Resistance Thermometry:

Principles and Applications of Resistance Thermometers and Thermistors

 Table of Contents

Abstract ......................................................................................................................................................................................................Page 3

Resistance Thermometers...................................................................................................................................................................Page 3
Resistance/Temperature Characteristics ................................................................................................................................Page 3
Platinum .......................................................................................................................................................................................Page 3
Copper ..........................................................................................................................................................................................Page 5
Nickel-Iron ...................................................................................................................................................................................Page 5
Nickel.............................................................................................................................................................................................Page 6
DIN Nickel ....................................................................................................................................................................................Page 6
Temperature Coefficient of Resistance ...................................................................................................................................Page 7
Comparison of Element Types ...................................................................................................................................................Page 8
Effects of Leadwire Resistance and Bridge Design .............................................................................................................Page 9
Two-wire Temperature Transmitters .................................................................................................................................... Page 11
Potential Sources of Error with Resistance Thermometers .......................................................................................... Page 12
Resistance Thermometer Types ............................................................................................................................................. Page 15
Where to Use Resistance Thermometers ............................................................................................................................ Page 18

Thermistors ............................................................................................................................................................................................ Page 19

Resistance and Temperature Characteristics of Thermistors ...................................................................................... Page 20
Thermistor Output Circuits ....................................................................................................................................................... Page 20
Potential Sources of Error with Thermistors ...................................................................................................................... Page 21
Where to Use Thermistors......................................................................................................................................................... Page 21
Thermistor Types .......................................................................................................................................................................... Page 22

Summary ................................................................................................................................................................................................. Page 23

Copyright 2011, Minco Page 2

 Abstract

Understanding the principles of resistance thermometry as they apply to resistance thermometers and
thermistors will help you achieve consistent and accurate readings from your temperature sensing instruments.
A resistance thermometer consists of a metallic element whose resistance increases with temperature. Their
designs range from helical-wound thermometers for laboratory use to industrial thermometers that consist of
several designs which allow it to conform closely to sensed surfaces. Two key benefits of the resistance
thermometer include accuracy and stability. A thermistor consists of a semiconductor material whose resistance
decreases as temperature increases. Key benefits of the thermistor are high resolution measurements over
limited ranges and low cost. This white paper describes resistance thermometers and thermistors in detail, and
will help you accurately calibrate your temperature reading instrument by performing calculations regarding
resistance/temperature characteristics and the temperature coefficient of resistance. Details are provided
regarding factors that can influence the temperature/resistance ratio such as element types, leadwire resistance,
electrical noise, vibration, self-heating, and exposure to temperatures at or beyond the endpoint of a specified
range.

Resistance Thermometers

Resistance thermometers may be called RTDs (resistance temperature detectors), PRTs (platinum resistance
thermometers), or SPRTs (standard platinum resistance thermometers). These thermometers operate on the
principle that electrical resistance changes in pure metal elements, relative to temperature.
The traditional sensing element of a resistance thermometer consists of a coil of small diameter wire wound to a
precise resistance value. The most common material is platinum, although nickel, copper, and nickel-iron alloys
compete with platinum in many applications.
A relatively recent alternative to the wire-wound RTD substitutes a thin film of platinum, which is deposited on a
ceramic substrate and trimmed to the desired resistance. Thin film elements attain high resistances with less
metal, thereby lowering cost.

Resistance/Temperature Characteristics

Resistance thermometers exhibit the most linear signal with respect to temperature of any sensing device. Small
deviations from straight line response, however, dictate the use of interpolating polynomials to calculate
resistance values between fixed temperature points.

Platinum

The resistance/temperature characteristic for standard platinum resistance thermometers, as defined by the ITS–
90 (International Temperature Scale of 1990), is a complex set of equations beyond the scope of this document.
A good reference for understanding the ITS–90 is NIST Technical Note 1265: Guidelines for Realizing the ITS–90, B.
W. Mangum and G. T. Furukawa, U. S. Department of Commerce, 1990.

Copyright 2011, Minco Page 3

Platinum generally follows the modified Callendar-Van Dusen equation over the range
–200 to 850°C (–328 to 1562°F):

t(°C) Equation

-200 to 0 [
R t = R 0 1 + At + Bt 2 + Ct 3 (t − 100 ) ]
0 to 850 Rt = R0 [1 + At + Bt ]
2

Rt = Resistance at temperature t (in °C),

R 0 = Base resistance at 0°C, and
A, B, and C are constants that describe a given thermometer.

The A, B, and C coefficients, which can be individually determined for high precision laboratory thermometers,
are entered into the Callendar-Van Dusen equation to generate resistance vs temperature tables.

Nominal values are:

TCR* = 0.003926 Ω/Ω/°C
A: 3.9848×10-3 B: -5.870×10-7 C: -4.0000×10-12
TCR = 0.003911 Ω/Ω/°C
A: 3.9692×10-3 B: -5.8495×10-7 C: -4.2325×10-12
TCR = 0.003850 Ω/Ω/°C
A: 3.9083×10-3 B: -5.775×10-7 C: -4.183×10-12
*Temperature Coefficient of Resistance

To determine temperature from a measured resistance for t above 0°C:

A 2 − 4 B 1 − t 
R
−A +
 R0 
t =
2B

For temperatures below 0°C, the equation is too complex to solve, so successive approximation is employed:
Rt
−1
R0
t1 = (initial approximation)
A + 100B

1 + At n + Bt n2 + Ct n3 (t n − 100 ) −
Rt
R0 This equation typically converges to
tn + 1 = tn −
A + 2Bt n − 300Ct 2
n + 4Ct 3
n
sufficient accuracy within 2 iterations.

Copyright 2011, Minco Page 4

Copper

The corresponding equations for copper elements are:

t(°C) Equation

-200 to -50 [
R t = R1 R 0 1 + A1 (t + 200 ) + B1 (t + 200 )2 ]
-50 to 150 R t = R 0 [1+ A2 t ]

150 to 260 [
R t = R 3 R 0 1 + A3 (t − 150 ) + B 3 (t − 150 )2 ]
Rt = Resistance at temperature t (in °C),
R 0 = Base resistance at 0°C, and
A, B, R1 and R3 are constants that describe a given thermometer.

Nominal values are:

TCR (Ω/Ω/°C) = 0.00427
R1: 1.17058×10-1 A1: 3.92313×10-2 B1: -7.45044×10-6
A2: 4.2743×10-3
R3: 1.641145 A3: 2.62628×10-3 B3: 2.43732×10-8

Note that copper resistance is linear with temperature from -50 to 150°C, and nearly linear over other ranges.

Nickel-iron

The equations for nickel-iron are:

t(°C) Equation

-200 to 0 [
R t = R 0 1 + A1t + B1t 2 + C 1t 3 ]
0 to 204 Rt = R0 [1 + A t + B t
2 2
2
+ C 2t 3 ]
Rt = Resistance at temperature t (in °C),
R0 = Base resistance at 0°C, and
A, B, and C are constants that describe a given thermometer.

Nominal values are:

TCR (Ω/Ω/°C) = 0.00518
A1: 4.68699×10-3 B1: 8.58992×10-6 C1: 0
A2: 4.59818×10-3 B2: 5.89404×10-6 C2: 0
TCR (Ω/Ω/°C) = 0.00527
A1: 4.63189×10-3 B1: 6.96196×10-6 C1: -1.72771×10-8
-3
A2: 4.63189×10 B2: 6.96196×10-6 C2: -5.71203×10-9

Copyright 2011, Minco Page 5

Nickel

Individual manufacturers have developed proprietary curves for nickel elements, working from direct
measurements, but there is no simple formula for direct calculation of resistance. Figure 1 shows the nonlinear
behavior of nickel in comparison to other elements.

This stepwise equation will closely

approximate a standard nickel curve:
R t = R 0 [ A + Bt + Ct 2 + Dt 3 ]
Rt = Resistance at temperature t (in °C),
R 0 = Base resistance at 0°C, and
A, B, C, and D are constants that describe a
given thermometer.

Figure 1: Linearity of metals used in resistance thermometers.

Nominal values are:

TCR (Ω/Ω/°C) = 0.00672
t(°C) A B C D
-1 -3 -6
-80 to -60 9.980384367×10 5.779005438×10 4.519218356×10 1.883007648×10-8
-60 to -30 9.995545058×10-1 5.854808892×10-3 5.782609262×10-6 2.584891485×10-8
-30 to 0 1.0 5.899358312×10-3 7.267589932×10-6 4.234870007×10-8
0 to 30 1.0 5.899358312×10-3 7.267589932×10-6 1.154640832×10-8
30 to 60 1.000118847 5.887473643×10-3 7.663745572×10-6 7.144678985×10-9
60 to 90 1.002329124 5.776959768×10-3 9.505643490×10-6 -3.088087226×10-9
90 to 120 9.940315172×10-1 6.053466667×10-3 6.432455728×10-6 8.294089672×10-9
120 to 150 1.007022904 5.728761999×10-3 9.138994624×10-6 7.759260700×10-10
150 to 180 8.918592090×10-1 8.032035898×10-3 -6.216164699×10-6 3.489850234×10-8
180 to 210 9.060247382×10-1 7.795943744×10-3 -4.904541625×10-6 3.246957072×10-8
210 to 240 1.103473241 4.975250849×10-3 8.527329303×10-6 1.114941068×10-8
240 to 260 1.437355995 8.017164189×10-4 2.591705610×10-5 -1.300325764×10-8

DIN Nickel

The equation for DIN nickel is:

[
R t = R 0 1 + At + Bt 2 + Dt 4 + Ft 6 ]
Rt = Resistance at temperature t (in °C),
R 0 = Base resistance at 0°C, and
A, B, D, and F are constants that describe a given thermometer.

Nominal values are:

A: 5.485 × 10-3 B: 6.65 × 10-6 D: 2.805 × 10-11 F: -2 × 10-17

Copyright 2011, Minco Page 6

Until recently, equations relating resistance to temperature were of interest mainly to researchers and designers
of linearizing analog circuits. Now many common instruments such as readouts, data loggers, and controllers
contain microprocessors. Interest in direct digital computation of temperature has increased accordingly. A few
considerations for the designer or programmer of digital instruments are:
1. Converting resistance to temperature requires finding the root of a third or fourth order polynomial. Instead,
many manufacturers use a lookup table, with resistance/temperature values stored in ROM (read only
memory). Choosing values sufficiently close together, on a 5°C interval for example, with linear interpolation
between will usually produce acceptable readings.
2. The use of resistance ratios, instead of absolute values, is recommended for measurements of the highest
accuracy. A ratio table sets the R0°C value to 1, and uses A, B, and C coefficients to calculate ratios from this
value. The absolute resistance at any given temperature is the product of R0 and the ratio. Resistance ratios
are preferred for their stability. Simple periodic checks of ice point resistance, with most recent R0 used to
calculate absolute resistances, will reduce or eliminate the need for complete recalibration and table
generation.

Temperature Coefficient of Resistance

Temperature Coefficient of Resistance (TCR) has many (R100°C − R 0° C )

R 0°C
definitions. For resistance thermometers, TCR is normally TCR (Ω / Ω / °C ) =
defined as the average resistance change per °C over the (100°C − 0°C )
range 0 to 100°C, divided by R0°C:
TCR values for the common elements are:
(12.897 − 9.035)
Copper: 9.035 = 0.00427 Ω / Ω / °C
100

(200.64 − 120 )
Nickel: 120 = 0.00672 Ω / Ω / °C
100

(917.33 − 604 ) (1387.21 − 908.4)

Nickel-iron: 604 = 0.00518 Ω / Ω / °C or 908.4 = 0.00527 Ω / Ω / °C
100 100

(138.50 − 100 )
Platinum: 100 = 0.003850 Ω / Ω / °C
100

(139.11 − 100 )
or 100 = 0.003911 Ω / Ω / °C
100

(139.26 − 100 )
or 100 = 0.003926 Ω / Ω / °C
100
In one sense, TCR expresses the sensitivity of the resistive wire used in the element, as it defines the average
temperature change of a hypothetical 1 Ω thermometer. But end-users normally use TCR to distinguish between
different resistance/temperature curves of the same element material, such as the three curves platinum.

Copyright 2011, Minco Page 7

Because all of these curves see widespread use, platinum TCRs must be properly specified to maintain
compatibility between thermometers and instruments.
There are four primary curves specified for platinum:
1. 0.003926 Ω/Ω/°C: Standard platinum resistance thermometers are the only PRTs that can achieve this TCR.
They must have high purity platinum wire (99.999% or better) wound in a strain-free configuration. The
stresses introduced in manufacturing lower the TCR of ordinary industrial models. Several manufacturers
offer industrial platinum thermometers with nominal TCR of 0.00392; TCRs around 0.003923 are achieved
regularly.
2. 0.003911 Ω/Ω/°C: This TCR is sometimes called the “U.S. Industrial Standard.” It is lower than laboratory
standards as the typical construction of high temperature ceramic elements impose strain on platinum wire.
3. 0.00385 Ω/Ω/°C: This is mandated by EN60751, ASTM E1137, and other national and international
specifications.
4. 0.00375 Ω/Ω/°C: Several manufacturers now offer thin-film 1000 Ω elements with 0.00375 TCR, intended for
low-cost applications.
There are few inherent advantages in specifying any particular TCR over another. Laboratory systems
traditionally use reference standards with the highest grade platinum, but industrial end-users may aim instead
for the greatest degree of standardization. In this case, 0.00385 TCR will be compatible with the greatest number
of manufacturers.

Comparison of Element Types

Platinum, with its wide temperature range and stability, has become the preferred element material for
resistance thermometers. Furthermore, advances in element construction have narrowed the price difference
between platinum and base metal thermometers. Nevertheless, nickel, copper, and nickel-iron do have benefits
for many applications and should be considered. The primary advantages of the four element types are
compared in Table 1.

Copyright 2011, Minco Page 8

 Resistivity Benefits Base resistance TCR Sensitivity
Element Temperature (Ω
Ω / circular (Ω
Ω/Ω
Ω/°C) (avg. Ω/°C,
type range mil foot 0 to 100°C)
at 20°C)
-259 to 1235°C
100 Ω• at 0°C 0.003926 0.392
(-434 to 2255°F)
-259 to 630°C
• Greatest range 100 Ω• at 0°C 0.00391 0.391
(-434 to 1166°F)
Platinum 63.8 • Best stability
-200 to 850°C • Good linearity 100 •Ω at 0°C 0.00385 0.385
(-328 to 1562°F)
-200 to 850°C
1000 •Ω at 0°C 0.00385 3.85
(-328 to 1562°F)
Copper -100 to 260°C 10.7 • Best linearity 10 Ω• at 25°C 0.00427 0.039
• Low cost
Nickel -100 to 260°C 41.5 120 Ω• at 0°C 0.00672 0.806
• Best sensitivity
604 Ω• at 0°C 0.00518 3.133
• Low cost
Nickel-iron -100 to 204°C 120.0 1000 Ω •at 70°F 0.00527 4.788
• Highest sensitivity
2000 Ω •at 70°F 0.00527 9.576

Table 1: Comparison of resistance thermometer element types

Effects of Leadwire Resistance and Bridge Design

Because an RTD is a resistance type sensor, any resistance in the extension wires between the RTD and control
instrument will add to readings. In some cases, one can compensate for this extra resistance with adjustments at
the instrument. However, this only compensates when the leads are at a constant temperature since variations
in ambient temperature alter copper leadwire resistance.
Table 2 shows resistance values of common copper leadwire sizes. To approximate error in an uncompensated
system, multiply the total length (in feet) of extension leads by the appropriate value in the table. Then divide by
the sensitivity of the RTD element from Table 1 to obtain an error figure in °C. For example, assume a 100 Ω
platinum element with 0.00385 TCR and 22 AWG leads, 100 feet long:
Total resistance = 200 ft × 0.0165 Ω/ft = 3.3 Ω
Approx. error = 3.3 Ω/(0.385 Ω/°C) = 8.6°C

Leadwire AWG Ohms/foot at 25°C Leadwire AWG Ohms/foot at 25°C

12 0.0016 22 0.0165
14 0.0026 24 0.0262
16 0.0041 26 0.0418
18 0.0065 28 0.0666
20 0.0103 30 0.1058

