Академический Документы
Профессиональный Документы
Культура Документы
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.
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
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.
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
vii
viii Practical Thermocouple Thermometry
Index 179
1
Introduction
1
2 Practical Thermocouple Thermometry
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.
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.
4 Practical Thermocouple Thermometry
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
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.
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
9
10 Practical Thermocouple Thermometry
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.
F = 1.8 C + 32
C = (F - 32)/1.8
K = C + 273.15
R = F + 459.67
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.
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),
12 Practical Thermocouple Thermometry
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'
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.
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:
contribution contribution
from from
conductor A conductor B
or
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,
14 Practical Thermocouple Thermometry
SAB = SA SB (2-5)
or
or
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
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.
16 Practical Thermocouple Thermometry
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:
or
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
Another setup for dealing with the reference junction temperature is shown in
Figure 2-5. Thermocouple loop analysis gives the following:
or
18 Practical Thermocouple Thermometry
or
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:
or
and
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.
or
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.
S AB V ( T 1 T 2 ) (2-23)
&
; '
&
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.
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
WIRE A
DC POWER
SUPPLY
COLD HOT
WIRE B
Figure 2-13. A Thermoelectric Heater/Cooler
WIRE A
WIRE B
Figure 2-14. A Thermoelectric Generator
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.
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 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:
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.
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.
References
1. The Theory and Properties of Thermocouple Elements, American Society
for Testing and Materials publication STP 492.
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.
5. The Use of Thermocouples in Temperature Measurement, American Society
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.
7. Kerlin, T. W., and Shepard, R. L., Industrial Temperature Measurement,
ISA, Research Triangle Park, NC, 1982.
8. Nicholas, J. V., and White, D. R., Traceable Temperatures, John Wiley &
Sons, New York, 1994.
9. Moffat, R. J., The Gradient Approach to Thermocouple Circuitry,
Experimental Technique, April 1984, pp. 23-25.
3
Measuring Temperature with
a Thermocouple
31
32 Practical Thermocouple Thermometry
ds
:
ds
<
E
Z
^
d
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.
EXAMPLE
SOLUTION
Therefore,
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.
EXAMPLE
SOLUTION
V = V(400oC 100oC)
That is, we can use the thermocouple tables (referenced to 0oC). Using the
table in Appendix C for Type J thermocouple gives
T = a 0 + a 1V (3-2)
Measuring Temperature with a Thermocouple 35
or
V = b0 + b1T (3-3)
where
T = temperature
V = thermocouple voltage (corrected for a 0C reference
temperature)
a0, a1, b0, b1 = constants
or
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.
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
36 Practical Thermocouple Thermometry
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
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.
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.
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.
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.
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.
Measuring Temperature with a Thermocouple 43
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
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.
46 Practical Thermocouple Thermometry
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:
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
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
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).
3.6.1 Basics
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).
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.
Time constant: The time required for the response to complete 63.2% of
its total change following a step change in the monitored temperature.
t
T ( t ) = T ( ) [ 1 e ] (3-9)
52 Practical Thermocouple Thermometry
where
T(t) = measured temperature change at time t
T() = final measured temperature change
= the time constant
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
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
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-
54 Practical Thermocouple Thermometry
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
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)
Measuring Temperature with a Thermocouple 55
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
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.
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.
56 Practical Thermocouple Thermometry
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)
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.
Measuring Temperature with a Thermocouple 57
Therefore,
where
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.
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).
58 Practical Thermocouple Thermometry
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
0.4
SENSOR NO. 1 - 300C, 147 BAR
T = 0.70 + 0.318 u -0.6
0.3
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
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,
where C4 and C5 are constants (assuming that sensor diameter is the only
variable under consideration).
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
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
60 Practical Thermocouple Thermometry
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).
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-
Measuring Temperature with a Thermocouple 61
loaded sensor and a thermowell with a matching tapered hole. This approach
ensures metal-to-metal contact throughout the tapered section.
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).
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:
1.4
Response 1.2
1
compensated
0.8 fast
0.6
0.4
0.2 slow
0
0 5 10
Time
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.
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
64 Practical Thermocouple Thermometry
References
1. The Use of Thermocouples in Temperature Measurement, American Society
for Testing and Materials, ASTM 470B Fourth Edition, 1993.
2. McGee, T. D., Principles and Methods of Temperature Measurement John
Wiley & Sons, New York, 1988.
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.
5. Kerlin, T. W., and Shepard, R. L., Industrial Temperature Measurement,
ISA, Research Triangle Park, NC, 1982.
6. Kinzie, P. A., Thermocouple Temperature Measurement, John Wiley &
Sons, New York, 1973.
7. Michalski, L., Eckersdorf, K., and McGhee, J., Temperature
Measurement, John Wiley & Sons, New York, 1991.
