A G R E AT E R M E A S U R E O F C O N F I D E N C E

How to Select the

Right Temperature
Sensor
By: Dale Cigoy
Title: Senior Applications Engineer
Keithley Instruments, Inc.
Cleveland, OH

can handle this temperature range, and

Introduction interfaces. Due to the scope of this article,
the type K is the most general purpose
The most widely measured physical such sensors will not be covered in detail.
thermocouple for such applications. On
parameter is temperature. Whether the other hand, a ceramic firing kiln can
in process industry applications or in Selecting a Temperature Sensor be many hundreds of degrees C. A type N
laboratory settings, accurate temperature There are several considerations would be good for the higher temperature,
measurements are a critical part of success. when selecting a temperature because of their stability and resistance
Accurate temperature measurements are sensor. First, consider to high temperature oxidation. On the
needed in medical applications, materials the type of application. What is the other end of the scale, superconducting
research in labs, electrical/electronic device to be measured? Is it ambient air devices achieve superconductivity at
component studies, biology research, temperature in a room or enclosure? An very low temperatures, around several
geological studies, and electrical product electronic component with plastic or metal Kelvin (or just above absolute zero or
device thermal characterization. packaging that may or may not have high – 273ºC), so the sensor must be able to
voltages present? Or an ingot of glowing withstand the sub-freezing climate in
There are many different types of sensors steel? Perhaps some part of an automobile which superconductors operate. Type E
available to measure temperature. The three engine such as an intake or exhaust port? thermocouples with their high output (about
most common are resistance temperature Some of these considerations can drive 68uV/C) would be good, because they are
detectors (RTDs), thermocouples (TCs), the choice of sensor due to environmental well suited to low temperatures. Table 1
and thermistors. Each of them have specific and safety factors, cost budget per sensor, gives a summary of TCs and their ranges.
operating parameters that may make it a and distance from sensor to instrument.
better choice for some applications than
others. Solid state sensors are also available The next consideration is the range of
for moderate temperature range applications expected measured temperature. An
and have the advantage of built-in signal automobile engine block, when fully
conditioning and ease of interfacing warmed up, can generate temperatures of
– in some cases with direct digital serial greater than 100ºC. Most thermocouples

How to Select the Right temperature Sensor June 2007 

 Thermocouple Type Materials Range One important property of thermocouples is their non-linearity;
J Iron(+) -40º to 760ºC that is, thermocouple output voltage is not linear with respect to
Constantan(-)
temperature. Consequently, to convert output voltage to temperature
accurately requires mathematical linearization.
K Chromel (+) -200º to +1200ºC

Alumel (-)

T Copper(+) -270º to 400ºC Thermocouples consist of two dissimilar metals joined (either
Constantan(-) welded or twisted) together at one end and open at the other. They
operate on the principle of the thermoelectric effect and can be
B Platinum 30% Rhodium (+) 20º to 1820ºC
thought of as the junction of two different metals producing a
Platinum 6% Rhodium (-)
voltage when a thermal difference exists between the two metals
E Chromel (+) -270º- 910ºC (also known as the Seebeck Effect). The voltage signal at the open
Constantan (-) or output end is a function of the temperature at the closed end. As
the temperature rises, the voltage signals increases.
N Nicrosil (+) -270º-1300ºC

Nisil (-)

S Platinum 13% Rhodium (+) -50º to 1760ºC Here’s what really happens. The open-end signal is a function of
Platinum (-) not only the closed-end temperature (the point of measurement)
but also the temperature of the open end. Only by holding T2 at
a standard temperature can the measured signal be considered a
Table 1: Thermocouple types
Table 1: Thermocouple types direct function of the change in T1. The open-end voltage, V1, is a
function of not only the closed end temperature (the temperature
Another consideration is the available area for the sensor to be mounted to adequately at the point of measurement), but also the temperature at the open
Another consideration is the available area for the sensor to be
sense the temperature to be measured. The device to be measured must have room
end (T2). The reason the voltage is developed is because different
mounted to adequately sense the temperature to be measured. materials produce different voltage for the same temperature
enough to handle the selected sensor mounting. For example, an integrated circuit is a
The device to be measured must have room enough to handle the difference. This is the reason for the two different metals. If
selected sensor mounting. For example, an integrated circuit is a
tiny electronic component, so the right sensor would depend on what parameter is to be
they were the same metals, then the voltage would be zero.
tiny electronic
measured, component,
the IC package, soframe
or the lead the or
right sensor
chip device would
itself. depend
Most sensors onin a
come
what parameter is to be measured, the IC package, or the lead frame
or chip device itself. Most sensors come in a variety of shapes and The industry standard for T2 is 0ºC. Most tables and charts make
sizes, and one is sure to fit the application. For the tiny electronic the assumption that T2 is at 0ºC. In industrial instrumentation, the
circuit of an IC, an electrically isolated RTD would probably be best difference between the actual temperature T2 and 0ºC is usually
because of the size, isolation, and accuracy of the RTD. corrected electronically within the instrument. This adjustment
is known as cold junction compensation or ice-point reference.

