PART 3 INSTRUMENTS
Temperature AND
Measurement APPARATUS
Copyright, O 1974, by
The American Society of Mechanical h g i n e e r s
Printed in the United States of America
The Scope of the work of Technical Committee No. 19 on Instruments andApparatus is to describe
the various types of instruments and methods of measurement likely to be prescribed in any of the
ASVE Performance Test Codes, Such details as the limits and sources of error, method of calibra-
tion, precautions, etc., a s will determine their range of application are given.
Only the methods of measurement and instruments, including instructions for their use, spec-
ified in the individual test codes are mandatory. Other methods of measurement and instruments,
that may be treated in the Supplements on Instruments and Apparatus, shall not be used unless
agreeable to all the parties to the test.
T h i s Supplement on Instruments and Apparatus, Part 3 on Temperature Zleasurement2 replaces
an older one published during the period from 1952-1961. Since that time the technhlogy of tem-
perature measurement h a s so changed and broadened that the earlier material has become obsolete,
This necessitated a complete revision on the Supplement resulting in the currently expanded and
more comprehensive document.
In accordance with the established policy of theAmericanSociety of Mechanical Engineers
concerning the inclusion of metric (SI or International System) units in all ASME publications,
this document includes an Appendix of appropriate conversion factors which will enable the user
to utilize both systems. These conversions are listed in the Appendix as they first appear through-
out the Supplement. Extensive use wasmade of the “ASME Orientation and Guide foruseof ‘Metric
Units, Third Edition” and The ASTM Metric Practice Guide E380-92.” These two publications
should be consulted for additional material concerning conversions from the US system to SI units.
This Editionwasapproved by thePerformance Test Codes Committeeon July 12, 1973. It
was approvedandadopted by the Council of the Society by action of theBoard on Codes and
Standards on May 29, 1974.
iii
R. F. Abrahamsen, Chairman
K. W. Woodfield, Secretary
R, P, Benedict, Fellow Engineer, STDE, Westinghouse Electric Corpo, Lester Branch Post Office
9175, Philadelphia, Pa, 19113
J. T , Callahan, Research Mechanical Engineer, Naval Ship Engineering Center, Philadelphia Divi-
sion, Applied Physics Department, Philadelphia, Pa. 19112
K. \V. hodfield, Professor of Mechanical Engineering, General Motors Institute, 1700 West Third
Avenue, Flint, Mi. 48502
K. C. Cotton, Chairman
J. H, Fernandes, Vice Chairman
TABLEOFCONTENTS
Chapter ages
1 GENERAL ...................................................................... 1
2 RADIATION THERMOMETERS .................................. 12
3 THERMOCOUPLE THERMOMETERS ............ 17
Section A, Thermocouples ........................................ 17
Section B, Instrumentation ........................................ 27
4 RESISTANCETHERMOMETERS ................................ 36
5 LIQUID-IN-GLASSTHERMOMETERS ............ 44
6 FILLED SYSTEM THERMOMETERS ............. 55
7 OPTICALPYROMETERS ............................................ 70
8 BIMETALLICTHERMOMETERS ................ 86
9 CALIBRATION OF INSTRUMENTS ............................ 91
APPENDIX ..................................................................... 134
CHAPTER 1, ,GENERAL
CONTENTS GENERAL
o
Par,
Scope
GENERAL:
Scope .........~..............,.......,,,..~...,,..~.,.,.....,.,.,....,.
Introduction
............................. 2
1
1 The purpose of this chapter is to present a sum-
TEMPERATURE SCALES ....................................... 4
INSTRUhfENTS ........................................................... 5 mary discussion of temperature measurement as re-
ACCESSORIES: lated to Performance Test Code work with particular
Wells ........................................................................ 8 emphasis on basic sources of error and means for
Other Accessories ............................................... 20
INSTALLATION ......................................................... 21 coping with them,
SOURCES O F ERROR:
Introduction
ConductionError
...........................................................
........................................................... 25
26
Introduction
Radiation Error............................................... 27 2 Measurement of temperature is generally con-
Heat Transfer at Low Velocity ........................... 28
AerodynamicHe'ating Effect ................................ 29 sideredtobeone of thesimplestandmostaccurate I
Heat Transfer at High Velocity ........................... 31 measurements performed in engineering. This is de-
GradientError ....................................................... 34 cidedly a misconception. Accurate temperature
DynamicError ....................................................... 36
CONCLUSIONS ........................................................... 38 measurement under some conditions is impossible
REFERENCES ........................................................... 39 with our present knowledge. Under many of the con-
Chapter
Range of Use, TRANSITION FROM
Type No. Deg F WELLTOATTACHMENT
SECTIONSHALLAVOID
(a) Radiation
Ambient
ther- and STRESS RAISING
DISCONTINUITIES
mometers 2 above
(6) Thermocouple
thermometers 3 -300 to "4500
(c) Resistance
thermome.ters 4 -450 to 4-1950
(d) Liquid-in-Glass
thermometers 5 -328 to tlllO
THERMAL RESPONSE
(e) Filled System PROVIDED THAT
thermometers 6 -400 t1200
to
(0 Optical
pyrometers 7 Above 1300
(g) Bimetallic
thermometers 8 -200 to 4- 800
FIG. 1.1 PERFORMANCETESTCODE
THERMOMETERWELLS
measurement, a s shown in TabIe 1.2 below. Kf = a constant obtained from Table 1.4
13 Table 1.2 is not all inclusive, but indicates The wake or Strouhal frequency i s given by:
that thermometer well design methods must care- V
fully balance these factors s o that accuracy is com- f, = 2.64 - (2)
B
promised a minimum when using a well of adequate where I
strength.
f, = wake frequency, cycles per sec
14 DesignProcedure. The purpose of this design
V = fluidvelocity,fps
procedure is to enable the user to determine if a B = diameter at tip (Fig. 1.11, in.
well selected for thermometry considerations is
strong enough to withstand specific application con- The ratio of wake to natural frequency ( f w / f n )
ditions of temperature, pressure, velocity and vibra- shall not exceed 0.8, and when this condition is
tion, meIl failures are caused by forces imposed by met, the Magnification Factor, relationship of dy-
static pressure,, steady state flow, and vibration. namic to static amplitude is given by:
Separate evaluations of each of the above effects
should be made in order to determine the limiting (fw/fn)Z r2
condition. This design procedure does not allow for ' =ï - = m ; ) z = G 2 (3)
effects due to corrosion or erosion. For r 5 0.8
15 The natural frequency of a well designed in where
accordance with Fig. 1.1 and of the dimensions F,,, -- magnification factor, dimensionless
given in Table 1.3 is given by the following equa- r = frequency ratio, ( f w / f n ) , dimensionless
tion :
16 Stress Analysis. The maximum pressure that
a thermometer well can withstand for a given mate-
rial at a given temperature shall be computed from
Mass
Velocity High Low
Increased heat transfer. Reduces impingement forces. Lower
Faster response. Karman trail vortex frequency.
Solution:
Step 1-Obtain the necessary data as follows:
L,,,= 46.8
- J0.3353
- -
(13,100 0.389 x 2400)
300 1 -b 0.658
= 7.7 in. > 4% in. (satisfactory)
Result: All the requirements for frequency and stress have been met, therefore, this thermom-
eter well is satisfactory for the intended service.
19 Thermometer wells as shown in Fig. 1-1, are Several schemes may be employed a s follows:
expected to satisfy 95 percent of the present well
problems. Increased temperature-pressure ratings (a) By surrounding the sensing junction with one
of central stations now await advances in met- or more coaxial tubes mounted in the direction of
allurgy, Such materials when available will also gas flow [8,9,10]. This arrangement screens the
increase the ratings of the thermometer wells. For sensor from radiation exchange with the surrounding
services where these thermometer wells are not surfaces.
now suitable, the use of a fixed beam type of well (S) By increasing the convective heat transfer
described in Ref. 171 is recommended. Interest rate from the gas to the sensor thereby minimizing
now centers on velocities of 300 fps or less. the effect of radiation losses. This may be accom-
Higher velocities give rise to consiilerable dif- plished by using a suction or aspirating type pyrom-
ferences between stagnation and static tem- eter [11,121. A special case of this principle is em-
peratures. For velocities in excess of 300 fps the ployed in the sonic-fIow thermocouple pyrometer in
fixed beam type of well is recommended. which the gas flow -over the sensor i s maintained a t
sonic velocity [13],
( c ) By covering the sensing junction with a low
Other Accessories emissivity radiation shield mounted directly onto
the junction [14]. This arrangement has been found
20 When it i s necessary to place a temperature
effective. in minimizing radiation losses where
sensing element in a gas or vapor a t a location
space limitations prevent the use of the coaxial-
where i t can "see" surfaces at materiaIly higher or
type screens.
lower temperatures than that of the medium in which
i t i s immersed, accessories may be used to minimize For a more complete discussion of this subject,
errors arising from radiation under such conditions. see Ref. [IS].
6
k
Y --
_" - "
. ~
~~ ~ ~ - - -~.
~
-a- ~~
i
-)
COPYRIGHT American Society of Mechanical Engineers "" ..~ "
B 7
T, = temperature a t point of attachment (e.g. equal to the normal total emissivity (cS) of the sen-
vessel wall), deg F sor. Also, since all the energy radiated by the sen-
sor is intercepted by the surrounding surfaces, the
L = immersion length of temperaturesensor, f t
configuration factor ( F A ) from the sensor to its sur-
m =: (hp/ka)'", ft" roundings i s equal to unity. A s a result, Eq. (7) may
be simplified as follows:
h = convectivecoefficient of heattransfer,
Btu/hr ft2 deg F
p = perimeter of temperaturesensor,ft
L d
k = thermalconductivity of temperaturesensor,
Btu/hr ft deg F Heat Transfer at L o w Velocity
a = conductioncross-sectionalarea of tempera-
28 Consider the case of a temperature sensor
ture sensor, ft2
exposed to a low velocity (i.e,, no aerodynamic
heating) gas stream with the sensor experiencing
Since the hyperbolic cosine of the mL product in-
both radiation and conduction effects, For the
creases as the product itself increases, it follows
steady-state condition between the flowing gas and
that the larger mL becomes the closer the indicated
the sensor, heat transfer by convection must equal
temperature, Ti, approaches the static temperature the rate of heit transfer by radiation and conduction.
of the gas, T s g , (i.e., the conduction error is re-
duced). As a consequence, any means of increasing This equilibrium condition may be written a s
mL product will result in a decreased conduction
follows:
error.
RadiationError
at the temperature sensing element, In the case ation effects, the temperature, Ti,indicated by the
where radiation, conduction, and aerodynamic heat- sensor may differ from the adiabatic temperature. In
ing occur simultaneously, the temperature indicated other words, all four temperatures Ti, T s g , T t and Ta
by the temperature sensor will be dependent upon generally have different values. For this case, the
the corresponding magnitudes of thes.e three heat applicable steady-state relation is a s follows:
transfer effects. If the moving gas stream is brought
to rest isentropically at the temperature sensor, the
resulting localized temperature of the gas stream is
called “total,” or “stagnation,” temperature, Tt.
The total temperature would be higher than the
L
static temperature because of the conversion of
where
kinetic energy to internal energy. These two temper-
atures are related in the following manner: he = effectiveconvective heat transfercoefficient, .
Btu/hr ft2OF.
Tt - T,, =: V ~ / ~ J R C C ~ (10) L201
The effective coefficient, her which. is primarily de-
where pendent upon flow regime, geometric configuration,
and orientation may be calculated through use of ap-
V = gas velocity, ft/sec propriate convection correlations. Further informa-
J = mechanical equivalent of h e a t t 7 7 8 f t 1bJBtu tion may be obtained from Refs. [22, 241. When cal-
culating the temperature error for the above case,
gc = dimensional constant = 32.1740 lb,ft/lbtsec2 the adiabatic temperature, being the unknown quan-
cP = specific heat at constant pressure, tity, is determined through use of Eq. (1% Equa-
Btu/lb, deg F tions (10) and (11)may then be used for determining
the static temp.erature of the gas.
30 Whenever the kinetic energy of the gas stream
is reduced, the conversion of kinetic energy to 32 At the relatively low velocity of 300 ft/sec the
internal energy is manifested by a localized rise in difference between the total and static temperatures
gas temperature a t the temperature sensor. This tem- is only 7 deg F, but a t a velocity of 1100ft/sec,
perature rise results in heat transfer from the local- approximately Mach 1 a t 40%, the difference intem-
ized region to the surrounding gas stream, as well perature increases to40 deg F. At avelocity of
a s to the sensor. If the sensor experiences no heat 22OOft/sec, approximately Mach 2 a t 40%, the dif-
transfer; an adiabatic condition, the temperature ference is 400 deg F, thus emphasizing the signif-
which the sensor then assumes is defined a s the icance of aerodynamic heating a t high velocities.
66
adiabatic temperature,” T,. For convenience, the 33 The standardized Performance Test Code wells
ability of a temperature sensor to “recover” the dimensioned in Table 1.3 are recommended for u s e
converted kinetic energy of the gas stream is de- with velocities of 300ft/sec or less. A recovery
fined in terms of a “recovery factor,” r, as follows: factor for air of 0‘65 should be used for these wells
at this velocity or lower. For higher velocities, a
r = (T, - TSg) / (Tt - T,& (11) DOI fixed 6eam type of well or temperature measuring
devices designed expressly for high velocity flow,
-
The rise in temperature, ( T t Tsg),of the s t a g such as stagnation thermocouple thermometers,
nated portion of the gas stream during isentropic should be used, L7, 8, 211.
slowing of the gas stream may be calculated by
means of Eq. (10). The recovery factor is primarily
dependent on geometric configuration, orientation, Gradient Error
and Mach number. For a more complete discussion of 34 In temperature measurements in a system where
this subject see Refs. [21, 22, 23, 261. there are velocity, density, or temperature gradients,
such gradients will be taken into account if the tem-
Heat Transfer at High Velocity perature measurement is used to determine the
31 In the case where aerodynamic heating occurs
and the temperature sensor has conduction and radi- * See footnote under Eq. 8.
energy of the medium. In such cases the value of the temperature of the sensor, in general, will be
the bulk temperature i s desired, which would be the out of phase with the temperature of the medium
temperature realized if the flow could be interrupted being measured,during a transient condition. In ad-
and the material thoroughly mixed without gain or dition to being out of phase it will also be of differ-
loss in energy (Le., an adiabatic situation). In prac- ent magnitude.
tice the desired result can be obtained by making 37 The ability of a temperature sensing element
s e t s of measurements at a sufficient number of
to respond to a change in temperature is typically
points to permit numerical integration.
given in terms of i t s "time constant." This is the
35 A number of measuring stations and sensing time required for the element to change in tempera-
elements shall be selected depending on the rela- ture an amount equal to 63.2 percent of the im-
tive magnitude of the gradients and the accuracy de- posed step-change. Therefore, a temperature sens-
sired, For each station, the local temperature, ing element having a short time constant for a given
density, velocity, and flow area shall be evaluated, physical situation will respond more rapidly to a
The value of the bulk temperature is related to the change in temperature than if i t had a long time con-
local quantities though the following equation: stant. A s a result, temperature error due to dynamic
response becomes more significant for elements
having long time constants, The response of a
temperature measurement system depends on the
following five maior parameters:
10
CONTENTS the energy per unit area per unit time radiated by
the body.
Par.
4 A selective radiation thermometer is one which
GENERAL:
Scope ........................................................................ 1 utilizes as an index of the temperature of a body the
Definitions .............................................................. 2 energy from only a narrow wavelength band or bands.
PRINCIPLES OF OPERATION ................................ 7
CLASSIFICATION: 5 A blackbody is one that absorbs all radiation
Description .............................................................. 13 incident upon it, reflecting or transmitting none.
Materials of Construction ...................................... 18
CHARACTERISTICS: 6 Emissivity is the ratio of the radiant energy
Range ...................................................................... 20
Sensitivity .............................................................. 21 emitted per unit time and per unit area by a body,
Precision ................................................................ 22 to that emitted by a blackbody at the same temper-
Accuracy .................................................................. 23 ature. Total emissivity refers to radiation of all
Response ................................................................ 24
ACCESSORIES ............................................................ 25 wavelengths, and monochromatic or spectral emìs-
APPLICATIONANDINSTALLATION: sivity refers to radiation of a particular wavelength.
Sources of Error .................................................... 28
EssentialConsiderations .................................... 34
Treatment of Data .................................................. 35 PRINCIPLES OF OPERATION
ADVANTAGESANDDISADVANTAGES:
Advantages ............................................................ 36 7 The operation of a radiation thermometer de-
Disadvantages ........................................................ 37
pends on the phenomenon that a body at elevated
temperature emits radiation, the intensity of which
bears a definite relation to the temperature of the
GENERAL body. The temperature of a body may be determined
from a mea'surement of the intensity of the radiation
Scope emitted. This measurement may involve the radia-
tion of all wavelengths emitted, or the radiation in
1 The purpose
. . of this chapter is to presen I t in- a restricted portion or portions of the spectrum.
formation which will guide the user in the selection, However, in general, the intensity of radiation de-
installation, and use of radiation thermometers. pends not only on the temperature of the source,
but also on the particular material constituting the
Definitions source. Thus, glowing carbon appears to the eye
approximately three times a s bright as glowing
2 A Radiation Thermometer consists of an optical platinum when both are at the same temperature.
system, used to intercept and concentrate a definite T h i s is technically expressed by the statement that
portion of the energy radiated from a body whose the emissive power or emissivity of carbon is ap-
temperature i s being measured; a temperature sensi- proximately three times that of platinum. A material
tive element, usually a thermocouple or a thermo-
having the highest theoretically possible emissivity
pile; and a measuring device, usually an emfmeas-
is known a s a blackbody radiator, or by a single
uring instrument. term "blackbody." By definition, the emissivity of
3 A total radiation thermometer is one which uti- a blackbody is unity. All materials have an emis-
l i z e s a s an index of the temperature of a body all sivity less than unity. A blackbody is.experimen-
c, X - 5
Wb = CLASSIFICATION
eC2/XT -. 1
Description
where
13 Radiation pyrometers have taken many forms,
Wb = radiant energy per unit time, per unit partly as a result of the preferences of the various
of wave length interval, by a black- designers, and partly because of the diversity of
body of unit area the uses to which the thermometers were tobe
CI,Cz = constants in the Planck radiation law adapted, At least theoretically, rate of radiation of
energy may be u s e d a s a measure of temperature
X = wavelength of radiant energy from the region of absolute zero up to the highest
e = base of natural or Napierian logarithms, temperatures found in the sun and stars. A single
measuring.instrument is not capable of making
measurements throughout this range because of the
10 As a consequence of the Stefan-Boltzmann varying amounts of energy available, the varying
law, it is possible to measure the temperature of a degrees of absorption of the emitted wavelength by
source by measuring the intensity of the radiation the media which transmit or reflect the radiation,
that it emits. This measurement is accomplished and the characteristics of the receiver itself.
by focusing energy radiated from a source at a uni-
form temperature, on an absorbing area, the re- 14 The most conspicuous and fundamental clas-
ceiver, which is heated by the incident radiation sification of radiation thermometers is that based
absorbed by it, The temperature of this receiver on their means of collecting the radiation and
rises until its rate of heat loss to its surroundings focusing it on the receiver. There are three types
by conduction, convection, and radiation is equal in use-the single mirror, the double mirror, and
to its rate of absorption of energy from the source. the lens.
In most radiation thermometers the rate of heat loss 15 Single Mirror Type, In the single mirror type,
from the receiver is such that equilibrium is reached radiation from the source enters the optical system
before i t s temperature is much above that of i t s through the aperture in the “front diaphragm.” It is
surroundings, even when the source is brightly in- reflected from a concave mirror at the other end of
candescent, an enclosing tube and is focused on the receiver of
13
ACCESSORIES
15
31 It is sometimes desired to measure the tem- absorption, and calibration. Corrections at other
perature of a body inside a furnace which is sealed than calibration temperatures should be determined
to maintain an atmosphere of hydrogen or other gas. by linear interpolation.
Windows provided for temperature measurements
introduce large errors, whether they are made of ADVANTAGES AND DISADVANTAGES
glass or quartz. Such errors must be determined ex-
36 Advantages. In the range of the thermo-
perimentally if possible and corrected. Closely re-
couples, the radiation thermometer has the follow-
lated to this type of error is the error caused by
ing advantages:
deposits of dirt or foreign material on the windows,
lenses, or mirrors of the radiation thermometer. (a) They are not subject to deterioration by fur-
This error can be eliminated by periodic cleaning. nace atmosphere and high temperatures.
( b ) There is no contact with hot body required,
32 Errors caused by fluctuations in ambient tem-
making it possible to measure the temperature
perature are sometimes enco-untered. Most manu-
of moving objects.
facturers provide ambient temperature compensation
(c) They can be used for measurement of surface
built into the radiation thermometer which conpen-
temperatures.
sates for these fluctuations to within stated limits.
(d) They can be used for measurement of high
33 When sighting into a closed-end tube, it i s im- temperatures.
portant that the bottom of the tube is focused on (e) They can be s o located a s not to be subject to
the diaphragm of the radiation thermometer, and vibration or shock.
that. its image is centered on the aperture in the ( f ) They can be used where fast response and
diaphragm. Otherwise the temperature reading may long, useful life are required.
be affected by variations in temperature along the
In the range of the optical pyrometer, the radia-
walls of the tube.
tion thermometer has the following advantages:
(a) Recording and/or controlling of temperatures
-is obtainable.
Essential Considerations
(b) The personal element does not enter into the
34 An examination of the site where the temper- measurement.
ature measurement is to be made will determine ( c ) Lower temperatures can be measured.
the focusing distance required. Manufacturers’
literature or representatives should aid in deter- 37- Disadvantages. In the range of the optical
mining the desired components. pyrometer or thermocouple thermometer, the disad-
vantages of the radiation thermometer are:
(a) It has a higher first cost.
Treatment of Data
( b ) When sighting froM a distance, large sources
35 The observed temperature readings should be are required.
corrected by adding corre.ctions for emissivity, (c) In general, the accuracy is lower.
CONTENTS ACCESSORIES 29
APPLICATION AND INSTALLATION:
Sources of Error ...................................................... 31
Section A, Thermocouples Essential Considerations .................................... 34
Treatment of Data .................................................. 40
Par. ADVANTAGES AND DISADVANTAGES:
Advantages .............................................................. 41
Disadvantages ........................................................ 42
REFERENCES ............................................................ 43
1
2 Section B, Instrumentafion
6
GENERAL .................................................................... 1
7
21
PRINCIPLESOF OPERATION ................................ 7
DESCRIPTION:
Potentiometer Circuits .......................................... 9
25 Reference Junction Compensation ........... 28
26 Automatic Null-Blancing Mechanism .......... 43
27 . Types of Indicating Potentiometers .......... 51
28 Reference Junction Apparatus .............. 67
SECTION A, THERMOCOUPLES
GENERAL cept where ioined together to form junctions. There
are necessarily two junctions to each thermocouple
. Scope corresponding to the two extremities of the thermo-
elements, The measuring junction is that which is
1 The purpose of this chapter is to present in- subjected to the temperature to be measured. The
formation which will guide the user in the selec-
reference junction is that which is a t a known tem-
tion, installation, and use of thermocouple ther-
perature. (Note: For practical reasons, the refer-
mometers. ence ìunction is usually formed in two parts, a s
shown in Fig* 3.%)
Definitions
4 Connecting Wires are a pair of electrical con-
2 A Thermocouple Thermometer is a temperature ductors which, connect the thermocouple to the emf
measuring system comprising a temperature sens- measuring device. These are generally of copper
ing element called a thermocouple which produces when the reference junctions of the thermocouple
an electromotive force (emf), a device for sensing are maintained a t some fixed temperature such as
emf which includes a printed scale for converting 324;' (ice point).
emf to equivalent temperature units, and electrical 5 Thermocouple Extension Wires are a pair of
conductors for operatively connecting the two. wires having such temperature-emf characteristics
See Fig. 3.1A.
relative to the thermocouple with which the wires
3 A Thermocouple is the temperature sensing ele- are intended to be used that, when properly con-
ment of a thermocouple thermometer comprising two nected to the thermocouple, the thermocouple refer-
dissimilar electrical conductors called thennoele- ence iunction is in effect extended to the other end
ments, electrically insulated from each other ex- of the wires,
MEASURING
I using any other material to form the junction. The
noble metals should be welded without using a
INSTRUMENT
flux. For torch welding base metals it is advanta-
geous to use a flux to minimize oxidation. Care
HTHERMOCOUPLE WIRES
should be exercised in the application of heat to
the ends of the thermoelements to avoid overheat-
ing. All traces of flux should be removed after the
welding process. Further details on the welding of
thermoelem-ents may be found in ISA Recommended
I d ¡ Practice RP 1.1-.7, and in the ANSI Temperature
I MEASURING
INSTRUMENT Measurement Thermocouples C 96.1.