8. Magison, E. C., Temperature Measurement in Industry, ISA, Research
Triangle Park, NC, 1990.
9. Nicholas, J. V., and White, D. R., Traceable Temperatures, John Wiley &
Sons, New York, 1994.
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
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.
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.
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.
A1, B1
A1, B2
A2, B1
A2, B2
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.
15
10
RESISTIVITY (Ohm-cm)
10
10
5
10
0
10
4.5.1 Introduction
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.
4.6.1 Sealing
type of sealant to use is the temperature at the back end of the sheath, not the
temperature at the junction.
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.
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.
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
80 Practical Thermocouple Thermometry
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.
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.
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.
Thermowells can fail for a variety of reasons. However, the most common
causes of failure are corrosion, erosion, mechanical failure and vibration.
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 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.
84 Practical Thermocouple Thermometry
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.
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.
Thermocouple Configurations 85
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.
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.
Thermocouple Configurations 87
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
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
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.
References
1. McGee, T. D., Principles and Methods of Temperature Measurement, John
Wiley & Sons, New York, 1988.
2. Michalski, L., Eckersdorf, K., and McGhee, J., Temperature
Measurement, John Wiley & Sons, New York, 1991.
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
89
90 Practical Thermocouple Thermometry
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.
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
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).
5.3.1 Type J
5.3.2 Type T
5.3.3 Type K
5.3.4 Type E
5.3.5 Type N
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
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
SOLUTION
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.
^dd
^
d
:
<
E
Z
^
d
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
96 Practical Thermocouple Thermometry
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.
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
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.
Thermocouple Types 97
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
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
to about 1500C) and BeO (which limits use to about 2000C and is
poisonous).
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
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 5-9. 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.
104 Practical Thermocouple Thermometry
References
1. The Theory and Properties of Thermocouples Elements, American Society
for Testing and Materials publication STP 492.
2. Kinzie, P. A., Thermocouple Temperature Measurement, John Wiley &
Sons, New York, 1973.
3. Magison, E. C., Temperature Measurement in Industry, ISA, Research
Triangle Park, NC, 1990.
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
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
106 Practical Thermocouple Thermometry
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
SHEATH
A B
INSULATOR
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.
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
110 Practical Thermocouple Thermometry
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).
or
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:
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.
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.
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
114 Practical Thermocouple Thermometry
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:
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.
through the seal at the thermocouples back end or through holes or cracks in
the sheath).
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
Thermocouple Degradation 117
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.
6.5.1 Cross-Calibration
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).
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 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
122 Practical Thermocouple Thermometry
12
10
THERMOCOUPLE RESPONSE
0
0 2 4 6 8 10 12 14 16 18 20
TIME (SEC)
explored fully, but it appears that this test has potential usefulness for
thermocouple diagnostics.
A B A B
RTD
Figure 6-6. An RTD/Thermocouple Self-Validating Sensor
124 Practical Thermocouple Thermometry
LIQUID
TEMPERATURE
MELTING
OR
FREEZING
SOLID
TIME
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.
3. Kinzie, P. A., Thermocouple Temperature Measurement, John Wiley &
Sons, New York, 1973.
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.)
126 Practical Thermocouple Thermometry
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)
127
128 Practical Thermocouple Thermometry
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
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
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 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.
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
Thermocouple Selection and Performance 131
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.
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).
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.
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.
136 Practical Thermocouple Thermometry
References
1. Magison, E. C., Temperature Measurement in Industry, ISA, Research
Triangle Park, NC, 1990.
2. Nicholas, J. V., and White, D. R., Traceable Temperatures, John Wiley &
Sons, New York, 1994.
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
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
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
Fahrenheit Scale 10
Fill Gas 107
First Order Approximation 51
Flow Rate 55
Gradient Thermocouples 85
Graphical Calibration Information 34
Grounded Junction Thermocouples 69
Immersion Sensors 51
Inhomogeneous 111 177
Installation Effects 46
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
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
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
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
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
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
ToC V(mv)
320 17.434
? 17.604
330 17.985
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,
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
138 Practical Thermocouple Thermometry
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
VT = VM VC
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.
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:
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.
140 Practical Thermocouple Thermometry
Solution: If the extension wire had been installed correctly, loop analysis
gives
where
T = process temperature.
Substituting known values results into the equation for cross-wiring gives
V(30 T) = 16.431
Therefore
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.
TR A
1
TM
2
FIG. D.1 B
Figure A-1. Properly Installed Thermocouple
142 Practical Thermocouple Thermometry
Note that in this correct setup, wire A is connected to readout terminal 1. The
thermocouple emf for this configuration is SAB(TMTR).
TR B
1
TM
2
FIG. D.2 A
Figure A-2. Thermocouple with Reversed Connections
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
or
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.
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.
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: