction is in direct contact with the monitored medium (see Figure 4-3).
The drawbacks of using exposed junction sensors are that they are vulnerable
to chemical attack, breakage and ground loops.
Sheathed, insulated junction thermocouples with a reduced diameter near the

tip shorten the heat transfer path between the monitored process and the
junction (see Figure 4-6). This design decreases the time constant, while
providing more structural strength than a sensor with smaller diameter along
its whole length.
Sheathed thermocouple time response can be improved by using end caps

made of metal with higher thermal conductivity than the typical sheath
material such as steel. Copper alloys may be used.
Grounded junction sheathed thermocouples generally have faster response

than insulated junction thermocouples with the same diameter. The common
grounded junction has wires welded directly into the end cap as shown in
Figure 4-4. Heat must flow only through a short metallic path to the
thermocouple junction. However, it is important to avoid excessive heating in
the process of welding the wires to the end cap in order to achieve the fastest
possible time response. If high heat is applied for an extended period during
welding, the wires can get hot enough to alter their metallurgical state for
some distance along the wires. This moves the effective junction away from
the end cap and increases the time constant.
Designs to improve response time for sensors in thermowells involve features

that improve the contact between the sensor surface and the thermowells
interior wall. Poor fits cause air gaps that increase heat transfer resistance
significantly.
The most common approach is to spring load the sensor into the thermowell.
Spring loading the sensor into the thermowell forces it to fit tightly. For
typical uniform-diameter sensors, spring loading forces good metal-to-metal
contact and improved heat transfer between the sensor tip and the bottom of
the thermowell. Additional improvement is achievable with a tapered, spring-
loaded sensor and a thermowell with a matching tapered hole. This approach
ensures metal-to-metal contact throughout the tapered section.
A second approach is to apply a roughened soft metal (such as gold or silver)

sleeve on the surface of the sensor. This approach has been used with
resistance thermometers in specialized applications where achieving fast
response was important enough to justify the increased cost. The idea is that
the sleeve material will conform and maintain good contact with the
thermowell wall. This approach improves sensor time response, but the
effectiveness decreases if the sensor is removed and re-inserted repeatedly,
thereby rubbing off the soft metal.
A third approach for sensors in thermowells is to insert a conductive grease or

fluid (such as oil) to fill the gap between the sensor and the thermowell. The
choice of filler material is important because it can increase response time if it
dries out; possibly creating a solid, low-conductivity residue between the
sensor and the thermowell or creating solid material at the bottom of the
thermowell that prevents full insertion of the sensor. This problem is common
and the use of filler material is generally discouraged.
A software approach to achieving faster response involves passing the signal

through a band-pass filter that amplifies the high frequency component of the
signal. The lower break frequency is set slightly below 1/ (where is the
sensor time constant) and the upper break frequency (or frequencies) is set
somewhat above 1/. This amplifies the faster component of the signal while
attenuating the very fast component (thereby reducing errors caused by
measurement noise). Figure 3-19 shows results for an unfiltered response and
a filtered response for a step change in monitored temperature. The results
shown are based on a sensor with a time constant of 2.0 seconds and a filter
with the following transfer function:
Gf = (1s+1)/((2s+1) (3s+1)) (3-23)
Figure 3-19. Response Compensation with Filters

where
Gf = filter transfer function
s = Laplace transform parameter
1 = reciprocal of the zero of the transfer function (set equal
to 3.0 in the example, giving a break frequency of 0.333
radians per second)
2 = reciprocal of the first pole of the transfer function (set
equal to 1.0 in the example, giving a break frequency of
1.0 radians per second)
3 = reciprocal of the second pole of the transfer function
(set equal to 0.5 in the example, giving a break fre-
quency of 2.0 radians per second).
For those unfamiliar with specification of a filter by means of its transfer

function, as in Equation 3-21, it is noted that this representation is standard
practice. The transfer function defines the hardware or software
implementation of the filter.
A hardware trick for achieving fast response involves using two sensors, one
of which has a faster response than the other. Using sensors with different
diameters can provide this difference in response characteristics. They are
mounted adjacent to one another, so they experience the same process
temperature. The response compensated sensor requires specialized readout
capability to process the sensor outputs as follows:
Rc = (n+1)Rf nRs (3-24)
where
n = an integer (typically a small number between 1 and 3).
Its value determines the speed and overshoot of a mea-
surement. Users can use simulation to select the best
value of n for a specific application.
Rc = response compensated measurement
Rf = response of the faster sensor
Rs = response of the slower sensor.
Figure 3-20 shows the response of a compensated sensor along with the
responses of the pair of sensors used. The example is for n=2 and for a fast
sensor time constant of 1.0 second and a slow sensor time constant of 2.0
seconds. The compensation equation for this example is as follows:
Rc = 3Rf 2Rs (3-25)

1.4
Response 1.2
1
compensated
0.8 fast
0.6
0.4
0.2 slow
0
0 5 10
Time
Figure 3-20. Hardware Response Compensation
That is, the compensated response is obtained by multiplying the fast sensors
response multiplied by three and subtracting the slow sensors response
multiplied by two.
The response of a compensated sensor system depends on the relative speed

of response of the two sensors and on the value selected for the parameter, n.
However, the method always provides a faster response than that obtained
with the fast sensor alone.
The response compensated sensor (software approach or hardware approach)

brings to temperature measurement a capability that is common to many
dynamic systems: the possibility of a trade-off between response and
overshoot. The examples in Figures 3-19 and 3-20 show that a response
compensated sensors faster early response could be used to provide quicker
indication of an unsafe or undesirable process temperature. A strong need to
have fast response would be needed to justify the added complexity
associated with the response compensated sensor. It should be noted that the
uncompensated sensor output is available as well as that of the compensated
sensor in both procedures. Therefore, the approaches permit traditional use of
the sensor (or sensors) as well as the response-enhanced output that might be
used for safety actions such as alarm, runback or shutdown.
3.6.3.5 Manufacturers Specifications of Sensor Time

Response
Manufacturers often quote time-constant values for their sensors. This value is
usually the result of a measurement with a single sensor in a single fluid
flowing at a single flow rate. From the discussion in preceding sections, it
should be clear that the time constant for a sensor installed in a process is
probably much different than the manufacturers value because the process
fluid and its flow rate may be quite different than those used in the
manufacturers tests. Differences of a factor of ten are common, and a factor of

one hundred is possible. Furthermore, manufacturers usually use water for
testing, but different manufacturers may use different flow conditions (i.e.,
flowing room temperature water, flowing heated water, boiling water).
Consequently, manufacturers values not only fail to provide in-process
performance, they also fail to provide a sound basis for comparing the
response characteristics for the sensors provided by different manufacturers.
References
for Testing and Materials, ASTM 470B Fourth Edition, 1993.
2. McGee, T. D., Principles and Methods of Temperature Measurement John
3. Burns, B. W., et al., Temperature-Electromotive Force Reference Functions
and Tables for the Letter-Designated Thermocouple Types Based on the ITS-
90, National Institute of Standards and Technology, NIST Monograph
175, 1993.
4. Bentley, J. P., Temperature Sensor Characteristics and Measurement
System Design, Journal Physics E. Science Instrumentation, 1984, pp.
430-439.
10. Danzig, J. A., Improved Transient Response of Thermocouple
Sensors, Rev. Science Instrumentation, 56(5), May 1985, pp. 723-725.
11. Pandey, D. K., Response Time Correlation for Chromel-Constantan
Thermocouples in Flowing Hot Air, Journal Physics E. Science
Instrumentation, Vol. 18, 1985, pp. 712-713.
4
Thermocouple
Configurations
The Main Points

Bare-wire thermocouples, which are usually made by the user, can
serve in applications where the monitored environment is compat-
ible with the thermocouple wire materials.
Sheathed thermocouples provide protection from contamination

by the monitored environment, but the sheath material must be
chosen for compatibility both with the monitored environment
and with the thermocouple wire materials.
4.1 General Considerations

Users have two choices: make thermocouples from thermocouple wire or
purchase thermocouples. If the choice is to make them, the thermocouple will
almost always be a bare-wire type.
4.2 Make-Your-Own Thermocouples

Simple, bare-wire thermocouples can be made easily, but the user must pay
attention to the mounting of the sensor and its compatibility with the
environment in which it will be used. Recommendations on thermocouple
types for various environmental conditions may be found in Section 5.3.
The fabrication of a thermocouple involves joining by soldering, brazing, or

welding. When the user makes a thermocouple he or she should twist the wire
two to three turns before joining. The soldering, brazing, or welding serves
two purposes: to increase the mechanical strength of the twisted junction and
65
to ensure good electrical contact between the wires. Thermocouples that are to
be used at low temperatures may be soldered (soft solder is suitable up to
about 200C, and hard solder is suitable up to about 600C). For welded
junctions, the wires should be cleaned prior to welding (abrasive cleaning
methods are often used). Some materials require the use of a flux in the
welding operation, and the heating method that is required depends on the
thermocouple type. Tungsten, tungsten/rhenium should be twisted, but not
soldered, brazed or welded. References 1 and 2 give recommended
approaches.
Since the presence of a third material (solder or braze) at the junction does not
affect the thermoelectric performance of the thermocouple (see Section 2.6),
from a thermoelectric standpoint the presence of the material added to the
junction is of no concern. Welding thermocouple wires is like any welding
operation except there is a special need to concentrate the heating on a small
region near the junction. The heating can cause metallurgical effects in the
wires. If this affects a significant length of wire, the wire will have an
inhomogeneous section at some point away from the junction in the
thermocouple wire, a section that may reside in a temperature gradient when
the thermocouple is being used. If this occurs, thermoelectric performance
will be affected. To combat this problem, the heating should be focused on the
junction, and the heating duration should be minimal. Special thermocouple
welders are commercially available that aid in achieving the proper welding
of the junction. After welding, the junction is usually buffed to obtain a small,
clean weld.
4.3 Sheathed Thermocouples

Sheathed thermocouples are all made from sheathed thermocouple stock, as
shown in Figure 4-1. The various types of sheathed thermocouples are
fabricated by sensor manufacturers using this stock as a starting point.
The insulator in the sheathed thermocouple stock is compacted metal oxide

powder. The most common insulator is magnesium oxide though other
materials such as aluminum oxide, beryllium oxide, thorium oxide, or
zirconium oxide are also used. Sheathed thermocouples with a metal oxide
insulator are called MIMS (mineral insulated, metal sheathed) thermocouples.
Most sheathed thermocouple stock is made by rolling a thin ribbon of metal

into a cylinder while simultaneously introducing insulating powder and the
thermocouple wires. The cylinder is then closed by welding the seam. The
cylinder is then drawn to reduce its diameter and compact the insulating
powder.
Thermocouple Configurations 67
Figure 4-1. Sheathed Thermocouple Stock
The integrity of the sheath is a key factor in determining the performance of

an MIMS thermocouple. The sheath is expected to isolate the thermocouple
wires from contamination by materials in the environment outside of the
sheath. If pathways exist in the sheath for contaminants to pass through,
degradation of sensor performance is possible. A metallic sheath consists of
an array of grains with distinct boundaries. The intergranular spaces provide
pathways through which contaminants can diffuse. In most cases, the seam
weld area is the part of the sheath most susceptible to providing pathways for
contaminants.
The problem of contamination through the sheath has led to greater emphasis
on designs that increase impermeability. One approach is seamless sheath
material that reduces contamination but increases cost. Another design is the
dual-layer sheath design, in which the sheath material is made of two
concentric layers as shown in Figure 4-2. The welds that close each cylinder
are placed on opposite sides since these are potential weak spots. The
microscopic effect of the layered design is to interrupt diffusion paths at the
interface between the layers.
4.3.1 Exposed Junction Thermocouples
An exposed junction thermocouple is made by cutting a desired length of

stock, removing a short section of sheath and insulator at one end, joining the
Figure 4-2. Dual-Layer Sheath Design
exposed wires at that end, and sealing or attaching a fitting at the other end. A
typical exposed junction thermocouple is shown in Figure 4-3.
Figure 4-3. Exposed Junction Thermocouple

4.3.2 Grounded Junction Thermocouple
A grounded junction thermocouple is made by cutting a desired length of

stock, welding a metal cap on one end (with the cap attached to the
thermocouple wires), and sealing or attaching a fitting to the other end. A
grounded junction thermocouple is shown in Figure 4-4.
Figure 4-4. Grounded Junction Thermocouple
4.3.3 Insulated Junction Thermocouple
The insulated junction thermocouple is made by cutting a desired length of

stock, removing the insulator over a small distance near one end, joining the
wires exposed by the removal of the insulator, backfilling the emptied region
with insulating powder, welding on an end cap, and sealing or attaching a
fitting to the other end. An insulated junction thermocouple is shown in
Figure 4-5.
Figure 4-5. Insulated Junction Thermocouple

Insulated junction thermocouples are sometimes made with reduced-

diameter sections near the tip (see Figure 4-6). This is accomplished by
swagging the end of a regular insulated junction thermocouple. This
reduction in the diameter causes the speed of response to be faster.
Figure 4-6. Reduced Tip Diameter Sheathed Thermocouple
4.3.4 Dual-Junction Thermocouples

The dual-junction thermocouple is made from four-wire stock (see Figure 4-7).
Figure 4-7. Thermocouple Four-Wire Stock

Dual-element sensors are produced in three configurations: grounded,

insulated/isolated, and insulated/common. These configurations are shown
in Figure 4-8. These sensors permit redundant measurements. The insulated/
isolated configuration permits two separate measurements. The other two
configurations permit up to four separate measurements (even though they
are not totally independent of one another). For example, for a sensor with
two wires of Type A (A1 and A2) and two wires of Type B (B1 and B2) that are
all joined together, the possible pairs for emf measurement are as follows:
A1, B1
A1, B2
A2, B1
A2, B2
The dual-element sensors have the advantage of offering multiple

measurements, and their other advantages and disadvantages are the same as
for the comparable single-element sensors.
It is possible (but uncommon) to make multiple-junction thermocouples with

four different wires, A, B, C and D. This permits six measurements as follows:
A, B
A, C
A, D
B, C
B, D
C, D
Note that degradation of any single wire affects three of the six
measurements. Sensor manufacturers have the ability to produce four-wire
sensors, but there is concern about compatibility of the sheath and the various
materials within the sheath, especially wire-to-wire chemical attacks.
4.3.5 Thermal Stress Considerations
All of the components in a sheathed thermocouple will expand and contract

as temperature increases and decreases. Since the materials have unequal
coefficients of thermal expansion, they will grow or shrink by different
amounts, and stresses can occur between the connected components. For
insulated junction thermocouples, the connections are wire-to-insulator and
insulator-to-sheath. Stress can occur as these components slide across one
Figure 4-8. Dual Element Thermocouple Configurations
another as they expand and contract. The situation is more problematical in

the case of grounded junction thermocouples. In addition to the wire-to-
insulator and insulator-to-sheath considerations, there is also the wire-to-end-
cap connection, and this is the dominant mechanism for creating stress. If the
wires and the sheath have significantly dissimilar thermal expansion
coefficients, then large stresses can occur as the temperature cycles and wire
breakage may result. Also, during a transient the sheath heats or cools faster
than the thermocouple wires since the sheath is in direct contact with the
monitored medium. Consequently, the sheath expands or contracts faster
than the wire and thereby creates stress on the wires. These stress-generating
effects are especially serious for grounded junction thermocouples.
4.3.6 Advantages and Disadvantages of the

Major Sheathed Configurations
Table 4-1 presents the advantages and disadvantages of the four major
sheathed thermocouple configurations.
Table 4-1. Advantages and Disadvantages of Standard Sheathed Thermocouple

Configurations
Thermocouple
Advantages Disadvantages
Configuration
1. Exposed junction Very fast response. Cannot tolerate high flows.
Wires subject to contamination.
Exposed insulator can absorb
liquids or gases and short
circuit.
2. Grounded junction Fast response. Wires protected Subject to wire breakage.
from environment. Least Subject to ground loops.
expensive sheathed type.
3. Insulated junction Wires protected from Slow response compared to
environment. Less subject to other types of sheathed
wire breakage than grounded thermocouples. More expensive
type. than exposed or grounded
thermocouples. Wires can be
contaminated by constituents of
sheath, insulator, or fill gas.
4. Dual-element Redundant measurement Disadvantages are the same as
capability. Other advantages are for comparable single-element
the same as for comparable thermocouples.
single-element thermocouples.
4.4 Insulators for Sheathed