12 Although most of the metals and alloys used 16 %h'en it is inconvenient to use an ice bath, a
a s thermoelements exhibit a relatively high degree thermally insulated block of copper, aluminum or
of mechanical stability, the majority of them are silver, or a stirred liquid bath may be substituted.
subject to change in calibration when exposed to In such case, the temperature of the reference junc-
contaminating and corrosive conditions. Protection tion must be measured with an auxiliary instrument
tubes which are impervious to such conditions are and taken into proper account.
therefore required if hiih accuracy and long life
17 Frequently some accuracy is sacrificed for
are required, Platinum versus platinum rhodium
convenience by eliminating the constant tempera-
thermocouples are particularly susceptible to con-
ture reference junction in favor of automatic refer-
tamination and should be protected by ceramic
ence junction compensation built into an indicating
tubes which are impervious to gases and vapors at
or recording instrument used to measure the emf de-
temperatures within the working range. T o provide
veloped by the thermocouple. Recently, automatic
additional mechanical strength, a refractory metal
reference junction controls, external from the emf
tube is often placed over the ceramic tube.
measuring instrument, have become available.
These provide either the ice point temperature or
13 Protecting wells are employed when the ther- various elevated temperatures, usually within
mócouple is used in liquids and gases at high +0.5%.
' pressure. They are made of metal, and may be
turned and drilled from bar stock or built up by 18 ThermocoupleExtensionWires. Where the
welding. They are usually fabricated of carbon highest accuracy is desired, thermocouples should
steel, stainless steel (18-8)or 14 percent chro- be long enough to connect directly to the instrument
mium iron. Thermocouple protection tubes for use or reference junction apparatus. This will eliminate
in molten metals, furnace atmospheres, salt baths errors which might be introduced by the use of ex-
and chemical processes must be selected to with- tension wires pot having temperature-emf character-
stand the particular conditions and hazards pre- istics identical to those of the thermocouple. M e r e
vailing in the installation, this is not feasible, extension wires from the ther-
mocouple reference temperature may be used. For
base-metal thermocouple installations, the exten-
14 In providing protection for the thermocouple, sion wires are either the same or nominally the
however, one must not lose sight of the fact that a same as the thermocouple materials. Extension
thermocouple can perform its function only when the wires must have the equivalent emf-temperature re-
conditions of heat transfer are such that the meas- Iation over the temperature range to be encountered
uring junction attains or at least approaches the in service.
temperature to be measured.
19 Because of the high cost of platinum and
15 ReferenceJunction. The most satisfactory platinim rhodium alloys, substitute materials are
and easily reproducible method of reference junc- generally used a s extension wire for thermocouples
tion control is realized by a n i c e bath, consisting made from these metals. A specially matched pair
of an intimate mixture of shaved ice and water. A of conductors, consisting of a copper wire and a
nickel-copper alloy wire, have found general use
wide-mouth Dewar jar provided with a cork will
as extension wires for the platinum v s platinum
serve to maintain the ice bath over several hours.
rhodium thermocouple over the temperature range
For each junction to be kept at the ice point, a
32 to 400%'. The copper wire is joined to the plati-
small glass tube, closed at the bottom and flared a t
num rhodium thermoelement and the copper-nickel
the top, is inserted through a hole in the cork,
alloy wire to the platinum thermoelement, These
Enough mercury is placed in each tube to fill it ap-
wires do not match the individual thermoelements,
proximately K in, The tubes are inserted s o that
but when used together they compensate reasonably
the mercury is several inches below the surface of
well over the range specified above. It is important
the ice-water mixture. The actual reference junc-
that the junctions of the extension wire and the
tion is made by inserting a thermoelement and a
thermoelement be a t the same temperature.
copper wire, each insulated in such a manner that
electrical contact between them is made only below 20 SwitchesandTerminalBlocks. All switches
the surface of the mercury. The copper connecting used with thermocouples should be of a rugged
wires are extended to the emf measuring instrument. construction and designed s o that both wires are
INSTRUMENTS A N D APPARATUS
slightly greater mechanical strength than the 24 SpecialPurposeMetals. Combinations of
conventional Platinum versus Platinum metals other than those listed above are sometimes
Rhodium thermocouples. used for special purposes, These include chrome1
versus stainless steel, nickel versus nickel molyb-
23 BaseMetals denum, and platinum rhodium versus gold palla-
dium. Each has advantages for particular applica-
(a) Copper-Constantan. (Type T) Constantan is an
tions,
alloy of approximately 55 percent Copper, 45
percent Nickel. The copper-constantan thermo-
couple is widely used in industrial and lab-
CHARACTERISTICS
oqatory applications over the temperature
range - 300 to +7OO0F. 25 RangeandAccuracy. The upper temperature
(¿J) Iron-Constantan. (Type J) This thermocouple limits for the various thermocouples depend on the
is probably the most widely used of all thermo- wire s i z e s and the environment in which the ther-
couples in industrial thermometry. I t is gen- mocouples are used. Table 3.1A lists the recom-
erally limited to the temperature range -200 mended upper temperature limits for thermocoupIes
+
to 1400%' but may be used up to 1800%' a t protected from corrosive or contaminating atmos-
a sacrifice of life. For the higher temperatures, pheres. The ranges of applicability and limits of
wire s i z e s of number 8 gage or larger are gen- error for thermocouples and extension wires of
erally employed. For temperatures up to 1400%', standard sizes are given in Table 3.2A. The cor-
iron-constantan thermocouples show good cal- responding values of temperature and emf for the
ibration stability in nonoxidizing atmospheres. various types of thermocouples are given in Table
3.3A. See National Bureau of Standards. Monograph
(C) Chromel-Alumel. (Type K) Chromel P i s an 125 for expanded reference tables of these thermo-
alloy of approximately 90 percent Nickel, 10 couples, emf versus temperature, and temperature
percent Chromium. Alumel has a composition versus emf, deg F and deg C.
of about 94 percent Nickel,3 percent Man-
ganese, 2 percent Aluminum, 1 percent Silicon. 26 Sensitivity. The sensitivity of athermocouple,
This thermocouple, usable over the temperature that is dE/dT, varies somewhat with temperature,
range -200 to 2300%', and higher for short The following table lists the average thermoelectric
time intervals, is more resistant to oxidation power for the conventional thermocouples.
than any other base-metal combination. It must
be protected against reducing atmospheres.
I
Alternate oxidizing and reducing atmospheres
Average dE/dT, Micro-
are particularly destructive. Both thermoele- volts per Deg F for
Thermocouple
ments are mechanically strong and are often the Range Specified
directly exposed to the temperature environ-
ment. Copper-Constantan 28.0 (32-650%)
Iron Constantan 32.0 (32-1400%')
(d) Chromel-Constantan.(Type E) This combina- Ghromel-Alumel 23.0 (322200%')
'. tion of thermoelements develops the highest Platinum-10% Rhodium
thermoelectric output of any of the conven- -Platinum G , 3 (1000-2650%')
tional thermocouples, namely, about 34 pV per Chromel-Constantan 42.0 (32-1400%')
deg F a t normal ambient temperature and in-
creasing to about 45 pV per deg F a t 1000°F.
This high output has led to the use of Chromel- 27 Precision. The precision of measurement at-
Constantan as sensing elements in thermopiles tainable with thermocouples depends primarily upon
for radiation detection and in differential the temperature range and upon the experimental
thermocouple systems. The thermocouple h a s techniques employed. A l l thermocouples show a
also found general application for temperature gradual drift in calibration with operating service.
measurements up to about 1400'F. It is char- Small diameter thermoelements (less than 0.010 in.
acterized by a high degree of calibration diameter) are particularly susceptible to change in
stability when used at temperatures not ex- calibration when used near their upper temperature
ceeding 1000%'. limit.
28 Response. The time response of a therrno- prevent stray currents in the conductor from enter-
couple to a temperature change is a function of the ing the thermocouple circuit and vitiating the read-
mass and geometry of the sensing junction and the ings.
mode or modes of heat transfer between the junc-
32 Inasmuch as the curves giving the relation
tion and its surroundings. The greater the surface
between emf and temperature are not, in general,
area to mass ratio of the junction, the greater will
linear, equal increments of temperature do not cor-
be the speed of response for a given heat transfer
respond to equal increments of emf. This should be
condition, "hen a thermocouple is encased in a
particularly observed in applying reference junction
well or protecting sheath, the response indicated
corrections. Many commercial instruments used in
by the probe will be governed by the combined ef-
conjunction with thermocouples are provided with
fects of heat transfer from the medium to the well
either manual or automatic means for compensating
and that of the well to the enclosed junction.
for deviations of the reference junction temperature
from that on which standard calibrations are based.
ACCESSORIES
33 In the event the emf measuring instrument. has
automatic reference junction compensation, ther-
29 The minor accessories used in conjunction mocouple extension wires should be used to con-
with thermocouples, such a s protection tubes, ex- nect the thermocouple to the instrument.
tension wires, switches and terminal blocks have
been discussed in earlier paragraphs of this chap-
34 EssentialConsiderations. Regardless of the
type of thermocouple or the techniques employed
ter.
in carrying out the measurements, there are certain
30 A detailed discussion covering instrumen- basic factors which must be considered. The pri-
tation employed to measure and record the emf sig- mary consideration is that the temperature indicated
nals from thermocouples is given in Section B, by a thermocouple is that of the measuring junc-
Thermocouple Thermometry Instrumentation. tion. The accuracy obtained in measuring the tem-
perature of any object or space usually depends
APPLICATION AND INSTALLATION upon how closely the measuring junction of the
thermocouple can be brought to the temperature of
31 Sources of Error. Experience has shown that
the object or space, or to some temperature which
there is a definite and reproducible relationship
is definitely related to that of the object or space.
between the difference in temperature of the meas-
uring and reference junctions of a thermocouple and 35 A small s i z e thermocouple measuring junction
the emf developed. If the reference junction is suitably imbedded in a solid or immersed in a
maintained a t some known temperature, such a s liquid will attain equality in temperature with the
3 2 9 (ice point), the emf developed by the thermo- substance and will, therefore, indicate the true
couple can be determined a s a function of the tem- temperafure of the solid or liquid to within the
perature of the measuring junction. Thus the device calibration accuracy of the instrument. However, in
may be calibrated and used for measuring tempera- many applications this may not be the case. If
tures. It is not necessary to maintain the reference under steady conditions there is a net exchange of
junction temperature, during use, the same a s dur- heat between the thermocouple junction and the
ing calibration, However, the reference junction substance, then a difference in temperature will
temperature in each case must be known. Tempera- exist between the two. The magnitude of this dif-
ture emf tables for thermocouples are usually based ference in temperature depends upon the rate of
on a reference junction temperature of 32%' (ice heat transfer and the thermal resistance between
point), the junction and substance. A s an illustration, s u p
The elements of a thermocouple must be electri- pose i t is desired to measure the temperature of a
cally insulated from each other, from ground and metal plate which is heated from within by some
from conductors on which they may be mounted, ex- means. The bare thermocouple measuring junction
cept a t the measuring ,junction. When a thermo- is brought into contact with the metal plate. The
couple i s mounted along a conductor, such as a junction will receive some heat from the plate by
pipe or a metal structure, special care should be thermal conduction. The junction will lose heat by
exercised to ensure good electrical insulation be- conduction along the thermocouple wires, and by
tween the thermocouple wires and the conductor to convection, conduction and radiation to' the sur-
24
i
.
i
(u) A relatively small signal output is produced Other references on the general subject of thermo-
requiring sensitive measuring equipment. couples:
(b) Knowledge of or compensation for reference 66
Thermoelectric Thermometry,” Wm. F. Roeser, Jour.
junction temperature is required. Applied Physics, vol. 11, no. 6, p. 388, 1940.
( c ) Subject to calibration changes with use. “Thermoelectric Thermometry,” Paul H. Dike, Leeds
& Northrup Co,, 1954.
“Thermoelectric Effects,” R. P. Benedict, Electrical
Manufacturing, p. 103, Feb. 1960.
66General Principles of Thermoelectric Thermometry,”
REFERENCES
-
D. I. Finch, vol. 3, part 2, of 66Temperature I t s
Measurement and Control in Science and Industry,”
Reinhold, New York, 1962.
43 Throughout the text Reference numbers are “Manual on the Use of Thermocouples in Temperature
enclosed in brackets, thus [lI. Measurement,” S T P 470.
..
”
“
i
INSTRUMENTS A N D APPARATUS
SECTION B, INSTRrJMENTATlON
GENERAL 8 The potentiometer, a s t h e term is used here,
serves a similar function in the measurement of
1 The basic principle of thermoelèctric thermom- voltage, and in fact may be called a “voltage bal-
etry is that a thermocouple develops an efm which ance,” the standard voltage being furnished by a
is a function of the difference in temperature of i t s standard cell, the “lever” being resistance ratios,
measuring (hot) and reference (cold) junctions. If and the galvanometer serving as the balance indi-
the temperature of the reference junction is known, cator. Since no current is drawn from the standard
the temperature of the measuring junction can be cell or the measured source at balance, the meas-
determined by measuring the emf generated in the urement is independent of external circuit resist-
circuit. The u s e of a thermocouple in temperature ance, except to the extent that this affects gal-
measurements therefore requires the use of an in- vanometer or balancing mechanism sensitivity.
strument capable of measuring emf.
DESCRIPTION
2 While there are two types of emf measuring in-
struments in common u s e in industry, millivolt-
meters and potentiometers, the former are rarely Potentiometer Circuits*
used in Performance Test Code work because of
their inherent limitations. 9 One-DialPotentiometer. Fig. 3.1B shows the
circuit of an elementary potentiometer having one
3 Recording potentiometers are widely used for measuring dial and a single range. L e t u s a s s u m e
industrial process temperature measurement and that it has a range of O to 100 mV.
control. However, in general, they are not well
suited to Performance Test Code work. Inaccura-
cies in charts are caused by printing limitations
and humidity effects. Practical limits on chart
widths and scale lengths result in inadequate
readability.
4 Indicating potentiometers are recommended for
all Performance Test Code work as they are avail-
able with the required readability and accuracy.
5 Thermocouples should be calibrated separately
and the calibration corrections should be applied
to the instrument readings.
6 Thermocouples should be Iong enough to con-
nect directly to the instrument or reference junc-
tion apparatus, This will eliminate errors which FIG. 3.16 SINGLE-DIALPOTENTIOMETER
might be introduced by the use of extension wires Range 0-100 m V
which may not have temperature-emf characteristics
identical to those of the thermocouple.
10 Slide-wire C and fixed resistance A are con-
nected in series with the battery and battery rheo-
PRINCIPLES OF OPERATION
stat. The resistance A is made a simple multiple
-7 Accurate measurement is usually a matter of of the emf of the standard cell SC, and the current
comparing an unknown quantity against a known through slide-wire C and fixed resistance A is
quantity or standard-the more direct the compari- selected s o that when standardized, the drop of
son the better. Accurate weighing, for example, is potential across resistance A is equal to the volt-
often accomplished by direct comparison against age of the standard cell which is nominally 1.0190 V#
standard weights using a mechanical balance. If
the measured weights are too heavy for direct com-
*The circuits shown in this section do not represent the
parison, lever arms may be used to multiply the actual circuits of the instruments listed in T a b l e 3.1B.
forces. They are intended only to illustrate basic principles,
28
29
reference junction temperature of 32 O F , must be 39 Slider P, can now be moved along slide-wire
added to convert to the equivalent value with-a ref- D to add or subtract a voltage to or from .the emf
+
erence junction temperature of 32%' (27.83 3.82 = generated by the thermocouple, If slider P, is
31.65mV). Referring to the temperature-emf table is moved toward resistor E, the voltage will be in-
for Type K thermocouple, it will be found that the creased and vice versa. With a suitable ~mmillivolt
temperature of the measuring junction is 1400%'. calibration on slide-wire D, connection P, can be
set to the correction required for any reference
35 "hile reference junction compensation can be junction temperature within i t s range.
made in this manner, i t is inconvenient when many
readings are being taken. 40 Since the main slide-wire in branch 1 and the
reference junction compensating slide-wire in
36 m e n using manual reference junction com- branch 2 are both calibrated in millivolts, they are
pensating facilities, the reference junction temper- suitable for u s e with any type of thermocouple,
ature must be measured accurately with a liquid-in-
glass thermometer, the equivalent emf found from 41 Method 2 for Manual Reference Junction Com-
the appropriate temperature-emf table based on pensation. Another method of obtaining manual
3 2 T , and the dials set to this value. reference junction compensation is shown in prin-
37 Method 1 for Manual Reference Junction Com- ciple in Fig. 3.5B. Here again the upper portion of
pensation. The principle of this method is shown
in Fig. 3.4B. if reference is made to Fig. 3.1B
BATTERI
WlEOST4T
showing the elementary circuit of the one-dial FIG. 3.58 SINGLE-DIAL POTENTIOMETER
Range 0-100 m V with manual reference
potentiometer, it will be seen that the upper sec-
junction compensator
tion of this circuit and that of Fig. 3.4B are iden-
tical. A second branch has been added in the cir-
cuit of Fig. 3.4B consisting of resistances B and the circuit is. identical with that of Fig. 3.15, that
E and slide-wire D connected in series, which in of the elementary one-dial potentiometer. In this
turn are connected in parallel with branch 1 and circuit, however, resistance B has been placed in
the battery circuit. series with the thermocouple. An auxiliary battery
38 For sake of illustration, let us assume that
,
and two rheostats C and C, supply current through
resistance B, and the drop of potential across it is
the resistance of each branch i s the same and,
added to the emf of the thermocouple.
therefore, currents i l and i 2 are equal, If the resist-
ance of (B + %D) i s made-equal'to resistance A in 42 Adjustment of this circuit for a given refer-
branch 1, and connection P, is moved to the slider ence junction temperature is made by short circuit-
on slide-wire D and placed in i t s mid-position, it ing the emf terminals and setting the main slide-
will be seen that for any specific value of emf, the wire scale to the millivolt equivalent of the refer-
two circuits will be in balance for the same posi- ence junction temperature and then adjusting rheo-
tion of slider P2 on the main slide-wire. s t a t s C , and C, until a balance is obtained. Fol-
30
-.
"
lowing this, the emf read directly from the main leads that of the line phase by 90 deg and the
slide-wire scale will be corrected for a reference motor rotates in a direction to increase the magni-
junction temperature of 3 2 9 and may be used di- tude of the known voltage.
rectly in the temperature-emf tables.
49 When the unknown voltage is lower than the
known voltage, the current from the power amplifier
Automatic Null-Balancing Mechanism lags the current in the line phase by 90 deg and
the motor rotates in the opposite direction to re-
43 Automatic null-balancing potentiometers fre- duce the magnitude of the known voltage,
quently employ an -electro-mechanical servomecha-
nism which detects the difference between an un- 50 Thus the motor is always driven in a direc-
known emf and known voltage and actuates a motor tion such that the difference between the unknown
to drive the slider on a slide-wire to a point where em€ and the known voltage is reduced to zero, a t
the two are equal. Fig. 3.6B shows a block diagram which point the motor c e a s e s to rotate, the instru-
of a typical system. ment being in balance.
Description
Range(s) mV
Three-Dial
Automatic
Balancing
0-70.1
1 Three-Dial
Balancing
Manual
0-16/0-160/
1 Two-Dial
Balancing
O-16/0-160/
Manual
Two-Diol
Automatic
Balancing
Two-Dial
Manual
Balancing
*Whichever is greater
32
tube to a depth of approximately one-half inch. In- 75 It is strongly recommended that a mercury-in-
sulated copper wires are attached to the ends of glass thermometer be installed in the bath, to indi-
the thermocouple wire from the measuring junction, cate the stability of i t s temperature. The bulb of
and each joint is placed in a test tube andbottomed the thermometer should be close to the bottoms of
s o that the junctions of the wires are submerged in the test tubes. Refer to Chapter 6, Liquid-in-Glass
the mercury, The assemblage of tubes is then S ~ O Z J J - Thermometers, for emergent-stem correction pro-
ly and gently pressed into the slush. cedure.
MERCURY-IN-GLASS 76 Multiple Ice Bath With MasterIceBath. When
n LTHERMOMETER
THERMOUHlPLE WIR the number of thermocouples used on an installa-
FROM MEASURING
JUNCTION tion is large enough to require a number of ice
&
M
-p
)p
ERW E TO
~- baths, it is recommended that a reference junction
_C ,-INSTRUMENT master ice bath be used to enable the operator to
readily determine whether a l l the baths are main-
taining a temperature of 32%’. Fig. 3.88’sliows in
pictorial and schematic form a master ice bath in-
stallation.
[THERMOCOUPLE
WIRE
FROM MEASURINGJUNGTION
FIG. 7
34
- -.
F”-’ ” -
?. ~~ ~
~ ~~~ ””
INSTRUMENTS A N D APPARATUS
formed from his own calibrated wire, Others are 8 1 ElectricalCompensation. A compensating
provided with many reference junction pairs brought circuit containing a source of current and a combi-
out to terminals which the user may connect into nation of fixed resistors and a temperature sensi-
h i s system. The error introduced into ,a system by tive resistor can be obtained which will have a
these devices may be a s s m a l l a s0.1c. variation of emf similar to that of the reference
junction of the thermocouple when its temperature
is allowed to change.
CONNECTINGCOPPER
THERMOCOUPLE
A Cu S ELECTOR
O SWITCH
2 (COPPER-)
B Cu
0 2- POLE
L
POTENTIOMETER
A A
Cu
.O N -THROW
N
B - Cu
9
MEASURING
JUNCTIONS
I 1
B Cu
u
A
v
Cu \ J
UNIFORM COPPERCONNECTING
TEMPERATURE WIRES
ZONE
V 'I
ICE BATH
(REFERENCEJUNCTION)
R , = R,(l + AT + BTZ)
GENERAL
where
Scope Rt = resistance of thermometer a t temperature, T
1 The purpose of this chapter is to present in- R, = resistance of thermometer a t 0°C
formation which will guide the user in the selec- A, B = constants, dependent on characteristics of
(u) The relation between resistivity and tempera- 11 Two factors establish the upper limit of the
ture is very simple; it holds over a wide tem- range of the standard precision platinum resistance
perature range. thermometer.
(b) Its resistivity is relatively high.
( c ) Its temperature coefficient of resistivity is (u) The protecting tube is Pyrex which yields to
satisfactory. s t r e s s e s at temperatures higher than about
(d) Physically, it is very stable. 500°C(932%').
(e) It is resistant to corrosion, (b) Mica crystals contain water of crystallization.
(f) It can be stress relieved by heating to high A t temperatures above 500°C (932%') water is
temperature in air. driven off and the crystalline structure is de-
(g) It can he drawn to very fine wires. stroyed; the mica forms which support the plati- .
num winding swell, distort, and crumble.