Thermocouples
By far the most commonly used temperature sensor is the Advantages
thermocouple or TC. The key reasons are that thermocouples are Thermocouples have many advantages over other types of
low cost, extremely rugged, can be run long distances, are self temperature sensors. For one, they are self powered, requiring no
powered, and there are many types of thermocouples available external power supply. They are also extremely rugged and can
to cover a wide range of temperatures. Low cost speaks for itself withstand harsh environments. Thermocouples are also inexpensive
in many applications. Ruggedness means they will last in many compared to RTDs and thermistors and come in a wide variety
different environments, including outdoors and with exposure to of types with wide temperature ranges. (Refer to Table 1 for
harsh factory environments. Metal-sheathed TCs are available Thermocouple ranges.)
to help protect them in harsh or corrosive environments, or
they can be run inside conduit piping. Different alloys allow
different range and sensitivity of measurement. Some common Disadvantages
types of TCs include J, K, T, E, R, S, B, and N, which refers to Thermocouples are non linear and require cold-junction
the type of material from which they are constructed (as in table compensation (CJC) for linearization. Also, the voltage signals
1). The type J, K, and T are the most common and are readily are low, typically in the tens to hundreds of microvolts, requiring
available in spools or pre-made forms. The ranges for all types careful techniques to eliminate noise and drift in low-voltage
of thermocouples can be found in NIST (National Institute of environments. Accuracies are typically in the range of 1-3%
Standards and Technology) reference tables at www.nist.gov. depending on wire alloy consistency and cold junction accuracies.

 June 2007 How to Select the Right temperature Sensor

Common Errors An improvement on the two-wire method is the three-wire method.
Avoiding some common mistakes when setting up and using Here again, a current is forced through the device and the resulting
thermocouples will yield better measurements. One common voltage is measured. However, a third wire provides compensation
problem is that the CJC is not configured or compensated properly for the lead resistance. This requires either a three-wire compensat-
or at all. This leads to inaccurate or nonlinear temperature ing measurement unit or actually measuring the contribution from
measurements. the third wire and subtracting it from the overall measurement.

A third technique is the four-wire method, which is generally found

Another mistake is not to use copper wire from the thermocouple in calibration labs for reference probes. Similar to the other two
connection to the measurement device. Normally the measurement methods, a current is forced through the RTD and the resulting volt-
devices (voltmeters, DMMs, etc.) have copper input terminals. Using age is measured. However, the current is forced on one set of wires,
another alloy (tin, aluminum, etc.) essentially introduces another while the voltage is sensed on another set. This method completely
thermocouple into the measurement. This is because any junction compensates for the lead resistance. The voltage is sensed at the
of dissimilar metals forms a thermocouple. On the measurement resistive element, not at the same point as the source current, which
device side, the voltmeter being used may not be sensitive or means that the lead resistance is completely out of the measurement
accurate enough for thermocouple measurements. To avoid the path. In other words, the test lead resistance is not part of the actual
problem, make sure the voltmeter is sensitive and accurate enough voltage measurement.
for the low-voltage signals (uV to mV) of the thermocouples. Some
proper shielding would also prevent any external noise. Surround As an example, consider the test lead resistance to be about
the sensitive circuit with a conductive shield and connect it to circuit 0.2Ohms and the RTD to have a resistance of 100 Ohms. In this ex-
or measurement LO for maximum effect. ample, the test-lead resistance is about 0.2% of the total resistance in
the test circuit. With the four-wire method, the test lead resistance is
not part of the measurement circuit and therefore only the resistance
RTDs of the RTD is measured. This eliminates the 0.2% error for a much
One of the most accurate temperature sensors is a resistance more accurate measurement.
temperature detector, or RTD. In an RTD, the resistance of the
device is proportional to temperature. The most common material Pros and Cons
for RTDs is platinum, with some RTDs being made from other RTDs have some advantages over other temperature sensors. For
metals such as nickel or copper. RTDs have a wide range of one, they are the most stable and most accurate of the different
temperature measurement. Depending on how they are constructed, temperature measurement devices. RTDs, like thermocouples, are
they can measure temperatures in the range of –270ºC to +850ºC. not linear. This means they also need some linearization, which is
generally done by using correction factors.