Table 2: Resistance of common copper leadwire sizes

Copyright 2011, Minco Page 9

Leadwire error can be significant, especially with small diameter leads or low sensitivity elements. Fortunately,
the use of a 3-lead system will reduce errors to a negligible level in most applications.
Figure 2 shows a 2-lead RTD connected to a typical
Wheatstone bridge circuit. The bridge, a resistive
network, translates the RTD’s resistance into an
electrical signal used by the monitoring or controlling
instrument. In this figure, ES is the supply voltage; EO is
the output voltage; R1, R2, and R3 are fixed resistors;
and Rt is the sensing element of resistance
thermometer. L1 and L2 are the resistances of the Figure 2: Uncompensated 2-lead bridge circuit
two leads.
With a balanced bridge, the voltage drops across the two upper arms, (1) R1
R1 R + L1 + L2
and (2) Rt + L1 + L2, are equal and E0 is zero. The fixed resistors R1, R2, and R3 are = t
specified so that the bridge ratios are equal at balanced condition: R2 R3

Maximum bridge sensitivity is realized if:

R1 = R2 = R3 = Rt + L1 + L2
This sensitivity is desirable, however, only in laboratory systems where the bridge is constantly rebalanced.
Typical industrial circuits will have unequal resistors.
The value of Rt at the temperature point of most interest will influence the selection of the bridge arm resistors.
Values should also be chosen to limit bridge currents and consequent self-heating in the thermometer. Bridge
resistors must be stable and insensitive to ambient temperature variations for best accuracy.
In a 2-lead bridge, the leadwire resistance L1 + L2 adds directly to readings. If leads are short enough or sensitivity
high enough, the offset may be acceptable. When long extension runs are required between the sensor and
instrument, or sensitivity is low, the person specifying the sensor should consider a 3-lead system. All resistance
thermometers with copper elements must have three leads to offset their low sensitivity.
Figure 3 shows a 3-lead compensating bridge circuit. L1
and L2 are now in two separate arms of the bridge, so
the measuring currents and voltage drops across them
are identical. The third lead, L3, forms part of the output
circuit and does not affect the bridge ratios or balance.
In fact, no current flows through L3 when the bridge is
in balance.
Figure 3: 3-lead compensating bridge circuit

Proper compensation with the 3-lead system depends on these conditions:

1. Lead resistances L1 and L2 should be equal. Most manufacturers match leads within ±5%. In most cases,
therefore, error with a 3-lead system is less than 5% of the error with a similar 2-wire system.
2. Leadwires should stay bundled together to ensure that ambient temperature changes act equally on all
leads.
3. Electronic circuitry connected to EO should have sufficient input impedance to prevent appreciable current
drain through L3. L3 normally acts only as a potential-carrying lead. Any current through it will cause errors
when the bridge is out of balance.

Copyright 2011, Minco Page 10

3-lead systems represent a practical compromise between accuracy over distance and the cost of extra leads.
Although well suited to most industrial areas, they may be affected by electrical noise and contact resistance at
the junction points.
4-lead circuits provide the same resistance compensation as 3-lead systems, but also relieve problems with
unmatched leads, contact resistances, and thermal EMFs. Thermal EMFs are spurious voltages introduced by the
thermocouple effect where two dissimilar metals make contact. Laboratory systems, used for the highest
precision measurements, are often variants of the Mueller bridge. The Mueller bridge is basically a switched 3-
wire system requiring two readings which are averaged to yield a true reading.
The need to throw switches and calculate the mean of
readings excludes Mueller bridges from use in
automated readout or control. An alternative, still
using 4-lead RTDs, is the constant current circuit of
Figure 4. Here a constant current source drives Is
through L1 and L4, and the potential produced by Rt
appears across L2 and L3. Circuitry connected to EO
must have an impedance high enough to prevent
appreciable current flow through L2 and L3. In this
circuit, the measured potential is unaffected by lead Figure 4: 4-lead system with constant current source
and contact resistance. Use of an AC current source
nullifies EMFs.

Two-wire Temperature Transmitters

Resistive networks like the Wheatstone bridge represent a passive solution to the problem of leadwire
resistance. The 2-wire temperature transmitter, in contrast, actively amplifies and conditions the RTD signal.
A transmitter, which mounts at or near the RTD location, converts the resistance reading to a current signal
proportional to temperature. This current travels over two extension wires to the control instrument. Unlike
voltage or resistance, current must be the same at both ends of a signal loop. This means that temperature
signals can be sent thousands of feet over two wires with no loss of accuracy from leadwire resistance or
electrical noise.
The standard process control transmitter produces a 4 to 20 mA signal proportional to temperature over a
specified range. The signal current also provides power for the transmitter’s electronics. Allowable resistance in
the signal loop depends on the voltage required by the transmitter at the 20 mA level. Hundreds of ohms—
thousands of feet of wire—are usually no problem.
The disadvantages of transmitters are price—typically about twice that of a resistance thermometer alone—and
the need to periodically recalibrate zero and span. On the other hand, cost savings result from the use of
inexpensive twisted-pair signal wires over long distances. Also, the linear current signal easily interfaces to
voltage input instruments through the use of a load resistor.
Most transmitters mount in a connection head attached to the resistance thermometer or in an instrument rack
nearby. Some types of transmitters mount in the same enclosure as the thermometer. This arrangement requires
careful attention to the problem of temperature rises induced by heat from the transmitter.