Platinum is subject to contamination by reducing
atmospheres and metallic vapors, In making precision platinum resistance thermom-
-eters for use at temperatures above 500°C(932%') a
10 PrecisionPlatinumResistanceThermometer.
high-temperature resistant mica must he used to
The precision platinum resistance thermometer is
make the winding form and the platinum winding
used to define the International Practical Temper-
must be enclosed in a quartz tube,
ature Scale from - 182.96"C (boiling point of oxy-
gen) to 630.74"c (melting point of antimony), 12 In low temperature calorimetry, a second form
(-297.3 to 1168.3oF). In accordance'with the defi- of precision platinum resistance thermometer is
nition of the International Practical Temperature frequently used in which the thermometer is part of
Scale, [1,2I* the platinum resistance thermometer the calorimeter, thereby eliminating the need for a
must meet the following specifications: connection head, In this form of thermometer the
standard thermometer resistor winding is placed in
The purity and physical condition of the platinum
a platinum tube; the tube is evacuated and filled
of which the thermometer is made should be such
with helium prior to sealing. Helium is used to in-
that the r a t i o s shall not be less than 1,3925 for crease the speed of response, by virtue of i t having
R, a higher thermal conductivity than air. This form of
T = lOO"C, precision platinum resistance thermometer has a
A paper by C.H. Meyers [3] describes in detail useful range of -269 t o +25OoC (-452 to t482%).
the construction of a strainfree precision platinum See Fig. 4.l4b).
resistance thermometer occupying a relatively
small volume in which a helically spiraled platinum 13 Another form of precision platinum resistance
wire is wound helically on a mica form. The mica thermometer is frequently used in calorimetry work
form consists of two crossed pieces of mica in the range of -5Q to -t-100°C (-58 to +212%'), In
notched to receive the helically wound platinurn this form, the platinum wire is wound on a flat mica
wire, The winding is annealed and then adjusted to form, notched to retain the platinum wire; the wind-
25.5 ohms -t 0,1 ohm at 0°C. Four- gold leads are ing is protected by flat mica sheets, and the assem-
welded to the platinum (two to each end of the bly is placed in a thin-walled meta1 tube. After
winding) and the assembly mounted in a pyrex tube *Throughout the text Reference numbers are enclosed in
7 mm OD. The leads are insulated with glass tub- brackets, thus[d.
closing the end, the portion of the tube over the 16 The differences in construction between the
resistance winding is fIattened to increase thè standard precision and industrial platinum resist-
speed of response. See Fig. 4.l(c). ance thermometers are tabulated below.
Standard Industrial
Component
Precision
(c) Its windings must be of fine wire to avoid mas- ability requires circuit adjustment upon replacement
siveness, and consequent slowness in response of units.
to temperature changes if it has the same re-
26 While the shape of the temperature-resistivity
sistance as platinum or nickel thermometers,
curves for the commonly used resistance thermom-
Copper resistance thermometers usually have a re- eter materials such as copper, nickel and platinum
sistance of 10 ohms a t 7 7 v . T h e s t a b i l i t y of cali- depart to only a minor degree from a straight line,
bration is excellent, much superior to that of nickel. the curve for a thermistor is exponential in shape.
Copper resistance thermometers can be depended
27 This extreme nonlinearity limits the useful
upon to maintain their accuracy over a long period
range of a calibrated instrument to relatively.nar-
of time provided the manufacturer’s temperature
row temperature spans unless specially designed
limitation is not exceeded. circuits are used to “linearize” their output.
20 In a common method of construction of both 28 Other Metals. To be useful as a resistor, the
copper and nickel resistance thermometers the in-
metal must have stable characteristics, not be sub-
sulated wire is wound bifilar on a metal bobbin and
ject to permanent change, and not have any critical
three leads are attached-two to one end of the
temperature; i.e., transformation point where crys-
winding, and one to the other end to compensate talline structure will change when heated to any
for ambient temperature changes. The bobbin is in- temperature within the intended operating range.
serted in a thin-walled metal tube, closed a t one
Pure metals are generally preferable. Alloys usual-
end. The bobbin must make intimate contact with ly have lower and less reproducible temperature
the inside of tube-by s o doing the speed of re- coefficient of resistivity than pure metals.
sponse is greatly increased. The protecting tube
over the nickel or copper resistance thermometer is CHARACTERISTICS
sealed.
29 Range. Precisionresistancethermometersare
21 Copper resistance thermometers intended for used for measuring temperatures from -269 to
insertion in the windings of electrical machinery 630.74”C (-452.2 to 1168.3v). Two precision plat-
are made of grids of copper imbedded in flexible inum resistance thermometers would be require-d to
plastic strips. cover this span-the thermometer described in Par.
10 plus a low temperature calorimetry thermometer
22 Ten ohm copper resistance thermometers are
useful over the range of -269 to t250°C (-452.2 to
used in pairs for wet bulb and dry bulb temperature
+482 v ) .
measurements in the determination of relative The industrial platinum resistance thermometer
humidity. can be used over ranges of O to 1064OC (32 to
1950oF), and -182.97 to 0°C (-297.3 to 3 2 T ) -
23 NonmetallicResistanceThermometer(Ther- the latter type is sealed to prevent moisture ac-
mistor). In general, thermistor resistance thermom- cumulation shorting out the leads.
eters are used where sensitivity, accuracy, speed Copper resistance thermometers can be used over
the range of -195.6 to 121.1OC (-320 to t250”F)
of response, ruggedness, small size, and calibra-
tion stability requirements exceed those attainable although the 100 ohm type is preferred for negative
with other types of thermometers, Thermistors have temperature measurements.
limited use, however, because of short ranges, non- Nickel resistance thermometers have a useful
cange of -40 to +121.loC (-100 to t300”F).
interchangeability and higher over-all cost.
24 The high resistance of thermistors minimizes 30 Sensitivity. Although the coefficient of re-
the effect of lead length variations while the high sistivity of platinum is lower than that for the base
temperature coefficient of resistivity permits the metal resistance thermometers, approximately 25
ohms in the resistance winding produces a sensi-
design of circuits having high sensitivity.
tivity of about O. 1 ohm per “C. Comparatively,
25 Unfortunately, thermistors cannot be obtained 10 ohms of copper produces a sensitivity of ap-
with a uniformity of resistance at a given tempera- proximately 0.022 ohm per deg F; 100 ohms of cop-
ture to better than about 20 percent, on a run-of- per, 0.215 ohm per deg F; and 100 ohms of nickel
mill basis. By selection, tolerances on the order of produces a sensitivity of 0.186 to 0.213 ohm per
2 percent c.an be realized, This lack of interchange- deg F.
Industrial
Precision 10 ohms Copper Nickel Thermistor
100 ohms 100 ohms
Sensitivity O.lohm/degC
0.22ohm/deg
F 0.22 ohm/deg F 0.22 ohm/deg F O. 186 ohm/deg F
0.213 ohm/deg F Varies with units
Precision M.O2 de F up
to 2 0 0 4
Accuracy 9 . 0 1 degCt3.0degFStd iO.5 deg F Std 20.5 deg F Std 9 . 5 deg F Std N.5 deg F Std
f1.5 deg F Spec d . 2 d e g F S p e c f 0 . 2 d e g F S p e c f 0 . 2 d e g F S p e c t0.2 degF Spec
Response-Bare 15 s e c 20 s e c 40 s e c 40 s e c
Fast
Response-w/well 30 sec 9 0 s e c 60
90 sec sec
Resistance 25.5 ohms
ohms
25 10 ohms 100 ohms 100 ohms
at 779' at 779' Varies with units
at 0% at 3 2 9 at 77%'
Linearity 70. 1°C/500C 70. l0C/5OoC excellent
excellent
excellent Sxponential
span SP an
Range -452.2 to -297.3 to to
-100
-325 to -100 to -100 to 5 0 0 v
1168.3%' 195oOF3009 3009 3009 -75 to 26OoC)
1-269 to (-182.96 to (-75 to (-200 to (-75to
530.74%) 1064OC) 150%) 15OoC) 140%)
L 4
FIG. 4.3 WHEATSTONEBRIDGE CIRCUIT FOR
FIG. 4.2 SCHEMATIC OF WHEATSTONE BRIDGE CIRCUIT INDICATORS OR RECORDERS
illustrates one method of accomplishing this result. S and S, are uniform slide-wires of equal length;
Three wires A, B, C connect the measuring instru- ,.
S has twice the resistance of S The contacts of
ment and the thermometer resistor (x). Of these, A the detector and battery leads are moved simulta-
and C should be identical in size, length, and ma- neously and by equal distance along the slide-wide
terial, and should be placed side by side through- to balance the bridge. The variable contacts are
out their length, s o a s to be alike in temperature. thus placed in the battery and detector circuits
The B wire, which is one of the battery wires, need where they can have no effect on the balance point
not be similar to the others, but it is common prac- of the bridge, while a one-to-one bridge ratio is
tice to form the three wires into a cable and make constantly maintained. A scale associated with the
them all alike. A and C are in the thermometer re- slide-wire may be graduated in degrees (C or F), or
sistor arm (x), and the variable resistance arm (r), in ohms. If the scale is graduated in degrees, a
respectively. Their resistance remains equal al- winding properly adjusted to match resistance and
though their temperature conditions may change, temperature coefficient of the material for which
and hence, with a one-to-one bridge ratio, such the bridge is calibrated must be used as the sensi-
changes have no effect on the bridge reading. tive element.
38 Unbalanced Bridge Method. This method is
37 It is desirable to have no variable contact re- shown schematically in Figs. 4.4 and 4.5. A, B, C,
sistances in the bridge arms, because the variations represent the terminals of the resistance thermom-
in bridge balance introduced a t the contacts may be eter resistor located at the point where the temper-
sufficient to seriously effect the reliabilityof the ature is to be measured. D and E are equal ratio
measurements. The effect of these variations, as arms. P is a fixed resistor of a value equal to that
41
’T
made equal to et by adjusting R,, i n which c a s e
R, = R, and R, is read directly from the calibrated
Indicofor dials of R,. It is essential that the current through
R, and R, remain constant during the two potentiom-
eter balances required. Consequently, in the second
method, it is necessary to check that the current
change resulting from adjustment or R, h a s been
insignificant. If r , is large compared with Rs and
Rt, this condition will be more easily satisfied.
The resistance measurement by means of this
FIG, 4.4 UNBALANCED BRIDGE METHOD-
method is independent of the lead resistance.
DC SUPPLY
Sources of Error
L A
I
DPDr
I perature from that of the gas.
A Essential Considerations
FIG. 4.5 UNBALANCED BRIDGE METHOD- 41 The proper value of current to be used in a
AC SUPPLY
given apparatus will usually be specified by the
manufacturer. Since the thermometer resistor is in-
39 PotentiometricMethod. Thepotentiometric sulated electrically, and to some extent, thermally,
method of resistance measurement is applicable to i t s temperature is raised by the measuring current
four-lead resistance thermometers only, Fig. 4.6. to something above that of i t s surroundings. There-
The current, adjusted to a suitable value by resis- fore, the current must be kept small enough to
tor r,, flows through a standard resistor (or resist- avoid causing a rise in temperature which is more
ance box) Rs and the thermometer resistor Rt. The than a small fraction of the limit of error of the
potential drops e, and et, through R, and Rt re- measuring equipment. To test the degree of com-
pliance with this condition, the thermometer may ADVANTAGES AND DISADVANTAGES
be placed in melting ice and the bridge balanced,
44 Advantages
using a measured bridge current. When a steady
reading is obtained, the current should be reduced (u) By proper selection, resistance thermometers
to half of its original value and, if necessary, the may be used to cover ranges extending from
bridge rebalanced. If a significant change in resist- -450 to 1950%’.
ance is observed, the current is again reduced by (6) High accuracy.
half and a new measurement made. The current (c) Excellent stability and reproducibility.
should be reduced to such a value that further re- (d) Interchangeable.
duction produces no observable change in the indi- (e) Can be matched to close tolerances for tem-
cation of the instrument. For the sake of sensitivity, perature difference measurements.
the bridge current used should be a s large a s per-
missible, 45 Disadvantages
42 The indicating or recording instrument may be (a) More mass than thermocouple.
located at a distance from the point at which the (6) Relatively slow response.
temperature is to be measured. Connection is made ( c ) Subject to mechanical damage if not properly
by means of a three or four conductor copper cable handled.
covered with rubber or lead. The conductors are (d)Higher cost than thermocouple.
usually 16 or 18 AWG copper individually insulated
with rubber or more temperature resistant material.
Each wire must be of the same length in order to
assure equality of resistance of the compensating REFERENCES
leads even though they may be subject to uniform
temperature conditions. 46 Throughout the text Reference numbers are
enclosed in brackets, thus 111.
Treatment of Data 111 ‘6The International Practice-Temperature Scale of
1963,” Comite International d e s Poids et Measures,
43 The observed temperature readings need not Metrologie vol. 5, No. 2 P. 35, Apr. 1969.
be corrected provided the resistance thermometer is [2] 661nternationalPractical Temperature Scale of 1968,9’
immersed, in accordance with manufacturer’s direc- Benidict, RP, L&N Technical Journal, Spring 1969.
tions, in the medium whose temperature is being 131 Published in 1932 a s National Bureau of Research,
measured, Corrections for drift in calibration may be Paper 508, (vol. 9, p. 807).
141 66WheatstoneBridges and Some Accessory Apparatus
evaluated by periodic checking of the resistance For Resistance Thermometry,” by E.F..hlueller, Sci-
thermometer a t the ice point and comparing these entific Papers of the Bureau of Standards, vol. 13,
checks to the original ice point resistance. p. 547, 1916.
GENERAL
7
Scope
Partial
Immersion Line-
1 The purpose of this chapter is to present infor- Liquid Level r
Definitions
PRINCIPLES OF OPERATION
CLASSIFICATION
Description
Materials of Construction
fast dyes. The working ranges of several liquids in 27 Precision of measurement of temperature with
common use are shown in Table 5.2. liquid-in-glass thermometers depends upon the
thermometer design and the application conditions
as well as the care exercised in reading.
TABLE 5.2 WORKING TEMPERATURERANGE
FOR LIQUIDS COMMONLY USED 28 Accuracy of temperature measurement is de-
pendent upon the same factors which affect preci-
Working Range Liquid
sion, and in addition the accuracy of calibration or
-38 to 115g°F Mercury standardization and periodic evaluation of secular
-56 to 250 F Mercury-thallium changes in the bulb glass. W i t h well designed ther-
-328 to 45OoF Organic liquids mometers the accuracy of calibration is a function
of range and graduation interval. National Bureau
of Standards certification tolerances for such ther-
23 Gases. When gas is used above the liquid, mometers are listed in Tables 5.3 and 5.4.
the gas should be inert to the liquid. Nitrogen and
29 The values shown in Tables 5.3 and 5.4 for
carbon dioxide are commonly used with mercury, 66
Tolerance in degrees” represent acceptable
while hydrogen is usually the choice with mercury-
limits of error of uncertified thermometers. “Accu-
thallium. For most organic liquids air is sufficient-
racy in degrees” is the limit of error to be antici-
ly inert. The pressure of the gas should be high
pated when corrections are applied, and when
enough to minimize the vaporization of the liquid
proper attention to such details as maintenance of
at any temperature in the operating range.
correct immersion, avoidance of parallax, etc,, are
exercised in the use of the thermometers. “Correc-
24 Metals. Various metals are used for bulb as- tions stated to” are the limits to which NBS certi-
semblies, although steel, stainless steel, and fication values are rounded off.
brass are the most common. Brass and aluminum are
generally used for the case. Heat transfer media are
30. With extreme care and attention to detail the
limits of accuracy listed in Tables 5.3 and 5.4
normally included in the bulb chamber. The most
might be made smaller, but usually it is preferable
common materials are mercury, copper dust, graph-
to use another type of measuring device, such a s a
ite, and silver dust. Selection is determined by the
platinum resistance thermometer, if a higher order
range of the thermometer and the material of the
of accuracy is required. Because of the uncertainty
bulb chamber.
of measurement of the temperature of an emergent
CHARACTERISTICS
column, partial immersion thermometers generally
cannot be exp-ected to give results of the same ac-
25 Range is determined by the application re- curacy a s otherwise equivalent design total immer-
quirements limited by the physical characteristics sion thermorfieters. The etched stem form is to be
of the materials of construction, Selection of ther- relied upon for results of the highest accuracy. The
mometers with extremely broad or narrow ranges or industrial and tube-and-scale forms are affected by
very fine degree of subdivision, will add signifi- heat conduction of the parts other than the g l a s s
cantly to the cost. Review of manufacturers’ stand- tube. Such heat transfer is difficult to measure ac-
ard listings will frequently be helpful in choosing curately and result in uncertainties in temperature
special ranges. measurement.
26 Sensitivity is determined by the cross-section 31 Response of liquid-in-glass thermometers is a
area of the capillary bore and the proportion of that function of thermometer design and use conditions.
area to the volume of the bulb, Application condi- In most cases, the response time-temperature rela-
tions frequently limit the length of scale and the tionship is exponential and a single value, most
size of bulb. Practical limitations of tubing and commonly the 6 3 percent response, i s used in evalu-
thermometer manufacture are significant factors in ation of t h i s variable.
establishing minimum bore diameters. Sensitivities
ACCESSORIES
far in excess of the precision or accuracy of meas-
urement are misleading and generally are costly, 32 Wells or sockets are the major accessories for
particularly if specially designed thermometers are liquid-in-glass thermometers. For a general discus-
being used. sion of wells refer to Chapter 1, Pars. 8 through 19.
48
32 up to 212
Above 212 up to 600
5 2 or1 E -0.5 0.2
0.5
0.2
0.2
i i:
32 up t o 600 0.5 -1.0 0.5
5
Above 600 up to 950 1 -2 0.5
32 up t o 600
Above 600 up to 950 1: 0.2-1.0
0.5 -1.0
0.2
0.2
Graduation
Temperature Range Tolerance Accuracy Corrections
Interval in
in Degrees i n Degrees in Degrees Stated to
Degrees
~~ ~~ ~~~~
32 up to 600
-3
1 -2
2 1
1
49
Sources of Error
I N SERVICE
34 Liquid-in-glass thermometers are subject to
variations in manufacture which necessitate the D U R I N G CALlBRATlOFl
determination and application of instrument correc-
tions for accurate results. Techniques and equip-
I
ment for such calibration are discussed in detail in
Chapter 9.
d-"- _""
--
""
"
-"-
-"
"" " -. I
"-
- _"
_" - -
""
-- "-
" "
_"
"
!
j ]
""
"
"
"
- ""
""
"- -
calibration. These should be taken into considera- _"" "-
""
""-
" --
_""
""
"_"
" "
""
"" -
error is the misuse of the emergent-stem correction. "- ""
"_
" ""
" - "
"
""
"
""
""-
""
- -"
""
""-
"
_""
_""
mometer calibrated for total immersion, This cor- -
r
""
50
52
( b ) PermanentChanges. A second type of change (b) Select the thermometer considering the folIow-
ing factors:
in thermometer glasses, known as the “secu-
(1) Range and graduation interval.
lar change,” results in a nonrecoverable de-
( 2 ) Immersion,
crease in bulb volume which may progress
(3) Sensitivity and response time.
with time even at room temperature, but which
is markedly accelerated at high temperatures.
(4)Form, i.e., etched stem, industrial etc.
This type of change is evidenced by an increase (5) If a well is to be used, it should be of the
in the ice-point reading. At low to moderate right material, shape, and size.
temperatures there may be a gradual change (6) The thermometer should be capable of
which will continue for years. With better giving results of the accuracy desired but
not appreciably better,
grades of thermometer glasses the change
will not exceed 0.1 deg C in many years, provided (c) Install the thermometer properly and use care
the thermometer has not been heated to tem- in reading, avoiding the following:
peratures above about 15Ooc. In addition, per- (1) External heat sources or sinks near the
manent changes in bulb volumes have some- thermometer which might affect its indica-
times been observed with thermometers which $ion.
53
Treatment of Data
54
Definitions
*-=
55
General Classification
Subclassification
( b ) With compensating means within the case only (b) Designed to operate with the measured temper-
(Class IB). See Fig. 6.3. ature below the temperature of the rest of the
thermal system (Class IIB). See Fig. 6.5.
9 Vapor PressureThermalSystem (Class II): A ( c ) Designed to operate with the measured temper-
thermal system partially filled with a volatile ature above und below the temperature of the
56
VOLATILE L
NONVO
LIQU
VOLATILE LIQUID^ M
I II
NONVOLATILE
LIQUID
VOLATILELIQUID
FIG. 6.5 VAPORPRESSURETHERMAL SYSTEM- FIG. 6.7
CLASS IIB VAPOR PRESSURE THERMAL SYSTEM-CLASS IID
57
Low temp limit -3OOOF -38OF. -65OF Hg-Th eutectic -40°F -4ooaF
High temp limit GOO’F 1200OF 600OF 600°F 1500’F
Note: Dimensions are to be used as a guide only and will vary in OD and length with manufacturer. Functional values also vary with manufacturer. For more detail,
see Table 6.2
INSTRUMENTS AND APPARATUS
r e s t of the thermal system (Class IIC). T h i s Description
type normally requires a larger sensitive por-
12 Bulb Size. The bulb size of the various ther-
tion than Class IIA or IIB. See Fig. 6.6.
mal systems varies greatly (approximately 100 to 1)
( d ) Designed to operate with the bulb temperature depending upon system class, temperature span
above, below and ut the temperature of the rest and capillary length. Table 6.2 provides a guide in
of the thermal system (Class IID). See Fig. 6.7. determining the size of the bulb. Considerable vari-
In this type the volatile liquid is confined to ation in bulb size exists among the various manu-
the sensitive portion and a second relatively facturers. The basic reasons for the variations in
non-volatile liquid is used to transmit the vapor bulb size are briefly described below. Typical mer-
pressure to the expansible device. cury filled thermometer bulbs are shown in Fig. 6.8.
10 Gas-Filled Thermal System (Class III): A ther- 13 Liquid-filled and mercury-filled systems (Class
mal system filled with a gas and operating on the I and C l a s s V), which operate on the principle of
principle of pressure change with temperature liquid expansion in the bulb, have a bulb internal
change. The system is usually compensated for am- volume which is inversely proportional to tempera-
bient temperature effects either: ture span. Therefore, larger temperature spans will
require smaller bulbs. Since the temperature span
(u)’With a second thermal system minus the bulb, of a liquid-filled system (Class I) or a mercury-
or equivalent means of compensation (Class filled system (Class V) may vary by 25 to 1, the
111.4). See Fig. 6.2. bulb size will vary accordingly. A few. manufacturers
(b) With compensating means within the case only have designed mercury thermometers in the 400 to
(Class IIIB). See Fig. 6.3. 12OOv range with 3/8 in. bulb diameters and 3 in.
bulb lengths,
MMEPERFORMANCETEST CODES
TABLE 6.2 APPROXIMATE BULB SENSITIVE 17 The Class IIC system bulb needs to accornrno-
DIMENSIONS date the entire capillary and Bourdon volume when
System
Class
1 1Fill
Outside
DiaKter, 1 1
Length,
In, Remarks
the bulb temperature becomes equal to the tempera-
ture of the capillary and Bourdon (See Fig. 6.6.)
The bulb size therefore is generally larger than that
of the Class IIA system; it is also dependent upon
IA & B Liquid 9/16 3 50 deg F span the capillary length.
IA & B Liquid 3/8 2-1/2 275 deg F or
greater span 18 The C l a s s IID system bulb must have an in-
IIA Vapor 9/16 4 ternal trap of such dimensions that the volatile
IIB Vapor 3/8 2 liquid will not enter the capillary under all values
IIC Vapor 9/16 6
IID Vapor 9/16 4 of ambient temperature (i.e., the trap must accom-
modate the non-volatile liquid expansion under all
HIA & B Gas 10 7/8 Based on 75 ft.
capillary values of ambient temperature). See Fig. 6.7.
,-FLANGE
f I
II
BULB J
EXTENSION
a. PLAIN BULB
UNION
CONNECTION
b. FLANGED BULB
WELL
KEY:
- BULB LENGTH
MEDIUM
I A
B - SENSITIVE PORTION
C - INSERTION LENGTH
i
D - IMMERSION LENGTH
c. THREADED WELL
( UNION BULB1
FIG. 6.8 TYPICAL MERCURY FILLED
obtained from the manufacturer. Several manufac- 22 Liquid systems (Class IB) generally are pro-
turers have been able to design very small bulb vided with overrange protection of 100 percent of
gas thermometers. Bulb s i z e in t h e s e d e s i g n s i s temperature span. Some thermometers are provided
3- to 3K-in. long and 3J8 in. in diameter. Tempera- with greater overrange protection, depending on the
ture ranges vary from -40 to 180oF to 400 to manufacturer,
1200OF.