Thermocouples
The insulators used in sheathed thermocouples are usually metal oxides. By
far, the most common insulator is magnesium oxide (MgO). Other less
frequently used materials are aluminum oxide (Al2O3), beryllium oxide
(BeO), thorium oxide (ThO2), and zirconium oxide (ZrO2). The physical form
is compacted powder. In the size reduction operations on the sheath,
insulator, and wire assembly, the insulating powder is compacted.
The resistivity of the most common insulator (magnesium oxide) is shown in

Figure 4-9. Two points merit mention: the resistivity is very high at low
temperature, and it decreases rapidly as temperature increases. The

temperature dependence has special relevance to a potential problem,
electrical shunting (see Section 6.3), which is caused by insufficient insulation
resistance.
15
10
RESISTIVITY (Ohm-cm)
10
10
5
10
0
10
0 500 1000 1500 2000 2500

TEMPERATURE (C)
Figure 4-9. Resistivity of Magnesium Oxide
Pure insulating materials generally are quite inert. Inertness is crucial in

thermocouple applications to avoid the migration of atoms between the
insulator and the thermocouple wires that may cause sensor decalibration (see
Chapter 6). However, insulators can become contaminated during handling
and manufacture. Moisture is readily absorbed by insulator powder. If the
insulator is exposed to high humidity during handling or manufacture,
moisture will be absorbed, and it can have a large effect on sensor
performance. Likewise, unclean conditions (possibly the contaminated hands
of workers who fabricate the thermocouples) can cause grease and other
contaminants to be absorbed by the insulator. Such problems could lead to
reliability problems with sensors from one manufacturer when virtually
identical sensors from another manufacturer perform well.
4.5 Surface Temperature Measurements

Thermocouples are sometimes mounted on surfaces in order to measure the
surface temperature. Installation details can have a significant effect on the
adequacy of measurements.
4.5.1 Introduction
Surface-mounted thermocouples may occur in three main configurations:
point measurement: surface insulated from the thermocouple

wires (see Figure 4-10).
Figure 4-10. Surface Point Measurement with Wires Insulated
point measurement: surface integral to thermocouple circuit (see

Figure 4-11).
temperature difference measurement (the differential thermocou-

ple, see Section 2.7)
Figure 4-11. Surface Point Measurement with Wires in Contact
In all surface temperature measurements, the installation should be handled

so as to minimize heat conduction along the wires or along the thermocouple
assembly (see Section 3.5).
4.5.2 Point Measurement: Surface Insulated

from Thermocouple Wires
Sheathed thermocouples must be used for surface temperature measurements
in applications where the environment at the surface being monitored is
unsuitable for bare thermocouple wires. In some cases, special configurations
such as tube skin thermocouples (see Figure 4-12) are useful. These
thermocouples may be held to the surface by a pressure fitting (possibly with
spring loading), by a clamp, or by welding them to the surface.
4.5.3 Point Measurement: Surface Integral to

Thermocouple Circuit
In this case, the bare thermocouple wires are attached directly to the surface or
to a fitting that is attached to the surface by bolting or riveting. In these cases, a
third material exists at the junction (the surface, the device used for
attachment, or both), but this material has no effect on the thermocouple
Figure 4-12. A Tube Skin Thermocouple
output if the temperatures at both transitions to thermocouple wires are the

same (see Section 2.6).
4.5.4 Temperature-Difference Measurement
For temperature-difference measurements along a surface, it may be

preferable to use a differential thermocouple (see Section 2.7) instead of using
two thermocouples and taking the difference in the two measurements.
4.5.5 Temperature Measurement on Moving

Surfaces
Thermocouples can be used to measure the temperature of moving, metallic

surfaces. The contact between the surface and the thermocouple elements is
made with brushes or rollers.
4.5.6 Thermal Shunting of Surface-Mounted

Thermocouples
Surface-mounted thermocouples are especially susceptible to errors caused by
heat conduction along the sensor. This phenomenon is a special case of the
stem loss phenomenon described in Section 3.5.1. A common configuration
for surface measurements is bonding or clamping the thermocouple junction

to a pipe or vessel wall. Usually, the thermocouple leads pass through an
environment (often ambient) whose temperature is significantly different than
the surface whose temperature is to be measured. This temperature difference
creates a driving force for heat transfer (from hotter regions to cooler regions).
When a thermocouple is attached to a surface, the wires themselves provide
heat transfer paths from the surface. Since the metal wires are good heat
conductors, this effect may be significant, at least for the small portion of the
surface where the thermocouple is attached. The consequence is that the
temperature profile and the temperature at the junction are changed because
of the thermocouple's presence. This problem is called thermal shunting.
To minimize the thermal shunting effect, one can install the sensor so the
location where the perturbation of the temperature profile occurs is distant
from the thermocouple junction. This may be accomplished by attaching a
length of insulated thermocouple wire (at least several inches) to the surface,
thereby moving the thermal perturbation to a point in the thermocouple loop
that is away from the junction. Also, if the thermocouple wires are
homogeneous, the presence of a large temperature gradient at the point where
the wires leave the surface will have no effect on the temperature
measurement.
4.6 The Back End of a Sheathed

Thermocouple
The nature of the back end of a sheathed thermocouple is a consideration in
sensor performance and in selecting sensors for a specific application. There
are three main considerations: sealing the back end, mechanical fittings, and
protecting wires and/or connections.
4.6.1 Sealing
The seal at the back end of a thermocouple is an important barrier to moisture

and other contaminants that could affect thermocouple performance. The
choice of the sealing material used depends primarily on the temperature that
it will experience. Up to about 200C (400F), epoxy is often used. At higher
temperatures (up to 950C [1750F]) a cement is used.
Generally, the high-temperature cement is less impermeable to moisture and

has a lower mechanical strength than epoxy but can withstand higher
temperatures. It should be noted that the temperature that determines the
type of sealant to use is the temperature at the back end of the sheath, not the
temperature at the junction.
4.6.2 Mechanical Fittings
The back end may have no fittings; it may have a quick-disconnect fitting; it
may have a fitting for attachment to a surface; or it may have a fitting that
permits installation in a thermowell or a protection tube.
4.6.3 Protection for Wires and/or Connections
The wires emerging from the back end of the thermocouple often need to be
protected from electrical shorting, chemical attack, or mechanical stress.
Moreover, a housing to contain connections is often used. The protection
commonly available on wires is insulation and braiding or flexible armor. The
environmental conditions around the wires will dictate the type of protection
needed from electrical shorting, chemical attack, or mechanical stress.
The transition to extension wire is often made in a connection head. These are
housings, usually made of iron, steel, aluminum, or nylon, with a cap that
either screws on or is hinged with a clip fastener. An example is shown in
Figure 4-13. Terminal blocks are mounted inside the connection head to
connect the thermocouple wires to the extension wires. Figure 4-14 shows a
typical configuration.
4.7 Thermowells and Protection Tubes

Thermowells and protection tubes serve three purposes in thermocouple
installations:3
1. protecting the thermocouple from failure because of high static

pressure and flow-induced mechanical stress.
2. protecting the thermocouple from chemical or erosive attack.
3. permitting the thermocouple to be replaced without opening the
monitored process or environment.
4.7.1 Typical Thermowells
Thermowells (also called pockets in Europe) are sturdy fixtures shaped and
bored from solid bar stock. The stem of a thermowell typically assumes one of
Figure 4-13. A Connection Head
Figure 4-14. Terminal Block inside a Connection Head

three standard shapes: straight, tapered and stepped (reduced diameter at

tip). Common process connections for thermowells include threaded, socket
weld, weld-in and flanged. Figure 4-15 shows typical thermowells.
Figure 4-15. Typical Thermowells
To improve sensor-to-process heat transfer, the bore of a thermowell is closely

matched to the diameter of its thermocouple. The most common and least
expensive thermocouple-thermowell assembly specification is to match a
0.250 inch probe to a 0.260 inch bore thermowell.
Sensors assembled with thermowells typically have spring-loaded fittings to

hold the sensor tightly against the end of the thermowell. This improves the
sensor-to-process heat transfer and reduces both the sensor time constant and
the stem loss error. Reducing the thermowell diameter near the tip also helps
reduce these problems.
For certain applications that require an inert or highly corrosion-resistant

thermowell, Teflon may be used as the thermowell material (for low-pressure
and low flow rate situations) or as a coating on metallic thermowells.
Temperature is a significant consideration in considering the suitability of a
coating for a given installation. For example, one should not install a Teflon
coated thermowell in a process with an expected operating temperature above
200C (the maximum recommended temperature for Teflon).
Thermowell sleeves are available for use in applications where velocity is not
a concern. They are applied to cover wetted portions of the sensor and
provide resistance to a corrosive environment.
4.7.2 Specifying Thermowells
Determining acceptable dimensions for a thermowell is often vital to a

successful installation.
The most commonly referenced dimension of a thermowell is its immersion

length, often described as its U length. The U-length is the length that the
thermowell extends from immediately below the process connection to its tip.
The U-length of a thermowell often plays a critical role in determining the
mechanical strength of a thermowell and its suitability for a given process
environment.
Extra length on the cold side of the thermowell over and above the material
necessary to fashion a connection with the sensor is called the T-length. This
extra length is typically added to extend the thermowell through insulation
covering a pipe or, in the case of weld-in wells, to provide ample room for a
full penetration weld into a heavy walled pipe.
4.7.3 Thermowell Failures
Thermowells can fail for a variety of reasons. However, the most common
causes of failure are corrosion, erosion, mechanical failure and vibration.
Corrosion and Material Compatibility: Care should be taken to ensure that

materials appropriate to the process are selected. The material must not only
be chemically compatible with the process, but must also be suitable for the
stress and pressure.
Erosion: Using unsuitable materials in environments with high speed small

particle impingement can lead to rapid deterioration of the thermowell shank.
An additional and often overlooked impact of erosion in applications where
the measurement is used for control is the reduction in the time constant of
the temperature signal and its effect on control system performance.
Mechanical: Breakage or bending can be caused by improper installation or a

force applied that exceeds the yield strength of the thermowell. Common
causes of mechanical failure due to installation error are failure to thread
screw-in sockets far enough and improper welding of the thermowell to the
process.
Fatigue due to Vibration: As a process fluid flows past a thermowell, the stem of
the well sheds vortices commonly referred to as the Von Karman vortex
effect.4 These vortices will, under certain conditions, cause the thermowell to
resonate in-line (in the direction of) or transverse (perpendicular) to the flow
of the process fluid. Either in-line or transverse resonances can cause a
thermowell to suffer catastrophic failure by shearing off or cracking open.
The above considerations can be assessed by application of the ASME 19.3TW

thermowell standard.
4.7.4 ASME PTC 19.3TW Thermowell Standard
With proper selection of materials and dimensions, thermowells can maintain

excellent strength characteristics in the face of high pressure and high
velocity.
The prediction of a thermowells ability to withstand pressure and flow

conditions is a stress analysis problem. Detailed methods are given in the
standard, ANSI/ASME PTC 19.3TW(2010), promulgated by the American
Society of Mechanical Engineers. It serves as the only US standard providing
explicit guidance as to a thermowells suitability for a range of process
conditions.
The 19.3TW standard provides methods and limits to account for and provide
a pass / fail result to gauge the effects of in-line wake frequency resonance,
transverse wake frequency resonance, dynamic (oscillating) stress, steady
state stress and pressure. The anticipated effects of corrosion and erosion on a
given thermowell installation can be addressed as well. Included within the
scope of this standard are threaded, weld-in, socket weld and flanged
thermowells. Not included within the scope of this standard are protection
tubes, thermowells with coatings, thermowells incorporating a weld at any
location other than the flange and thermowells incorporating velocity, or
support collars.
Often the characteristics that cause a thermowell to perform well for

measuring temperature conflict with the characteristics that provide good
mechanical strength. Thicker, shorter thermowells are typically more resistant
to failure than longer, thinner thermowells. Conversely, longer thinner
thermowells are better for measuring temperature. The thermowell designers
job is to keep the well long and lean enough to enable its sensor to provide an
accurate and responsive temperature indication while ensuring that the well
is not so long and thin that its design poses a risk to its own mechanical
integrity or the operation of its installed sensor. The ASME thermowell
standard provides an objective means of balancing these two competing
interests.
Software is available to select thermowells that meet the requirements in the

19.3TW standard.5 The user specifies the features of the planned application
and the software provides pass-fail indications and related information.
Appendix E shows a typical application of the software.
4.7.5 Protection Tubes
In contrast to thermowells, which are shaped and drilled from solid bar stock,
a protection tube is a comparatively thin walled ceramic tube or welded metal
pipe. Figure 4-16 shows typical protection tubes.
Figure 4-16. Typical Protection Tubes
Metal protection tubes are typically fabricated from pipe and welded at the
tip. This style of protection tube finds particular favor in-low velocity,
extremely corrosive or high temperature applications where more expensive
alloyed metal types such as Inconel or Hastelloy are required.
Ceramic protection tubes are often fabricated from Alumina or Mullite for
applications where the temperature limits exceed the capabilities of affordable
metals. Although ceramic protection tubes are capable of withstanding
considerably higher temperatures than their metal counterparts they are also
much more fragile and prone to cracking due to thermal shock. Users of
ceramic protection tubes should be sure to pre-heat these products to 900 F
before inserting into a hot process environment as the failure to take this
precaution can cause the protection tube to crack due to thermal shock.
The air surrounding the thermocouple probe in a protection tube causes the
time response to be slower than in a thermowell where the gap is typically
smaller.
4.8 Special Applications of

Thermocouples
This section deals with special applications of thermocouples that provide
unique measurement capabilities. Some of the methods have seen only
limited applications, but they are included to show some of the possibilities
for non-traditional implementations of thermocouples.
4.8.1 Gradient Thermocouples
Special thermocouples permit the measurement of temperature gradients.

Creating multiple junctions along the length of the thermocouple provides
this capability. Measurement at n points along the thermocouple requires n
junctions involving a minimum of (n + 1) wires.
4.8.2 The Heat-Seeking Thermocouple
A special thermocouple design permits a measurement that provides the

temperature at the hottest point along the thermocouple circuit. The basis for
the design is the strong temperature dependence of the resistivity of
magnesium oxide that is used as an insulator in typical sheathed
thermocouples (see Figure 4-9). For example, MgO resistivity decreases by a
factor of about 100,000 when the temperature increases from 0oC to 500oC. A
region of reduced resistance between parallel thermocouple wires creates an
effective junction at that point. The situation may be viewed as a string of
parallel junctions in which the one in the region with the highest temperature
and lowest resistivity dominates the measurement. Appendix B shows the
relationship for the case of two parallel junctions. Each junction contributes
emf in proportion to the product of the wire-to-wire electrical conductance

(reciprocal of resistance) at the location in question, and the emf that would be
produced if the junction at the location in question were the only junction. If
increased temperature causes the conductance at some location to be much
higher than anywhere else along the sensor, then the temperature at that
location is provided by the thermocouple.
Basically, the heat seeking thermocouple exploits a phenomenon (low

resistivity at high temperature) that is usually considered a problem in high
temperature measurements.
4.8.3 Pulsed Thermocouples
A pulsed thermocouple6 may be used to estimate the temperature above the

maximum useful temperature of the sensor. The thermocouple must start at
some temperature below its damage threshold, then be exposed temporarily
to the high temperature environment. One way to achieve the desired
transient is by using a cooling jacket for the sensor that is turned off to start
the transient. Cooling is resumed before the sensor reaches its damage
threshold. Another way is to move the thermocouple quickly from a cooler
environment outside of the hot medium into the hot environment. The sensor
is extracted before reaching its damage threshold.
Data processing provides an estimate of the temperature that would be

indicated if the sensor were able to tolerate the process conditions. Consider
discrete sampling with a sampling interval, t, during the increasing portion
of the transient. Let T(n) represent the nth data point, let represent the time
constant of the sensor and let Tf represent the process temperature sought in
the measurement. The incremental response for sample (n + 1) in terms of the
response at sample n for an exponential transient is as follows:
T(n) = T(n 1) * exp(t/) + Tf * (1 exp(t/)) (4-1)
Since (exp(t/)) is a constant, we may write
T(n) = A*T(n 1) + B (4-2)
and
Tf = B/(1 A) (4-3)
This shows that a plot of T(n) vs. T(n 1) is a straight line and Tf is available
from the observed slope and intercept.
Measurement noise is always a problem in measurements such as this. The

effect of noise on the estimate of Tf can be minimized by using a least-squares
fitting of a straight line to data for T(n) vs. T(n 1).
EXAMPLE
Figure 4-17 shows calculated results for a measurement with noise having a
maximum amplitude of five percent of measurements. Figure 4-18 shows the
data analysis for this example. The values obtained for A and B by least
squares analysis are 0.937 and 0.0525, giving an estimate of the final
temperature of 0.989 (true value = 1.0 in this example).
Response/Final Response
1.2
1
0.8
0.6
0.4
0.2
0
0 2 4 6
Time/Sensor Time Constant
Figure 4-17. A Pulsed Thermocouple Transient