Copyright 2011, Minco Page 11

 Potential Sources of Error with Resistance Thermometers

Resistance thermometer systems are susceptible to three types of errors: The inherent tolerances built into the
thermometers, gradients between the thermometer and the medium to be sensed, and errors introduced along
the path between the sensor and readout or control instrument. Some sources of error are electrical; others
result from the mechanical construction of the thermometer.
Potential sources of error include:
Interchangeability and Conformity:
Conformity specifies the amount of resistance a thermometer is allowed to deviate from a standard curve (such
as the curve produced by the Callendar-Van Dusen equation). Conformity has two components: a tolerance at
the reference temperature, usually 0°C, and a tolerance on the slope or TCR. Figure 5 shows that a resistance
thermometer conforms most closely to its curve at the reference temperature, while the resistance fans out
above and below this reference. For example, IEC 751, Class B, requires calibration within 0.12 Ω (0.3°C) at 0°C,
but allows TCR to deviate from nominal 0.00385 by ±0.000012 Ω/Ω/°C. Thus, tolerance spreads to 0.8°C at 100°C,
1.3°C at 200°C, and on up to 3.8°C at 700°C. Interchangeability between two thermometers is no more than twice
the value of their conformity.

Figure 5: Effects of R0 and slope tolerance on total tolerance.

Commercial platinum resistance thermometer elements are available with extremely tight tolerances, to within
0.01 Ω (0.026°C) in some cases. When interchangeability is an overriding consideration, you may consider other
means to achieve it. For example, manufacturers may alter their calibration procedures to fix the reference
temperature—and tightest tolerance—at a point other than 0°C. Or if the difference between two thermometers
is more important than absolute temperature, matched pairs—measured to agree within a certain tolerance—
may be less expensive than calibrating each thermometer within a small range of nominal.
It is important to note that conformity and interchangeability specifications only denote the relative accuracy of
two otherwise identical thermometers mounted side by side in the same environment. They do not include
errors acting equally upon both thermometers.

Sensitivity:
The resistance change per degree change in temperature is a function of base resistance and TCR (Temperature
Coefficient of Resistance). Although a thermometer with higher sensitivity is not necessarily more accurate, a
larger signal simplifies output electronics and is less susceptible to leadwire effects and electrical noise. In
addition, a larger resistance produces the same voltage output with less measuring current, which helps to limit
self-heating of the thermometer element.

Copyright 2011, Minco Page 12

Insulation Resistance:
If the sensing element and leads are not completely insulated from the case, a shunting effect occurs in which
the case becomes a parallel resistor and lowers apparent readings. In most industrial thermometers, with
specified insulation resistances in the 100 megohm range, error approaches zero. The manufacturer must take
care to seal water-absorbing materials.
The shunting effect decreases with low-resistance elements, which accounts for the use of 25.5 Ω PRTs in
laboratory measurements.

Self-Heating:
A resistance thermometer is a passive resistance sensor; it requires a measuring current to produce a useful
signal. Because this measuring current heats the element wire above the true ambient temperature, errors will
result unless the extra heat is dissipated.
Self-heating is most often expressed in mW/°C, which is the power in milliwatts (1000 I2R) required to raise the
thermometer’s internal temperature by 1°C. The higher the mW/°C figure, the lower the self-heating. As an
example, assume a 5 mA measuring current is driven through a 100 Ω platinum RTD at 100°C. Self-heating is
specified as 50 mW/°C in water moving at 3 ft/sec.
The amount of heat generated is: 1000 mW × (0.005 A)2 × (138.5 Ω) = 3.5 mW
The self-heating error is: (3.5 mW) / (50 mW/°C) = 0.07°C
The generated heat increases with higher sensor element resistance (when a constant current measurement
device is used), or with increasing measuring current. The resulting error is inversely proportional to the ability of
the thermometer to shed extra heat; which, in turn, depends on thermometer materials, construction, and
environment. The worst self-heating occurs when a high resistance is packed into a small body. Thin film
elements, with little surface area to dissipate heat, are an example. Self-heating also depends on the medium in
which the thermometer is immersed. Error in still air may be over 100 times greater than in moving water.

Time Constant:
A time constant indicates the responsiveness of a resistance thermometer to temperature change. A common
expression is the time it takes a thermometer to reflect 63.2% of a step temperature change in moving water.
Response speed depends on the mass of the thermometer and the rate at which heat transfers from the outer
surface to the sensing element. A rapid time constant reduces errors in a system subject to rapid temperature
changes.

Repeatability:
The degree of accord between two successive readings with a thermometer is its repeatability. Loss of
repeatability results from permanent or temporary changes to the resistance characteristics of the element and
may be caused by exposing the thermometer to temperatures at or beyond the endpoints of its specified range.
A repeatability test cycles the thermometer between low and high temperatures; any changes to R0°C are noted.
A typical repeatability rating for an industrial platinum resistance thermometers is ±0.1°C.

Stability:
Stability is long term drift in thermometer readings. A typical specification would limit drift to 0.1°C per year for
rated operation. Normal service at points well within the temperature rating typically cause much less drift. Drift
is a consequence of the element material, with platinum being the most stable; encapsulating materials which
could contaminate the element; and mechanical stress placed on the element by expansion of winding bobbins
or other supporting structures.