23 In vapor pressure thermal systems (Class II),
overrange protection is generally more limited than
20 Overrange Protection. Overrange protection is in other systems because of the increasing rate of
defined as the maximum temperature to which the vapor pressure rise with temperature rise. Typical
bulb of a filled system may be exposed indefiniteIy vapor pressure temperature relationships are shown
without damage to the system. It is usually ex- in Fig, 6.9. A specific overrange temperature for
pressed in per cent of temperature span above the each range offered is usualIy specified by the
upper limit of the range. A summary of extent of manufacturer. T h i s is generally appreciably less
overrange protection for filled system thermometers than 100 percent of temperature span. If the upper
of the various types is specified in Table 6.1. l i m i t of range is near the critical point of the fluid
21 The overrange protection of liquid gystems fill, the overrange protection may be extended be-
(Class IA) varies with capillary length. Generally, cause of the fill being a vapor above the critical
i t i s in the region of 100 to 200 percent of the point. Under some limited conditions, it is possible
to fill the system so that all of the liquid will be
temperature span for short systems, For long.. sys-
.
exhausted from a bulb at a bulb temperature above
tems, because the capillary volume generally ap-
the instrument range, in which case the safe over-
proaches the bulb volume, the capillary volume
range temperature can be increased,
change with ambient temperature change neutralizes
the overrange possibilities of the Bourdon, thus re- 24 The nonlinear vapor pressure-temperature rela-
ducing the overrange Protection to essentially zero tionship is an advantage where the user desires
for systems 200 f t long, more reading sensitivity toward the top of range.
TEMPERATURECF1
FIG. 6.9 VAPORPRESSURE-TEMPERATURECURVES
-.
62
of the gas employed, which for commonly used forms Bourdon motion into pointer or pen motion.
helium is -451OF. The upper temperature is usually The backlash in mechanical gearing and links usu-
limited to lOOOoF, but gas systems have been made ally is greater than 0.1 percent; thus, the sensi-
to operate successfully up to 15OOoF. tivity may be on the order of 0.25 percent of range
36 Range. The minimum range of the organic span. In the designs where mechanical linkages are
liquid system (Class I) is limited by maximum bulb reduced or eliminated, sensitivity may be affected
s i z e to approximately 25 deg F. Because of non- by highly viscous oil which is put on the Bourdon
linearity of expansivity and compressibility of the coil to damp shock and vibration effects. Sensi-
organic liquids employed, the maximum range is tivity in these designs is no better than above.
frequently limited to 200 to 400 deg F, because of 41 Accuracy. Filledsystemthermometersare
differences in manufacture s o that a specified ac- normally regarded a s 1.0 percent instruments. This
curacy may be met with linear dials or charts. means that under most environmental conditions of
37 The range of a vapor system (Class II) is case or capillary ambients the error will not ex-
ceed 1percent of temperature span. However, many
limited by low and high temperatures of -430 and
hOOOF, respectively, for the reasons discussed in instruments are calibrated to higher accuracy and
Par. 34. However, the nonlinear vapor pressure- in indoor applications the maximum error is fre-
temperature relationship is accentuated by greater quently specified a s 0.5 percent of temperature
ranges and tKe range is therefore normally limited span. Accuracy may be only 2 or 3 percent when
to approximately 250 deg F. thermometers are used in environments where the
case and capillary temperature vary considerably
38 The mínimum range of a mercury system from usual room temperature (e+, case or capil-
(Class V) is limited by maximum bulb size to ap- lary temperature can be a s low as 60% and a s
proximately 50 deg F. F o r amercury system the bulb high a s 16OOF in some power plant applications).
s i z e is larger for a particular comparative range The reduction of accuracy is caused by the in-
than it is for an organic liquid system, because the ability of the compensation devices to completely
expansion rate of mercury is l e s s than that of or- compensate for ambient temperature changes.
ganic liquid by a factor of approximately six. The Direct reading thermometer c a s e s , which are at-
maximum range is limited only by the upper useful tached directly to the bulb, are exposed to heat
temperature, generally lOOOOF, and the freezing conducted along the thermometer stem and also to
point of -38OF. Mercury filled system thermometers heat radiated by unlogged pipes. Capillaries
are made, however, with ranges as low a s -40 to wound around boiler casings are also subjected to
180 deg F and a s high a s 400 deg F to 1200 deg F. high ambient temperatures. Thermometers used
39 Because the pressure within a gas system outdoors could be subjected to very low ambient
(Class III) essentially follows Charles' Law, (i*e. temperatures. Accuracy is also affected by
absolute pressure is proportional to absolute tem- mechanical backlash and mechanical and fluid
perature) it is characteristic of this system that the friction.
shorter the range the higher will be the internal 42 TemperatureCompensation. Sincethecapil-
system operating pressure. This condition limits laries and Bourdons as well as the bulbs of thermal
the minimum range to approximately 120 deg F. The systems are filled with actuating fluid, these por-
maximum range is limited only by the upper tem- tions of the system are sensitive to ambient tem-
perature, usually lOOOOF, although longer ranges perature. Therefore, system errors will result be-
generally require larger bulbs to provide an ade- cause of ambient temperature variations unless
quately linear output. Small bulb and large bulb compensation means are employed.
gas thermometers capable of operating up to 1200%
are available. A nonlinear scale is used in some 43 The vapor-filled system (Class II), as an ex-
designs to eliminate the nonlinear output effect. ception, is not subject to errors from the fluid fill.
The only temperature error observed in this system
is of small magnitude; i t is caused by change of
40 -Sensitivity. The Bourdon of a filled system
elastic modulus of the Bourdon material with tem-
will respond to the smallest measurable change in
bulb temperature. Therefore, the output for small perature, and is usually ignored.
temperature changes is affected only by friction or 44 T h e liquid-, gas- and mercury-filled systems
loose fits in the mechanical apparatus which trans- (Class I, III and V) are provided with full compen-
45 Because the capillary error of a gas system 49 A bulb will respond faster if the following
(Class III) is reduced as the bulb size is increased, three fundamental design factors are employed:
the Class IIIA system which has full compensation (u) Increase the external area relative to the in-
is rarely built. Gas-filled systems are therefore ternal volume.
generally limited to the Class IIIB type, where only ( b ) Lower the heat capacity.
the case is compensated, ( c ) Increase the thermal conductivity of the bulb
walls and internal fill.
46 The mercury system with full compensation
(Class VA) is frequently supplied with a single The gas system (Class III) is frequently the most
capillary which is continuously temperature com- favorable because the bulb can usually be made
pensated along its entire length (see Fig. 6.3). with a relatively thin wall and the heat capacity of
This is achieved by employing a capillary with a the internal gas is almost negligible. However, the
precision bore enclosing a precision drawn Invar large bulb size frequently required tends to offset
wire s o that the expansion of the Invar wire and this natural advantage* The small bulb gas thermom-
mercury equals the expansion of the surrounding eters are somewhat slower in response than conven-
capillary. tional larger bulb gas thermometers or mercury
filled thermometers.
47 The liquid, gas, and mercury systems with
case compensation only (Classes IB, IIIB and VB) 50 The vapor systems (Class ITA, IIB, and IIC)
are frequently employed because of the simplicity have almost equally favorable response because
of construction, The capillary bore s i z e i s reduced the heat capacity of the volatile fluids employed is
to a point where system response is not seriously low and the thermal conductivity high. This is
affected in order to minimize the capillary temper particularly true for small temperature changes be-
ature error. The Bourdon of these systems is com- cause under these conditions condensation and
pensated by means of a bimetallic strip (see Fig. evaporation of the fill will take place on the inter-
6,3). These systems are employed when the cap- nal bulb walls. The (Class IID) vapor system has
illary length may be sufficiently short or when the somewhat slower response because of the presence
ambient temperature range is sufficiently small that of the internal bulb trap and in some cases, also
the capillary error may be ignored. In practice, the because of the increased viscosity of the non-
capillary error of a liquid-filled system (Class IB) volatile liquid fill.
servo, the feedback force being generated by the up or down, which may be corrected by a simple
transmitter pressure. Similarly, in an electrical screw adjustment. However, a severe shock could
transmitter the filled system force is balanced by cause a permanent s e t in the Bourdon or misalígn-
a force which is generated by an electrical current. ment of the linkage. In this case, the change in
In other electrical transmitters the Bourdon motion calibration would not necessarily be uniform over
directly operates the core of a differential trans- the entire range.
former for a-c output or the force from a Bourdon
56 ConductionandImmersionError. The bulb of
actuates a strain gage for d-c output.
a filled system thermometer must be completely im-
mersed in the medium in which the temperature is
APPLICATION AND INSTALLATION being measured. If this is not done, a significant
portion of the filling medium volume can be a t a dif-
Sources of Error ferent temperature than that which i s being meas-
ured. Errors due to improper immersion can be ex-
55 Zero Shift Error. Filledsystem thermometers tremely large. Actually, the bulb should be im-
are subject to mechanical abuse during shipment, mersed s o that not only the filling fluid reservdir
which may cause an error in the calibration. The is immersed but also a sufficient amount of the
user, therefore, should check the instrument cali- bulb extension to prevent heat conduction to or
bration and make corrections. A calibration error from the sensitive portion. The amount of extra im-
associated with shipment is usually confined to a mersion varies a s the heat transfer and temperature
"zero" shift, in which the entire range is shifted environment varies. For a thermometer with a 3-in.
h-"---g
"""""
10"-
8 -
6-
4-
2
2- f z
n
3 loi
i$ 86-
t;
L
4- w
k 2-
$
D"
8-
6-
K
ik 4-
2-
10-
sensitive portion length and being used to measure user to specify this elevation or depression to the
temperature a s high a s looo°F, for example, the manufacturer s o that he will calibrate the instru-
bulb should be immersed about 5 in. Heat transfer ment accordingly. The vapor system (Class IIC)
from unimmersed portions of the thermometer should will have liquid within the capillary only over part
be reduced when measurements are being made in a of the range span and is, therefore, not recommend-
medium having low heat transfer capabilities, The ed when the bulb and case are at appreciably dif-
entire sensitive portion should be immersed in the ferent elevations. The instrument cannot read cor-
flowing fluid when thermometers are used in forced rectly for bulb temperatures both above and below
convection applications. the capillary case temperature.
57 CapillaryImmersion Error. Thecapillary of 61 Barometric Error. This error is essentially
all system types except vapor systems (Class II) is non-existent for systems operating on the volu-
temperature sensitive. Dual capillary systems fre- metric principle, í.e., for the liquid-filled system
quently are used in liquid systems (Class IA) and (Class I) and the mercury-filled system (Class V).
compensated capillary is used in mercury systems Vapor systems (Class II) and gas systems (Class
(Class VA). These compensating means are im- III) operating on the pressure principles are sensi-
perfect and the instrument output reading will vary tive to barometric pressure changes by the ratio
with length of capillary immersion. If the immersion of barometric pressure change to the internal pres-
length is greater than 8 in., the immersion length sure change corresponding to the range. These
should be specified to the manufacturer or the in- systems therefore are designed to havea minimum
strument should be adjusted by the user under the pressure change of 100 psi for the range of the
conditions of the application. Small bulb liquid thermometer. Since the maximum barometric pres-
filled thermometers are more affected by capillary sure change is approximately 50.4 psi, this error
immersion than larger bulb designs, e.gL, on a mer- will be equal to or less than 0.4 percent of range.
cury filled thermometer with a 3-in. by 3/8-in, bulb
and a range of 400 to 1200°F, no more than 2 in.
of capillary should be immersed.
58 Bulb Elevation Error. When the Bourdon ele-
vation of a liquid or mercury system (Class I or V>
is changed relative to the bulb, a pressure head
caused by the column of the fluid fill is generated
within the system. This pressure redistribution
causes a small volume change of the fluid and of
the bulb and capillary thereby causing a system
error. If the bulb is to be elevated more than 25 ft
above the case, it is desirable for the manufacturer
to know the above elevation to increase the pres-
sure of the system s o that the bulb pressure will
not drop to zero after installation.
59 The elevation error is nonexistent in a g a s
system (Class III).
60 If the Bourdon is above the bulb in a vapor
system (Class IIA or C l a s s IID) the pressure with-
in the Bourdon equals the vapor pressure in the
bulb minus the liquid pressure head in the capil-
alary. This means that the bulb elevation error is
equal to the ratio of the liquid head to the internal
vapor pressure change across the temperature span.
This further confirms that i t is advantageous for
the manufacturer to provide systems having a rela-
tively large internal pressure. If the bulb elevation
relative to the case is 20 ft, it is advisable for the FIG. 6.12 PREFORMED CAPILLARY BULB
67
68
CONTENTS Definitions
70
I
1832OF
viewed. The outline or detail of the reference sec-
tion of the filament is indistinguishable from the
surrounding field and "disappears" when the cur-
rent in the lamp is properly adjusted. The value of
the current in the lamp may be measured by means
RADIATION
+ MEASURED BY of a milliameter, the scale of which is ordinarily
PYROMETER graduated in terms of temperature, or, alternatively,
a potentiometric measurement of the current may be
made; in models employing a built-in potentiometer,
I I I I I
1 2 3 4 5 6
the potentiometer scale is graduated in terms of
WAVELENGTH, h ()un) temperature, Standardized absorption glass filters
are interposed between the target and the lamp,
FIG. 7.1 PLANCK'SBLACKBODYRADIATION
DISTRIBUTION FUNCTION, SHOWING
thus permitting a wide range of temperature to be
SPECTRAL BAND UTILIZED BYAN measured without requiring high filament tempera-
AUTOMATIC OPTICAL PYROMETER tures. Optical pyrometers of this type are available
A T 0.65 W. covering the temperature range L400 to 18,OOOT;
[The spectral bandwidth for a disappearing fila- however, the majority of applications are below
ment optical pyrometer a t the same wavelength 4500%', with applications above 7000v being rare.
is somewhat greater.]
14 ConstantRadiance ComparisowLampType.
cally arranged as illustrated in Fig. 7.2. An objec- The essential elements of the instrument are illus-
tive lens focuses a real image of the target in the trated in Fig, 7.3. In this type of instrument, the
plane of a standard lamp filament. Both image and lamp filament operates at a constant radiance ob-
b
OBJECTIVE MICROSCOPE
APERTURE OCULAR
APERTURE
STOP
4 MEASURING
INSTRUMENT
WEDGE
OBJECTIVE
LENS 7 GROUND
GLASS SPOT
/ I
FILTER
AXIS OF ROTATION OF
OPTICAL WEDGE
FIXEDINDEX TO INDICATE 1
TEMPERATUREONTHESCALE
tained by setting the current through the lamp to a image of the target. The temperature is read direct-
standard value by means of a rheostat and milliam- ly from a scale attached to the wedge. A red filter
meter, The radiance of the target is matched against in the eyepiece restricts the wavelengths used to a
that of the lamp filament by means of a polarizing narrow band, Pyrometers of this type are available
prism, an iris diaphragm, or an absorbing device covering the temperature range from 1400 to 5200%'.
such as a gray glass circular wedge, interposed
between the target and the lamp. A circular glass 15 Self-BalancingVariableRadiance Comparison-
wedge i s used to reduce the luminance of the light Lamp Type (Automatic Qptical Pyrometer). The
emitted from the target to match the luminance of a essential elements of the instrument are illustrated
spot on a ground glass screen (illuminated by the in Fig. 7.4 and Fig. 7.5. Operation of the instru-
lamp) i n the field of view of the eyepiece. The ment [3] is similar in principle to that of the disap-
lamp current is adjusted to the fixed value speci- pearing filament optical pyrometer; it differs in that
fied for the particular lamp? and the instrument is detection of radiation is accomplished with a photo=
focused on the target and the wedge is rotated detector (usually a photomultiplier tube) rather than
until the image of the spot disappears against the by eye, the lamp current is adjusted by an electron-
SMALLAPERTURE
INFIRSTSURFACE
MIRROR DEFINES
TARGETAREA
OCULAR
- ROTATABLE POLARIZING
FILTER
H--."---
ERECTING
=m-
POLARIZING FILTER
""
LENSE
74
ic null-balancing system rather than manually, and integrated signal controls the lamp current, driving
the spectral bandwidth employed is usually sub- it up or down to achieve a zero amplitude square
stantially narrower than in the disappearing fila- wave from the photomultiplier, at which time the
ment type. A further difference is that the pyrometer lamp current is said to be in null-balance, The
lamp is not mounted in the plane of the target image, pyrometer has the same range .and is calibrated in
resulting in the use of two separate optical trains, essentially the same manner a s the conventional
one for pyrometer lamp radiation and one for target disappearing filament opticalpyrometer,. i.e., the
radiation; the lamp filament is therefore not neces- pyrometer lamp current is determined a s a function
sarily at the same radiance as the source, although of blackbody target temperature.
the radiant flux arriving at the detector through one Because of the null-balance operation of the elec-
optical train is equal to that arriving at the detector tronics system, the calibration repeatability is al-
through the other optical train.
most totally independent of normal aging effects
and other variations in the electronic components,
16 In the automatic optical pyrometer, the modu-
and is determined almost entirely by the stability
lator alternately passes radiation from the target
of the pyrometer lamp calibration.
and then from the pyrometer standard lamp at some
frequency such a s 90 Hz or 400 Hz; if the photo- 17 Other Automatic Pyrometers. Theautomatic
multiplier receives unequal amounts of radiant optical pyrometer permits a large number of pos-
energy from the two sources, its response is a sible design variations not available in manually
square wave signal, the phase of which (with re- operated optical pyrometers, stemming from the
spect to the modulator driver) determines whether use of photodetectors other than the eye, together
the lamp current is too high or too low. This signal with appropriately selected spectral bandpass
is synchronously demodulated, then integrated. The filters.
OPTICAL
FILTER- /"-MODULATOR
-i ___L__
PREAMPLIFIER PHOTOMULTIPLIER
LAMP
1
REFERENCE AMPLIFIER
MILLIAMMETER
L
i
I
LAMP DRIVER RECORDER
OUTPUT
18 Some of the practical variations, such a s the over the next few years. In the interim, some in-
use of a wavelength significantly different from struments in this category should be considered
0.65 p, may involve the basic instrument design suitable for ASME Performance T e s t Code work.
a s described above, but require extensive considera- Their use in that capacity may yield a significant
tion of the sources of error discussed following advantage over either manually operated or auto-
Par. 47, especially of the transmissive properties matic optical pyrometers operating a t about 0.65 pn,
of the medium between the pyrometer and the tar- but such use should be undertaken only after care-
get. Water vapor and carbon dioxide absorption in ful consideration of their suitability to the particu-
certain spectral regions of the infrared must be lar problem a t hand; such consideration should be
taken into account. Spectral bandwidth a s w e l l a s primarily with respect to the measurement accuracy
wavelength then becomes inportant. Special ad- associated with any particular application.
vantage derives from the ability to select a spectral
region in which the target emissivity is known to Materials of Construction
be high, and in being able to make measurements
22 Since the optical pyrometer is an optical in-
a t lower temperatures by operating in the infrared
strument, i t is essential that all compcnents be of
region of the spectrum.
high quality and properly aligned and assembled.
19 Some automatic pyrometers used in the infra- The lenses and other transmitting materials must
red region of the spectrum u s e a standard s.ource of be free from imperfections which will cause distor-
radiance (such a s a lamp or a blackbody operated tion or scattering of the light rays. Particular at-
at a fixed temperature) a s a reference, but also de- tention should be given to the pyrometer lamp to
pend in part for their accuracy upon the stability of insure that the tungsten filament is of uniform
active electronic components, Other automatic py- luminance over that portion in the field of view
rometers use no reference standard of radiance, but that is used a s a radiance reference. Solid-state
depend for their accuracy entirely upon the stability electronic components used in automatic or semi-
of active electronic compqnents, including the automatic pyrometers should not be subjected to
photodetector. temperature above approximately 122%’.
76
~ ~ _ _ -~
_ ~
P
" 7
" ~. - -i
77
~ ~ ..
\ ~. ~
~ ~~ ~ ~
are available for certain commercially available actual temperature that is important, such as in
units, including: ( a ) emissivity cornpensatink glass some manufacturing processes, i t is often adequate
filter to be used for direct temperature measurement to use the radiance temperature without correction,
(requiring no emissivity correction) of molten iron To determine when this can be done i t is neces-
or steel or any other material having an emissivity sary to have some understanding of the factors that
of 0.4; ( b ) special short-focus objective lens for influence emissivity. Similar information is neces-.
measuring objects of l e s s than 1/32-in. diameter. sary when corrections are to be applied to convert
For some models, temperatures of targets whose radiance temperature to temperature. This informa-
diameters are a s s m a l l a s 0.005 in. may be meas- tion i s presented in the secti0.n on “Radiation from
ured. In general, for disappearing filament optical the Surface of Real Materials” (Par. 41).
pyrometers, precision suffers when the target width
35 f‘racticalBlackbodies. A completelyenclosed
is less than about four times the filament width, a s
cavity in any opaque material at a uniform tempera-
seen through the optical pyrometer [21.
ture contains blackbody radiation characteristic of
APPLICATION AND INSTALLATION
the temperature of the cavity walls, but independ-
ent of the materials of their construction. A small
32 General. Opticalpyrometersareordinarily hole, or aperture, in the enclosure will emit radia-
calibrated to read correctly when sighted on a black- tion that very closely approximates that of a black-
body and, in general, this is the preferred way in body at that temperature.
which to use them, ”hen the temperature of a non-
blackbody is to be determined, it is often possible 36 When the cavity is a t room temperature, the
to simulate blackbody radiation by creating a cavity aperture will appear visually to be very black. How-
in the body, such as by drilling a hole in its sur- ever, a small fraction of light incident upon the
face, and viewing the radiation emerging from the aperture from outside will be reflected back out the
hole, Information regarding design criteria and ef- aperture‘after a number of reflections from the
fectiveness for simulated blackbodies of this type cavity walls; since the reflectance of the aperture
is presented in the following section on “Practical is slightly greater than zero, its absorptance (and
Blackbodies” (Par. 35). therefore its effective emissivity) must be slightly
less than unity (Par. 43).
33 Many furnaces approximate blackbody condi-
tions very satisfactorily, although there is no con- 37 Anything that will reduce the reflectance of
venient quantitative method for estimating their ef- the aperture will increase its effective emissivity,
fective emissivity.” In a perfect blackbody, the This can be done by increasing the number of
details of the inside of the furnace vanish and a times an incident ray is reflected by the cavity
piece of steel, for example, that is being heated walls before it emerges from the aperture (usually
cannot be distinguished from the background, If the by making the dimensions of the aperture small
objects in the furnace can be distinguished, but compared to those of the cavity), and by construct-
only on close observation, and if much of the de- ing the cavity walls of a low reflectance material
tail is lost after they have been in the furnace for s o that more radiation is absorbed at each reflec-
some time, i t is not likely that the temperature tion,
measurement will be seriously in error. If in error 38 The nhmber of reflections a ray will make be-
at all, the observed temperature will be too high fore finally finding its way back out through the
when the furnace walls are of higher radiance than aperture will depend upon the general shape of the
the materia! being heated and too low when the cavity and on the detailed charaçter of the surface
walls are of a lower radiance. The latter condition roughness of the cavity walls. Reflection from a
is possible if the heat supply is variable or if it is perfectly smooth surface is;described as being
shut off and the furnace allowed to cool. specular, while the reflection from a perfectly
34 When blackbody conditions cannot be simu- rough surface is described as being diffuse. Real
lated, it is necessary to account for the effect of materials have surfaces that are characterized by
emissivity. Where i t ‘ k repeatability rather than a mixture of specular and diffuse components of re-
flected radiation. Effective emissivities near unity
*Since the emissivity of a blackbody is unity by defini- are more easily attainable in cavities with specu-
tion, it is a contradiction in terms to speak of a black- larly reflecting than with diffusely reflecting wall
body with emissivity less than unity. The most common
solution to this problem, in the case of a simulated materials; however, considerable care must be
blackbody, is to speak of its “effective” emissivity. taken when the interior surface of the cavity is
78
specular, because under these conditions the ef- E = emissivity of materials forming the blackbody
fective emissivity is significantly- directional, and interior surface
seriously large errors may result from viewing the
S = area of aperture
cavity from the wrong direction.