Value of Data Point n+1
7.00E-01
6.00E-01 y = 0.937x + 0.0525
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
0 0.5 1
Value of Data Point n
Figure 4-18. Pulsed Thermocouple Data Analysis

4.8.4 Radially-Insulated Thermocouples
Some applications require the insertion of a sensor through a heated region to

reach the environment whose temperature is to be monitored. An example is a
process that uses a heated jacket around a vessel or pipe. This situation occurs
in the plastics industry and elsewhere.
A heated region along a thermocouples stem is a special case of the stem loss
problem described in Section 3.5.1. For a sensor that passes through a hot
zone, the accuracy of a measurement depends on the heat transfer resistance
between the measuring junction and the monitored process vs. the heat
transfer resistance between the junction and the temperature in the heated
jacket. To measure accurately, the junction-to-process heat transfer resistance
must be much smaller than the junction-to-heated jacket heat transfer
resistance.
Use of a standard metal-sheathed thermocouple can suffer from very large

errors if used in measurements requiring insertion through a heated zone.
Special sensors with extra insulation in the section exposed to the heated zone
provide much reduced junction-to-heated zone heat transfer and much
improved accuracy over standard sheathed sensors.
References
1. McGee, T. D., Principles and Methods of Temperature Measurement, John
3. Masek, J. A., Guide to Thermowells, Instrumentation & Control
Systems, April 1979, pp. 3943.
4. Blevins, J. D., Flow-Induced Vibrations, Krieger, 2001, Malabar, FL.
5. JMS Southeast software, SwiftyCalcTM Software, available at
http://www.jms-se.com.
6. Glawe, G. E., H. A.Will, L. N. Krause, A new Approach to the Pulsed
Thermocouple for High Gas Temperature Measurements, Proceedings
of the Twenty-Second International Instrumentation Symposium, ISA, May
25-27, pp. 329335, Pittsburgh, PA.
5
Thermocouple Types
The Main Points

Nine thermocouples have been standardized by the American Society
for Testing and Materials (ASTM). Five are base-metal thermocouples,
three are noble-metal thermocouples and one is a refractory-metal
thermocouple.
Manufacturers also provide non-standard thermocouples that carry
their own letter designation.
Thermocouples that satisfy non-U.S. standards are also available.
These thermocouples satisfy country-specific standards or the
international standard of the International Electrotechnical
Commission (IEC)
Thermocouples for very high temperatures (up to 2315C) usually
consist of wires made of tungsten and tungsten-rhenium alloys.
Thermocouples are standardized on the basis of their emf-versus
temperature relationships, not their compositions. Consequently,
thermocouples of the same Type from different manufacturers may
have different compositions.
5.1 The ASTM Thermocouples

The nine thermocouples that have been standardized by the American Society
for Testing and Materials (ASTM) have been given letter designations J, T, K,
E, N, C, S, R, and B. The requirement for a thermocouple to be designated by
one of these letter types is its emf vs. temperature behavior, not its
composition. Consequently, there is some variation in the composition of any
given type of thermocouple supplied by different manufacturers. Because of
these differences in composition, there may by some differences in calibration
and degradation of performance.
89
The older ASTM thermocouples were developed around the beginning of the
twentieth century. The latest to be approved by ASTM are the Type N and
Type C. The nominal composition of each thermocouple type is given in Table
5-1. It is common practice to designate the individual wires for each type
according to the polarity of the thermocouple. If the voltage on a wire is
positive when the open end is at a lower temperature than the junction, the
wire is said to be the positive thermoelement, and the other is said to be
negative thermoelement. The thermocouple letter designation is followed by
the letter P for the positive thermoelement and the letter N for the
negative thermoelement, when it is desired to identify the separate wires.
Table 5-1. Nominal Chemical Composition of Thermoelements
JN,
KP, RN,
TN,
JPb TP EP KN NP NN RP SP SN BP BN CP CN
ENa
Element Nominal Chemical Composition, %
Iron 99.5
Carbon b
Manganese b 2
Sulfur b
Phosphorus b
Silicon b 1 1.4 4.4
Nickel b 45 90 95 84.4 95.5
Copper b 55 100
Chromium b 10 14.2
Aluminum 2
Platinum 87 90 100 70.4 93.9
Rhodium 13 10 29.6 6.1
Magnesium
Tungsten 95 74
Rhenium 5 26
aTypes JN, TN, and EN thermoelements usually contain small amounts of various elements for
the control of thermal emf, with corresponding reductions in the nickel or copper content, or
both.
bThermoelectric iron (JP) contains small but varying amounts of these elements.
Thermocouple Types 91
5.2 EMF versus Temperature for ASTM

Thermocouples
The nominal emf-versus-temperature relationships for the ASTM
thermocouples are shown in Figure 3-1 and are given in the tables in
Appendix C. As seen in Section 3.2, it is essential to have equations that relate
emf and temperature for use in readouts and computer-based systems.
Equations may be developed to provide the temperature for a specified emf,
or the emf for a specified temperature. Both forms are given in Appendix D.
All the metals used in standard thermocouples except iron, copper and
platinum are alloys. The composition of the alloy, the purity of the single-
component metals, and the metallurgical state of the metals cannot be made
identical in every manufacturing run, and these factors can affect the
thermoelectric properties of the materials. Thermocouple wire manufacturers
test their product to determine whether its emf-versus-temperature relation is
close enough to nominal performance (within tolerances).
5.3 Chemical Compatibilities

The materials used for thermocouple wires are attacked by certain chemicals.
Since this attack will contaminate or destroy the wires, chemical compatibility
must be considered when selecting thermocouples for use in a specific
environment. This is primarily a consideration for bare or exposed junction
thermocouples. Compatibility information for each thermocouple Type is
presented in the following sections.1-3
5.3.1 Type J
Type J may be used in oxidizing, reducing, or inert atmospheres for

temperatures between 0 and 760C. Oxidation and attack by sulfurous
atmospheres affect the iron wire above about 540, making Type J unsuitable
in this situation. The iron leg is susceptible to rust in moist environments and
to rust and embrittlement at temperatures below 0C.
5.3.2 Type T
Type T may be used in oxidizing, reducing, or inert atmospheres for

temperatures between 200C and 370C. Type T is often used for low-
temperature measurements. The upper limit on the temperature of use is due
to the rapid oxidation of copper at higher temperature.
Type T thermocouples are the most common selection for cryogenic

applications. However, one recently published report4 concluded that Type T
thermocouples do not conform to the ASTM standard accuracy limits of error
between temperatures of 100oC and 200oC. The report recommends
revising the standard for temperatures between 100oC and 200oC. As of the
publication date for this book, the standard has not been revised. For
applications at temperatures below 100oC in which accuracy is critical, users
should request an initial calibration report from the thermocouple supplier,
including a calibration at the liquid nitrogen temperature.
5.3.3 Type K
Type K may be used in oxidizing or inert atmospheres for temperatures

between 200C and 1260C. Type K thermocouples should not be exposed to
atmospheres that are reducing or alternately oxidizing and reducing. They
also should not be used in reducing atmospheres that also have low
concentrations of oxygen present, since this causes green rot, a preferential
oxidation of the positive element. Type K is also unsuitable for use in
sulfurous atmospheres and in a vacuum. Type K is susceptible to a
troublesome metallurgical transformation (see Section 6.1.2).
5.3.4 Type E
Type E may be used in oxidizing and inert atmospheres for temperatures

between 200C and 900C. Like Type K, Type E thermocouples are
unsuitable for reducing environments, alternating oxidizing and reducing
environments, environments with a low concentration of oxygen, and
vacuum environments. Type E experiences the same metallurgical
transformation as Type K (see Section 6.1.2).
5.3.5 Type N
Type N may be used in oxidizing and inert atmospheres for temperatures

between 200C and 1260C. Type N was developed as a replacement for
Type K. It is less susceptible to green rot and does not undergo the
troublesome metallurgical transformation that affects Type K thermocouples
(see Section 6.1.2). Type N should not be used in reducing environments or in
a vacuum.
5.3.6 Types R and S
Types R and S may be used in oxidizing or inert atmospheres for

temperatures between 0C and 1480C. They are unsuitable for use in
reducing environments. These thermocouples are affected by metallic and

nonmetallic vapors. Types R and S should be placed in alumina, rather than
metallic protection tubes, because metals from a metallic sheath can diffuse to
the thermocouple wire and contaminate it. Types R and S are less useful than
Type B at high temperature, because they experience greater grain growth,
making the wire more susceptible to chemical attack.
5.3.7 Type B
Type B may be used in oxidizing or inert atmospheres for temperatures
between 870C and 1700C. The lower temperature limit is due to the very
small Seebeck coefficient for Type B at lower temperatures. Type B should not
be used in reducing environments. It is affected by metallic and nonmetallic
vapors and should be placed in alumina, rather than metallic protection tubes.
Type B is less susceptible to grain growth at high temperatures than are Types
R or S.
5.3.8 Tungsten-Rhenium
Tungsten-rhenium thermocouples are used for high-temperature

measurements (up to 2315C). They are suitable for reducing or inert
atmospheres. Manufacturers offer three different tungsten-rhenium
thermocouples, each with different proportions of tungsten and rhenium.
However, the American Society for Testing and Materials has included in its
thermocouple standard only the thermocouple with 95 percent tungsten and 5
percent rhenium in one leg, and 74 percent tungsten and 26 percent rhenium
in the other leg. This thermocouple is designated Type C.
5.4 Tolerances on ASTM Thermocouples

Thermocouple tolerances are shown in Table 5-2. Note that the tolerances on
thermocouples are given as a temperature uncertainty over part of the range
and as a percentage of temperature over the remainder of the range.
Table 5-2. Tolerances for New ASTM Thermocouples

Tolerance (Reference Junction at 0C)
Temperature Range
Standard Tolerance Special Tolerance
Thermocouple C or percentage of C C or percentage of C

C
Type (whichever is greater) (whichever is greater)
T 0 to 300 1 or 0.75% 0.5 or 0.4%
J 0 to 700 2.2 0.75% 1.1 or 0.4%
E 0 to 900 1.7 or 0.6% 1 or 0.4%
K or N 0 to 1250 2.2 or 0.75% 1.1 or 0.4%
R or S 0 to 1450 1.5 or 0.25% 0.6 or 0.1%
B 670 to 1700 0.5% 0.25%
C 0 to 2315 4.4 or 1%
T -200 to 0 1 or 1.5%
E -200 to 0 1.7 or 1%
K -200 to 0 2.2 or 2%
EXAMPLE
What is the manufacturing tolerance for a Type K thermocouple (standard

limits of error) at 1000C?
SOLUTION
T = 0.0075 1000 = 7.5C
That is, two Type K thermocouples could differ by 15C (27F) when
measuring a temperature of 1000C (1832F), and each could still be within
tolerance. Tolerances provide the maximum departures from the nominal
calibration. Manufacturers strive to obtain the nominal calibration in their
products, and, consequently, it is much more likely for a new thermocouples
calibration to be close to nominal than for it to be at the maximum departure
from nominal allowed by the tolerance. Therefore, the 15C difference just
cited is possible but not likely.
The tolerance inherent in thermocouples may permit excessive temperature

measurement errors for some practical measurement problems. A user can
improve the situation somewhat by purchasing special-grade thermocouples
(with smaller tolerances and higher costs). A user might also consider
calibrating each new thermocouple individually. However, these steps should
be taken only after careful consideration, given the probable calibration shifts
that will occur after the thermocouple is put to use (see Chapter 6).
5.5 Seebeck Coefficients for ASTM

Thermocouples
Since the Seebeck coefficient is the measure of the sensitivity of the
measurable emf to changes in the monitored temperature, it is of interest in
itself. Figure 5-1 gives the Seebeck coefficient-versus-temperature curves for
all nine ASTM thermocouples. The curves do not show all of the small peaks
and valleys in the curves, but they do show the main features. Table 5-3 gives
approximate Seebeck coefficients over different temperature ranges. From
these, we can see that the Seebeck coefficients range from zero to about 80
V/C and that they generally depend strongly on temperature.
^dd

^

d

:

<

E

Z

^

d
Figure 5-1. Seebeck Coefficients for Standard Thermocouples
The Seebeck coefficient curves are self-explanatory, but one thermocouple, the
Type B, has a unique feature that warrants special mention. Its Seebeck
coefficient is near zero at usual ambient temperatures. This eliminates the
need for reference junction compensation. Recall from Equation 3-1 that the
pertinent formula is as follows:
0 0
V ( 0 C T2 ) = V ( 0 C T 1 ) + V ( T1 T2 ) (5-1)
Since V(0C T1) is negligible for T1 values up to at least 40C, the expression
reduces to
0
V ( 0 C T2 ) = V ( T1 T2 ) (5-2)
This means that the tables or formulas for temperature versus emf can use the
measured emf directly without correction for reference temperatures other
than 0C.
Table 5-3. Nominal Seebeck Coefficients

Thermocouple Type
Temperature E J K N R S T B C
C Seebeck Coefficient-Microvolts/C
-190 27.3 24.2 17.1 11.2 ... ... 17.1 ... ...
-100 44.8 41.4 30.6 20.9 ... ... 28.4 ... ...
0 58.5 50.2 39.4 26.1 ... ... 38.0 ... ...
200 74.5 55.8 40.0 32.9 8.8 8.5 53.0 2.0 17.1
400 80.0 55.3 42.3 37.1 10.5 9.5 ... 4.0 19.0
600 81.0 58.5 42.6 38.9 11.5 10.3 ... 6.0 19.5
800 78.5 64.3 41.0 39.2 12.3 11.0 ... 7.7 19.2
1000 ... ... 39.0 38.5 13.0 11.5 ... 9.2 18.4
1200 ... ... 36.5 33.1 13.8 12.0 ... 10.3 17.2
1400 ... ... ... ... 13.8 12.0 ... 11.3 16.0
1600 ... ... ... ... ... 11.8 ... 11.6 14.8
1800 ... ... ... ... ... ... ... ... 13.6
2000 ... ... ... ... ... ... ... ... 12.2
2200 ... ... ... ... ... ... ... ... 10.4
5.6 Temperature Limits for ASTM

Thermocouples
Each ASTM thermocouple Type has a range of temperatures over which it can
be used. The range is determined by the magnitude of the Seebeck coefficient
and by the onset of excessive decalibration caused by contamination or
metallurgical alterations. Generally, the Seebeck coefficient becomes small at
low temperatures, and, most often, excessive contamination and metallurgical
alteration occur at high temperature.
The recommended range of use depends on wire size for bare thermocouples.
This is because contamination is primarily a surface phenomenon, and small-
diameter wires are more strongly affected (possibly even destroyed) by
contamination than are larger wires. The wires inside sheathed
thermocouples are protected from the process environment, so smaller wires
may be used than for bare thermocouples of the same type.
Temperature limits for protected thermocouples or for bare thermocouples in

recommended environments are given in the American Society for Testing
and Materials, The Use of Thermocouples in Temperature Measurement.5
5.7 Color Coding of ASTM Thermocouples

To facilitate the identification of thermocouples, each ASTM Type has a color
code that is used by all U.S. manufacturers. For U.S. thermocouple wire and
thermocouple extension wire, the insulator on the negative lead is always red.
(Note that this is different from the normal color coding for electrical
systems.) The color codes for the positive leads are shown in Table 5-4.
The connectors for quick-disconnect thermocouples are also color coded. The
colors are the same as those used for the positive leads, as shown in Table 5-4.
The overall color for the insulation on all base-metal ASTM thermocouple
wire is brown. There is no standard overall color for the insulation on ASTM
refractory metal or noble-metal thermocouples.
Table 5-4. Color Codes for Positive Wire Insulation (The Negative Wire Insulation Is
Always Red)
Color (positive element)
Type
Thermocouple Wire Insulator
J White
E Purple
T Blue
K Yellow
N Orange
R None Established
S None Established
B None Established
C None Established
5.8 Compensating Extension Wires for