Copyright 2011, Minco Page 13

Shock and Vibration:
Mechanical shock and vibration can alter thermometer readings or cause complete failure. In fact, stability and
ruggedness are somewhat exclusive. A laboratory thermometer designed for maximum stability contains an
unsupported element which is far too fragile for industrial use.
The elements of most industrial resistance thermometers are fully supported by a bobbin or packing material,
and therefore stand up quite well to extreme environments. More likely to suffer are leadwire transition points,
which should be properly immobilized. A typical RTD will meet a specification allowing shock of 100 Gs of 8
milliseconds duration and vibration of 10 to 2000 Hz at 20 Gs.

Packaging and Thermal Transfer:

Sheaths and other structures surrounding resistive elements should maximize heat transfer from the sensed
medium, minimize heat transfer from ambients which can alter readings, and provide necessary protection of
the elements. Proper materials and construction can dramatically improve reading accuracy.
One strategy practicable only with wire-wound resistance thermometers—versus thermistors, thermocouples,
and solid-state devices—is temperature averaging. An element may be wound to average temperature over
lengths of up to 100 feet.

Copyright 2011, Minco Page 14

 Resistance Thermometer Types

Examples of commonly available resistance thermometer types, with an emphasis on the design features which
take advantage of the benefits listed above, and which avoid the sources of error, are presented in the following
paragraphs.

Standard Platinum Resistance Thermometers for Laboratory Use

NIST specifies the standard platinum resistance thermometer (SPRT) as the standard interpolating instrument
used to define temperatures from -259.35 to 961.78°C. According to the ITS–90, a standard platinum resistance
thermometer must meet one of the following criteria:
R29.7646°C R −38.8344°C
R 0.01°C ≥ 1.11807 or R0.01°C ≤ 0.844235
This requires very high purity platinum wound in a nearly strain-free manner.

Figure 6 shows a classic “bird cage“ strain-free element,

in which the element wire is supported by insulating disks.

Figure 6: “Bird-cage” platinum element

Figure 7 shows a less expensive helical wind. The highest

accuracy primary standards encapsulate the element
within a glass or quartz sheath, although stainless steel or
Inconel may be substituted for secondary, or transfer,
standards. The element area must be hermetically sealed
to exclude oxidizing agents and helium-filled if used at
cryogenic ranges.

Figure 7: Helical-wound strain-free element

The unsupported coils of a strain-free element are quite susceptible to damage from shock and vibration. Even a
slight tap on a table can invalidate an expensive calibration.

Industrial Resistance Thermometer Elements

Resistance thermometers could not have migrated from the laboratory to the industrial plant without the
invention of rugged, low-cost elements. The need to encapsulate the resistive element for protection from shock
and corrosive environments, without unduly straining the element, have led to a variety of technologies for
constructing elements.

Copyright 2011, Minco Page 15

Figure 8 diagrams a common ceramic element. The
element wire, commonly platinum, is wound around a
ceramic bobbin, welded to leadwires, and coated with
glass. The manufacturer must carefully match materials
to prevent thermal expansion strains on the wire. Figure 8: Industrial resistance thermometer

Higher temperatures are possible with the element

construction of Figure 9. Here, a coil of platinum wire
is passed through bores in a ceramic tube. The bores
are then filled with a ceramic or alumina powder to
cushion the coils. Figure 9: High temperature element

Copper, nickel, and nickel-iron may replace platinum as element materials for lower temperatures, generally
below 260°C. Moderate temperatures also allow the use of organic materials in element construction, enabling a
wide variety of styles. Resulting benefits include lower cost and faster time response.
Thin-film elements (Figure 10), more recently developed, are LEADED
comparable to wire-wound ceramic elements in performance,
but lower in price. They consist of a flat substrate with a thin
-OR-
film of deposited platinum, laser-trimmed to proper resistance.
Thinfilms can have high resistances, commonly 1000 ohms SURFACE MOUNT
or even 10,000 ohms, without the expense of extra platinum Figure 10: Thin-film element
wire. Naked elements exhibit very fast time responses,
although they will respond more slowly in probes because of difficulties transferring heat to the flat element.
Also, the user should be careful to limit measuring current as the small size and high resistance of thin films
invite self-heating.

Industrial RTD Probes

The encased probe is the standard resistance thermometer configuration for industrial process control and
machinery protection. Most probe cases are stainless steel or Inconel to withstand high temperatures, although
other materials offer advantages at intermediate
ranges. For example, the tip-sensitive probe of Figure
11 has a copper-alloy tip which conducts heat 20 times
better than stainless steel. This design improves
thermal contact with sensed surfaces and reduces Figure 11: Tip-sensitive probe
errors from conduction along the sheath.
Standard probe diameters range from 0.125 to 0.250 inch. Smaller probes respond faster when directly
immersed, but larger probes may fit more snugly in standard thermowells. Probe lengths range from a few
inches to ten feet or more.
Figure 12 shows the construction of a high
temperature probe. The element fits in the tip,
surrounded by high temperature powder or
cement. Extensions leads, normally un-insulated,
extend back from the element and are
encapsulated by powder, cement, or bored ceramic
spacers. External leads, often insulated with Teflon Figure 12: High temperature probe
or glass braid, are potted with cement at the entry
point to seal against moisture.

Copyright 2011, Minco Page 16

The alternative construction of Figure 13 places the
element, potting, and lead transitions within a
module at the tip of the probe. This design allows
the user to cut the probe to required lengths.
Temperature is limited to the rating of the external
leadwire insulation: 260°C (500°F) for Teflon, up to Figure 13: Cut-to-length probe
550°C (1022°F) for insulations such as woven
mica/glass.