S = area of interior surface of blackbody cavity,
39 While there is as yet no simple formula for including the aperture
accurately expressing the effective emissivity of
an aperture in a cavity of arbitrary shape in terms C¿ = the solid angle of radiation emerging from
of easily determined parameters, various formulas the cavity aperture, having its apex at the
or graphs have been devised [6-101 that are appli- intersection of the viewing axis with the
cable in most special cases of practical signifi- back wall of the cavity.
cance, The method of DeVos L61 is generally con-
sidered to be valid and of great generality, and is 41 Radiation from the Surface of Real Materials.
commonly u s e d a s a reference against which other A body made of an actual material may be desig-
formulas .are evaluated; however, it is mathemati- nated as a real body, to distinguish it from a black-
cally very cumbersome and is not recommended for body. The interior of an opaque real body is totally
absorbing, and when a t a uniform temperature must
routine engineering applications unless high ac-
cdracy is mandatory. A review of the subject from therefore radiate a s a blackbody. When blackbody
the viewpoint of radiation thermometry h a s recent- radiation from the interior approaches the surface,
part of i t is reflected back into the interior; the re-
ly been made by Bedford [71.
mainder passes through the surface and is emitted.
40 Gouffe'sMethod Esl. Gouffe'smethod for The fraction that is emitted is defined to be the
computation of effective emissivity for cavitiesof emissivity of the surface. The fraction that is re-
arbitrary shape assumes perfectly diffuse reflection flected is defined to be the reflectance, which has
from the cavity walls; it is exact for spherical the same value for radiation approaching the sur-
cavities (with diffusely reflecting walls), but face from either side,
yeilds effective emissivity values that are slightly
low for cavities of other shapes. The error tends 42 The spectral emissivity ex is the fraction by
to increase the further the cavity shape departs which blackbody radiation is reduced in the process
of being emitted from the surface. A surface at ab-
from that of a sphere, but is small enough for a
wide variety of cavity shapes to justify the use of solute temperature T o and having an emissivity eA
the method for many common applications. I t is will appear to the pyrometer (having a narrow spec-
presented here for use as a guide in estimating ef- tral bandwidth) to be a blackbody a t a lower temper-
fective emissivity because it is concisely formu- ature T,; the relationship between T , and To is as
lated for easy application; more nearly exact follows Ill:
values of effective emissivity may be obtained
from the references cited above [9,1O] for certain TO - UT, = he/czloge CA ,T o > Tr (4)
commonly used cavity shapes and for wall mate-
rials having both specular and diffuse components where
of reflectance,
To = absolute temperature of the target in kelvins
€0 +
= c; (1 JG) (31 T, = absolute temperature of the target as in-
e dicated by the pyrometer, called the spec-
where e l = tral radiance (luminance) temperature, in
- (s/S)I + (.s/S)
e [1 kelvins
árid k =
a .
(1 - e ) Ks/S) - (F)] , a smallnega- Xe = mean-effective-wavelength,inmeters
tive number, tend-
= 0.014388 mK ,
ing to zero as the
cavity shape ap- EX = spectralemissivity of thetargetsurface
proaches that of
a sphere.
43 The blackbody rádiation incident upon the
60 = effective emissivity of blackbody aperture surface from the interior must either be emitted or
internally reflected, for every wavelength; this may under consideration should be used, rather than
be expressed a s follows: published values such a s t h o s e in Tables 7.1 and
7.2; an often used method is to drill a small hole
CA + RA = 1 (5) in the surface of the material in question to provide
a blackbody cavity, From an optical pyrometer de-
where CA = spectral emissivity of the surface
termination of the radiance temperature of both the
R,+ = spectral reflectance of the surface blackbody cavity and the surface of the material
adjacent to it in the temperature range of interest,
44 From the above expression it can be seen that the spectral emissivity may be calculated, using
a good emitter is a poor reflector, and vice versa. Eq. ($)-or the A-value may be calculated, using
Thus, carbon has a high emissivity and a low re- Eq. (6), (Par. 55).
flectance, while platinum has a low emissivity and
a high reflectance. Anything that affects reflect- 47 Sources of Error, The sources of error in op-
ance must have a corresponding effect on emissivity.
tical pyrometry may be broadly grouped into three
Since reflectance is wavelength dependent and categories: (a) those associated with the pyrom-
slightly direction and temperature dependent, s o is
the emissivity. A material that reflects and emits a
constant fraction at all wavelengths is said to be a TABLE 7.1 SPECTRAL EMISSIVITY OF MATERIALS,
SMOOTH SURFACE, UNOXIDIZED
“graybody”; like a blackbody, a perfect graybody
is an idealization that can be experimentally real- Wavelength = 0.65 ptn (red light)
ized only as an approximation. Actual materials (Roeser andWensel, National Bureau of Standards)*
may be considered to be gray only in restricted
spectral regions.
Material I Solid I Liquid
0.37
but its emissivity will be increased by a like
Manganese 0.59 0.59
amount. Emissivity is thus seen to be dependent Molybdenum 0.37 0.40
upon the state of surface roughness of the radiating Nickel 0.36 0.37
body. Tables 7.1. and 7.2 list the spectral emis- Palladium 0.33 0.37
Platinum 0.30 0.38
sivities of the more common engineering materials
(Refs, 11-14 contain an exhaustive compilation Rhodium 0.24 0.30
Silver 0.07 0.07
and evaluation of emissivity data on a large number
of materials), and Table 7.3 presents corrections
Tantalum
Thorium
0.49
0.36
...
0.40
that must be added to indicated temperatures to Titanium 0.63 0.65
correct for the effect of emissivity for pyrometers Tungsten
Uranium
0.43
0.54
...
0.34
operating at 0.65 p. A more complete tabulation Vanadium 0.35 0.32
of corrections may be found in NBS Monograph 30 Yttrium 0.35 0.35
[IS]. In using Tables 7.1 and 7.2 it is necessary Zirconium 0.32 0.30
to take into account the state of surface roughness, Steel 0.35 0.37
Cast Iron 0.37 0.40
which will tend to increase the emissivity over the
values listed in the table. The extent of oxidation
Constantan
Monel
0.35
0.37
...
...
will also influence the emissivity, an effect that . Chromel P (90 Ni-10 Cr) 0.35 ...
often causes difficulty in practice, where heated 80 Ni-20 Cr
60 Ni-24 Fe-16 Cr
0.35
0.36
...
...
materials exhibit surface oxidation changing with
time.
Alumel (95 Ni;Bal. AI, Mn, Si)
90 Pt-IO Rh
0.37
0.27
...
...
46 When possible, a measured value of the spec- *From the “Handbook of Chemistry and Physics,” Chemical
tral emissivity of the particular piece of material Rubber Publishing Company.
eter, (b) those associated with the media between radiance measured by the pyrometer, and a r e usu-
the pyrometer and the source, and (c) those asso- ally correctable,
ciated with the source. 49 The mean-effective-wavelength varies some-
TABLE 7.2 SPECTRAL EMISSIVITY OF OXlDES WITH what from one model to another, but for optical
SMOOTH SURFACES pyrometers it is usually between 0,636 p and
Wavelength = 0.65 p (red light) 0.662 p. For the type shown in Fig. 7.2, it is
typically assumed to be about 0.65 Pm. A curve of
(Roeser and Wensel, National Bureau of Standards)*
the mean-effective-wavelength as a function of the
1 Probable
~~. target temperature is ordinarily available from the
Value for the
Range of Oxide Formed manufacturer, Variations among filters in disappear-
Materiai Observed on Smooth ing filament optical pyrometers cause variations of
Values Metal a s much a s -Ilpercent in the mean-effective wave-
Aluminum oxide 0.22 to 0.40 0.30 length [16], relative to values supplied by the
manufacturer, giving rise to corresponding uncer-
Beryllium oxide
Cerium oxide
Chromium oxide
0.07 to 0.37
0.58 to 0.80
0.70
0.60 to 0.80
...
0.35
tainties in computed emissivity and window trans-
Cobalt oxide ........ 0.75
mission corrections; this will usually be the domi-
nant uncertainty in the value of the mean-effective-
Columbium oxide 0.55 to 0.71 0.70
Copper oxide 0.70 to
0.60 0.80 wavelength, since the uncertainty due to differing
Iron oxide 0.63 to 0.98 0.70 visual responses among observers will rarely ex-
Magnesium oxide 0.10 to 0.43 0.20 ceed 0.2 percent.
Nickel oxide 0.85 to 0.96 0.90
Thorium oxide 0.20 to 0.57 0.50 50 Size-of-SourceEffect. Radiation from outside
Tin oxide
Titanium oxide
Uranium oxide
........
0.32 to 0.60
........ 02O‘
0.30
of the target area but from the immediate neighbor-
hood of the target is found to influence the pyrom-
Vanadium oxide ........ 0.70 eter indicatión at least to a small extent; this is
Yttrium oxide
Zirconium oxide
........
0.18 0.43
0.60
0.40
called the size-of-source effect. For visual and
automatic optical pyrometers, the effect is most
Alumel (oxidized)
Cast Iron (oxidized)
........
to
........ 0.87
0.70 noticeable for small targets. It is caused primarily
Chrome1 P (90 Ni-10 Cr) by the scattering of radiation within the pyrom-
(oxidized) ........ 0.87 eter optical system, and (in visual optical pyrom-
CrNi-2080 (oxidized)
60 Ni-24 Fe-16 Cr (oxidized)
........
........ 0.90
0.83
eters) by heating of the pyrometer lamp filament by
the incident radiation; on the upper temperature
55 Fe-37.5 Cr-7.5 A l
(oxidized)
70 Fe-23 Cr-5 Al-2 CO
........ 0.78 ranges of visual optical pyrometers and some
(oxidized)
Constantan (55 Cu-45 N í )
........ 0.75
models of automatic optical pyrometers; small
transmission changes in the range filters can oc-
(oxidized) ........ O. 84 cur if the filters are heated somewhat by absorbed
Carbon Steel (oxidized)
Stainless Steel (18-8)
........ 0.80 radiation.
(oxidized)
Porcelain
........
0.25 to 0.50
0.85
... 51 In the automatic optical pyrometer, location
of the range filters behind the mirror aperture (one
*From the “Handbook of Chemistry and Physics,” Chemical Rubber of the two possible locations of F, in Fig. 7.4) re-
Publishing Company. duces the filter heating effect to a negligible level
[31; in this configuration, an automatic optical py-
48 Sources of error associated with the pyrom- rometer with clean optical surfaces has a size-of-
eter have to do primarily with the stability of cali- source effect usually not greater than 0.6 deg F a t
bration of the reference lamp, the determination of the gold point (1948.0%’), tending to indicate a
the mean-effective-wavelength, the spectral trans- higher temperature a s the source area is increased.
mittance characteristics of the absorption glass If the extraneous source area is,at nominally the
filter, and the accuracy of calibration; except for same temperature as the target, the magnitude of
the mean effective wavelength, discussed below, the effect is proportional to the square of the
these were discussed in the section on “accuracy.” absolute (target) temperature and to the mean-ef-
Most of the other errors can be treated in terms of fective- wavelength; the effect is therefore in-
the extent to which they reduce (or increase) the herently larger in infrared sensing pyrometers.
~~~-~ ~ ~
52 Very few data are available on the magnitude same as that of a window of spectral transmittance
of this error in disappearing filament optica€py- 7 ~ For
. any measured value of T, the value of To
rometers; the pyrometer lamp filament is heated may be obtained if the mean-effective-wavelength
slightly by the radiation of the target image, and he, the spectral transmittance TA, and the second
no special precautions have been taken to mini- radiation .constant cz are known. I t is usually more
mize the effect of scattered radiation. The small practical to experimentally determine the value of
amount of available data [4] suggests that the ef- A by direct measurement of T and T,; A is very
fect in a disappearing filament pyrometer is of the nearly constant (it varies slightly because the
order of 2 to 4 deg F (at the gold point) difference mean-effective-wavelength varies slightly with
between viewing very small and very large targets. temperature), and may thus be used to relate other
values of T and To. The window transmittance,
53 The size-of-source effect can cause a very
and thus the A-value, are dependent upon the di-
large error, especially in pyrometers operating in
rection of the transmitted radiation. The transmit-
the infrared, if the area adjacent to the target is a t
tance is highest in the direction normal to the
a much higher radiance temperature than the target,
surface.
or if a much higher temperature source in the back-
ground (behind the target) lies near the line of 56 A tabulation of the difference between T and
sight. T o appears in Table 7.4 L151 for the case of a thin
piece of clear window glass, Although surface re-
54 The effect can be almost entirely eliminated flectance accounts for almost all of the reduction
for a particular application by calibrating the py- in transmittance €or very thin windows, the trans-
rometer against a target of the same s i z e and mittance of a window tends to decrease with in-
shape, and at the same distance as is to be used creasing thickness due to absorption; it is usually
in the intended application. advisable to determine the A-value experimentally,
55 Windows and AtmosphericAbsorption. Sup- especially for windows that are thick or if reduced
pose a pyrometer sighted on a blackbody indicates transmittance is visually detectable.
a temperature ' T o (expressed on the Kelvin scale), 57 The effect of atmospheric transmission is
but indicates a lower temperature T when viewing analogous to that of window transmission. However,
the same blackbody through a window having a if atmospheric attentuation is not visually apparent,
transmittance TA. The relationship between the the atmosphere is sufficiently transparent that the
temperatures indicated with and without the window correction is negligible for any pyrometer using
in place i s given [Il to a close approximation by only visible red wavelength radiation. Where it is
not negligible, atmospheric transmission is likely
1/To - 1 / T z= &/C, logerx = -A, To> T (6)
to be so variable as a function of time a s to render
Note the similarity between Eqs. (4)and (6); the computed corrections impractical. The most com-
role of the spectral emissivïty CA is exactly the mon practice in such cases is to sight the pyrom-
TABLE 7.3 EMISSIVITY AND TRANSMITTANCE CORRECTIONS
For Addition To Observed Temperatures for Optical Pyrometer
Using Red Light at Wavelength = 0.650 P m
Spectral E m i s s i v i f y or Transmittance
Indicated
Temperature O. 1 0.2 0.3 0.4 0.6 O .51.0 0.9 0.8 O. 7
700°C llO°C 740c 54OC 4 lo€ 3l0C 22OC 16% IOOC 5% O
ao0 135 91 66 so 37 27. 19 12 6 O
900 163 1o9 80 60 45 33 23 14 7 O
1000 19 4 130 95 71 53 39 27 17 a O
1100 229 152 Ill a3 62 45 31 19 9. O
1200 266 177 128 96 71 52 36 22 10 O
1300 308 20 3 147 I10 81 59 41 25 12 O
1400 352 232 168 124 92 67 46 29 13 O
2500 40 1 262 189 140 104 76 52 32 15 O
1600 45 3 295 212 157 117 a5 58 36 17 O
1700 510 330 237 176 130 94 65 40 19 O
1800 570 368 263 195 144 104 72 44 21 O
1900 634 408 29 1 215 159 115 79 49 23 O
2000 704 450 321 236 174 126 86 53 25 O
82
_-
-.
85
GENERAL
Scope
Definitions
86
~~ ~ ~ ~ ~ ~
L
INSTRUMENTS AND APPARATUS
threads, Insertion length can be distinguished from terna1 standard pipe thread connections. The bulb
immersion length with the following example. In a diameter varies from approximately 1/8 to 3/8 in.
fully filled pipe flow application, immersion length depending on the model, bulb length, and manu-
is the portion of the thermometer bulb exposed to facturer. Bulb lengths from approximately 2-1/2in.
the flowing fluid, However, a relatively stagnant . to 5 ftareavailable.
fluid column surrounds the remainder of the ther-
mometer bulb between the inner pipe surface and B b a b o r a t ~ ~ory Test Type Thermometer. These
the base of the threads. The Sensitive Portion thermometers are characterized by higher accura-
Length is the distance from the thermometer bulb cies than the industrial type and the absence of
tip to the upper end of the bimetallic element. See threaded connections. Bulb diameters and lengths
Fig. 83, are available -in the sizes given for Industrial
thermometers.
PRINCIPLES OF OPERATION
CLASSIFICATION-
FIG, 8.3 NOMENCLATURE
Description (Courtesy of Scientific Apparatus Makers Association)
5 IndustrialTypeThermometer. T h e s e thermom-
eters are generally supplied with 1/2or 3/4 ln. ex-
r
One method of increasing accuracy is to increase
the number of coils in the bimetal element thus in-
creasing the angular motion for a given temperature
change.
87
atures should be
Do not apply
91
92
with R ,oo/Ro tightened from 1,3920 to 1.3925. las establishing the relation between temperature
(e) Range 1 is now divided into four parts: and the indications of instruments calibrated by
P a r t 1 extends from the triple point to-the boil- means of values assigned to the defining fixed
ing point of hydrogen. points. These fixed points are defined by speci-
P a r t 2 extends from the boiling point of hydrogen fied equilibrium states, each of which is under a
to the triple point of oxygen. pressure of 1 standard atmosphere, except where
P a r t 3 extends from the triple point to the boil- noted differently. The defining fixed points and
ing point of oxygen. the exact numerical values assigned to them are
Part 4 extends from the boiling point of oxygen given in Table 9.1,
to the freezing point of water.
( f ) In Range 1, the Callendar-Van Dusen equation 10 InterpolationMeans. The means available for
interpolating temperatures led to the division of the
i s no longer used, but interpolation is by a new
scale into four parts, using three different instru-
reference function
ments for interpolation a s detailed in Table 9.2.
-t
i=t
20
2 A . ( h !VrnF
8 1
li , K
The reference resistance ratio represented by
‘(9.1) 11 It should be understood that temperature scales
and their units are arbitrary considerations. IPTS-68
is a document not a device. While the IPTS-68 is
WREF i s defined by the basic scale having 100 units or degrees between
the ice point and the steam point, the Fahrenheit
WREF = W,, - A V scale having 180 units or degrees between these
two points is commonly used in engineering prac-
where Wnl is the measured resistance ratio,R,/’R,, tice. The relationship between these two s c a l e s is
and A W is a deviation defined by a specific poly- given in Chapter 1.
nominal interpolation equation, one being given
for each part of Range 1. METHODS OF CALIBRATION
(g) In Range 2, the Callendar equation i s modified Fixed Points
by a correction term s o that interpolated values
of temperature will conform more closely with 12 A precise method of calibration is that of de-
thermodynamic temperatures; Le., termining the reading of an instrument at one or
more of the defining fixed points which are listed
t s 8= t”+A t (9.2) in Table 9.1.
where t” i s temperature by the Callendar equa- TABLE 9.1 SUMMARY O F FIXED-POINT VALUES
tion; and At is a correction term given by
K
(9.3)
t.p. hydrogen 13.81 -259.34
b.p. hydrogen, 25/76 atm -256.108 17.042
(h) In Range 4, the second radiation constant, C,,
-252.87
b.p. hydrogen 20.28
is changed from 0.01438 meter kelvins to
27,102 -246.048 b.p. neon
0.014388 meter kelvins,
t.p. oxygen -218.789 54.361
8 By design, IPTS-68 h a s been chosen in such a b.p. oxygen 90,188 -182.962
way that temperatures measured on it closely ap- f.p. water 0.0 273.15
I‘ proximate thermodynamic temperatures; Le. differ- t.p. water +0.01 273.16
ences are within the limits of the present accuracy b.p. water 1O0 373.15
of measurement. f.p. zinc 692.73 419.58
b.p. sulfur 717.824 444.674
9 FixedPoints. IPTS48 is based on 11 repro-
f.p. silver 916.93 1235.08
ducible temperatures (defining fixed points), to
f.p. gold 1064.43 1337.58
which numericalvalues are assigned, and on formu-
93
where Rb is the resistance, at temperature t, of the than 0.65 mm in diameter. Before calibration, the
platinum resistor between the branch points formed wires of the couple are annealed in air €or an hour
by the junctions of the current and potential leads a t about 2000%’. The wire of the thermocouple is
of a standard resistance- thermometer. The constant, mounted so a s to avoid all mechanical constraints
R, is the resistance at 0°C (32%’), and the constants, in the region where steep temperature gradients are
A and B , are to be determined from measured values likely to occur.
of Rt at the steam and zinc points.
29 The primary standard platinum-r h od’lum versus
22 From the oxygen point to 0°C (32oF), the temp- platinum thermocouple is used in the calibration of
perature, t, is defined by Eq. (9.1). other thermocouples and also in the checking and
23 For easier use in temperature measurement, calibration of optical pyrometers below a tempera-
the interpolation formulae, for the range 0°C (32 F) ture of 2750°F.
to the freezing point of antimony are in Eqs. -(9.2) 30 Standard platinum-rhodium versus platinum
and (9.3)‘ thermocouples should be calibrated periodically by
24 Standard platinum resistance thermometers the National Bureau of Standards or equally quali-
should be calibrated periodically by the National fied laboratory.
Bureau of Standards or equally qualified agency. 31 OpticalPyrometer. Unlike the standardplati-
25 StandardPlatinum-VersusPlatinumRhodium num resistance thermometer and the standard plati-
Thermocouple. From the freezing point of antimony num-rhodium versus platinum thermocouple, there is
a t 1167.3p to the gold point at 1947.97%’, the no device such a s an optical pyrometer, which when
standard platinum 90 percent-rhodium 10 percent constructed according to a s e t of specifications,
versus platinum thermocouple is used. has a calibration approaching that of the Interna-
tional Practical Temperature Scale.
26 The temperature, t, is defined by the formula:
32 While Planck’s formula leads to
E = a + bt + ctz,
”
J, - exp (c2 / h TAJ -1
where E is the electromotive force of a standard
JA^ exp (cp I’ X T ) - 1
thermocouple of platinum-rhodium and platinum
alloy when one junction i s a t 32%’ and the other is in which
a t the temperature t. The constants, a, b, and c, J , and JA^ = the radiant energies per unit wave-
are to be calculated from the measured values of F length interval at wavelength, X,
at the freezing point of antimony and at the silver emitted per unit time by unit area
and gold points. of a blackbody a t temperature T ,
and at the gold point TA^, respec-
27 The platinum wire of the standard thenno- tively.
couple is annealed and of such purity that the ratio
c p = 1.4388 cm K
R , , , o~/& OFis greater than 1.3920. The alloy wire X = a wavelength of the visible spec-
shall consist nominally of 90 percent platinum and trum, cm
10 percent rhodium by weight. When one junction e = base of Naperian logarithm.
i s a t 3 2 v , and the other at the freezing point of
antimony (1167,3’%’), silver, or gold, the completed
which defines IPTS-68 above the gold point,
thermocouple has electromotive forces, in micro-
a narrowband optical pyrometer must be calibrated
volts, such that
a t a number of points using this formula, in order to
produce a means for extrapolation above the gold
EA, = 10,300 ? 50pV
point.
EA^ - EA^ -. 118.3 -I-0.158 EA^ - 10,300) i: 4pV
33 Calibration is done by sighting the pyrometer
EAu - = 476.6 + 0.631 EA^ - 10,300) ? 8pV on a blackbody maintained at the gold point. Rotat-
ing sectors of known transmission characteristics
28 Satisfactory standard thermocouples have been are then interposed between the blackbody and the
made of wires not less than 0.35 mm and not more pyrometer, and calibration at points higher than the
96
F-
97
- Power
SUPPlY
46 In all test baths a properly located well or
other suitable provision should be made for using a
platinum resistance thermometer a s the ultimate
primary standard.