ASTM Thermocouples
Extension wire may be used to connect thermocouple wire to thermocouple
readout instrumentation (see Section 2.5). Generally, it is used only when the
readout is located far from the thermocouple. Using the lower cost extension
wire reduces the total cost of the installation.
For base-metal thermocouples (J, T, K, E, and N), the composition of extension

wire is nominally the same as that for the corresponding thermocouple
material. It comes from production runs that satisfy normal thermocouple
tolerances at lower temperatures but not necessarily at higher temperatures.
Totally different material is used for noble-metal and refractory-metal

thermocouple extension wire than for thermocouple wire. This is because the
wire for noble-metal and refractory-metal thermocouples is expensive. Base-
metal alloys are used that have relative Seebeck coefficients close to those of
the noble-metal or refractory-metal thermocouples at lower temperatures.
Thermocouple extension wire is designated by the letter X following the

ASTM designation for the thermocouple type. For example, extension wire for
a Type J thermocouple is designated JX, and the positive and negative wires
are designated JPX and JNX. Tolerances for extension wires are shown in
Table 5-5. It should be noted that the tolerances for base-metal extension wire
are the same as for thermocouple wire, but the temperature ranges are
smaller.
Table 5-5. Tolerances for New ASTM Extension Wire

Tolerance
Thermocouple
Temperature Range Standard Special
Extension
C Tolerance Tolerance
Wire Type
C C
TX -60 to 100 1.0 0.5
JX 0 to 200 2.2 1.1
EX 0 to 200 1.7 1.0
KX 0 to 200 2.2 1.1
NX 0 to 200 2.2 1.1
RX 0 to 200 5
SX 0 to 200 5
BX 0 to 200 4.2
Ba 0 to 100 3.7
CX 0 to 200 Initial calibration tolerance
0.110mV
Like thermocouple wire, compensating extension wire is color coded. The

color code is shown in Table 5-6.
Even though the tolerances for thermocouple wire and compensating

extension wire for base-metal thermocouples are the same in the range of use
recommended for the compensating wire, every effort should be made to
minimize the temperature difference along the extension wire. For noble-
metal thermocouples, care should be taken to ensure that both transitions to
extension wire are at the same temperature. Since the thermocouple wires do
not match extension-wire Seebeck coefficients on a wire-by-wire basis,
different temperatures at the connections can cause an incorrect output.
Table 5-6. Color Coding for ASTM Extension Wire (The Negative Wire Insulation Is
Always Red)
Color
Type Positive Element Insulation Overall Insulation

JX White Black
EX Purple Purple
TX Blue Blue
KX Yellow Yellow
NX Orange Orange
RX Black Green
SX Black Green
BX Gray Gray
CX Green Red
5.9 High-Temperature Thermocouples

The maximum useful temperature for base-metal ASTM thermocouples is
1260C (Type K or N), and for noble-metal ASTM thermocouples it is 1700C
(Type B). For higher temperatures, thermocouples must be made from
refractory metals. The best candidates for metals that can withstand these
high temperatures are tungsten, rhenium, molybdenum, tantalum, and
iridium and their alloys. Thermocouples employing nonmetals, notably
graphite, have also been developed. Thermocouples based on tungsten and
tungsten-rhenium have reached the greatest level of development, use, and
standardization (Type C is an ASTM tungsten-rhenium thermocouple).
Unofficial letter designations have been established by some thermocouple

vendors to facilitate the identification of the tungsten-rhenium
thermocouples.
Chemical compatibility with other materials is a special concern at the high

temperatures where tungsten-rhenium thermocouples are used. These
thermocouples operate well only in vacuum, hydrogen, or inert atmospheres.
Since bare thermocouples are not suitable in many frequently encountered
high-temperature environments, sheathed configurations are usually used.
The sheath materials often used are molybdenum, tantalum, and alumina
(maximum temperature of 1800C). Insulators include MgO (which limits use
to about 1500C) and BeO (which limits use to about 2000C and is
poisonous).
5.10 International Standards

Thermocouple standards have been formulated in other major industrial
countries. For example, German standards bear the prefix DIN followed by a
number; British standards bear the prefix BS followed by a number; Japanese
standards bear the prefix JISC followed by a number; and French standards
bear the prefix NFC followed by a number. In addition, the International
Electrotechnical Commission (IEC) develops international standards that are
intended to supersede the various national standards and replace them in
routine use throughout the world.
Because of the widespread growth of international commerce, it is common to

find equipment containing thermocouples that were produced outside the
country in which the equipment is being used. IEC standardssuch as IEC
60584are increasingly gaining worldwide use.
Because of national differences in the color coding of thermocouples, there

remains ample opportunity for confusion. In recent years, most major
industrialized countries except the United States have adopted the IEC color
coding for thermocouples. Consequently, users may find thermocouples with
U.S. color coding, thermocouples with color codings that were in use in
countries before they adopted IEC color coding, color coding for countries
that continue to use their own coding and IEC color coding. The United States
may find it necessary to adopt IEC coding in the future, but thermocouple
users need to exercise caution for the time being. The various color codes are
shown in Table 5-7.
Table 5-7. Color Coding for Thermocouples (The Three Entries Are for Overall
Insulator Color [First Color], Positive Wire Insulator Color [Second Color],
and Negative Wire Insulator Color [Third Color])
Type U.S.a Britishb Germanb Japaneseb Frenchb IEC
T Brown Blue Brown Brown Blue Brown
+ Blue White Red Red Yellow Brown
Red Blue Brown White Blue White
J Brown Black Blue Yellow Black Black

+ White Yellow Red Red Yellow Black
Red Blue Blue White Black White
E Brown Brown Black Purple Purple Purple

+ Purple Brown Red Red Yellow Purple
Red Blue Black White Purple White
K Brown Red Green Blue Yellow Green

+ Yellow Brown Red Red Yellow Green
Red Blue Green White Purple White
N Brown Orange None None None Pink

+ Orange Orange Pink
Red Blue White
R or S None Green White Black Green Orange

+ White Red Red Yellow Orange
Blue White White Green White
B None None Gray Gray None None

+ Red Red
White White
C None None None None None None

+

Notes:
a. Still in use (2012).
b. Superseded by IEC color coding.
c. The outer jacket is blue if rated as intrinsically safe.
The non-IEC thermocouples most commonly encountered are two German

DIN-designated thermocouples. Type L (also referred to as J DIN) is similar to
Type J, and Type U (also referred to as T DIN) is similar to Type T. The
differences in nominal calibration between Type J and Type L are within 5C
from 100C to 500C, increase to about 12C at 200C, and climb to 20C at
900C. The difference between Type T and Type U are smaller (around 5C
over the whole useful range of 200C to 900C). It is apparent that in low-
accuracy applications, Type J may replace Type L and Type T may replace
Type U, but the user must be sure that the inaccuracies introduced by doing
this are acceptable.
A relatively new thermocouple that has received acceptance by the IEC is the
Type A tungsten rhenium thermocouple. This thermocouple is notable for its
significantly higher acceptable temperature range (1000oC to 2500oC) than the
Type C thermocouple (0oC to 2315oC). To date, the Type A thermocouple has
not yet been recognized by the ASTM, but shows promise as the only
thermocouple suitable for exceeding 2315oC.
This book has largely addressed tolerances for ASTM thermocouples and
extension wire tolerances, since this is the most commonly referenced
standard in the United States. However, readers in the U.S. and elsewhere
may encounter thermocouples built to the IEC standard. Tolerances for
thermocouples and thermocouple extension wire under the ASTM standard
and the IEC standard differ significantly both in substance and form. Tables
5-8 and 5-9 show IEC tolerances for thermocouples and extension wire.
Table 5-8. IEC Tolerances & Classes for New Thermocouples

Tolerance values () and temperature limits of validity
Thermocouple
Type Class 1 Class 2 Class 3
Limit of Error: 1.5C or 0.4%[t] 2.5C or 0.75%[t] 2.5C or 1.5%[t]

over
Temp range for:
Type J -40C to 800C -40C to 750C

Type K -40C to 1000C -40C to 1200C -200C to 40C
Type N -40C to 1000C -40C to 1200C -200C to 40C
Type E -40C to 750C -40C to 900C -200C to 40C
Limit of Error 0.5C or 0.4%[t] 1C or 0.75%[t] 1C or 1.5%[t]
Over
Temp range for
Type T -40C to 350C -40C to 350C -200C to 40C
Limit of Error 1C or [1+.003(t-1100)] 1.5C or 0.25%[t] 4C or 0.5%[t]
Over
Temp range for
0C to 1600C
Type R or S 0C to 1600C
Type B 600C to 1700C 600C to 1700C
Limit of Error 1%[t]
Over
Temp range for
Type C 426C to 2315C
Limit of Error 1%[t]
Over
Temp range for
Type A 1000C to 2500C
Table 5-9. IEC Tolerances & Classes for New Thermocouple Extension Wire &
Compensating Cable. IEC 60584-3-2007
Measuring Tolerance
Extension Wire
Thermocouple Junction Class 1 Class 2
Temperature Range
Wire Type Temperature Tolerance Tolerance
C
C V (C) V (C)
JX -25 to 200 500 85 (1.5) 140 (2.5)
TX -25 to 200 300 30 (0.5) 60 ( 1.0)
EX -25 to 200 500 120 (1.5) 200( 2.5)
KX -25 to 200 900 60 (1.5) 100 ( 2.5)
NX -25 to 200 900 60 (1.5) 100 ( 2.5)
KCA 0 to 150 900 100 ( 2.5)
KCB 0 to 100 900 100 ( 2.5)
NC 0 to 150 1000 100 ( 2.5)
RCA 0 to 100 1000 30 ( 2.5)
RCB 0 to 200 1000 60 ( 5.0)
SCA 0 to 100 1000 30 ( 2.5)
SCB 0 to 200 60 ( 5.0)
Note 1: Cable temperature range may be restricted to lower values because of cable
insulation temperature limits (e.g., many PVC insulations should not be used at temperatures
> 105C).
Note 2: Copper extension wire may be used for Type B thermocouples with a maximum
deviation of 40V ( 3.5C when the measuring junction is at 1400C).
Note 3: The stated IEC tolerances in C are approximate equivalent tolerances based upon
specific V tolerances at the stated maximum measuring junction temperature. Because emf-
temperature relationships are non-linear, the error in C will often be larger at lower
temperatures.
Note 4: Extension Cables are designated by letter type followed by X (e.g., JX).
Compensating Cables are designated by letter type followed by C (e.g., NC). Subsequent
letters designate differing tolerances by temperature range (e.g., RCA 0 to 100; RCB 0 to
200). Extension cable is manufactured from the same nominal composition as the
corresponding thermocouple. Compensating cable is manufactured from conductors having a
different composition than the corresponding thermocouple type.
References
1. The Theory and Properties of Thermocouples Elements, American Society
4. Dowell, D. L., A Critical Look at Type T Thermocouples in Low-
Temperature Measurement Applications, International Journal of
Thermophysics, Vol. 31, 2010, pp. 15271532.
5. The Use of Thermocouples in Temperature Measurement, American
Society for Testing and Materials 470B, Fourth Edition, 1993.
6
Thermocouple Degradation
The Main Points

Changes in the chemical composition or the metallurgical state of
thermocouple wires can and do occur when thermocouples are put to
use. This can and does cause thermocouple decalibration with use, and
this decalibration can be significant.
An inhomogeneous section in a thermocouple will cause a
measurement error only if it resides in a temperature gradient.
The recalibration or calibration testing of used thermocouples is not
recommended except for noble-metal thermocouples where it may be
economically advantageous to attempt restoration of the calibration
through reannealing.
In sheathed thermocouples, the proper selection of the thermocouple
system components (wires, insulators, metallic sheath materials, gas
contained in voids inside of thermocouple sheaths) can have a major
effect on decalibration tendencies.
Degraded thermocouples can indicate temperatures that are either too
high or too low.
Diagnostic tests can help identify the existence of thermocouple
problems.
6.1 Basics
For a thermocouple to perform properly, the Seebeck coefficient for each wire
must vary only in accordance with the nominal temperature dependence for
the specific thermocouple type. That is, there should be no changes in the
Seebeck coefficient from the as-built condition as a result of changes in the
chemical composition or metallurgical state of the wires. Unfortunately, such
changes do occur.
105
6.1.1 Chemical Composition Changes
Thermocouple wires are made of single-component metals or alloys. They can

interact chemically with other materials around them. For bare-wire
thermocouples, this environment is the process in which the thermocouple is
installed. For sheathed thermocouples, it is the sheath material, insulator,
sealant, moisture and any trace contaminants built into the sensor, and gas
that was present in the void space inside the sheath when the sensor was
sealed, and any contaminants that diffuse through seals or sheaths.
Chemical attack involves the migration of atoms to and from the wires. Since
this is a surface phenomenon, the alteration will usually develop as a band at
the surface of the wire. When this occurs, the wire consists of parallel
conductors, each with a different Seebeck coefficient. The net Seebeck
coefficient is a weighted sum of the Seebeck coefficients for the two segments.
The weighting factor is the conductance of the wire segment. That is, the
absolute Seebeck coefficient for two parallel conductors (a and b) is as follows:
aS a + b S b
S = ---------------------------------- (6-1)
+ b a
where
= 1/resistance
Since wire contamination usually begins at the surface and works its way
inward, the severity of its effect on thermocouple performance depends on the
wires diameter. If contamination penetrates to a depth that is a small fraction
of the wire diameter, then its effect on the thermocouples performance would
be less than in a situation where a large fraction of the wire cross section is
affected. Consequently, small-diameter wires should not be used in cases
where contamination is expected.
It should be noted that the loss or gain of atoms from a surface depends on the
condition of the surface. Certain compounds at the surface inhibit the
migration of atoms through the surface in either direction. Oxide layers on
aluminum or stainless steels are examples of this behavior. This is called
passivation. A problem that may (and does) occur in thermocouples is the
destruction of this protecting layer by its chemical interaction with
components of the environment around the wires. This exposes the
underlying metal and makes it susceptible to chemical attack. Thus, some
contaminants that do not affect thermoelectric performance directly may be a
concern because they facilitate attack by materials which otherwise would be
unable to attack the wires.
Thermocouple Degradation 107
Chemical reactions are generally a larger concern at higher temperatures than

at lower temperatures. Reactions that are insignificant or extremely slow at
low temperature can become significant and fast at high temperature. For
example, the platinum-rhodium alloys used in types R, S, and B are quite inert
at low temperatures, but the rhodium component becomes increasingly
susceptible to oxidation at high temperatures. Also, rhodium tends to
vaporize at high temperature and migrate. This effect can cause composition
changes in both wires and changes in calibration.
In a sheathed thermocouple, the sources of migrating atoms and migration

paths include the following (see Figure 6-1):
Wire to wire. Atoms can disengage from one wire and migrate to the
other, thereby altering the Seebeck coefficient of each.
Sheath to/from wire. Atoms can disengage from the sheath metal and
migrate to one or both of the wires or vice versa. This usually results in
preferential alteration of the Seebeck coefficient for one of the wires.
Insulator contaminants to wire. Pure insulator material for common
thermocouple insulators does not react significantly with other sensor
constituents, but trace amounts of contaminants in the insulator can.
These contaminants may be introduced during the manufacture of the
insulator or the fabrication of the thermocouple. Also, chemical
reactions between the insulator and certain contaminants can release
ions that were components of the insulator, and these ions can
contaminate the thermocouple wire.
Moisture. Some insulators are very hygroscopic. Water absorbed in
the insulator can lead to wire contamination and to major reductions
in insulation resistance. A possible consequence is electrical shunting
(see Section 6.3). Moisture can also destroy the protective oxide layer
on some metals and expose the metals to chemical attack.
Fill gas. Thermocouple insulators are granular with small void spares
between the grains. These voids are filled with the gas present during
manufacture. Some gas constituents can interact chemically with wire
metal.
Diffusion through sheath. Sheath metal consists of grains that are stacked
together. The boundaries between grains provide pathways for the
diffusion of contaminants through the sheath.
6.1.2 Metallurgical Changes
Thermocouples can undergo metallurgical changes as well as chemical

changes. The Seebeck coefficient depends on the crystal structure of the metal
that comprises the thermocouple wires. This structure is affected by cold
SHEATH
A B
INSULATOR
Figure 6-1. Sources of Migrating Atoms and Migration Paths
working or heat treating. Both of these phenomena can occur in