Probe Assemblies

A wide variety of mounting fittings and accessories aid probe installation. Selection depends on the nature of
the medium being sensed and cost requirements.
Direct immersion of a probe into a liquid requires a fitting with a pipe thread, which may be adjustable or
welded on the probe. Figure 14 shows a typical assembly, with one thread for mounting the probe and another
for a connection head. Connection heads provide a transition between probe leads and external signal wires.

PROBE
FITTING

CONNECTION HEAD

Figure 14: Typical probe assembly

Mounting in a solid material is best accomplished with a spring-loaded holder, which may be fixed or adjustable.
Spring loading provides good contact of the probe tip against the bottom of the hole and dampens potentially
damaging vibration.
When liquids are particularly corrosive, under high pressure, or fast-flowing, a thermowell may be necessary. A
thermowell is a tube, closed at one end, which protects the probe and allows its removal without breaking the
liquid seal. Many materials and styles are available to match application requirements. Thermowells drilled from
solid bar stock provide the highest pressure ratings, but welded models cost less. Figure 15 shows a typical
thermowell assembly, including a spring-loaded holder for enhanced thermal response and reliability.

PROBE
THERMOWELL
SPRING-LOADED HOLDER
CONNECTION HEAD

Figure 15: Thermowell assembly

Copyright 2011, Minco Page 17

Flexible Resistance Thermometers

The encased probes described above do not adapt well to sensing flat surfaces. Unlike thermocouple junctions,
which can be welded directly to metal surfaces, resistance thermometers present a certain amount of bulk; and
heat losses to ambient air may affect readings. Small flat elements, such as thin films, may mount on surfaces,
but fragile element and leadwire connections make installation difficult.
Figure 16 shows a flexible resistance thermometer with a
wire-wound sensing element sandwiched between
insulating layers. It conforms closely to sensed surfaces, and
has thin insulation to readily transmit heat to the sensing
element. The wire element may be wound to nearly any size
to average out temperature gradients, and the flexible
construction can withstand extreme shock and vibration.
Specification of the covering insulation depends on the
environment seen by the sensor. Polyimide is popular for
aerospace and medical use, as it withstands both chemicals
and vacuums. Silicone rubber features a higher temperature Figure 16: Flexible resistance thermometer
rating, to 220°C (428°F).

Special Purpose Resistance Thermometers

Resistance thermometers readily adapt to most process control and thermal equipment designs. The user may
specify cases with axial leads for circuit board mounting, flat packages for clamping to surfaces, miniature cases
for embedment into metal blocks, and any sheaths and fittings which can be produced by a machine shop. In
addition, wire windings may be configured to sense over large areas.

Where to Use Resistance Thermometers

In summary, resistance thermometers offer the greatest benefits relative to other thermometer types in these
situations:

• Accuracy and stability are the foremost goals of the application

• Accuracy must extend over a wide temperature range

• Area, rather than point, sensing improves control

• A high degree of standardization is desirable

Copyright 2011, Minco Page 18

 Thermistors

A thermistor operates through electrical resistance changes in semiconductors rather than pure metal. The base
material is a mixture of metal oxides pressed into a bead, rod, disk, wafer, or other shape. The bead, with
embedded leadwires, is sintered at high temperatures and often coated with epoxy or glass. Beads may be quite
small—down to 0.01” (0.25 mm) diameter in some cases.
The design, construction, and characteristics of thermistors vary widely among manufacturers.
Typical properties are:
1. Thermistors exhibit very large resistance changes, but usually in a direction opposite to resistance
thermometers; resistance drops as temperature rises. This is called negative temperature coefficient of
resistance (NTC).
2. Base resistances, commonly specified at 25°C, range from thousands to millions of ohms. Thermistor
sensitivity dwarfs that of resistance thermometers.
3. Resistance/temperature curves deviate widely from linearity, except over narrow ranges.
4. Thermistors tend to drift more than resistance thermometers, although they stabilize over time.
5. Temperature ranges are moderate, with 300°C (572°F) the common upper limit.
The combination of nonlinearity, high sensitivity, and instability has generally limited thermistors to high-
resolution measurements over limited ranges. A classic example is medical thermometry. Physicians are only
concerned with a small range around 37°C (98.6°F) and thermistors can be chosen to provide a large, fairly linear
signal in this area. One-point calibration is simple and sufficient.
Some manufacturers of thermistors offer special models with these characteristics:
1. Positive temperature coefficient (PTC) models, which are used more for current limiting in electronic circuits
than for temperature measurement; as current increases through the bead, self-heating drives up resistance
dramatically, throttling the current.
2. Thermistors which are closely matched, and therefore interchangeable, over a specified range.
3. Linearized thermistors or thermistor sets which produce a highly linear output over a limited range.
4. Glass-coated thermistors, specially aged, which may maintain excellent stabilities over moderate
temperatures, e.g. ±.005°C per year over the range 0 to 100°C.
As a general rule, a thermistor acts as a single-purpose sensing device. The designer must relate temperature
ranges, resistance/temperature characteristics, and output circuits for each application. The reward is a high
degree of resolution and accuracy with a relatively inexpensive system.

Copyright 2011, Minco Page 19

 Resistance and Temperature Characteristics of Thermistors

Figure 17 traces the resistance curve of a typical thermistor, referenced to a platinum resistance thermometer.
The thermistors resistance/temperature change is far more dramatic, although less linear and with a negative
slope. In general, the log of a thermistor’s resistance is proportional to the inverse of its temperature. The
following equation, called the Steinhart and Hart equation, is valid for ranges up to 100°C in size:
1 = a + a log (R ) + a log (R )3
T 0 1 e T 3 e T

where:
T is the absolute temperature in K (t°C + 273.15)
RT is the zero-power resistance at temperature T, and
a0, a1, and a3 are coefficients which describe a given
thermistor.