Voriable
Resistance
(b) Comparators, VariableTemperature. Forstand- mation point of solid carbon dioxide, liquid
ardization at other than fixed points, the vari- nitrogen is used a s the cooling medium for
able temperature-type comparators are used. an isolated bath filled with some low-boiling
Figs. 9.5, 9.6,9.7,9.8, and 9.9, illustrate vari- hydrocarbon such a s isopentane. A suitable
ous types of such units for use in the range bath for this purpose is illustrated in Fig. 9.5
-256 to +lOOO°F. All of these units basically and is described in detail in ASTM Method
consist of a well-stirred, insulated liquid bath E77-70.
provided with suitable controls for maintaining
The evacuable flask in Fig. 9.5 is positioned
the temperature either constant or uniformly in-
in the larger container by three cork wedges
creasing! A s with the ice point equipment, a
strung on a wire. It must extend above the
viewing telescope is necessary for precision of
nitrogen container enough so that nitrogen
one-tenth of a scale division on liquid-in-glass
may be added without spilling into the iso-
thermometers. The liquids used in the com-
pentane. The aluminum block is completely
parators are chosen to fulfill the following re-
covered with isopentane and the space be-
quirements: low viscosity, nonflammability,
tween the two Dewar flasks filled with liquid
nontoxícity, and freedom from other offensive
nitrogen. A loose fitting cardboard shield
characteristics in the various temperature
with holes is placed over the top to reduce
ranges in which they are to be used.
condensation and convection. When the
(1) Comparators for the range from -256 to system has cooled to a few degrees above
-103OF. For comparison in the range from the desired temperature, the inner flask is
-256 to -103%', which i s below the subli- evacuated by the pump to stabilize the tem-
Mefol Flask
*Holder
J
FIG. 9.5 COMPARATORFOR TEMPERATURE RANGEFROM -256 TO - 1 0 3 O ~
100
I
RingSegmtnt showing
Ball GrooveandMethod
of Holding to Plate
2Required -Brass Spring Clip
26 Req. Phos. Bronze
Handle
Brass Plate Holes for 13
'I L Thermometers
i I
eodv
Ring
SupportingPools
'?harmomator CIIps
200°F
Pyrex jar suitably supported and equipped illustrated in Figs. 9.8(a) or 9.8(b) may be
with heaters, cooling coil, and stirring used.
equipment. .A commercially available bath Suitable high-flash point oils should be used.
provided with controIs to maintain constant A t the higher temperatures great care must
or very slowly rising temperature throughout be taken to avoid dangerous flash fires which
the testing range and with a top cover hav- may occur, particularly on removal of ther-
ing a large opening is shown in Fig. 9.7, mometers or thermometer holder. - A solid
For liquid-in-glass thermometer testing an cover plate to replace the thermometer hold-
insert equipped wíth two rotary holders made er as well a s adequate fire protection equip-
a s shown in Fig. 9.7 is provided. .Each of ment of the carbon dioxide type, should be
these holders is a flat plate restingon a provided.
number of ball bearings s o that the plate
(5) Comparators for the Range from 450 to 1000°F.
may easily be spun to bring any particular
A satisfactory comparator for the range 450
thermometer into view. Positive heat dis-
to 1000°F is a s a l t bath. . A type having a
tribution to all parts of the bath and thorough
tube arrangement immersed in the pot of a
circulation are obtained by a unique assembly
commercial heat treating bath is illustrated
of cooling and heating coils in a cylindrical
housing which also surrounds the stirring in Fig. 9.9. .Welded or riveted pots should
propeller, The bath should be filled with under no circumstances be used. -External
distilIed water. An alternative design is il- electrical heating is commonly practiced,
lustrated in Fig. 9.8(a), . although gas fired units may be purchased
and have been used successfully.
(4) Cbrnparators for the Range from 190 to 70OoF. Care should be exercised, particularly above
In the range 190 to 700% a comparator a s 752%, to avoid bringing any organic matter
101
pvrr
Thrrmomrftr
(o) Stirred Liquid Bath with Two coaxid (6) StirredLiquid Bath with ! h o Tuber Con-
Tubes. nected at Top and Bottom.
FIG. 9.8 ALTERNATIVE DESIGNS O F COMPARATORSFOR TEMPERATURE RANGES FROM 190 to 700°F
or low boiling liquid, such a s water, in con- avoided by using steel wells immersed in
tact with the molten salt, since dangerous the molten salt into which the thermometers
fires or explosions may occur. The bath or thermocouples are placed. Thin walled
should be heated up slowly at the start to steel tubes closed at one end are suitable
avoid the formation of pockets of molten for this purpose.
salt which have a tendency to blow out the Comparators with molten tin as the bath
solid mass with disastrous results. The liquid have been used successfully in this
bath should be covered while heating until temperature range. Details of design, con-
approximately 25 percent of the salt has struction, and use may be obtained from the
melted. National Bureau of Standards.
A salt comprised of 50 percent sodium ni-
trate and 50 percent sodium nitrite is sug-
Mueller Bridge
gested.
..
At thelowertemperaturesthethermometers 47 When calibratingresistancethermometers by
or thermocouples may be immerseddirectlycomparison with astandardplatinumresistance
in thesalt,but at thehighertemperatures thermometer, a bridge of the Mueller type is com-
attack of theglass may occur. This can be monly used. The circuit i s showninsimplified
102
1 ”
A
IO x / O R
taiIed instructions for checking the self-consistency “Methods of Testing Thermocouples and Thermo-
of calibration of Mueller bridges. couple Materials,” dated February 6, 1958, and the
calibration methods appearing in Chapter 8, of the
Laboratory Standard Potentiometer
“Manual On The Use of Thermocouples In Tempera-
50 Laboratory High Precision Potentiometers are ture Measurement”, STP 470, American Society for
used in the calibration of all types of thermocouples Testing and Materials.
and thermocouple materials. They should be cali-
brated periodically by the National Bureau of General Considerations
Standards or other laboratory similarly qualified.
58 GeneralMethods. The calibration of a thermo-
51 Refer to Chapter 3, Section B of this Supple- couple consists of the determination of its electo-
ment for principles of operation and types of Labora- motive force (emf) a t a sufficient number of known
tory Potentiometers available, temperatures so that, with some accepted means of
interpolation, its emf will be known over the entire
RADIATION THERMOMETERS
temperature range in which i t is to be used. The
52 Calibration of a radiation thermometer consists process requires a standard thermometer to indicate
in measuring the emf which it produces when fo- temperatures on a standard scale, a means for meas-
cused on a source, the apparent temperature of which uring the emfof the thermocouple, and a controlled
has been measured with a standardized radiation environment in which the thermocouple and the
thermometer which follows the same power law as standard can be brought to the same temperature.
the radiation thermometer under test. Some of the more commonly used techniques for ac-
complishing such calibrations will be discussed in
53 It is not necessary that the source be a black this section.
body to make this calibration valid, but it shouId
be nonselective in its emissivity; that is, have the 59 WorkingStandards. Any one of several types
same emissivity throughout i t s spectrum. It is, of thermometers, calibrated in terms of the IPTS,
therefore, permissible to make use of a method of may be used a s a working standard for the calibra-
calibration which has been found to be convenient tion of thermocouples. The choice will depend upon
in routine calibrations. the temperature range covered, whether a laboratory
furnace or a stirred liquid bath is used, the accuracy
54 The source is a furnace which is maintained expected of the calibration, or in c a s e s where more
a t a constant temperature; for example 2400%
than one type will suffice, the convenience or pref-
Lower temperatures are simulated by interposing a erence of the calibrating laboratory.
rotating sector disk between the source and the
radiation thermometer, and the apparent temperature 40 ResistanceThermometers. The standard plati-
measured by means of a standard radiation thermom- num resistance thermometer is the most accurate
eter. To make this method of calibration valid, the standard for use from approximately -253OC
two radiation thermometers being compared must (-423%’) to 630.5OC(1167%’). In c a s e s where an
follow the same power law, because the spectral uncertainty approaching O. 1°C is necessary at tern-
distribution of the radiation from a source a t 2400T. peratures below -56OC(-69OF) or above about
is quite different from that for a black body a t the 200°C (392v) there are few alternatives to the
measured apparent temperature. use of resistance thermometers as standards.
55 In all cases the dimensions of the furnace
61 Liquid-in-Glass Thermometers. Thistype of
opening should be such that the field-of-view re-
quirements of the radiation thermometer are satis- thermometer may be used from approximately -183OC
(-297%) to 40Ooc (752?), or even higher with
fied.
special types, Generally, the accuracy of these
56 Manufacturers will supply data for checking thermometers is less below -%OC (-69V), where
equipment if requested. organic thermometric fluids are used, and above
300°C(572%’) where instability of the bulb glass
THERMOCOUPLE THERMOMETERS
may require frequent calibration. Specifications for
57 Muchof this material in this section is based ASTM liquid-in-glass thermometers are given in
upon National Bureau of Standards Circular 590, ASTM El-71.
- 104
62 Types E and T Thermocouples. Either of these electrically in air. The entire thermocouple is sup-
types of thermocouples may be used down to a tem- ported between two binding posts, which should be
perature of -183°C (-297%’) or lower, but the at- close together, s o that the tension in the wires and
tainable accuracy may be limited by the accuracy stretching while hot are kept at a minimum, The
of the emf measurements and the inhomogeneity of temperature of the wire is most conveniently deter-
the wire at low temperatures. The stability of the mined with an optical pyrometer. The ordinary port-
larger sizes of wire is greater than that of smaller able type of opticaI pyrometer is very satisfactory
wires under the same conditions. Twenty-four gage for this purpose. A s commonly used, the magnifica-
wire is a useful compromise between the lesser tion is too low for sighting on an object as small as
stability of smaller wire and the greater thermal the wires of noble-metal thermocouples, but this is
conduction (greater required depth of immersion) of easily remedied by lengthening the telescope tube
larger wire. ‘Recommended upper limits are 425°C or using an objective lens of shorter focal length.
(797%’) for the Type E and 2OO0C(392%’) for Type T.
67 There are some questions as to the optimum
63 Types R and S Thermocouples. The Type S or temperature and length of time a t which such thermo-
Type R thermocouple is the most satisfactory work- couples should be annealed to produce the most
ing standard for use in the range from 630.5OC constant characteristics in later use, and as to
(1167°F) up to about 1200°C (2192%’). I t s u s e may whether annealing for more than a few minutes is
be extended down to room temperature if it is de- harmful or beneficial. Most of the mechanical
sired to use the same standard over a wide.range, strains are relieved during the first few minutes of
but i t s s e n s i t i v i t y f a l l soff appreciably as tempera- heating at 1400 to 1500°C (2552 to 2732”F), but i t
tures below 200°C(392%’) are reached. Twenty-four has been claimed that the changes in the thermal
gage wire is most commonly used for these stand- emf of a couple in later use will be smaller if the
ards. wires are heated for several hours before calibra-
64 HighTemperatureStandards. The IPTS above tion and use. The principal objection to annealing
1064.43°C (1946’F) is defined in terms of ratios of thermocouples for a long time at high temperatures,
radiant energy, the ratios usually being measured aside from the changes. in emf taking place, is that
by means of an optical pyrometer. .The optical py- the wires are weakened mechanically a s a result of
rometer, sighted on a blackbody cavity built into grain growth. .It h a s been found that annealing at
the calibration furnace, therefore, can serve a s a temperatures much above 150OOC(2732%’) produces
working standard for all temperatures above 1064.43OC. rapid changes in the emf and leaves the wire very
weak mechanically. The National Bureau of Stand-
On the other hand, thermocouples, calibrated on the
ards has adopted the procedure of annealing Types
optical pyrometer scale, can themselves be u s e d a s
standards. The Type B thermocouple is useful up R and S thermocouples for 1hr a t 1450°C (2642%’).
to abqut 1600°C (2912%’). Tungsten rhenium alloys 68 It has not been demonstrated conclusively that
can be used to higher temperatures, but the optical T y p e s R and S thermocouples after contamination
pyrometer is more commonly used. can be materially improved in homogeneity by pro-
longed heating in air, although it is logical to sup-
65 Annealing. Practically all base-metal thermo- pose that certain impurities can be driven off or,
couple wire is annealed or given a “stabilizing through oxidation, rendered l e s s detrimental.
heat treatment” by the manufacturer. Such treat-
ment is generally considered sufficient, and seldom 69 Measurement of Emf. One of the factors in the
i s it found advisable to further anneal the wire be- accuracy of the calibration of a thermocouple is the
fore testing. accuracy of the instrument used to measure the emf,
Fortunately, in most instances, an instrument is
66 Although new platinum-rhodium versus plati- available whose performance is such that the accu-
num thermocouple wire a s sold by some manufactur- racy of the calibration need not be limited by the
ers is already annealed, it has become regular prac- accuracy of the emf measurements. -For work of the
tice in many laboratories to anneal all Types R and highest accuracy it is advisable to u s e a potenti-
S thermocouples, whether new or previously used, ometer of the type in which there are no slidewires
before attempting an accurate calibration. This is and in which all the settings are made by means of
usually accomplished by heating the thermocouple dial switches. However, for most work, in which an
accuracy of 5 pV will suffice, slidewire potentiom- 73 After reasonable homogeneity of one sample
eters of the laboratory type are sufficiently-accurate. of wire h a s been established, it may be used in
Portable potentiometers accurate within 40 to 1OopV testing the homogeneity of similar wires by welding
are also available. the two together and inserting the junction into a
70 Homogeneity. The emf developed by a thermo- heated furnace. .The resulting emf at various depths
couple made from homogeneous wires will be a of immersion may be measured by any convenient
function of the temperature difference between the method. Other similar methods have been described
measuring and the reference junction. If, however, for detecting inhomogeneity.
the wires are not homogeneous, and the inhomoge- 74 Tests such as those described above will in-
neity is present in a region where a temperature dicate the uncertainty in temperature measurements
gradient exists, extraneous emf’s will be developed, due to inhomogeneity in the wires. . F o r example, if
and the output of the thermocouple will depend upon a difference in emfof lOpV is detected along either
factors in addition to the temperature difference be- element of a platinum-rhodium couple by heating
tween the two junctions. The homogeneity of the various parts of the wire to 6OO0C(1112%’), measure-
thermocouple wire, therefore, is an important factor ments made with it are subject to an uncertainty of
in accurate measurements. the order of 1 deg to 500°C, or 2 deg at 12OOOC.
Similarly, if an emf of 10 pV is detected along an
71 Thermocouple wire now being produced is usu- element of a base-metal couple with a source of
ally sufficiently homogeneous in chemical compo- heat at 100°C, measurements made with it are sub-
sition for most purposes. Occasionally inhomoge- ject to an uncertainty of the order of 0.2 deg at this
neity in a thermocouple may be traced to the manu- temperature. The effects of inhomogeneity in both
facturer, but such cases are rare. More often it is wires may be either additive or subtractive and, as
introduced in the wires during tests or use. It usu- the emf developed along an inhomogeneous wire de-
ally is not necessary, therefore, to examine new pends upon the temperature distribution, i t i s evi-
thermocouples for inhomogeneity, but thermocouples dent that corrections for inhomogeneity are imprac-
that have been used for some time should be so ticable if not impossible.
examined before an accurate calibration is attempted.
75 GeneralCalibration Methods. T h e temperature-
72 While rather simple methods are available for emf relation of a homogeneous thermocouple is a
detecting thermoelectric inhomogeneity, no satis- definite physical property and therefore does not
factory method has been devised for quantitatively depend upon the details of the apparatus or method
determining it or the resulting errors in the measure- employed in determining this relation. Consequent-
ment of temperatures, Abrupt changes in the thermo- ly, there are numerous methods of calibrating ther-
electric power may be detected by connecting the mocouples, the choice of which depends upon the
two ends of the wire to a sensitive galvanometer type of thermocouple, temperature range, accuracy
and slowly moving a source of heat, such as a required, size of wires, apparatus available and
Bunsen burner or small electric furnace, along the personal preference. However, the emfof a thermo-
wire. This method is not satisfactory for detecting couple with its measuring junction at a specified
gradual changes in the thermoelectric power along temperature depends upon the temperature difference
the length of the wire, Inhomogeneity of this nature between its measuring and reference junctions.
may be detected by doubling the wire and inserting Therefore, whatever method of calibration is used,
it to various depths in a uniformly heated furnace, the reference junction must be maintained constant
the two ends of the wire being connected to a a t some known temperature and this temperature
galvanometer a s before. If, for example, the must be stated as a necessary part of the calibra-
doubled end of the wire i s immersed 1 0 in, in a tion results.
furnace with a sharp temperature gradient S O that 76 Thermocouplecalibrationsarerequired with
two points on the wire 2 0 in. apart are in the tem- various degrees of accuracy ranging from 0.1. to 5 or
perature gradient, the emf determined with the gal- 10 deg C.- For an accuracy of 0.1 deg, agreement
vanometer is a measure of the difference in the with the IPTS and methods of interpolating between
thermoelectric properties of the wire at these two the calibration points become problems of prime im-
points. portance, but for an accuracy of about 10 deg com-
106
paratively simple methods of calibration will usually a calibration accurate to 2 or 3 deg C in the range
suffice. The most accurate calibrations in the room temperature to 1200°C (2192%') is sufficient.
range -183OC (-297%) to 300°C (572%) are made Other thermocouples can be calibrated by compari-
by comparing the couples directly with a standard son with such working standards almost as accu-
platinum-resistance thermometer. in a stirred liquid rately a s the calibration of the standard is known.
bath. In the range 300 to 630.5OC (572 to 1167v) However, i t might be pointed out that outside the
(and below if a platinum-resistance thermometer and range 630.5 to 1063OC any type of thermocouple
stirred liquid bath is not available) thermocouples suitable for the purpose, and calibrated to agree
are most accurately calibrated at the freezingor with the resistance thermometer or optical pyrom-
boiling points of pure substances, Between 630.5 eter in their respective ranges, has a s much claim
and 1063OC (1167 and 1945OF), Type S thermocouple to yielding temperatures on the IPTS as the Type S
calibrated at 630.5OC and the freezing points of thermocouple. In fact, at the Lower temperatures
gold and silver, serves to define the IPTS, and certain types of base-metal thermocouples are defi-
other types of thermocouples are most accurately nitely better adapted for precise measurements.
calibrated in this range by direct comparison with
a standard thermocouple calibrated as specified. 78 The calibration of thermocouples then may be
Other thermocouples may be calibrated just as ac- divided into two general classes, depending upon
curately at the fixed points a s t h e T y p eS tharmo- the method of determining the temperature of the
couple, but interpolated values at intermediate measuring junction, (1) calibration at fixed points
points may depart slightly from the IPTS. Above and (2) calibration by comparison with standard in-
1063°C (1945v), the most basic calibrations are struments such a s thermocouples, resistance ther-
made by observing the emf when one junction of the mometers, etc.
thermocouple is in a blackbody furnace, the temper-
79 In order to obtain the high accuracies referred to
ature of which is measured with an optical pyrom-
eter. However, the difficulties encountered in above and usually associated with calibrations at
bringing a blackbody furnace to a uniform tempera- fixed points, it is necessary to follow certain pre-
ture make the direct comparison of these two types scribed methods and to take the special precautions
of instruments by no means a simple matter. described in detail in the following paragraphs, but
for an accuracy of about 5 deg C the more elaborate
77 Although the Type S thermocouple serves to apparatus to be described need not be employed.
define the IPTS oniy in the range 630.5 to 1063OC,
this type of thermocouple calibrated at fixed points 80 CalibrationUncertainties. Theseveralfac-
is used extensively both above and below this tors which contribute to the uncertainties in the
range a s a working standard in the calibration of emf versus temperature relationship for a particular
other thermocouples. ' F o r most industrial purposes thermocouple as determined by calibration may be
Calibration Uncertainty
107
grouped into two kinds; those influencing the ob- ferent techniques are given in Tables 9.3, 9.4, 9.5,
servations at calibration points, and those arising 9.6 and 9.7. The estimates assume that reasonable
from any added uncertainty a s a result of interpola- care is exercised in the work. More or less accu-
tion between the calibration points. Errors from rate results are possible using the same methods,
either of these sources of uncertainty can be mate- depending upon soundness of the techniques used,
rially reduced, within limits, through u s e of well de-. While excessive care is a waste when relatively
signed equipment and careful techniques; hence the crude measurements are sufficient, i t should be em-
required accuracy should be clearly understood phasized that inadequate attention to possible
when choosing calibration facilities. sources of error is more often found to be the prac-
tice than the converse. In the following some of
81 Estimates of the accuracies attainable in the the important considerations associated with the
calibration of homogeneous thermocouples by dif- various calibration methods are briefly emphasized.
Calibration Uncertainty
I I I I
Calibration Uncertainty
Temp.
Range of Observed
At Of Interpolated
OC Values,
Standard"
Points,
Calibration
Points
TY Pe
I
Deg C Deg C
E -1% to 425 About every 100 deg C PRT o. 1 o. 2
Il About every 50 deg C Il II o. 1
Il II
E or T 0.2 0.2
-56 to 200 li LIG o. 1 o. 1
T -196 to 250 About every 100 deg C PRT o. 1 0.2
II 50 deg C II o. 1 o. 1
11 50 deg C0.2 E or T 0.2
-56 t o 200 50deg C 0. LIG 1 o. 1
108
"
.-
Y
-
COPYRIGHT American Society of Mechanical Engineers
Licensed by Information Handling Services
ASME
~- PTCx19.3
~ 7q 1 : 0 7 5 9 b 7 0 005,2503 7 m
.!
INSTRUhqENTS AND .APPARATUS
TABLE 9.6 TUNGSTEN-RHENIUM T Y P E THERMOCOUPLES
(Maximum calibration uncertainties for range 1000 to 2000 C
using melting points by wire or disc method)
Calibration Uncertainty
At Observed
Points I O f lnterpolafed
Values'
Nickel 1453
(1453OC) t3.5 deg C to t4.0 deg C
1552OC
Palladium 1552
(1552OC) f3.0 deg C f4.0 deg C
l?69OC
Platinum 1769
(1769OC) k3.0 deg C k7.0 deg C
2io"oo~c
Rhodium
(1960OC) t5.0 deg C
Calibration Uncertainty
a Using difference curve from reference table with calibration points spaced
every 200 deg C.
40Ir60Rh vs. Ir, 50 Ir50Rh VS. Ir, or 60Ir40Rh vs. Ir.
c W YS. 74W26Re, 97'83Re VS. 74W26Re, or 95We VS. 74W26Re.
70Pt30Rh VS. 94Pt6Rh.