thermocouples during manufacture and/or routine use.
Possible metallurgical changes in thermocouple materials include the

following:
Local segregation of components or impurities during manufacture

Thermal relief
Mechanical cold working
Structural ordering/disordering
Recrystallization
Cold working is a change in crystal structure caused by work-induced

deformation at low (ambient) temperature. Cold working naturally occurs
during the initial fabrication of thermocouple wires. Thermocouple material is
initially produced as ingots. The metal is subjected to repeated drawing and
swaging operations, each reducing the wire diameter that existed in the
previous stage. This process cold works the metal. To ensure that the final
product is free of defects due to variations in the amount of cold work, the
wires may be annealed. Cold work can also occur during the routine use of
thermocouples by clamping, squeezing, or bending the wires.
The crystals in a metal form when molten metal cools and solidifies. Metals
maintain the as-formed crystal structure until they are affected by mechanical
working or heat treating. Some thermocouple metals experience
recrystallization or structural ordering/disordering at some temperature
below the maximum temperature of use. These cause changes in the Seebeck
coefficient.
Thermocouple wires can experience metallurgical changes that are reversible

as the thermocouple experiences changes in its operating temperature. The
positive leg of the type K and E is susceptible to this type of behavior. In this
case, the positive wire experiences a metallurgical change at about 550C.
There is uncertainty about the underlying physical cause of the observed
effect. Some experts think it is an order/disorder effect (the equilibrium state
of the crystal lattice is ordered below about 550C and disordered above
550C),1 and others think it is a phenomenon called a spin-cluster mechanism.2
Whatever the underlying physics, there is no doubt that the effect occurs. We
will avoid the argument over underlying causes by asserting that the positive
thermoelement has one equilibrium condition below about 550C, state A,
and another equilibrium condition above about 550C, state B. Furthermore,
the metal can exist in state B at temperatures below 550C. Quenching from a
higher temperature traps the metal in state B. Metal trapped in state B
spontaneously transforms to state A at temperatures below 550C, but the
time required to complete the transformation depends on the temperature. At
room temperature, the time required is very long (essentially infinite); at
300C it takes about forty weeks; at 350C it takes four days; at 400C it takes
seventy-five minutes; at 450C it takes one minute; and at higher temperatures
up to about 550C (where state B becomes the stable condition), the transition
from state B to state A is virtually instantaneous.3
This behavior has serious thermoelectric consequences. This is illustrated by

three hypothetical scenarios. In each case, the temperatures are selected
arbitrarily in order to illustrate a point.
Wire annealed above 550C and used to measure temperatures below 250C.
In this case, the annealing causes the positive thermoelement to be in
the unstable state B, but the temperatures are too low for the
transformation to state A to take place in a short time. In this case, the
calibration remains in the as-built condition for an extended period of
time.
Wire annealed above 550C and used to measure temperatures above 550C.
This case is more complicated. At the start of the first heat-up, all of the
wire is in state B. As a result of the heat-up, the junction and some
portion of the wire adjacent to the junction will reach temperatures
above 550C and will be in state B. Since the open end will usually be
at room temperature, there will be a portion of the wire adjacent to the
open end that is also in state B. This is due to the quenching that has
trapped the state B condition. Since one end of the thermocouple
circuit is at or near room temperature and the other end is above
550C, there must be an intermediate section where the temperature is
300C to 500C. Some of the positive thermoelement in this zone will
be in state A. Consequently, from open end to junction, the condition
will be state B/state A/state B. The state A section is inhomogeneous
compared to the other sections. This can cause measurement errors
(see Section 6.1.3). Furthermore, the length of the portion in state A
will grow with time, creating a nonstationary inhomogeneity.
Wire annealed above 550C is used to measure temperatures in the range
250C to 550C. In this case, all of the wire is initially in state B. The
portion operating at temperatures between 250C to 500C
spontaneously converts to state A over a period of time that is shorter
than the typical durations of use of industrial thermocouples (quite
slow at the lower end of the temperature range and quite fast at the
higher end). Consequently, from open end to junction, there will be a
transition from state B to state A. Furthermore, the transition will
gradually move toward the end where the wire is at the lower end
(250C) of the temperature range. This creates a nonstationary
inhomogeneity even when the temperature profile along the wire is
unchanging. This can cause measurement errors (see Section 6.1.3).
6.1.3 Thermoelectric Consequences of Chemical

or Metallurgical Changes
If the chemical or metallurgical changes occur homogeneously along the

whole length of a thermocouple wire, the net result is equivalent to replacing
the wire with another that has a different Seebeck coefficient. In essence, this
amounts to changing from the original thermocouple to one with a different
calibration. The thermocouple would continue to provide an emf that
increases as temperature increases, but the temperature indications will be
wrong. However, thermocouple wires usually experience a variation in the
environmental conditions that cause chemical or metallurgical changes along
parts of the circuit. For example, the transition in the type K positive
thermoelement described in the previous section can cause this condition to
exist.
In the case in which chemical or metallurgical changes occur only along a

portion of the thermocouple wire, the Seebeck coefficient is unchanged except
over the length of wire where the chemical or metallurgical changes occurred.
A simplified depiction of the situation is shown in Figure 6-2, where the
changes in Seebeck coefficient occur abruptly. Thermocouple loop analysis
gives the following:
V = SA(T2 T0) + SA(T3 T2) + SA(T1 T3) + SB(T0 T1) (6-2)
Figure 6-2. Abrupt Changes in Seebeck Coefficient of Thermocouple Wire
Algebraic manipulation gives
V = SA(T1 T0) SB(T1 T0) + SA(T2 T3) SA(T2 T3) (6-3)
or
V = SAB(T1 T0) + SAA(T2 T3) (6-4)
The first term is the voltage that would have been produced if the
thermocouple had not undergone attack. Consequently, the second term is the
error caused by the inhomogeneous region. If the relative Seebeck coefficient
between the unaffected wire and the altered wire is nonzero, then a
measurement error will occur if T2 is not equal to T3. That is:
An inhomogeneous section in a thermocouple wire will cause a measurement

error only if it resides in a temperature gradient.
This is a very important result. We have seen that process or environmental

conditions usually cause alterations only along some portion of a wire. This
makes the measurement error dependent on the temperature profile along the
wires. One consequence of this is that it confounds any attempts to
recalibrate used thermocouples. This is because in a calibration facility it is
impossible to duplicate the temperature profile that the thermocouple system
will experience when it is being used in a process.
Since recalibration is not a suitable option for determining the accuracy of a

used process thermocouple, then how can thermocouple accuracy be
checked? Rather than send a used thermocouple to a calibration laboratory,
the user should remove the used thermocouple and replace it in the process
with a new, unused thermocouple or a resistance thermometer. If the
indicated temperature with the replacement sensor agrees satisfactorily with
the temperature indicated previously with the used sensor, then the used
sensor can be re-installed and continue use until the next accuracy check.
Some users choose to replace thermocouples periodically rather than

performing accuracy checks. This approach is effective if the user has
ascertained that unacceptable decalibration does not occur during a
thermocouples period of use in the process.
The error caused by the development of an inhomogeneous section in a

thermocouple circuit may be positive or negative. As can be seen in Equation
6-4, the polarity depends on the relative Seebeck coefficient between the
affected and unaffected segments (since SAA = SA-SA , SAA can be positive or
negative). Also, the polarity depends on the temperature difference, T2-T3,
across the affected region, and this can be positive or negative.
6.2 Decalibration Tendencies of ASTM

Thermocouples
The possible mechanisms for the decalibration of thermocouples outlined in
the previous section occur to varying degrees in the various ASTM
thermocouples. In this section, the factors that influence decalibration for
these thermocouples are presented so as to provide part of the rationale for
selecting thermocouples for specific applications. Bare-wire thermocouples
and sheathed thermocouples will be considered separately. In the case of
bare-wire thermocouples, the environment that determines possible chemical
changes in the thermocouple is the process or environment into which the
sensor is installed. For sheathed thermocouples (except the exposed junction
configuration), it is the environment inside the sheath and along the wires
outside of the sheath that is important. The metallurgical effects on sensor
performance are caused mainly by the temperatures to which the
thermocouple and extension wires are exposed, as well as the mechanical
operations on the wires, and these can occur with bare or sheathed
configurations.
It might be presumed that because of the long and extensive use of

thermocouples, the decalibration mechanisms would be well known for all of
the standard types. Much is known, but the knowledge base is inadequate for
making unequivocally optimal choices for thermocouples in every
application. This is especially the case for sheathed thermocouples where the
contamination of the wires by other sensor constituents has received close
scrutiny only in recent years. The greatest attention has been given to Type K
and the thermocouple developed to overcome problems with Type K, Type N.
The main reason for this attention is probably that the K and N thermocouples
are the base metal types that can serve at the highest temperatures of all the
standard base metal thermocouples.
6.2.1 Bare-Wire Thermocouples
The literature on thermocouples provides guidelines on the compatibility or

incompatibility of the ASTM thermocouples with the various environments in
which they might be used. This information was summarized in Section 5.3.
In some cases, the chemical basis for the incompatibility has been studied and
reported, but in others the literature provides information only on the effect,
not the cause.
6.2.2 Sheathed Thermocouples
Type K is the most widely used sheathed thermocouple, and it has

experienced significant decalibration in use as a result of the problem with the
positive thermoelement discussed in Section 6.1.2. For these reasons, it has
received intensive study, and a new type, Type N, has been developed to
reduce the problems experienced with Type K. Sheathed thermocouples have
come to be viewed as a system (consisting of thermocouple wires, sheath
material, insulator, and fill gas) as a result of work on the Type K and Type N.
Sensor design based on these considerations is called integrated design of MIMS
(mineral insulated, metal sheathed) thermocouples.4
Types K, E, and N employ nickel and chromium alloys with possible minor
inclusions of aluminum, cobalt, copper, iron, carbon, silicon, manganese, and
magnesium. The Type N composition was selected to eliminate the
temperature and time-dependent metallurgical effects in Type K and to
improve oxidation resistance.5 The main changes in Type N are the
elimination of manganese and aluminum, an increase in the amount of
chromium and silicon, and the addition of magnesium.
The nominal Type E thermocouple uses the same material for the positive
thermoelement as Type K. There is very little published information about
Type E degradation compared to Type K, but they both share the component
that is known to cause problems in Type K. Consequently, one would expect
the Type E to experience the same problems seen in Type K.
Both the Type K and Type N have been found to be susceptible to the
migration of manganese from the sheath to the thermocouple wire (though no
results have been reported, it is presumed that the Type E suffers this problem
too since it employs the same type of thermoelement).6 Work on types K and
N has led to a greater appreciation of the importance of selecting sheath
materials that are compatible with the thermocouple wires. New metals for
use as sheath materials for Type K and Type N have been developed and are
available with various trade names from sensor manufacturers.
The development of special sheath material for Type N raises three important
questions:
1. If trace materials such as manganese in sheath metal cause Type K or

Type N to decalibrate, what about the compatibilities of other types
and standard sheath metals for which similar studies have not been
done?
2. Should Type K and Type N (and possibly Type E) thermocouples with
sheaths made of metals containing manganese be abandoned?
3. Which external environments are compatible and which are
incompatible with the new sheath materials; and for cases where
environmental conditions make the new sheath material unsuitable,
what is an acceptable alternative?
Complete answers and, in some cases, even partial answers are not yet
available for these questions.
Another issue for sheathed thermocouples is the fill gas that occupies the
voids in the insulator. Because special gas atmospheres are not used in the
manufacture of most thermocouples, the fill gas is usually air. It has been
found that the use of inert gas as the fill gas in Type K or Type N improves
sensor performance.7 Air as fill gas in Type K thermocouples can cause green
rot (see Section 5.3), which is the preferential oxidation of chromium when
only a limited amount of oxygen is available. Green rot removes chromium
from the alloy and changes its Seebeck coefficient.
6.3 Insulation Resistance

Adequate insulation resistance is required to maintain electrical isolation
between wires and from wires to the sheath in sheathed thermocouples.
Inadequate resistance leads to a problem called electrical shunting. It is more
likely to occur in very-small-diameter thermocouples (because the insulator
thickness between conductors is small), in very-high-temperature
applications (because the resistivity of insulators decreases markedly as
temperature increases, as shown in Figure 4.9), and in cases where moisture is
present in the insulator (because of manufacturing inadequacies or leakage
through the seal at the thermocouples back end or through holes or cracks in
the sheath).
If the wire-to-wire insulation resistance is too low, the thermocouple circuit is

as shown in Figure 6-3. In this case, the thermocouple wires are bridged along
their length as well as at the junction. Since every bridge is an effective
junction that involves the same two wires, the thermocouple will read
correctly if the whole affected region and the junction are at the same
temperature. In the common case where there is a temperature variation
along the sensor, the indicated temperature will be between T1 and T2.
Figure 6-3. Thermocouple Circuit with Low Wire-to-Wire Insulation
In the second type of electrical shunting, the problem is inadequate wire-to-

sheath resistance. The equivalent circuit is shown in Figure 6-4. In this case,
the sheath material enters the thermocouple circuit as a parallel branch for
each arm of the thermocouple. This will alter the Seebeck coefficient for each
arm of the thermocouple and cause the OCV to differ from its calibration.
6.4 Mechanical Problems with

Thermocouples
The proper electrical performance of a sheathed thermocouple may be
affected by mechanical problems. Three main mechanical problems may arise:
wire continuity interruption (breakage), sheath failure, and seal failure.
Wire continuity interruption can occur in sheathed thermocouples primarily

because of thermal stresses (see Section 4.3.5) and in bare-wire thermocouples
primarily because of chemical attack on the wires. Since thermocouple wires
Figure 6-4. Thermocouple with Low Wire-to-Sheath Insulation
(either in sheathed or bare-wire thermocouples) are generally constrained to

maintain their positions in spite of breakage, the broken ends may fail to
touch, continue to touch, or touch intermittently. If broken wires continue to
touch, the thermocouple will continue to operate normally (but the high
contact resistance may increase the loop resistance significantly.) If the broken
wires fail to touch, the thermocouple will cease to provide a signal, and the
failure will be apparent. If the broken wires touch intermittently, then a
fluctuating signal will be observed.
Sheath failure may occur as a result of flow-induced stress on sensors installed

in flowing fluids. Usually, this will be a catastrophic failure resulting in the
cessation of the thermocouple's operation. Selecting sensors with adequate
mechanical strength is a stress analysis problem that can be addressed by
standard techniques. Detailed procedures are available in the literature.8
Diffusion paths may exist in sheath materials along the grain boundaries in
the metal (especially for metal with large grains caused by manufacturing
procedures or conditions of use). Small holes (pinholes) or cracks may also
exist in thermocouple sheaths as a result of manufacturing flaws or corrosion,
permitting ingress of fluid or vapor, contamination of the insulator, electrical
shunting (see Section 6.3), and chemical attack on the sensor wires.
The seal at the back end of the thermocouple is intended to isolate the sensor
internals from the atmosphere around the back end. Ideally, the seal would be
impervious to gases and liquids and would provide leak-tight contact with
the wires passing through the seal as well as with the housing. If this
imperviousness or contact is not achieved in manufacture or if either or both
are lost due to aging, then a path is opened for contaminants to enter the
sheath, which may possibly lead to electrical shunting and/or chemical attack
on the wires.
6.5 Thermocouple Diagnostics

Ideally, we would have tests which could reveal the accuracy of a
thermocouple unequivocally by measurements made at the open end. No
such universal test exists, but there are still things that can be done. Some
techniques require the processing of information from multiple sensors, some
require special measurements on standard thermocouple configurations, and
some require the use of special sensors designed specifically to facilitate
diagnosis.
The formulation of special tests on standard thermocouples requires that

potential degradation processes be identified, then measurements that detect
manifestations of these changes be developed. The potential manifestations of
degradation processes that may be exploited for diagnostic tests are as
follows:
Loop resistance (wire breakage or thinning or electrical shunting)

Insulation resistance (moisture in sheath)
Sudden shifts in OCV (electrical shorting, inhomogeneity developing
or experiencing a new temperature profile)
Noise and transients (wire breakage with intermittent touching)
Loop current step response (inhomogeneities in the thermocouple
circuit)
Diagnostic tests involving multiple sensors and measurements on individual

sensors are discussed in the following sections.
6.5.1 Cross-Calibration
Cross-calibration is the comparison of measurements from redundant sensors

exposed to the same process conditions.9 The redundant measurements may
be made with sensors of the same type or with different sensor types (diverse
sensors). Cross-calibration with multiple sensors of the same type suffers from
the possibility of common mode degradation. That is, the sensors can
experience the same degradation when exposed to the same process
conditions, and comparisons will show no difference because all are in error
by the same amount. On the other hand, if multiple sensors of different types
(diverse sensors) are cross-checked, then identical degradation mechanisms

are usually implausible and common-mode problems are eliminated. Some
arrangements for cross-calibration are as follows:
1. Multiple identical thermocouples installed close to one another in the process.