The particular coefficients of the equation depend solely

on the composition of the thermistor material and vary
widely. There are no published standards to guarantee
compatibility, although some popular types may be
available from more than one source.

Figure 17: Resistance/temperature curve: thermistor

vs. resistance thermometer

Thermistor Output Circuits

The lack of linearity in thermistor curves has spurred the development of many types of compensating circuits.
Two of the simpler compensating circuits to linearize voltage and resistance are described below.
Figure 18 shows a simple voltage divider circuit. The output voltage is given by:
E O (t ) 1
= = F (t )
ES 1 + sr (t )
where:
E 0 (t ) = output voltage at temperature t,
E s = source voltage,
R = a fixed resistance,
R (t 0 ) = reference, or base, resistance of the thermistor,
s = R (t 0 ) / R , and Figure 18: Voltage linearizing thermistor
circuit (Courtesy Thermometrics, Inc.)
r (t ) = R (t ) / R (t 0 ) (resistance at temperature t referenced to base
resistance)

F(t) is approximately linear over a selected temperature range. The user should specify base resistance R(t0) and
fixed resistance R so that the curve for s is centered over the temperature of interest. The output voltage tracks
temperature in a reasonably linear fashion over a limited range.

Copyright 2011, Minco Page 20

For a constant-current type circuit, in which a current source is driven through the thermistor and voltage drop
measured, a linear resistance is necessary. An ordinary shunt resistor in parallel with sensing element, as in
Figure 19, yields this formula:

R = R1 [1 − F (t )]
where:
R = overall resistance of the circuit
R1 = a fixed resistance and
F (t ) is the function defined above.

Figure 19: Resistance linearizing thermistor circuit

Again, response is fairly linear over limited spans. (Courtesy Thermometrics, Inc.)

Complex circuits, with Wheatstone bridges or other resistive networks, can further linearize thermistor output
when necessary. Most employ matched sets of thermistor beads.

Potential Sources of Error with Thermistors

Most of the sources of error listed for resistance thermometers apply as well to thermistors. Thermistors remove
one worry: Leadwire resistance is normally negligible in relation to their high resistance. Three or four-wire
circuits are rarely required for lead compensation, although they may be needed for linearization with matched
sets.
Self-heating can present far more of a problem with thermistors. High resistance generates more internal heat
and there is little surface area for dissipation. The same current through a thermistor and RTD will produce self-
heating around 100 times greater in the thermistor. The user should carefully read self-heat specifications and
limit current accordingly.
Because of self-heating error, the reference resistance of thermistors is specified at a zero-power value. This is an
extrapolation from a series of measurements at different sensing currents. It gives the ideal resistance assuming
no current and no consequent self-heating.

Where to Use Thermistors

Thermistors offer the greatest benefit in these situations:

• The application requires high resolution over a narrow span

• Probes are to be disposable, or may be frequently and easily recalibrated

• Low cost is a primary consideration

• The application is an OEM device produced in sufficient volume to justify the design of special
linearizing circuits

• Point sensing or miniaturization is desirable

Copyright 2011, Minco Page 21

 Thermistor Types

Most manufacturers offer beads in the configurations of Figure 20: a

bare bead, a glass-encapsulated bead, and an assembly with extension
leads. These simple elements may be assembled into the same types of
cases as RTDs. Bead thermistors are inherently point sensing devices
and may fit very small areas, although most environments require
some protection for the fragile bead and leadwire transitions.
Nearly all temperature sensing applications employ bead type
thermistors. Others, such as rods, chips, or flakes, mount on circuit
boards for temperature compensation or current limiting.
Figure 20: Thermistor types

Copyright 2011, Minco Page 22

 Summary

The superior sensitivity and stability of resistance thermometers and thermistors, in comparison to
thermocouples, give them important advantages in low and intermediate temperature ranges. In addition,
resistive devices often simplify control and readout electronics. Resistance thermometers are specified primarily
for accuracy and stability from cryogenic levels to the melting points of metals. They are accurate over a wide
temperature range, may be used to sense temperature over a large area, and are highly standardized. While
normally less stable than resistance thermometers, thermistors offer lower cost and higher sensitivity over more
limited ranges. Thermistor sensors can be relatively small and are well suited to product or equipment
applications where quantities of use justify the design of special readout circuits. As is evident throughout this
paper, both sensors can vary in terms of element types, leadwire resistance, electrical noise, vibration, self-
heating, and exposure to temperatures at or beyond the endpoint of a specified range.

Copyright 2011, Minco Page 23

Worldwide Headquarters European Headquarters Asia Pacific Headquarters ISO 9001:2000 / AS9100B
7300 Commerce Lane Usine et Service 20 Science Park Road © Minco 2011
Minneapolis, MN 55432 USA Commercial, Z.I. #02-31 Teletech Park
Tel: 1.763.571.3121 09310 Aston, France Singapore Science Park II Flex Circuits
Fax: 1.763.571.0927 Tel: (33) 5 61 03 24 01 Singapore 117674 Thermofoil™ Heaters
sales@minco.com Fax: (33) 5 61 03 24 09 Tel: (65) 6511 3388 Sensors
www.minco.com Fax: (65) 6511 3399 Instruments