OC OF
Boiling point of helium -268.935 -452.083
Boiling point of equilibrium hydrogen -252.883 -423.189
Sublimation point of carbon dioxide -109.3
-78.5
Freezing point of mercury -38.86 -37.95
Freezing point of water 0.00 32.00
Triple point of benzoic acid 122.36 252.25
Freezing point of indium 156.61 313.90
Freezing point of tin 231.91 449.44
Freezing point of bismuth 271.37 520.47
Freezing point of cadmium 321.03 609.86
Freezing point of lead 327.43 621.37
Freezing point of antimony 1166.9
630.5
Freezing point of aluminum 1220.2
660.1
Freezing point of copper 1083.0 1981.0
Freezing point of palladium 1552.0 2826.0
Freezing point of platinum 1769.0 3216.0
110
following conditions are fulfilled: (1) the thermo- by optical pyrometry. This method is not well
couple is protected from contamination; (2) the adapted to metals that oxidize rapidly, and if used
thermocouple is immersed in the freezink-point with materials whose melting temperature is altered
sample sufficiently far to eliminate heating or cool- by the oxide, the metal should be melted in a neutral
ing of the junction by heat flow along the wires and atmosphere.
protection tube; (3) the reference junctions are
maintained a t a constant and reproducible tempera-
90 CalibrationUsingComparisonMethods. The
ture; (4) the freezing-point sample is pure; and calibration of a thermocouple by comparison with a
(5) the metal is maintained at essentially a uniform working standard is sufficiently accurate for most
temperature during freezing. purposes and can be done conveniently in most in-
dustrial and technical laboratories. The s u c c e s s of
88 Techniques for achieving these conditions this method usually depends upon the ability of the
are well developed. Manyof the metals listed in observer to bring the measuring junction of the ther-
Table 9.8 are available commercially in high purity mocouple to the same temperature a s the actuating
( c a 99.999 percent or better) and can be used as- element of the standard, such a s the measuring
suming the freezing point temperatures given in the j-unction of a standard thermocouple or the bulb of a
table. It is essential, however, that protection resistance or liquid-in-glass thermometer. The ac-
tubes and crucibles be chosen of such material that curacy obtained i s further limited by the accuracy
the pure metals will not be contaminated. Copper of the standard. Of course, the reference junction
and silver must be protected from oxygen contamina- temperature must be known, but this can be controlled.
tion, and it is also advisable to protect a1umin.w The method of bringing the measuring junction of
and antimony; this is done by using covered cru- the thermocouple to the same temperature as that of
cibles and covering the freezing point metals with the actuating element of the standard depends upon
powdered graphite. The choice of a suitable furnace the type of thermocouple, type of standard, and the
is a l s o important. The furnace must provide uniform method of heating.
heating in the region of the freezing point sample,
and have adequate controls to bring the sample 91 LaboratoryFurnaces. Thecalibration pro-
slowly into its freeze, Complete units consisting cedure consists of measuring the emf of the thermo-
of freezing point sample, crucible, and furnace are couple being calibrated at selected calibration
available commercially. Freezing point standards points, the temperature of each point being meas-
ured with a working standard. The number and
of tin, lead, zinc, aluminum, and copper may be
choice of calibration points will depend on the type
purchased from the National Bureau of Standards.
of thermocouple, the temperature range covered, and
89 MeltingPoints. The emfof a thermocouple a t the accuracy required.
the melting point of a metal may be determined with 92 Platinum-RhodiumVersusPlatinumThermo-
the same apparatus as that described above for couples. Thermocouples employing platinum and
freezing points, but the use of the freeze is usually platinum rhodium alloys are seldom used for accu-
more satisfactory. Melting points are used to ad- rate measurements below 30006 (572%’) because the
vantage, however, when only a limited amoun-t of sensitivity of these thermocouples decreases rapid-
material is available or a t high temperatures where ly at low temperatures.
experimental techniques with freezing points are
difficult. To apply this method, a short length of 93 These thermocouples are usually calibrated
a t temperatures up to 120OOCby comparison with
metal whose melting point is known is joined be-
either a Type S or Type R working standard in elec-
tween the end of the two wires of the thermocouple
trically heated furnaces. Above K3IO0C (2192OF)
and placed in an electrically heated furnace the
the Type B thermocouple is a preferred working
temperature of which is slowly raised. When the
standard because of its greater stability at high
mdting point of the metal is reached, the emf of the
thermocouple remains steady for a few minutes and temperatures. . T h i s thermocouple may be used to
then drops to zero as the fused metal drops away 160OOC(2912%’) or higher.
from the junction. With good technique the method 94 One method for the comparison of two such
can give results with no greater uncertainty than thermocouples is based upon the simultaneous
that with which the IPTS is realized above 1063OC reading of the emf of the standard and the test ther-
111
mocouple without waiting for the furnace to stabi- or a photoelectric pyrometer. If the test thermo-
l i z e a t any given temperature. The measuring junc- couple is inserted into the back of a blackbody
tions are maintained always a t c l o s e to the same cavity built into the furnace, a pyrometer may be
temperature by welding them into a common bead or used directly as the working standard. Alterna-
by wrapping them together with platinum wire or tively, the Type B thermocouple can be used a s the
ribbon. A separate potentiometer is used to meas- working standard after it has been calibrated against
ure each emf, one connected to each thermocouple, a pyrometer.
and each potentiometer is provided with a reflect-
98 The thermocouples are insulated and protect-
ing galvanometer. The two spots of light are re- ed by suitable ceramic tubes. It is essential that
flected onto a single scale, the galvanometers being good insulation be maintained between the two po-
s e t in such a position that the spots coincide at the tentiometers and thermocouple circuits except a t
zero point on the scale when the circuits are open, the point where the junctions are welded together.
and therefore, also when the potentiometers are set The reference junctions are maintained at a known
to balance the emfof each thermocouple. Simul- temperature.
taneous readings are obtained by setting one poten-
tiometer to a desired value and adjusting the other 99 Variations of the two potentiometer method
so that both spots of light pass across the zero of may be used to automate the calibration process
when the thermocouple being calibrated and the
the scale together as the temperature of the furnace
standard thermocouple are of the same type. If the
is raised or lowered.
emf of the standard is read with one potentiometer
95 By making observations first with a rising and and the emf difference between the standard and
then with a falling temperature, the rates of rise and the unknown are read with the second potentiometer
fall being approximately equal, and taking the mean the calibration data may be recorded automatically.
of the results found, several minor errors such as
those due to differences in the periods of the gal-
100 If two potentiometers are not available for
vanometers, etc., are eliminated or greatly reduced. taking simultaneous readings, the furnace may be
brought to essentially a constant temperature and
96 This method i s particularly adapted to the the emf of each thermocouple read alternately on
calibration of thermocouples a t any number of se- one instrument.
lected points. -For example, if it is desired to deter-
termine the temperature of a thermocouple corres- 101 When thethermocouplesarecalibrated by
ponding to 10.0 millivolts, this emf is s e t up on the welding or wrapping the junctions together, both
potentiometer connected to the thermocouple, the would be expected to be close to the same temper-
emf of the standard thermocouple observed as de- ature even when the temperature of the furnace is
scribed above, and the temperature obtained from changing. If i t is necessary or advisable to cali-
the emfof the standard. If i t is desired to deter- brate the thermocouples without removing them from
mine the emfof a thermocouple corresponding to the protection tubes, then the junctions of the ther-
1000°C (1832%'), the emf of the standard correspond- mocouple being tested and that of the standard
ing to this temperature is s e t up on the potentiom- should be brought as close togetheras possible in
eter connected to the standard and the emf of the a uniformly heated portion of the furnace. . In this
thermocouple being calibrated is observed directly case it is necessary that the furnace be brought to
with the second potentiometer, To reduce the time approximately a constant temperature before taking
required to calibrate by this method the furnace observations.
should be so constructed that it will heat or cool 102 There are a number of other methods of heat-
rapidly. Fast response is obtained in one furnace
ing and of bringing the junctions to approximately
design.which employs a nickel-chromium tube as the the same temperature, for example, inserting the
heating element. thermocouples properly protected into a bath of
97 A similar furnace using a silicon carbide tube molten metal or into holes drilled in a large metal
a s the heating element can be used to extend the block. .The block of metal may be heated in a muffle
calibration range upwacd. A t temperatures above furnace or, if made of a good thermal conductor such
1063OC (1945T) the IPTS is defined in terms of as copper, may be heated electrically. Tin, which
ratios of radiation usually measured with an optical has a low melting point, 232OC ( 4 5 0 v ) , and low
112
volatility, makes a satisfactory bath material. The appear to be good. More than one base-metal ther-
thermocouples should be immersed to the same mocouple may be welded together and the hole
depth with the junctions close together. Ceramic drilIed in the composite junction. The thermocouple
tubes are sufficient protection, but to avoid break- should be clamped in place so that the junctions re-
age by thermal shock when immersed in molten main in contact. - If two potentiometers are used for
metal it is preferable to place them inside of seconde taking simultaneous readings, the temperature of
ary tubes of iron, nickel-chromium, graphite, or simi- the furnace may b e changing a s much as a few de-
lar material, In all of these methods, particularly grees per minute during an observation, but if a
in those cases in which the junctions of the thermo- single instrument is used for measuring the emf, the
couples are not brought into direct contact, i t is furnace temperature should be maintained practi-
important that the depth of immersion be sufficient cally constant during observations. . Rhen wires,
to eliminate cooling or heating of the junctions by insulators, and protection tubes are large, tests
heat flow along the thermocouple and the insulating should be made to insure that the depth of immer-
and protecting tubes. T h i s can be determined by sion is sufficient.
observing the change in the emf of the thermocouple 104 StirredLiquidBaths. A t temperaturesbelow
a s the depth of immersion is changed slightly. If 62OoC (ll48T) stirred liquid baths provide an effi-
proper precautions are taken, the accuracy yielded cient medium for bringing a thermocouple and a tem-
by any method of heating or bringing the junctions perature standard to the same temperature.
to the same temperature may be a s great as that ob=
tained by any other method. 105 Water, petroleum oils, or other organic
liquids, depending upon temperature range, are com-
103 BaseaMetalThermocouples, Themethods of
monly used bath media. Molten s a l t s or liquid tin
testing base-metal thermocouples above room tem- are used at temperatures higher than are suitable
perature are generally the same a s t h o s e j u s t de-
for oil. Base-metal thermocouples, either bare wire
scribed for testing noble-metal thermocouples with
or insulated, may be accurately calibrated in such
the exception, in some cases, of the methods of
baths. Usually no special preparation of the ther-
bringing the junctions of the standard and the ther-
mocouple will be required other than to insert it to
mocouple being tested to the same temperature and the bottom of a protection tube for immersion in the
the methods of protecting platinum-rhodium stand- liquid bath. Borosilicate glass tubing, such as
ards from contamination. . One arrangement of bring-
Pyrex glass, is convient for use up to 538OC
ing the junction of a platinum-rhodium standard to (1000%’). Vitreous silica or ceramic tubing may be
the same temperature a s that of a large base-metal used to 620°C (1148%’). The tube should be closed
thermocouple for accurate calibration is to insert at the immersed end and of an internal diameter
the junction of the standard into a small hole (about such as to permit easy insertion of the thermocouple
0.06 in. in diameter) drilled in the hot junction of or thermocouples to be calibrated, but no longer than
the base-metal thermocouple, The platinum-rhodium necessary. -Unfavorable heat transfer conditions in
standard is protected by ceramic tubes to within a an unnecessarily large diameter tube will require a
few hundredths of an inch of the hot junction, and
greater depth of immersion in the bath than would a
the end of the ceramic tube is sealed to the thermo-
close fitting tube. If a bare wire thermocouple is
couple by Pyrex glass or by a small amount of
being calibrated, the wires must be provided with
kaolin and water-glass cement. This prevents con-
electrical insulation over the length inserted in the
tamination of the standard thermocouple, with the
protection tube. Sheathed thermocouples may be
exception of the small length of about 0.1 in,; which
immersed directly in the bath liquid in c a s e s where
is necessarily in contact with the base-metal ther-
the sheath material will not be attacked by the
mocouple. If the furnace is uniformly heated in this
liquid. -Salt baths for use at high temperature must
region (and i t is of little value to make such a test
be provided with suitable wells into which the ther-
unless it is) contamination at this point will not
mocouple protection tubes and standard thermom-
cause any error. If the wire of the standard becomes
eters may be inserted for protection from the molten
brittle at the junction, this part of the wire may be
salt.
cut off and enough wire drawn through the softened
s e a l to form a new junction. The seal should be 106 The standard thermometermaybe a thermo-
examined after each test and remade if it does not couple standard inserted in the protection tube with
113
the thermocouple being calibrated, or i t may be a the thermocouple being tested, then remove this
liquid-in-glass thermometer or resistance thermom- thermocouple and insert the standard thermocouple
eter immersed in the bath close to the thermocouple to the same depth.
protection tube. The choice of a standard thermom-
110 If desired, comparisons can be made, prefer-
eter will be governed principally by the degree of
ably by either of the first or second methods at
uncertainty which can be tolerated.
several temperatures, and a curve obtained for each
107 Fixed Installations. Afterthermocouples permanently installed thermocouple showing the
have been used for some time a t high temperatures, necessary corrections to be applied to its readings.
i t is difficult-if not impossible, to determine how Although testing a thermocouple a t one temperature
much the calibrations are in error by removing them yields some information, it is not safe to assume
from an installation and testing in a laboratory fur- that the changes in the emfof the thermocouple are
nace. The thermocouples are usually heterogeneous proportional to the temperature or to the emf. For
after such use and in such a condition that the emf example, i t h a s been observed that a thermocouple
developed by the thermocouples depends upon the which had changed in u s e by the equivalent of 9 deg
temperature distribution along the wires. If pos- c at 315'C (16 deg F at 599%') had changed only
sible, such a thermocouple should be tested under the equivalent of 6 deg C a t llOO°C (11deg F a t
the same conditions and in the same installation in 20 12OF).
which i t is used. Although it i s not usually pos- 11 1It may be thought that the method of check-
sible to obtain a s high a precision by testing the ing thermocouples under working conditions is un-
thermocouple in place a s is obtained in laboratory satisfactory because, in most furnaces used in in-
tests, the result is far more useful in the s e n s e of dustrial processes, large temperature gradients
being representative of the behavior of the thermo- exist and there is no certainty that the standard
couple. The calibration is accomplished by com- thermocouple is at the same temperature as the
paring the thermocouple with a thermocouple stand- thermocouple being tested. This objection, how-
ard. ever, is not serious, because if temperature gradi-
ents do exist of such a magnitude as to cause much
108 In this case, as in the calibration of any
difference in temperature between two similarly
thermocouple by comparison methods, the main ob-
mounted thermocouples located close together, the
jective is to bring the hot junction to the same tem-
reading of the standard thermocouple represents the
perature as that of the thermocouple being tested.
temperature of the fixed thermocouple a s closely a s
One method i s to drill a hole in the furnace, flue,
the temperature of the latter represents that of the
etc., at the side of each thermocouple permanently
furnace.
installed, large enough to permit insertion of the
checking thermocouples. The hole is kept plugged, 112 Another advantage of checkingthermocouples
except when tests are being made. The standard in the same installation in which they are used is
thermocouple is inserted through this hole to the
that the thermocouple, extension wires, and indi-
same depth a s the thermocouple being tested with
cator are tested as a unit and under the conditions
the hot junction ends of the protection tubes as of use.
close together as possible. .Preferably a potenti-
ometer should be used with the standard thermo- 113 InterpolationMethods. An experimentalther-
couple. mocouple calibration consists of a series of voltage
109 In many installations the base-metal thermo- measurements determined a t a finite number of
couple and protecting tube are mounted inside an- known temperatures. If a t e s t thermocouple were
other protecting tube of iron, fire clay, carborundum, compared with a standard temperature instrument at
or some other refractory which is permanently ce- 100 temperatures within a 10 deg F r s g e , there would
mented or fastened into the furnace wall. Frequent- be little need for interpolation between the calibra-
ly there is room to insert a small test thermocouple tion points. However, if from 4 to 10 calibration
in this outer tube alongside of the fixed thermo- points are all that can be afforded in a given range
couple. A third method, much less satisfactory,.is of interest, then what is needed to characterize an
to wait until the furnace, flue, etc.; have reached a individual thermocouple is a continuous relation, by
constant temperature and make observations with means of which temperatures can be approximated
O 5 10 15 20 25 30
Voltage ( E couple) mV
with a minimum uncertainity from voltage measure- calibrating the thermocouple also at the freezing
ments at intermediate levels, Efforts to obtain such point of zinc and using an equation of the form
a continuous relation appear thwarted from the start c = a + bt 3- ct 2 + d t 3 , the temperature range can be
because of the small number of discrete calibration extended down to 400°C without introducing an un-
pointsavailable.However,interpolationbetween certainty of more than 0.1 deg C in the range 630.5
the calibration points is possible since the emf to 1063.0"C. By calibrating the thermocouple at
changes only slowly and smoothly with temperature. freezing points of gold, antimony, and zinc and
using an equation of the form e = a t bt t c t 2 , a
114 One can present raw calibration data directly
calibration is obtained for the range 400 to llOO°C,
in terms of temperature T and voltage E,,,k, on a
which agrees with IPTS to 0.5 deg C. However, in
scale s o chosen that the information appears well
general, this practice of directly representing
represented by a single curve (see Fig. 9.11) or by
thermocouple characteristics does not yield results
a simple mathematical equation. For example, for
within the required l i m i t s of uncertainty.
the highest acduracy in the range 630.5 to 1063.0"c
with the Type S thermocouple, the method is that 115 A better method is based on the use of dif-
prescribed in the NBS Circular No. 590. An equa- ferences between observed values and values ob-
tion of the form e = a + bt + c t 2 , is used where a, b, tained from standard reference tables. Such refer-
and c are constants determined by calibration at the ence tables are presented in ASTM Standard E230-
freezing points of gold, silver, and antimony. By 72, Temperature-Electromotive Force (EMF)
115
P
3
U
Lu
I
cc
e
Y
o
U
C
?!
n
.-
Y
Y
B
m
-
4.
O
>
O 5 10 15 20 25 30
Voltage (E couple) mV
>
E
.o
h
W
Tables for Thermocouples. The data of Fig. 9.11 minimized by u s e Qf the least squares technique,
are replotted in Fig, 9.12 in terms of differences one starts the search for the most probable inter-
from the proper reference table. The maximum polation equation by passing a least squares equa-
spread between points taken at the same level tion of the first degree through the experimental
(replication), but obtained in random order with data. A check is then made to ascertain whether
respect to time and level (randomization) is taken all experimental points are contained within the un-
as the uncertainty envelope. This information, certainty envelope which is centered on the linear
taken from Fig. 9.12 is plotted in Fig. 9.13, and interpolation equation (see Fig. 9.15). One pro-
constitutes a vital bit of information about the ceeds,accordingtotheresults of theforegoing .
particular thermocouple and the calibration system. check, to the next highest degree equation, stop-
In lieu of an experimental determination of the un- ping a t the lowest degree least squares equation
certainty, one must rely on judgment or on the cur- which satisfies the uncertainty requirements. For
rent literature for this information. the example given here, a third degree interpolation
equation is required (see Fig. 9.16). By obtaining
116 Usually, only a single set of calibration
voltage differences from the least squares fit of any
points is available. .Typical points would be those
s e t of calibration points, the uncertainty in the
taken from one run shown in Fig. 9.11 or 9.12 aed
thermocouple difference characteristic will be with-
these are shown in Fig. 9.14 together with four of
in one-half the uncertainty interval. Generally, the
the many possible methods for representing the
form of the uncertainty envelope and the degree of
thermocoupledifferencecharacteristic.Although
the most probable least squares interpolation equa-
at first it appears that the most probable relation
tion are strongly dependent on the amount of cali-
characterizing a given thermocouple is sensibly in-
bration data available and on the temperature range
determinate from a s i n g l e s e tof calibration points,
i t is an important fact that all experimental points under consideration, It is recommended that the
must be contained within the uncertainty interval number of distinct calibration points available
when the uncertainty interval is centered on the should be at least 2 (degree -t. 1). The factor 2 is
most probable interpolation equation. arrived a t from numerical analysis reasoning. A
distinct calibration point is defined arbitrarily as
117 Making use of this principle, together with one which is separated, temperature-wise, from all
the fact that overall experimental uncertainties are other points in t h e s e t b y a smuch a s one-tenth the
/-i
I I
> 0.01
E
-2a
h
0
3
-0.01
C
?!-0.03
LC
Y
n
-0.04
-cO
O
> -0.05
> 0.01
E
-n
h
O
3
O
Uncertainty envelope
w -0.01
Y
I
e
t (does not include all
experimental data;
therefore, most probable
!Y. -0.02 i n t e r p o l a t i n g equation
o h a s n o t been obtained)
C
?!-0.03 F i r s t degree least
.-
Y
u
n
squares interpolating
equation
-0.04
-
O
c
Experimental Data (Run 2)
-0.05
O 5 10 15 20 25 30
Voltage (E couple), mV
> -0.01
E
Third degree least
-an-
h
o squares interpolating
J equation
;-0.01
I
Experimental Data (Run 2)
cc Uncertainty envelope
e -0.02
(includes all experimental
!Y. data; therefore, the third
B degree interpolating
U
equation best represents
-0.03 this calibration run)
u
Y
6
O -0.04
m
-O
+
O
> -0.05
O 5 10 15 20 25 30
Voltage (E couple), mV
118
119
1 .
~ . . ~
122 If the thermocouple is very short, resulting as standard Pt 27. This standard is spectrochemi-
in the reference junctions being near the furnace cally pure, has been thoroughly annealed, and al-
and subject to considerable variations or uncertain- though i t may not be the purest platinum that has
ty in temperature, it is usually more convenient to been prepared, it serves as a satisfactory standard
use thermocouple leads to transfer the reference to which the thermal emf of other materials may be
junctions to a region of more constant and lower referred to. However, nothing prevents any other
temperature than to measure the temperature of the laboratory from setting up a laboratory standard for
reference junctions near the furnace. The thermo- their own use. In order that the various 1abo;atories
couple leads of base-metal couples are usually and manufacturers may specify and express values
made of the same materials a s the thermocouple of thermal emf on a common basis, a common and
wires, but in the case of platinum-rhodium versus ultimate standard is necessary.
platinum couples a copper thermocouple lead i s 124 Platinum is u s e d a s a working standard for
connected to the platinum-rhodium wire and a COP- testing thermocouple materials in some laboratories.
per-nickel thermocouple lead to the platinum wire. I t is generally more convenient to u s e a working
Thermocouple leads, for any of the couples dis- standard of the same material a s that being tested.
cussed here, are available from ali of t.he pyrometer In any case, the thermal emf of a material against
instrument manufacturers. Although the temperature- the standard P t 27 is the algebraic s u m of the emf
emf relation of the copper versus copper-nickel of the material against the working standard and the
thermocouple lead wire is practically the same a s emf of the working standard against the standard
that of platinum-rhodium versus platinum couples, Pt 27 (the law of intermediate metals), . When plati-
the individual thermocouple lead wires are not num is used a s a working standard in testing some
iden tical thermoelectrically with the couple wires other material, the thermal emf measured is large.
to which they are attached. Therefore, the two T o obtain the thermal emfof the material against
junctions where the leads are attached to the the standard P t 27, the relatively small emf of the
couple should be kept at nearly the same tempera- platinum working standard against the standard
ture. This is not necessary in the case of base- Pt 27 is added to the large measured emf. 'When the
metal couples when each thermocouple lead and working standard is of the same kind of material a s
couple wire to which it is attached are of the same that being tested, the thermal emf measured i s small.
material, (The thermocouple leads are then referred To obtain the- thermal emfof the material against
to a s thermocouple extension wires.) the standard Pt 27 in this case, the relatively large
_emf of the working standard against the standard
Calibration of Thermocouple Materials
Pt 27 is added to the small measured emf.
123 Thermocouples are ordinarily made up to de- 125 Except in the case of constantan, two
samples of a similar material which will develop
velop a specified emf at one or more temperatures,
more than 0.25 pV/deg F against one another are
and in order to select and match materials to do this,
exceptional. In most cases the value is less than
a convenient method of testing each element is re-
0.1 pV/deg F. Even in the case of constantan, the
quired. One method of accomplishing this i s to de-
thermal emf between two extreme samples does not
termine. the thermal emf of the various materials
exceed 1.5 pV/deg F. Therefore, in determining the
againkt some stable and reproducible material. At
difference between two samples of a similar mate-
low temperatures copper is sometimes used for this rial, it is not necessary to measure the temperature
purpose, but platinum appears to be the most satis- accurately.
factory because it can be used at any temperature
up to its melting point, can be freed from all traces 126 The average thermal emf/deg F of platinum
of impurities, and can be readily annealed in air. against other thermocouple materials is specified in
Two samples of platinum, both of which are spec- Table 9.11. It is seen that in measuring the thermal
tro-chemically pure, may differ slightly in thermal emf of these materials directly against platinum
emf, but the same is true of any other metal. To working standards, it is necessary to measure an
avoid the ambiguity that might arise from this fact, emf which changes by a large amount for a small
the thermal emfof thermocouple materials tested a t change in temperature, An accurate measurement of
the National Bureau of Standards (since 1922) is the emf corresponding to a given temperature, there-
referred to an arbitrary piece of platinum designated fore, requires an accurate measurement of the tem-
120
TABLE 9.11 AVERAGE THERMAL* EMFIDEG F OF Some of the precautions that must be observed to ob-
PLATINUM AGAINST OTHER tain accurate results are discussed in the following
THERMOCOUPLE MATERIALS
paragraphs.
I l '
Average 129 Platinum. Thethermal emf of thermocouple
Change in platinum against the standard P t 2 7 is usually less
Thermal Emf
Material ture, With than 20 pV a t 2192v and in testing one sample of
OF Temperature, platinum against another it is not necessary to meas-
pV/Deg F
ure the temperature of the hot junction to closer
1832 6.4
than 9 0 9 to obtain a comparison accurate to 1pv.