The obvious problem is common-mode degradation. The advantage of
this arrangement is that it eliminates the installation or connection
errors that might occur if the installer has to deal with different sensors
in the same location. Also, one readout with appropriate switching is
sufficient.
2. Multiple diverse thermocouples located close to one another in the process.
This approach uses two thermocouples of different types. It usually
eliminates the common-mode problem (both thermocouples may have
decalibrated, but decalibration by the same amount is implausible).
3. Multiple diverse sensor types located close to one another in the process. This
approach uses a thermocouple and a totally different type of sensor
(such as a resistance thermometer or a thermistor).
4. Multiple identical thermocouples located in a single sheath. The
configuration that is commercially available is the dual-element
thermocouple (see Section 4.3.4). It is available in three configurations,
two of which permit two measurements with the third allowing up to
four (though they are not all independent).
5. Multiple diverse thermocouples located in a single sheath. Four-wire
configurations could, in principle, be manufactured with four
different types of wire. Of course, this magnifies the compatibility
problem since all of the wires must be compatible with all of the other
wires as well as the sheath material. The authors are not aware of any
commercially available sensors of this type.
6. Multiple diverse sensor types located in a single sheath. Dual-element
temperature sensors are commercially available that have a
thermocouple along with a resistance thermometer or a thermistor.
This approach is a convenient way to accomplish a redundant, diverse
measurement, but it does mean that installation and maintenance
personnel will have to keep up with more wires (typically, two from
the thermocouple and three from a resistance thermometer or two
from a thermistor). The materials compatibility issue arises here, but it
is probably not as large a concern as for the dual, diverse
thermocouple just described.
6.5.2 Analytic Redundancy

Analytic redundancy is a term for techniques that use measurements of
different quantities to check for inconsistencies that indicate possible
measurement errors. An array of techniques have been developed. In general,

they use theoretical process models or correlations that were learned during
normal operation of the process and the sensors. In their article Local
Sensor Validation, S. K. Yung and D. W. Clarke provide detailed information
on analytic redundancy techniques as well as a guide to the pertinent
literature.9 It will suffice here to give a simple example to illustrate the
concept.
EXAMPLE
A boiler that is designed to produce saturated steam is fitted
with a thermocouple and a pressure sensor. Since there is a
unique steam temperature at any pressure for saturated steam,
the pressure measurement can provide an implied temperature
to compare with the thermocouple indication. This is a very
simple example of model-based analytic redundancy (in this
case, the model is the saturation temperature versus the
saturation pressure information contained in the steam tables).
6.5.3 Loop Resistance
The electrical resistance of a thermocouple circuit depends on the length of the

thermocouple wire, its type and diameter, the length of the extension wire, its
type and diameter, the temperatures along the total circuit, and the contact
resistance at any connections or breaks. The normal loop resistance is strongly
dependent on the specific installation, but changes in measured loop
resistance may be used to indicate wire thinning (possibly due to chemical
attack on the wires), increased contact resistance (possibly due to broken but
touching wires), or loose or corroded connections. Electrical shunting will also
cause changes in loop resistance. If the thermocouple wires experience
electrical short circuiting, the loop resistance will decrease.
Loop resistance measurements are complicated by the presence of a

thermoelectric emf on the wires. Resistance measurements are made by
imposing a small, constant current on the circuit, then measuring the voltage
drop across a known resistance. If the thermocouple experiences a
temperature difference during the resistance measurement, the thermoelectric
emf that is produced by the temperature difference affects the resistance
measurement. This effect reverses when the wires connected to the resistance-
measuring instrument are reversed. Consequently, averaging two
measurements--the second with the leads reversed relative to the first--will
provide the true resistance of the loop.
6.5.4 Insulation Resistance
Moisture in the insulator causes reduced insulation resistance and electrical

shunting, which degrade thermocouple performance. Insulated junction
thermocouples connected to metallic pipes, vessels, and the like can be tested
for insulation resistance by measuring the resistance between a thermocouple
wire and some point that is electrically common with the structure in which
the thermocouple is installed. A well-insulated thermocouple should have an
insulation resistance of at least 100 megohms at room temperature.
A megohmeter should be used when measuring insulation resistance. These

instruments use a much higher applied voltage than a standard ohmmeter.
Since resistance measurements depend on measuring the voltage drop across
a fixed resistor, good accuracy depends on having a suitably large current
flow in the circuit. The higher applied voltage is used to provide this current.
6.5.5 Sudden Shifts
An inexplicable sudden change in sensor output is a clear warning to process

operators. Like other process sensors, thermocouples can demonstrate this
behavior. Possible causes for sudden sensor output changes include electrical
shorting (possibly due to moisture ingress) or changes in process conditions
that cause a change in the temperature difference across an inhomogeneous
section. The shift may be noticed, not noticed but recorded, or not noticed and
not recorded. In the first two cases, the operator has information (either
obvious or available through investigation) that should create concern about
measurement accuracy. In the future, computerized measurement systems
may monitor for such sudden shifts.
6.5.6 Noise and Transients

Fluctuations (noise) in the sensor output may be a manifestation of the
intermittent contact of a broken wire. The frequency with which contact is
made and broken will depend on the frequency of any vibration in the process
to which the sensor is attached. The speed of the vibratory changes is
generally much faster than the speed of process temperature changes or the
speed of response of the sensor. This problem can therefore be identified by
detecting rates of change that are too fast to be caused by real changes in
measured temperature. These rapid changes in output would generally not be
observable in typical thermocouple readout instrumentation. Instrumentation
that can be employed to detect high-frequency fluctuations includes
oscilloscopes, recorders, and spectrum analyzers. Any of these could be used
to detect rapid changes in the signal.
6.5.7 Loop Current Step Response
A technique called loop current step response has been developed that permits
the measurement of the time constant for a thermocouple installed in a
process.10 This technique involves passing a direct current through the
thermocouple (causing Joule heating), then monitoring the thermoelectric emf
during cooldown after the current has ceased. Experience with the loop
current step response method suggests that it may also be useful for detecting
anomalies in the thermocouple circuit.
The local heat produced by a current passing through a conductor depends on

the electrical resistance of the conductor in that locality. The rate of local heat
generation will be higher in the portions of the thermocouple circuit where
the electrical resistance is higher (that is, any portions of the circuit with
smaller-diameter wires and regions with significant contact resistance such as
broken but touching wires).
The temperature rise due to heating in a region depends on the heat transfer
resistance in that region. Locations with thick insulation or low or stagnant
flow around the wires will experience larger temperature rises. If the
thermocouple circuit is composed of two homogeneous wires, the transient
emf during cooldown depends only on the junction temperature. A typical
cooldown curve is shown in Figure 6-5. If the circuit contains
inhomogeneities, the transient emf during cooldown depends on the
temperatures at the inhomogeneities as well as at the junction. Consequently,
departures from a normal cooldown curve may be useful in detecting
thermocouple inhomogeneities.
The magnitude of the temperature rise during the application of the electrical
current may also provide diagnostic information. The temperature rise due to
Joule heating increases as the heat transfer resistance increases. The causes for
increases in heat transfer resistance might be a poor fit between the
thermocouple and a thermowell or the fouling of the surface of an immersion
sensor. Since the thermocouple emf cannot be measured while the heating
current is being applied, the cooldown transient must be extrapolated back to
its initial value.
Anomalous spikes in the cooling curve have also been observed.11 These have
not been studied thoroughly, but it is suspected that they are due to abnormal
contact resistances or abnormal insulator properties.
The diagnostic possibilities for the loop current step response test arose as
troublesome artifacts that complicated the initial purpose of the test; that is,
measuring the sensors time constant. These possibilities have not nearly been
12
10
THERMOCOUPLE RESPONSE
0
0 2 4 6 8 10 12 14 16 18 20
TIME (SEC)
Figure 6-5. A Loop Current Step Response Cooldown
explored fully, but it appears that this test has potential usefulness for
thermocouple diagnostics.
6.6 Self-Validating Sensors

A self-validating (or smart) sensor has the ability to provide evidence that
decalibration has occurred. There are three approaches for self-validation.
These involve the use of multi-wire thermocouples, thermocouple/resistance
thermometer sensors and sensors with embedded melt/freeze capsules.
Practical implementations require appropriate read-out instrumentation as
well as appropriate sensors. Self-validating sensor implementation faces
several significant hurdles. These are as follows:
Inertia: Practitioners have been using temperature sensors in the

same way for decades and switching to a new, unfamiliar technol-
ogy requires that it have proven benefits.
Installation concerns: Installation errors occur even with simple

two-wire thermocouples. Some self-validating sensors have addi-
tional wires and require careful attention to set-up procedures.
Readout instrumentation: Switching to self-validating sensors

requires junking existing instrumentation and installing new
instrumentation.
Cost: Self-validating sensor/readout/wiring systems cost more

than traditional temperature measurement systems, requiring veri-
fication that the benefits justify the cost.
The consequence of these issues is that implementation of self-validating

sensors has been slow, even though the technology has been available for
many years.
The multi-wire thermocouple described in Section 4.3.4 provides up to six

temperature measurements from a single sensor. Consider a sensor with four
different wires labeled A, B, C and D. Potential measurements include the
following possible wire pairs: AB, AC, AD, BC, BD, and CD. Disagreement
between these measurements suggests decalibration of some part of the
measurement system. Using four different wires is uncommon because of
possible wire-to-wire contamination.
Manufacturers provide sensors containing two thermocouples and a

resistance thermometer. Resistance thermometers located far from readout
instrumentation require use of three or four wires to permit compensation for
lead wire resistance. Traditionally, the lead wires are made of copper. In a
dual thermocouple/resistance thermometer, thermocouple wires are used
instead of copper, as shown in Figure 6-6. This configuration permits three
separate temperature measurements, one from the resistance thermometer
and two from the thermocouples.
A B A B
RTD
Figure 6-6. An RTD/Thermocouple Self-Validating Sensor
Sensors with embedded melt/freeze capsules12,13,14 exploit the temperature

plateau of the capsule material during an increasing or decreasing
temperature of the monitored medium. The plateau occurs because
temperature remains constant during melting or freezing (as shown in
Figure 6-7). The plateau occurs at the known melt/freeze temperature of the
capsule material. Comparison with the temperature indicated by the sensor
provides a one-point calibration check.
LIQUID
TEMPERATURE
MELTING
OR
FREEZING
SOLID
TIME
Figure 6-7. Temperature Variation in a Melt/Freeze Episode
The special self-validating sensors require readout instrumentation that can

interpret the calibration information and report the result. A commercially15
available readout instrument provides three indications: calibration
acceptable, calibration questionable and calibration unacceptable. If the self-
validating concept catches on, future technology may even provide readout
instruments that can report measurement uncertainty and residual life
estimates.
References
1. Kollie, T. G., et al., Temperature Measurement Errors with Type K
(Chromel vs. Alumel) Thermocouples due to Short-Ranged Ordering
in Chromel, Rev. Sci. Inst., Vol. 46, No. 11, November 1975, pp. 1447
1461.
2. Pollock, D. D., Thermocouples, Theory, and Properties, CRC Press, Boca
Raton, FL, 1991.
4. Bentley, R. E., Thermoelectric Behavior of Ni-Based ID-MIMS
Thermocouples Using the Nicrosil-Plus Sheathing Alloy, Temperature:
Its Measurement in Science and Industry, Vol. 6, Part 1, James F.
Schooley, Editor, American Institute of Physics, New York, 1992.
5. Majesko, G., Brady, W. R., and Burley, N. A., Introduction to Type N
Thermocouples, Plant Engineering, June 27, 1985, pp. 8890.
6. Bentley, R. E., and Morgan, T. L., Ni-Based Thermocouples in
Mineral-Insulated Metal-Sheathed Format: Thermoelectric Instabilities
to 1100C, Journal Physics E: Science Instrumentation, 19, 1986, pp. 262
268.
7. Burley, N. A., N-Clad-N, a Novel, Integrally Sheathed Thermocouple:
Optimum Design Rationale for Ultra-High Thermoelectric Stability,
Temperature: Its Measurement in Science and Industry, Vol. 6, Part 1,
James F. Schooley, Editor, American Institute of Physics, New York,
1992.
8. Gibson, I. H., Optimal Selection of Thermowells, ISA Transactions,
34, 1995, pp. 209216.
9. Yung, S. K., and Clarke, D. W., Local Sensor Validation, Measurement
and Control, Vol. 22, June 1989, pp. 132141.
10. Kerlin, T. W., Temperature Sensor Response, Measurement and
Control, 14, 2, April 1980, pp. 122130.
11. Hashemian, H. M., and Petersen, K. M., Measurement of Performance
of Installed Thermocouples, Paper No. 91113, Proceedings of the 37th
International Instrumentation Symposium, San Diego, CA, May 59, 1991.
12. Ruppel, F. R. Development of a Smart Temperature Measurement
System Based on a Self-Calibrating Thermocouple, Master of Science
Thesis, The University of Tennessee Nuclear Engineering Department,
1988. (Also published as Oak Ridge National Laboratory report
ORNL/TM-11127, April, 1989.)
13. Bernhard, F., et al. Application of Self-Calibrating Thermocouples

with Miniature Fixed-Point Cells in a Temperature Range From 500oC
to 650oC in Steam Generators, pp. 16041608, Proceedings. XVII
IMEKO World Congress, June 2227, 2003, Dubrovnik, Croatia.
14. Bernhard, F. In-Situ Calibration of Inhomogeneous Thermocouples
by Integrated Miniature Fixed-Point Cells, pp. 320337 of the
Proceedings of the Estonian Academy of Science, Vol. 14, Number 4,
2007.
15. Barberree, Daniel A. Dynamically Self-Validating Contact
Temperature Sensors, ISA Temperature Symposium, Chicago, Il,
October 2124, 2002.
7
Thermocouple Selection and
Performance
The Main Points

The proper selection of thermocouples can make the difference
between good performance and unacceptable performance.
7.1 The Factors in Selecting a

Thermocouple
The performance requirements that should determine whether a
thermocouple is suitable for a particular application are as follows:
Required accuracy (see Sections 5.4 and 6.2 for information on initial
tolerances and degradation tendencies)
Required service life (see Chapter 6 for information on degradation
tendencies)
Cost (see Section 7.2)
Compatibility with existing instrumentation (see Section 3.3 for
information about instrumentation characteristics)
Speed of response (see Section 3.6 for information on measuring
transient temperatures)
The environmental conditions that affect sensor selection are as follows:
Operating temperature and temperature cycling (see Section 5.3 for

information on allowable operating temperatures and Section 4.3.5 for
information on stress-induced failures caused by temperature cycling)
127
Chemicals present in the environment (including water or water

vapor) (see Sections 6.1 and 6.2)
Flow rate of liquids or gases past the sensor (see Sections 4.7 and 6.4)
Pressure (possibly dictating the use of a thermowell, see Section 4.7)
Metals present in the process (bare sensors) or in the sheath (sheathed
sensors) (see Sections 6.1 and 6.2)
The choices available to the user in striving to select the best alternative are as
follows:
Thermocouple type
Sheathed or bare
Insulated or grounded
Wire diameter (for bare sensors)
Sheath material
Sensor diameter (for sheathed sensors)
Fill gas (for sheathed sensors)
Sensor length
Sensor configuration
Connectors
Thermowell or protection tube characteristics
Information can be found in Chapters 1 through 6 of this book. Additional

discussion of sensor selection may be found in references 1 through 3. It
should be noted that one cannot simply look in some table to determine the
best choice for a temperature measurement problem.
7.2 Costs
Users will need to consult suppliers to obtain current prices and associated
instrumentation for thermocouples, but some information will be provided in
this chapter to help you make cost comparisons. First, lets look at the
breakdown of costs for all of the components of the measurement system: the
sensor, the readout or transmitter, and the wiring. A breakdown for sheathed
base-metal thermocouples is as follows:
Thermocouple Selection and Performance 129
Component of Measurement System Percentage of Total Cost

Sensor 5 to 20
Readout or transmitter (per channel) 30 to 90
Wiring 0 to 60
These relative costs vary widely, but one important point is clear: total cost
may very well depend much more on other components in the measurement
system than on the sensor.
The cost of base-metal ASTM thermocouples of the same configuration are

about the same as for all sheathed thermocouples with 304 SS or Inconel
sheaths, and the cost per sensor is negligible for make-your-own bare sensors.
Consequently, cost usually is not a determining factor in choosing among the
various base-metal thermocouple types. However, special materials (such as
special alloy sheaths for type N) and special manufacturing methods (such as
fabrication in an inert environment to eliminate air, moisture, and
contaminants in the insulator) can improve performance but will also increase
costs.
Generally, the high-temperature thermocouples (typically tungsten and

tungsten-rhenium) cost five to ten times as much as comparably configured
ASTM base-metal thermocouples. The sheath material and insulator material
selected contribute significantly to this cost increase. The cost of noble-metal
ASTM thermocouples (R, S, B) depends heavily on the price of platinum and
thus differing from base-metal thermocouples, where the cost of the
thermocouple wires themselves is usually a negligible part of the total cost.
Typically, a sheathed, noble-metal thermocouple costs about ten times as
much as a comparably configured base-metal thermocouple.
The relative costs of the alternatives to thermocouples are also pertinent. Over
certain portions of the temperature range of thermocouple application the
user might consider a resistance thermometer or a thermistor. The relative
costs are shown in Table 7-1.
Table 7-1. Cost Comparison for Sheathed Sensors

Sensor Type
Cost Relative to
(Basis: -inch outside Temperature Range (C)
Thermocouple Cost
diameter, 12 inches long)
Base-metal ASTM sheathed 1 270C to 1300C
thermocouple
Sheathed platinum resistance 2 to 4 220C to 800C

thermometer
Sheathed thermistor 1 1/2 to 3 80C to 150C

Wiring costs may be a dominant cost consideration when comparing

thermocouples against their main competitor over a large fraction of the
applicable temperature range, namely, the platinum resistance thermometer.
The wiring for a platinum resistance thermometer is copper (as compared to
compensating extension wire for thermocouples). For comparable wire size
and insulation material, copper wire is 20 to 70 percent of the cost of
thermocouple extension wire. However, resistance thermometers require
three or four wires, compared to two for thermocouples. The lower wiring
cost associated with certain installations for resistance thermometers can
offset the lower costs of thermocouple sensors.
7.3 Thermocouple Accuracy

Earlier chapters of this book have shown that users should be concerned
about the accuracy of new thermocouples and of older thermocouples after
they have been used in industrial environments. The tolerances for
thermocouples apply only for new sensors. Departures from these tolerances
can start at the initial moment of use. The tolerances for ASTM thermocouples
(see Table 5.2) are the maximum allowable departures from the nominal
calibration for new thermocouples. Consequently, one would expect a
distribution of probabilities for errors within the tolerance band. A
hypothetical, but plausible, distribution is shown in Figure 7-1. The most
likely (highest-probability) error is zero, but errors out to the tolerance limits
are possible. Consequently, errors smaller than tolerance values are likely but
not assured. In applications where temperature measurement accuracy is
crucial the tolerances must define acceptable uncertainties in spite of the low
probability of encountering errors as large as the tolerances.
It is useful to examine the tolerances of all of the ASTM base metal

thermocouples as a function of temperature to determine the best that one can
do in terms of initial tolerances.Choosing the thermocouple with the smallest
tolerance in every temperature range for standard tolerance thermocouples
gives tolerances of 1.0C (at 0C) to 9.4C (at 1250C). Thus, one can
conclude the following:
Maximum errors due to thermocouple tolerance should be 1C to 10C for new

thermocouples.
When sensors are put into use, the tolerances no longer apply. Decalibration is
a concern. It is generally caused by chemical or metallurgical effects that
introduce thermoelectric inhomogeneities. Errors larger than 50C have been
verified. Errors this large are not common, but their possibility suggests that
Figure 7-1. Hypothetical Distribution of Errors
thermocouple performance should be checked carefully (see Section 6.5) in

applications where measurement accuracy is crucial.
The accuracy of the whole measurement systemsensor plus readoutis the

quantity of ultimate interest. Manufacturers report achievable accuracies of
0.2C to 3C for readouts. This includes emf measurement, reference junction
compensation, and conversion from corrected emf to temperature. However,
it should be noted that drift occurs in electronic systems, and periodic
checking of calibration should be performed. When decalibration is
encountered, adjustments should be made in those instruments that are
provided with an adjustment (possibly zero and span) capability. Of course,
human error (improper adjustment) is also a potential source of error.
A general assessment of achievable measurement accuracy with

thermocouple systems is as follows:
The best accuracy that can be expected for thermocouple measurement
systems is a few degrees Celsius, and the worst is a few tens of degrees Celsius
except in rare cases in which even larger errors are encountered.
7.4 Thermocouple Durability

Thermocouples can experience catastrophic failure as well as the non-
catastrophic degradation discussed in the previous section. A common failure
is wire breakage, sheath failure or thermowell failure caused by chemical
attack, thermal stress, flow-induced stress or pressure. If a thermocouple
assembly undergoes catastrophic failure, the signal disappears and there is at
least no danger of relying on incorrect measurements. Statistics on the life
spans of industrial thermocouples are hard to find in the literature. One 1973
British source indicates that the average life span of thermocouples in a
sample of 663 chemical process installations is two and a half years.4
Thermocouples can also fail or become inaccurate in ways that are not
immediately obvious, as in total loss of a signal. These are insidious and
potentially costly or dangerous. Decalibration due to creation of
inhomogeneities while in use is common. Insulation failure also occurs when
wires are exposed to unsuitable environments, and if electrical shorting
results, large measurement errors are possible. Appendix F documents some
real-world events caused by the failure of thermocouples, thermowells and
wiring.
Thermocouple failure is always a nuisance and sometimes is an event with

production, cost, and/or safety consequences. Some suggestions for reducing
failures are as follows:
Use insulated junction thermocouples rather than grounded junc-

tion or exposed junction thermocouples.
Purchase thermocouples from suppliers whose products have a

better record of back-end seal and sheath integrity. It may be nec-
essary to purchase initially from several suppliers and then evalu-
ate the thermocouples in the process. Careful purchasers may even
elect to visit the manufacturing facilities of potential suppliers to
evaluate their manufacturing and quality control procedures.
Make sure that all components of the measurement system are

compatible with the chemical and thermal environment where
they will be used.
Use available software to select thermowells that can tolerate the

environment in which they will be placed.
Anticipate possible calibration shifts due to growth of inhomoge-

neities as the thermocouple operates in a process. One may choose
to replace sensors regularly early in the life of a process. This could
lead to a performance-based replacement schedule or to switching

to a different type of thermocouple that demonstrates better per-
formance.
Establish realistic estimates of the inaccuracy of temperature mea-

surements that can be tolerated without unacceptable cost, produc-
tion or safety consequences.
7.5 Options for Achieving Performance

Requirements
Selecting a thermocouple for a given application depends on the factors
outlined in Section 7.1.
7.5.1 Required Accuracy
The accuracy tolerances of new thermocouples are given in Section 5.3, and
possible decalibrations during use are discussed in Chapter 6. The achievable
accuracy of instrumentation is given in Section 3.3. The overall assessment is
that the measurement uncertainty with new thermocouples and
instrumentation is 1C to 12C or 13C. The best accuracies are achievable
with noble-metal thermocouples and more sophisticated instrumentation (at
higher costs). Careful annealing (to ensure homogeneity along the
thermocouple length) and the calibration of individual sensors can be
employed to minimize errors in new thermocouples (this is most appropriate
for noble-metal thermocouples).
During use, errors due to chemical or metallurgical effects can introduce

thermoelectric inhomogeneities and associated measurement errors. Errors of
tens of degrees Celsius are not uncommon. The diagnostic methods described
in Section 6.5 may be helpful in detecting errors. When unacceptable errors in
base-metal thermocouples are identified, the sensors should be replaced.
Because of their value, decalibrated noble-metal thermocouples should be
carefully reannealed, and the calibration should be reconfirmed before they
are returned to service.
7.5.2 Required Service Life

The required service life of thermocouples ranges from a few seconds (for
disposable thermocouples) to as long as possible. Service life is determined by
the time before a thermocouple undergoes catastrophic failure or reaches
unacceptable inaccuracy. Conflicts between accuracy requirements and

service life requirements occur frequently because sensor degradation can
accumulate with use. The diagnostic tests described in Section 6.5 provide a
useful but imperfect means for assessing accuracy during operation and for
extending service life with increased confidence in sensor accuracy.
The literature provides little information on service life experience with

thermocouples in industry. General guidelines on achieving maximum
service life are as follows:
Bare thermocouples
1. Follow the compatibility guidelines given in Section 6.2 for
selecting thermocouple types.
2. Use heavy-gauge wire unless extremely fast response is required.
3. Use the joining procedures described in Section 4.2.
Sheathed thermocouples
1. Use sheath materials that are compatible with the thermocouple
materials. (The state of knowledge required for this is deficient but
is likely to increase. Users should follow new literature, consult
experts, and demand guidance from manufacturers.)
2. Use insulated thermocouples rather than grounded junction
thermocouples.
3. Perform an in-house comparative evaluation of thermocouples,
and use thermocouples from the manufacturer whose sensors
perform the best. Since the capabilities of manufacturers can
change, this evaluation should be ongoing.
7.5.3 Overall Cost

Thermocouple measurement system costs were addressed in Section 7.2.
Systems involving base-metal thermocouples generally cost hundreds of
dollars (not thousands) per channel in 2011. For higher accuracy, noble-metal
thermocouples can be used, and for measurements at very high temperatures
refractory metal thermocouples can be used. In both cases, the cost of the
sensor component of the measurement system increases significantly
(typically, by a factor of ten).
Of course, it is desirable to get the required measurement system performance

at the lowest cost, especially when the application requires many separate
sensors and/or frequent sensor replacements. But the largest cost may well be
lost productivity or safety caused by the inadequacies of the temperature
measurement system. Ideally, the user would know such quantities as the
value of lost production per degree of temperature measurement inaccuracy,

the value of lost production per failure of a temperature sensor, and the
increased risk per degree of temperature measurement inaccuracy. These
measures would facilitate the rational evaluation of the cost of improving
measurements versus the cost of tolerating suboptimal measurements, but
they are difficult to obtain.
7.5.4 Compatibility with Existing

Instrumentation
Measurement systems in industry can often be improved by installing new,

improved sensors (for example, the type N was developed as an alternative to
the type K), but the user must weigh the benefits versus the total cost of
replacement. The total cost may include wiring changes and the replacement
of instrumentation (transmitters, controllers, recorders, etc.).
7.5.5 Speed of Response
Users can improve the response speed of measurement systems by using

sensors and/or thermowells with smaller diameters (including sensors or
thermowells with reduced-diameter tips). Since this approach generally
results in reduced mechanical strength, the user may have to weigh the trade-
off between speed of response and durability. There are also ways to increase
the response speed using software or special hardware configurations (see
Section 3.6.3.4), but these are used infrequently.
7.6 The Future

In the future, users can expect to see thermocouples that are more accurate,
more resistant to decalibration, and more durable. The knowledge base
needed by manufacturers to build better thermocouples is growing. This
includes evolutionary improvements in the familiar thermocouples and
revolutionary advances leading to smart thermocouple measurement
systems with self-diagnostic capabilities.
Technology is also available (but not as yet implemented) that permits the
development of computer software that could provide expert advice on the
optimal choices for the thermocouple to be used in a specific application.
These advances are likely to occur and will be most beneficial to those users
who understand thermocouple operating principles.
References
3. Smith, J., Matching Temperature Sensors with Process Tasks,
Instrumentation and Control Systems, Vol. 67, April 1994, pp. 77-82.
4. Lees, F. P., Some Data on the Failure Modes of Instruments in the
Chemical Plant Environment, The Chemical Engineer, September 1973,
pp. 418-421.
INDEX
Index Terms Links
Accuracy 42 46 133
American National Standards Institute,
(ANSI) 5
American Society for Testing and Materials,
(ASTM) 5 15 89 93 177
Analytic Redundancy 118
Analytical Calibration Information 34
ASME PTC 19.3TW Thermowell Standard 83
ASTM Thermocouples 89
Color Coding 97
Compensating Extension Wires 97
Decalibration Tendencies 112
Seebeck Coefficients 95 98
Temperature Limits 96
Tolerances 93
Tungsten-Rhenium 93
Type B 93 153 159 164
Type C 93 154 160 165
Type E 92 149 157 162
Type J 91 147 156 161
Type K 92 148 157 162
Type N 92 150 158 163
Type R 92 152 159 164
Type S 92 151 158 163
Type T 91 148 156 161
This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links
Bare Sensors 51 65 113

Benedict, Robert 26
Casing 46
Celsius Scale 10
Chemical Changes
Thermoelectric Consequences 110
Chemical Compatibilities 91
Chemical Composition Changes 106
Chemical Composition of, Thermoelements 90
Chemical Stability 15
Compatibility 135
Correct Temperatures 10
Correlations 56
Cost 15 128 134
Cross-Calibration 117
Data Processing 15
Decalibration 42
Degradation 105
Differential Thermocouple 20
Diffusion through Sheath 107
Dual-Element Sensors 71
Dual-Junction Thermocouples 70
Ductility 15

Index Terms Links
Electrical Shunting 114

Electromotive Force (EMF) 12 31
EMF versus Temperature 91
Equations for 34
Environmental Conditions 127
Exposed Junction Thermocouples 67
Fahrenheit Scale 10
Fill Gas 107
First Order Approximation 51
Flow Rate 55
Gradient Thermocouples 85
Graphical Calibration Information 34
Grounded Junction Thermocouples 69
Hand Held Calibrator 41

High-Temperature Thermocouples 85 99
Immersion Sensors 51
Inhomogeneous 111 177
Installation Effects 46

Index Terms Links
Instrumentation 36 42
Insulated Junction Thermocouples 69
Insulation Resistance 114 120
Insulator Contaminants to Wire 107
Integrated Circuit Sensors 3
Internal Heating 49
International Electrotechnical Commission,
(IEC) 5 100 103
Joule Heating 121

Junction Construction Effects 19
Kelvin Scale 10
Kinetic Energy Dissipation 49
Law
Homogeneous Metals 26
Intermediate Materials 27
Successive or Intermediate,
Temperatures 27
Thermocouple Thermometry 28
Lead Wire Effects 16
Linear EMF vs. Temperature 34
Loop Analysis 9 12 45
Loop Current Step Response 121
Loop Resistance 119

Index Terms Links
Make-Your-Own Thermocouples 65
Metallurgical Changes 107
Thermoelectric Consequences 110
Metallurgical Stability 15
Migrating Atoms 108
Migration Paths 108
Moisture 107
Multiple Thermocouple Circuits 23
Negative Thermoelement 90
Noise 120
Nominal Calibration Data 34
Nonlinearity 35
Open-Circuit Voltage (OCV) 12 31 37

Open-End Compensation 36
Parallel Thermocouple Circuits 23 145

Physical Temperatures 10
Polynomial Equation Coefficients 155
Positive Thermoelement 90
Protection Tubes 79
Pulsed Thermocouples 86

Index Terms Links
Radially-Insulated Thermocouples 88
Radiative Exchanges 49
Ramp Response 53
Rankine Scale 10
Recorder 40
Resistance Temperature Detectors 3
Response 55 59
Scale
Celsius 10
Fahrenheit 10
Kelvin 10
Rankine 10
Seebeck Coefficient 12 15 18 21 26
105 115
Self-Validating Sensors 122 124
Sensor Calibrator 41
Sensor Diameter 55 57
Sensor Time Response 63
Sensors 36 48
Alternatives 2
Bare 51
Sheathed 51
Series Thermocouple Circuits 23
Service Life 133
Sheathed Thermocouples 51 66 112
Advantages 73
Disadvantages 73
Insulators for 73

Index Terms Links
Sheathed Thermocouples (Cont)

Mechanical Fittings 79
Protection for Wires and/or,
Connections 79
Sealing 78
Smart Sensors 122
Specifications 63
Speed of Response 135
Spin-Cluster Mechanism 109
Standards 5
Stem Loss Effect 46
Step Response 52
Sudden Shifts 120
Surface Insulated from Thermocouple,
Wires 76
Surface Integral to Thermocouple Circuit 76
Surface Temperature Measurements 75
Surface-Mounted Sensors 50
Tabular Calibration Information 34

Temperature 31 59 160
Controller 40
Equations 34
Gradient 11
Logger 40
Measurement 1 9 25
Measurement on Moving Surfaces 77
Scales 9
Temperature-Difference Measurement 77
Thermal Shunting 48 78
of Surface-Mounted Thermocouples 77

Index Terms Links
Thermal Stress 71
Thermistors 3
Thermocouples 31
Accuracy 130
ASTM 89
Bare-Wire 65
Configurations 65
Diagnostics 117
Dual-Junction 70
Durability 132
Exposed Junction 67
Extension Grade Wires 19
Failure 132
Gradient 85
Grounded Junction 69
Heat-Seeking 85
High-Temperature 99
Instrumentation 36
Insulated Junction 69
Loop Analysis 13
Make-Your-Own 65
Mechanical Problems 115
Parallel 145
Pulsed 86
Radially-Insulated 88
Selection and Performance 127
Series 23
Sheathed 66 113
Tables 32 147
Thermometry 1
Tube Skin 77
Types 15 89
Thermoelectric Circuits 25
Laws 26

Index Terms Links
Thermoelectric Performance 19
Thermoelectric Power 13
Thermoelectricity 25
Thermopile 23
Thermowells
Failures 82
Specifying 82
Standards 83 167
Time Constant 51 57
Transient Temperatures 50
Transients 120
Transmitter Designs 38
Tube Skin Thermocouples 77
USB Device 41
Wireless Transmitter 39
Zone Box 44

Appendix A
Hypothetical Thermocouple
Problems and Solutions
Hypothetical problems and their solutions appear below. Readers who wish
to become skilled in analyzing, designing and trouble-shooting thermocouple
applications will benefit by working through these problems and solutions.
Problem (how to interpolate between values in thermocouple tables): The

emf referenced to the ice point for a Type J thermocouple is 17.604 mv.
Interpolate between provided data in Appendix C to estimate the measuring
junction temperature.
Solution: From the Type J table, we obtain the following:
ToC V(mv)
320 17.434
? 17.604
330 17.985
The difference in emf for temperatures of 320oC and 330oC is 17.98517.434 =

0.551.
The difference in emf for temperatures of 320oC and the unknown tempera-
ture is 17.60417.434 = 0.170. Therefore, the measured emf is 0.170/0.551 or
30.9 percent of the emf variation for temperatures between 320oC and 330oC.
For this small range of temperature assume that temperature variations are
proportional to measured emf variations. Therefore,
T (for 17.604 mv) = 320 + 10x0.309 = 323.09oC
Thus, the estimated temperature is 323.09oC. This should be close to the true
value, but there is a small uncertainty because of the assumed proportionality.
137
Problem (effect of using incorrect readout instrument): It has been

discovered that a Type J thermocouple is incorrectly connected to a readout
designed for a Type K. The indicated temperature is 420oC. The readout is in a
location where the ambient temperature is 20oC. What is the true temperature
at the measuring junction?
Solution: For the Type K readout to indicate 420oC, the following relation
applies for the emf from the thermocouple and from the compensation in the
readout,
VM = VC + VT
where
VM = emf used by the Type K readout to provide the indi-

cated temperature of 420oC
VC = compensating emf produced by the Type K readout for
a temperature of 20oC
VT = emf produced by the Type J thermocouple.
To obtain the true measuring junction temperature, it is necessary to find the

actual emf produced by the Type J thermocouple. That is:
VT = VM VC
The following values of VM and VC are found in the thermocouple tables:
VM = 17.243 (from the Type K table for an indicated temperature of 420oC)
VC = 0.796 (from the Type K table)
Therefore, the emf produced by the Type J thermocouple is 17.2430.796 =

16.447. To use the Type J table to find the measuring junction temperature, we
must compensate for the open end at 20oC rather than 0oC as assumed in the
table. From the Type J table, the compensating emf is 1.019 mv. Therefore, the
emf referenced to the ice point for this Type J thermocouple is 16.447 + 1.019 =
17.466 mv.
The Type J table in Appendix C shows that this emf indicates a temperature
between 320oC and 330oC. We must interpolate to estimate the actual
temperature. The emf corresponding to 320oC is 17.434 mv and the emf
corresponding to 330oC is 17.986 mv. The measured emf is 5.8 percent of the
Hypothetical Thermocouple Problems and Solutions 139
difference in emf for this ten degree span. Therefore, the estimated actual
junction temperature is 320.58oC.
Problem (reference junction compensation): It has been said that Type B

thermocouples need no reference junction compensation because the relative
Seebeck coefficient is nearly zero for temperatures from 0oC to 40oC.
Part A. Evaluate the temperature measurement error if no reference junction

compensation is applied and if the measuring junction is at 300oC and the
open end is at 20oC.
Part B. A technician heard that Type B thermocouples do not need reference

junction compensation and assumed that this is also true for Type N. Evaluate
the temperature measurement error if the measuring junction is at 300oC and
the open end is at 20oC and no reference junction compensation is applied.
Solution to Part A: The table for Type B in Appendix C shows that the emf for
a Type B thermocouple operating between 0oC and 20oC is 0.003 mv. The
emf for a Type B with the measuring junction at 300oC and the open end at
0oC is 0.431 mv. Thermocouple loop analysis gives the following:
V(0oC 300oC) = V(0oC 20oC) + V(20oC 300oC).
Therefore, the emf from the thermocouple operating between 20oC and 300oC
is equal to 0.431(.003) = 0.434 mv. This is the emf that would be used to
convert to measuring junction temperature if no reference junction
compensation is applied. The Type B table shows that a temperature of 300oC
corresponds to an emf of 0.431 mv and 310oC corresponds to an emf of 0.462
mv. Interpolating gives an indicated temperature of 300.97oC when the emf
used is 0.434 mv. Therefore, failure to apply reference junction compensation
causes an error of +0.97 oC.
Solution to Part B: The table for Type N in Appendix C shows that the emf for
a Type N thermocouple operating between 0oC and 20oC is 0.535 mv. The emf
for a Type N with the measuring junction at 300oC and the open end at 0oC is
9.341 mv. Therefore, the emf that would be used in converting to junction
temperature is 9.2310.535 = 8.696 mv, if no reference junction compensation
were applied. The Type N table shows that a temperature of 280oC
corresponds to an emf of 8.637 mv and 290oC corresponds to an emf of 8.968
mv. Interpolating gives an indicated temperature of 281.78oC when the emf
used is 8.696 mv. Therefore, failure to apply reference junction compensation
causes an error of 18.28oC.
Problem (Calibrator voltage): A calibrator being used to calibrate a Type K

instrument is at a temperature of 20oC. What will be the voltage produced by
the calibrator when a temperature of 350oC is specified?
Solution: The pertinent thermocouple loop analysis relation is as follows:
V(0oC 350oC) = V(0oC 20oC) + V(20oC 350oC).
Using values from the Type K table in Appendix C, we obtain
14.293 = 0.796 + V(20oC 350oC)
V(20oC 350oC) = 13.497.
That is, the calibrator must generate an emf of 13.497 mv.
Problem (effect of cross-wiring extension wire): It was discovered that

Type K extension wire had been cross-wired. The indicated process
temperature was 410oC. Recent measurements show that the temperature
where cross-wiring occurred is 30oC, and it is believed that this temperature
has remained unchanged since the thermocouple was put into service. The
temperature at the readout is 20oC and this temperature also remains fairly
constant. Estimate the true process temperature when the readout indicates
410oC.
Solution: If the extension wire had been installed correctly, loop analysis
gives
V(0 T) = V(0 20) + V(20 30) + V(30 T)
where
T = process temperature.
For cross-wiring, we obtain
V(0 T) = V(0 20) - V(20 30) + V(30 T).
Note the minus sign.

Substituting known values results into the equation for cross-wiring gives
16.820 = 0.796 (1.203 0.796) + V(30 T)
V(30 T) = 16.431
Therefore
V(0 T) = 1.203 + 16.431 = 17.634
The Type K table shows an emf of 17.243 mv for 420oC and 17.667 mv for
430oC. Interpolating to obtain the temperature corresponding to an emf of
17.634 gives a measuring junction temperature of 429.22oC. Therefore, the
wiring error caused a measurement error of approximately +19.22oC.
It may seem odd that a cross-wired section that experiences only a 10oC
temperature gradient could cause an error of approximately twice that
temperature. Note, however, that cross-wiring causes subtraction from the
measured emf rather than adding the emf contribution from the cross-wired
portion. Failure to include the proper emf causes half of the error and
subtracting that emf causes the other half. So it is logical that cross-wiring
causes an error that is around twice the contribution from the cross-wired
section.
Problem (reversed connections to readout instrument): A Type N

thermocouple is connected to the wrong terminals on a readout instrument.
The indicated temperature is 240oC. What is the true temperature if the
ambient temperature at the readout is 20oC?
Solution: Figure A-1 applies if the thermocouple were installed properly:
TR A
1
TM
2
FIG. D.1 B
Figure A-1. Properly Installed Thermocouple
Note that in this correct setup, wire A is connected to readout terminal 1. The
thermocouple emf for this configuration is SAB(TMTR).
Figure A-2 applies for the improperly installed thermocouple.
TR B
1
TM
2
FIG. D.2 A
Figure A-2. Thermocouple with Reversed Connections
Note that in this incorrect setup, wire B is connected to readout terminal 1.

The thermocouple emf for this configuration is SBA(TMTR), which is equal to
SAB(TMTR).
Therefore, the emf produced by the improperly connected thermocouple is

the negative of the emf produced by a properly installed thermocouple. Now
we must consider the contribution of the reference junction compensation
provided by the readout. It is equal to SAB(TRT0), where T0 is 0oC. This
compensation is the same regardless of an error in hookup. The compensated
voltage is SAB(TMTR) + SAB(TRT0).
Now lets use these results to solve the problem. To indicate 240oC, the
compensated emf had to be 4.277 mv (from Appendix C). The compensating
emf is 0.525 mv (also from Appendix C). Therefore
4.277 = -V(TM TR) + 0.525
or
V(TM TR) = 4.277 + 0.525 = 4.802
This value is the emf produced by the thermocouple. If properly connected,

this value would be added to the compensating voltage (0.525) to give a
compensated emf of 5.327 mv. According to Appendix C, this compensated
emf corresponds to a temperature slightly above 180oC.
Note that this result shows that a cross-connected thermocouple does not give
an indication that is simply the negative of the true temperature, as might be
assumed.
Problem (misconception about cold junction compensation): A practitioner

believes (incorrectly) that thermocouple cold junction compensation involves
adding the temperature of the open end to the temperature indicated by the
emf produced by a thermocouple. Estimate the error of this approach if the
open end is at 20oC and the measuring junction is at 300oC for a Type N
thermocouple.
Solution: The emf produced by a Type N operating at 300oC is 9.341 mv, if the
open end is at 0oC (see Appendix C). The emf produced by a Type N
operating at 20oC is 0.525 mv, if the open end is at 0oC. Therefore, the emf
produced by the thermocouple operating between 20oC and 300oC is
9.341-0.525=8.816 mv. According to Appendix C, the temperature
corresponding to 8.816 mv (if the open end were at 0oC) lies between 280oC
and 290oC. Interpolating gives an estimated temperature corresponding to
8.818 mv of 285.41oC.
Following the practitioners approach, we add the open end temperature to

obtain an estimate of the temperature of the measuring junction of 305.41oC.
Therefore, the error caused by this incorrect compensation for the open end
temperature is 5.41oC.
Problem (implication of calibration uncertainty): Process yield for a

chemical operation has been found by testing to be optimum at 410oC (as
measured with a standard grade Type K thermocouple) and that the yield
drops approximately 0.05 percent for each degree above or below the
optimum. Evaluate the need for re-optimization when the thermocouple is
replaced with a new standard grade, Type K.
Solution: The tolerance on new, standard grade Type K thermocouples is

+/- 2.2oC for a measurement at 410oC. Therefore, the maximum difference for
two new thermocouples is 4.4oC. The actual difference for new thermocouples
is probably less than 4.4oC, but the initially installed thermocouple may have
experienced some degradation when the optimization was performed.
Therefore, a change of 4.4oC is a reasonable estimate of the result of changing
thermocouples. The estimated change in yield due to this measurement
change is 4.4*0.05 = 0.22%. The value of lost yield versus the cost of re-
optimization would determine the advisability of re-optimization.
Problem (incorrect extension wiring): A practitioner uses the same extension

wire for both legs of the circuit, rather than the appropriate extension wire
pair. The thermocouple is Type N, the junction temperature is 400oC, the
temperature at the transition from thermocouple wire to extension wire is
100oC, and the temperature at the readout is 20oC. What temperature will be
indicated by the readout?
Solution: The segment of the circuit containing the extension wire contributes
no voltage because both wires have the same Seebeck coefficient. The voltage
from the thermocouple may be determined as follows:
V(100oC 400oC) = V(0oC 400oC) - V(0oC 100oC)
Substituting values from Appendix C gives the following:
V(100oC 400oC) = 12.974 2.774 = 10.200 mv
The compensation voltage produced by the readout is 0.525 mv. Therefore,

the voltage used in conversion to temperature is 10.200+0.525 = 10.725.
Appendix C shows that the indicated temperature will be slightly less than
340oC.

Thermocouple Measurements

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Thermocouple Measurements

Caricato da

Copyright:

Formati disponibili

PRACTICAL

Copyright 2012 International Society of Automation (ISA)

All rights reserved.

Printed in the United States of America.

Library of Congress Cataloging-in-Publication Data in process.

The descriptions of thermocouple principles, the tools needed to analyze

One might argue that internet information now makes a book on

Chapter 3Measuring Temperature with a

Chapter 4Thermocouple Configurations 65

4.7 Thermowells and Protection Tubes 79

Chapter 5Thermocouple Types 89

Chapter 6Thermocouple Degradation 105

Chapter 7Thermocouple Selection and Performance 127

Appendix AHypothetical Thermocouple Problems and

Appendix BCircuit Analysis for Parallel

Appendix CThermocouple Tables (Temperature in C) 147

Appendix DPolynomial Equation Coefficients 155

Appendix EThermowell Selection 167

Appendix FCase Studies 171

Appendix GQuick Reference Thermocouple Guide 177

1.1 The Thermocouple

This approach is a proven technology for temperature measurement in

This short book focuses on the practical aspects of thermocouple

Figure 1-1. Schematic View of a Thermocouple

Figure 1-2. Thermocouple in Practical Applications

temperature. This simplicity of implementation is both a blessing and a curse.

1.2 The Competition

Sensor Useful Temperature Range

The competitors to thermocouples for process measurements have different

Table 1-1 summarizes the relative advantages and disadvantages of

Table 1-1. Comparison of Thermocouples and RTDs

Noncontact temperature sensors are also available. They provide

International commerce involves the movement of products across national

Thermocouple standards define the nominal performance and tolerances for

Standards serve a crucial role in industrial temperature measurement. They

1.4 Key References

13. Magison, E. C., Temperature Measurement in Industry, ISA, Research

The Main Points

Voltage is produced along the portions of the thermocouple wires

Voltage for an ideal thermocouple is related to the temperature dif-

Thermocouple loop analysis is simple and can explain all the

For temperature measurement, the quantity of interest is the open-

It does not matter how thermocouple wires are joined (twisted,

2.1 Temperature Scales

her measurements to conform to a temperature scale that is universally

The relationships between the scales are as follows:

It is customary to refer to the temperatures as degrees C, degrees F,

The temperature scales are revised periodically because scientists are

Scientific experts meet regularly to evaluate new results in their effort to

2.2 What Causes the Thermocouple

Consider first a single conductor in a temperature gradient. The conductor

One might be tempted to conclude that the existence of a potential difference

DISTANCE ALONG CONDUCTORS

2.3 The Seebeck Coefficient and

V = S (T1 T0) (2-1)

The Seebeck coefficient (also called the thermoelectric power) is the

V = SA(T1 T0) + SB(T0 T1) (2-2)

This is algebraically the same as

V = SA(T1 T0) SB(T1 T0) (2-3)

V = (SA SB)(T1 T0) (2-4)

Consequently, Equation 2-4 may be written as follows:

V = SAB(T1 T0) (2-6)

This is the fundamental relation in thermocouple thermometry.