Platinum-10 percent rhodium
7.3
Platinum-13 percent rhodium 1832 The reference-junction temperature need not be ac-
Chrome1 1652 17.5 curately controlled. The platinum standard (i.e.,
Alumel 1652 4.8 the wire previously compared with the standard
Iron 1112 6.4 P t 27) is welded to the wire being tested to form a
Constantan 1112 26.0 couple and the emf measured a t one or more temper-
Constantan 212 20.8 atures by any of the methods described for calibrat-
Copper 5.2 2 12 ing platinum-rhodium thermocouples. The wires
should be carefully insulated and protected. Meas-
* Complete tables giving the average thermal emf of urements a t two temperatures, about 1112 and 21929,
platinum-10 percent rhodium, and platinum-13 percent are sufficient to develop the emf at any temperature
rhodium against platinum are given in NBS Ciicular 561. as the emf is small and practically proportional to
The average thermal emf of chromel and of alumel
against platinum are given in NBS Research Paper 767, the temperature.
copper and constantan against platinum in NBS Research
Paper RP 1080, of iron and constantan against platinum 130 In many laboratories the platinum standard
in NBS Research Paper 2415. and the platinum element of the couple used to
measure the temperature are one and the same. The
perature of the junctions. The necessity for this sample or wire being tested is then welded to the
accurate measurement of temperature, however, is junction of the couple and the emf of the couple and
avoided when the measurements are made by using that between the two platinum wires are measured
a working standard of material similar to that being simultaneously with two potentiometers or alter-
tested, since in this case the emf developed is small nately with one instrument. Simultaneous readings
and changes very little even for large changes in of these electromotive forces should notbe madë
temperature. In the latter method, the accurate with a millivoltmeter or with a current flowing in
measurement of temperature is not entirely avoided, either circuit, because one wire is common to both
but merely shifted to the laboratory that determines circuits and in this case the potential difference
the thermal emf of the working standards against the measured by one instrument is influenced by the
standard P t 27. current flowing in the other circuit, .However, this
objection is not encountered in the method described
127 The small thermal emf of a platinum working above in which the platinum standard is not the
standard against the standard P t 27 a t any tempera- same wire as the platinum of the thermocouple.
ture can be determined as accurately as the emf can
131 Platinum-Rhodium Alloy. The testing of
be measured. These standards are subject to change
platinum-rhodium thermocouple wire directly against
during u s e but, if properly used and occasionally
checked, can be relied upon to about 2 pV a t 1832°F. platinurn is exactly the same a s th-e calibration of
The thermal emf of working standards of other mate- platinum-rhodium thermocouples. .Platinum against
platinum-10 percent rhodium develops 6.5 pV/deg F
rials is determined and certified a t the National
and platinum against platinum-13 percent rhodium
Bureau of Standards to the equivalent of t 2 deg F a t
develops about 7.3 pV/deg F a t 1832OF. Therefore, in
high temperatures. order to determine the thermal emf of a sample of
128 In any event the testing of a thermocouple platinum-rhodium against platinum to 220 pV, i t is
material is essentially the determination of the emf necessary to measure the temperature to k2.7 deg
of a thermocouple in which the material being tested F. Such an accuracy in temperature measurements
is one element and a working standard the other. i s obtained only with a very homogeneous and ac-
121
curately calibrated couple in a uniformly heated into a hole drilled in the junction formed by
furnace, but if the emfof one sample of wire is welding the material to platinum, This brings
known with this accuracy, it may be used to deter- the junctions to the same temperature.
mine the emfof other samples without the necessity In the use of platinum or platinum-rhodium for
of accurately measuring the temperature. For ex- testing thermocouple materials, the wires are
ample, the thermal emf per degree of any sample of used a large number of times before checking
platinum-10 percent rhodium against any other or scrapping. Base-metal thermocouple wires
sample rarely exceeds 0.03 pV/deg F (50pV a t used for testing similar materials should not be
1832T':). Therefore, if the thermal emf of one used more than once if the highest accuracy i s
sample against platinum is known to +2OpV a t required, because there is a slight change in
1832OF, the emf of other samples against the same these materials when heated to a high tempera-
platinum can be determined to about the same ac- ture and if they are used repeatedly, the wires
curacy by comparing the samples of platinum- become heterogeneous. The procedure then is
rhodium and measuring the temperature of the hot to select a coil of wire and test if for homoge-
junction to 18 or 36 deg F. The same applies for neity by taking several samples from different
platinum-13 percent rhodium. parts of the coil, welding them all together, and
measuring the emf between the various samples.
132 The working standard used to determine the
If the coil is sufficiently homogeneous as found
thermal emfof the platinum-rhodium may be a sample
from such tests, one or more samples may be
of platinum, of platinum-rhodium, or either element
taken from it and the thermal emf determined as
of the -thermocouple used in measuring the tempera-
accurately as necessary by. comparison with a
ture. Platinum-10 percent rhodium against platinum-
standard, the emf of which is known9 against
13 percent rhodium develops about 0.89 pV/deg F
the standard P t 27. The average value for the
a t 1 8 3 2 V so that if the thermal emf of one of these
thermal emf of the few selected samples from
materials against platinum is knolvn to +20 pV a t
the coil against the standard P t 27 will apply
1832%', the thermal emfof the other against the
for the remainder of the coil with sufficient
same platinum can be determined to k30 pV by com-
accuracy for most purposes, Any sample from
paring the two and measuring the temperature to
this coil may then be used a s a working stan-
211 deg F.
dard for testing similar materials. The accuracy
133 A number of wires can be welded together with which the temperature must be measured
and tested by any of these methods. depends upon the difference between the
standard and the material being tested. In case
134 Rase-MetalThermocoupleMaterials of some materials that have been well stan-
dardized, the differences are small enough that
(u) At High Temperatures. In testing base-metal ar, accuracy of 90 deg F is sufficient. Seldom,
thermocouple materials (alumel, chromel, con- if ever, should it be necessary to measure the
stantan, copper, and iron) the procedure i s very
temperature closer than 18 deg Fe
much the same a s in calibrating base-metal
thermocouples. Although such thermal-emf (b) At Low Temperatures. Annealedelectrolytic
measurements are ultimately referred to platinum, copper is very uniform in its thermoelectric
it is not necessary to measure each sample di- properties and is often used as a standard for
rectly against platinum. When the measurements thermoelectric testing at temperatures below
are made against platinum (and this must fre- 572T'. The thermal emf of other materials
quently be done), the platinum wire should be against either copper or platinum may be de-
sealed through the end of a glazed porcelain termined very accurately by using a stirred
protecting tube with pyrex glass, leaving about liquid bath or fixed points. The steam point is
1 cm of the wire exposed for welding to the an excellent one for this purpose.
base-metal wire or wires. The largest uncer- Table 9.12 specifies the thermal emfof an-
tainty in the measurements arises from the un- nealed electrolytic copper against National
certainty in the determination of the tempera- Bureau of Standards standard P t 27 and may be
ture of the junction. The junction of a standard used to convert values of the thermal emf of any
platinum-rhodium thermocouple may be inserted material against one of these standard materials
122
"
I
-
-
Electromotive Force
Platinum
Chromel Alumel Constanton Iron Copper
Versus
Versus Versus Versus Versus Versus
Platinum-
Platinum Platinum Platinum Platinum Platinum
Rhodium"
I I I I
mV mV mV mV
I ! I I
v,
W
-I
a
3
O
V
O
z
W
I
I-
(3
zI-
U
Co
zlFI d
ò ò
3
W
Cr Fr
M brJ
a 5
O O
I-
m O
U o,
O h
2 al
> al
Y
Y
s
O
Øa Øa
4 4
n
a
v,
O
H
m
O
d
m
n I I
O O O
I Y Y
l- 0.1 0.1
u1 m m
I
a a,:
m.
a
e
124
m
x
h
Y
2
)I
O
CI
fi
.e
O
a
M
fi
'Pl
d
.d
O
m
"
El
O
L
El
c*
u1
CI
.rlfi
O
a
a
x
tz
125
.. c
126
peratures may then be calculated. In the case of because the heating of the bulb, which is necessary
organic liquid-filled thermometers the coefficient of to drive the liquid into the expansion chamber, may
expansion of the liquid should be obtained by ex- overheat the glass and either break the bulb as a
periment or from the manufacturer in order to perform result of the pressure of the gas or destroy the ac-
these computations. This method has the advan- curacy of the thermometer by expanding the bulb.
tage that the standard may be selected to have Thermometers that have a contraction chamber be-
greater sensitivity than the thermometer being test- low the lowest graduation are likely to develop
ed, thus increasing the accuracy of measurement. separations either in the chamber or above it. I t is
A second method, which is the one best suited to frequently possible to join such separations by
large-quantity testing, involves comparison of the cooling the thermometer s o that the separated por-
thermometers with standards similar in all details tion as well a s t h e main column both stand in the
of constructions above the immersion point, but chamber. Tapping the tube against the hand, or
differing below the immersion point to the extent of tapping the bulb on a pad of paper will usually
including an auxiliary ice point scale. cause the separated portions of the liquid to join,
For separations which are more difficult to join, it
VISUAL INSPECTION may be necessary to cool the bulb in dry ice to a
point low enough to bring all of the liquid into the
146 General. Thermometersshouldbeinspected
bulb itself. By softly tapping on a pad or against
for gas bubbles in the bulb or liquid column, glob-
u l e s of liquid in the stem, foreign matter, and glass the hand it usually is possible to bring the liquid
faults. together in the bulb. .Then the bulb should be al-
lowed to warm up slowly. . The liquid should emerge
147 Gas Bubbles. Gasbubblesarereadilyde- into the bore with no separation.
tected and are more likely to occur in shipment than
during service. No method h a s been discovered that 149 In organic-liquid-filledthermometersdistilla-
will entirely prevent such displacement of the gas. tion may occur, with subsequent condensation of the
If bubbles are observed in the bulb, they can gener- colorless parent liquid in the upper part of the ther-
ally be removed by cooling the bulb with dry ice or mometer. -Such thermometers should always be in-
other convenient coolant until all the liquid is drawn spected for these separations which can be repaired
into the bulb. Gentle tapping of the thermometer by the procedures described above. If the thermom-
while held upright will cause the bubbles to rise to eter has an expansion chamber thatis observed to
the surface, It is very important that, if the bulb is be filled with liquid, the column can be reunited by
cooled in this process below the freezing point of very careful heating of the chamber to drive the
the liquid, care should be exercised to warm the liquid into the bore whence it can be rejoined to the
stem sufficiently during the melting process s o that main body as described above,
no solidification occurs in the stem; otherwise the
bulb may burst or the capillary split internally be-
150 Organic liquids a s u s e d in thermometers, in
cause of the expansion forces generated in the bulb.
contrast to mercury, wet the glass, -Sufficient time
148 If gas bubbles are observed in the stem, should always be allowed for drainage to occur,
several different ways are suggested for joining the particularly when using or standardizing such ther-
columns, depending on the construction of the ther- mometers below 32%. I t i s frequently a good prac-
mometer and the type of separation. .If a small por- tice to immerse only the bulb of the thermometer
tion of the liquid has separated at the top of the until drainage has taken place, because drainage is
column and the thermometer is provided with an ex- appreciably retarded if the capillary is also cooled.
pansion chamber, the liquid usually can be joined
by carefully and slowly heating the bulb until the 151 Globules of Liquid in the stem which result
separated portion is driven into the expansion from mechanical separation can normally be rejoined
chamber. -When the column itself flows into the by heating the bulb until the liquid column merges
chamber, the separated portion usually will join with the globules. .If such globules appear to unite
with the main column. . A slight tapping of the ther- and then reappear upon cooling the bulb, this is in-
mometer against the palm of the hand will facilitate dicative of oxidation of the mercury, or the presence
this joining. This method should not be employed of obstructions in the bore, and therefore, the ther-
for high-temperature thermometers (above 500%), mometer should be rejected.
152 ForeignMatter in the bore can sometimes be pearance of the tested and untested sections of
detected with the unaided eye, but it is desirable the scale portion. Burning out, loosening, or
and generally convenient to use a magnifying glass chalking of the pigment shall be cause fpr re-
of low power for this examination. The most com- jection.
mon types of foreign matter which should be cause
for rejection are glass chips, particles of dirt or T e s t for Permanency of Range
lint, oxide of mercury either red, yellow, or black,
products of glass weathering commonly called white 156 The test for permanency of range is designed
to determine the adequacy of the stabilizing heat
deposit, and stones or iron spots traceable to faulty
treatment accorded the thermometer bulb during manu-
glass fabrication.
facture. .An inadequately stabilized bulb will under-
153 Where a specific gas is specified to be used go shrinkage with time which may be significant,
as filling above the liquid, any other gas present particularly at higher temperatures.
may be treated as foreign matter. .The most common
example is in the use of air instead of nitrogen in 157 Determine the ice point of the thermometer to
mercury-in-glass thermometers, which may have be tested and place it in a permanency of range com-
been introduced by accident or in violation of the parator of the type illustrated in Fig. 9.2. Heat for
specifications. The presence of air can readily be 24 hr with the thermometer bulb a t l e a s t a tthe tem-
detected by heating the gas filled portion of the perature specified, or, in the absence of specifica-
thermometer and a short section of the liquid column tions, to not over 7OO0F for normal glass bulbs or
to a temperature of 650 to 700%’, using the same 900%’ for borosilicate bulbs. Allow it to cool slow-
type of equipment a s illustrated in Fig. 9.1. Under ly, and after 72 hr again determine the ice point.
these conditions oxidation of the mercury will occur The difference in ice point readings is a measure of
and will normally be evidenced by the production of quality of the heat treatment accorded the thermom-
crystals of red oxide of mercury after being heated eter in manufacture. If upon subsequent standard-
10 to 12 hr. ization the thermometer is within the prescribed
tolerances, the thermometer shall be considered to
154 GlassFaults may be of various types, Any
have passed the test for permanencyof range.
stones or striae that distort the bore or its appear-
ance is sufficient cause for rejection. Strains in
Calibration
the glass as observedwith a polarized light strain
gage near enlargements in the stem or bore, or at 158 General Considerations. Liquid-in-glass
the top of the thermometer, are detrimental if S O thermometers may be calibrated by observing read-
severe that fire cracks may later occur. Strains ings at fixed points, such a s the ice point, and by
near the bulb are indicative of incomplete glass comparison,with secondary sta,ndard thermometers
stabilization and are particularly objectionable in selected in accord with the criteria discussed in
thermometers for use above 320v. The test for Pars. 35 to 38, inclusive.
permanency of range will normally serve to reject
high range thermometers in which this defect is (a) Reading Thermometers to Avoid Parallax.
most significant. The error due to parallax may be eliminated by
(1) carefully observing the thermometer in a
155 Test for Permanency o f Pigment
manner which will assure that the reflection of
(a) The test for permanency of pigment is designed the scale can be seen in the mercury thread and
to determine the ability of the pigment material (2) by adjusting the line of sight s o that the
used to fill the thermometer graduations and graduation of the scale nearest the meniscus
other markings to withstand the exposure con- exactly hides its own image; the line of sight
ditions encountered in use without being oblit- will then be normal to the stem a t t h a t point.
erated. When reading thermometers, one must realize
( b ) Place any convenient portion of the scale sec- that the scale lines are of appreciable width
tion of the thermometer to be tested in an oven and compensate for this in arriving at the tem-
of the type shown in Fig. 9.1, Heat for 3 hr at perature, The best practice is to consider the
approximately 500%. Allow i t to cool slowly. position of the lines as defined by their middle
Inspect the thermometer for differences in ap- parts.
128
(b) Depth of Immersion for Total Immersion Ther- ings taken a t least 1min apart should agree within
mometers. Although by definition total immer- one-tenth of a division.
sion thermometers should have the entire index
immersed, both in use and in standardization, 161 Alternatively, some of the ice may be heaped
i t i s frequently inconvenient to do s o . If any around the stem above the ice point and a deep nar-
portion of the index is exposed, the emergent row channel formed to permit observations of the
mercury column correction should be deter- meniscus which is thus kept well below the general
mined, No correction need be applied if it is level of the ice. Observations may then be made a s
found to be less than one-fifth of the standard- described above without, however, raising the ther-
ization error; otherwise the proper correction mometer.
should be applied. In some instances total im-
mersion thermometers may be used under con- 162 Record the readings and comparewithpre-
ditions of complete immersion. A significant vious readings. - If the readings are found to be
error may be introduced unless the thermometer higher or lower than the reading corresponding to a
is standardized under the same conditions a s previous standardization, readings at all other tem-
those in use. This is due to the effect of in- peratures will be correspondingly increased or de-
creased. If the method involving the taking of the
creased gas pressure above the column produc-
ice point immediately after heating to a specified
ing a distortion of the bulb.
( c ) Dept of Immersion for Partial Immersion Ther- temperature is used, a note should appear in the
mometers. Thermometers of this type shall be table of corrections for the thermometer, a s follows:
standardized by one of the methods described N0TE:The tabulated corrections apply for the condition
in Pars. 38 ( b ) and 38 ( c ) inclusive, using the of immersion indicated, provided the i c e point reading
appropriate comparators. If the thermometer is taken after heating to ..... ...,.
for not l e s s than 3 min i s
of the general purpose type, it should be stand-
ardized by immersion to the specified depth and 163 If the ice point reading (taken in not less
the readings compared with those of the second- than 2 min and not more than 5 min after removal of
ary standard. the thermometer from the heated bath) is found to be
higher (or lower) than stated, all other readings will
159 CalibrationatIcePoint. Select clear pieces be higher (or lower) to the same extent.
of ice, preferabIy made from relatively pure water.
Discard any cloudy or unsound portions, Rinse the 164 CalibrationatOtherFixedPoints. If the
ice with distilled water and shave or crush into range of the thermometer is such that the ice point
small pieces, avoiding direct contact with the hands is not included, but one of the other fixed points is
or any chemically unclean objects. Fill the Dewar included, i t may be used to advantage. For descrip-
vessel with the crushed ice and add sufficient dis- tion of the equipment and techniques best suited to
tilled and preferably precooled water to form a the purpose, reference may be made to the publica-
slush, but not enough to float the ice. Insert the tions of the National Bureau of Standards and the
thermometer, packing the ice gently about the stem, Massachusetts Institute of Technology. The steam
to a depth sufficient to cover the 32OF graduation. point, the triple point of benzoic acid, and the sul-
A s the ice melts, drain off some of the water and fur point are particularly worthy of mention.
add more crushed ice. - F o r thermometers graduated
in fractional degrees the thermometer should be 165 CalibrationatTemperaturesOtherThan
held a t room temperature for at l e a s t 72 hr, unless Fixed Points. Determine the ice point of- t h e sec-
it is specified that the ice point be taken immedi- ondary standard and the thermometer to be tested a s
ately after heating to a specified temperature. For described in Par. 127. If the secondary standard is
thermometers graduated in single degreesor large of the liquid-in-glass type, it should be held at
subdivisions, this waiting period may be omitted, room temperature for at least 72 hr before this de-
termination, unless the ice point was originally de-
160 Raise the thermometer a few millimeters termined immediately after heating to a specified
after at least 3 min have elapsed, tap the stem temperature. T h e thermometer to be tested should
gently, and observe the reading. Successive read- be treated in similar manner.
130
nace, the other end of the furnace being left open s o optical pyrometers compared are of the same type,
that the end of the couple is in sight. employing the same band of the spectrum in each,
the readings of both will depart from the true tem-
186 After heating the furnace to the desired tem-
perature of the filament by the same amount. The
perature, i t should be held as nearly constant as
,standard pyrometer having been originally calibrated
possible, and with the optical pyrometer sighted
to read black-body temperatures, the pyrometer un-
upon the end of the thermocouple tube, readings
der check will be calibrated in the same terms.
should be taken on each a s nearly simultaneously
a s possible. Continue to take readings until equal- 190 A s the tungsten filament pyrometer lamps
ity or a constant difference between the standard are never intentionally operated at filament temper-
thermocouple and the optical pyrometer is estab- atures higher than 2 7 0 0 9 , they undergo practically
lished. no deterioration due to evaporation of the tungsten
187 When, a s is usual, a range shifting device is with resultant increase of filament resistance and
a part of the optical pyrometer there will, in general, blackening of the bulb. Lamps which have ha¿
be a considerable overlap between the two or several years of service in industrial plants have shown no
ranges. .In such a case, it is well to check a t a appreciable change in calibration. 'Lamp life is
usually limited only by mechanical defects in the
point lying within the overlap, so that the reading
base or by breakage. Occasional checking is de-
may be applied to each scale and a check secured
upon the accuracy of the range shifting medium. sirable, however, to eliminate errors due to dust or
metal fragments on lenses, screens, and bulbs,
188 For checking at points beyond the range of a
platinum-platinum-rhodium thermocouple, the same
procedure a s outlined in the foregoing may be fol- BIMETALLIC THERMOMETERS
lowed, substituting a standard optical pyrometer and 191 Bimetallicthermometers are usually Cali-
a fire clay target for the standard thermocouple. The brated by comparing their readings with those of
fire clay target is mounted in the furnace at the primary or secondary standard liquid-in-glass thep
place normally occupied by the thermocouple and mometers, usjng variable temperature comparators
the standard and unknown alternately sighted into described in Par. 46 ( b ) and its subparagraphs.
the same end of the furnace and read under condi-
tions as described above. 192 Calibration at fixed points is generally not
necessary because the accuracies involved are in
189 A ribbon filament tungsten Iamp is a very the range of 0.5 to 10 deg F.
convenient means of checking optical pyrometer
readings up to 4172%'(23OO0C) i n laboratories or 193 If a well is provided with the thermometer, it
plants, where it is possible to maintain an optical should not be used during calibration.
pyrometer calibrated by the National Bureau of
Standards or other qualified laboratory. The fila- 194 Before calibrating a bimetallic thermometer
ment should be at least 0.050 in. wide, and long it should be checked for proper operation by sub-
enough to insure a uniform temperature over i t s jecting it to variations of temperature within its
central portion. A shallow notch in the filament range and observing that the pointer moves freely
marks the point at which settings are to be made. without sudden movements from point-to-point,
A flat window in the side of the bulb is desirable
195 The bulb should be immersed at least 2 in.
for viewing the filament, but i t is not absolutely
in the liquid bath of the comparators. A check
indispensable. While focusing the standard pyrom-
should be made to determine that the immersion is
eter on the flat filament, the current through the
sufficient by increasing it to determine'if the read-
latter is adjusted to develop the desired pyrometer
ing changes.
reading. The pyrometer to be checked is now sub-
stituted for the standard instrument and a reading 196 T h e errors of the instrument should be deter-
made with it, the current in the flat filament being mined at a minimum of five points on i t s s c a l e while
held constant at the value to which it was adjusted the temperature of the bath(s) is varied in.ascending
during the standard pyrometer reading. While the and descending order, successively. Sufficient time
tungsten filament is not a black body, if the two should be allowed at each check point to achieve
132
133
APPENDIX
Conversion Factors
.Fahrenheit
degree
Celsius
degree ("C) LC ( t -32)/1.8
~
degcee Fahrenheit
kelvin (KI t K = ($ t 459.67)/1.8
Fahrenheit
degree
Rankine (R) t ~ ( =t p t 459.67)
degree Celsius kelvin (K) t~ = t c t 273.15
foot/second
metre/second (m/s) 3.048 OOO* E -O1
metre foot (m) 3.048 OOO* E -01
inch metre (m) 2.540 000" E -02
pound-force/inch2 (psi) pascal (Pa) 6.894 757 E t 0 3
pound-mass/foot3
kilogram/metre3 (kg/rna) 1.601 846 E t o 1
Btu in./s-ftz
deg F watt/rnetre-kelvin (W/'m*K) 5.192 204 E t o 2
Btu/h*ft2
deg F watt/metrez-kelvin (w/m2.K> 5.678
263 E +O0
metre2 foot2 (m.) Ç . B O 304* E -02
watt Btdhour (W) 2.930 711 E -01
*Relationships that are exact in terms of the base units are followed by an asterisk.
The factors are written as a number greater than one and less than ten with six or less
decimal places, The number is followed by the letter E (for exponent), a plus or minus
symbol, and two digits which indicate to power of 10 by which the number must be
multiplied to obtain the correct value.
134
PTC
PTC
19.12,
19.13,
Measurement of Time, , . . .. .. .. .. .. .. .. . . . .. .. . . .. ..
Quality,beakDetectionendMeasurement).
e ,o
, ,, (1970)
(1958)
(196 1)
PTC
PTC
19.14,
19.16,
LinearMeasurements. , . . . . . . . . . . . . .. .. . ..
Density Determinations of Solids and Liquids. ,,,,
( 1958)
(1965)
PTC
PTC
19.17,
19.90, Smdke-Density Determinations. , . .. .. .. .. . .. .. .. .. . ..
Determination of Viscosity of Liquids.
, ,
,
,
(1965)
(1971)