RTDs require an external stimulus, usually a current source, One of the drawbacks is that RTDs are more expensive than
to function properly. However, the current generates heat in thermistors and thermocouples. They also require a current source.
the resistive element, which causes an error in the temperature They have a small delta R, which means there is a low resistance-to-
measurements. The measurement error is calculated by the formula: temperature change. For example, to change one degree Celsius, the
RTD might change by 0.1Ohm.
Delta T = P x S
where T is temperature, P is the I-squared power generated, When using RTDs, several common occurrences are often not taken
and S is degrees C/milliwatt. into account, the biggest of which is self-heating. Self-heating of the
RTD with the test current could result in measurement inaccuracy.
There are several techniques for measuring temperature with an If measuring low temperature (below 0ºC), the heat generated from
RTD. The first is a two-wire method. This method works by forcing the RTD could derate the expected temperature. Also, if there is no
current through the RTD and measuring the resulting voltage. The compensation for the test leads, even more error can be introduced
benefit is that it is a simple method using only two wires, making it into the measurement. Using the four-wire method helps eliminate
easy to connect and implement. The main drawback is that the lead this type of error. Another common mistake is not selecting the
resistance is part of the measurement, which can introduce some proper RTD temperature range. Trying to measure outside of the
error into the measurement. RTD temperature range can result in greater errors or even sensor
damage.

How to Select the Right temperature Sensor June 2007 

 Thermistors Summary
Another common temperature sensor is a thermistor. Like an RTD, There are several sensors from which to choose for measuring tem-
a thermistor changes resistance as temperature changes. The therm- perature. The thermocouple, RTD, and thermistor are three of the
istor offers higher sensitivity than RTDs, meaning that the thermis- most common sensors used today. (Table 2 offers a comparison of
tor resistance will change much more in response to temperature these three sensor types.) Thermocouples are the most widely used
changes than an RTD. sensors. They have the widest measurement range and are used for
flame temperatures, gas/electric ovens, and heating systems. Therm-
Most thermistors have a negative temperature coefficient, mean- istors are mainly used in human environment temperatures (from
ing that the resistance decreases when the temperature increases. 0ºC to 30ºC), such as freezers and incubators, whereas RTDs are the
Thermistors are also less linear than RTDs and require a correction most accurate and are used almost exclusively in calibration/stan-
factor. The Steinhart-Hart equation, which describes the resistance dards applications. Each sensor has it own advantages and disad-
change of a thermistor as related to temperature, is used to help vantages. Selecting the appropriate sensor is important in obtaining
Figure 3 – Four-wire resistance measurement setup.
approximate individual thermistor curves. (A good example of the accuracy and reliability of the temperature measurement.
calculation of the curves is at http://www.reed-electronics.com/tm-
world/software/stinhart.zip.) Table 2:

Thermocouple RTD Thermistor

The equation is: Range -200º to 2000ºC -250ºC to 850ºC -100ºC to 300ºC
Accuracy > 1ºC 0.03ºC 0.1ºC
Thermal Response Fast Slow Medium
Cost Low High Low to moderate
1/T = A + B x (lnR) + C x (lnR) squared Long term stability Low High Medium

where: Table 2:
T is degrees in Kelvin
R is the thermistor resistance
A, B, and C are curve fitting constants determined through a cali-
bration process ln is the natural log function (log to the base e)

Pros and Cons

Thermistors have some advantages over other sensors. For one,
they are simple to set up and operate using just a standard two-wire
measurement method. They also have a fast response time. Because
thermistors can be made very small, they can respond to tempera-
ture changes quickly.

On the down side, the non-linear properties of thermistors require

linearization. They also have a limited temperature range and are
not as rugged as TCs or RTDs. Because thermistors are semicon-
ductors, they are more likely to have de-calibration issues at high
temperatures. Thermistors also require a current source and like
RTDs have self-heating characteristics.
Figure 1: Typical thermocouple setup showing cold junction compensation
Common Mistakes
Not allowing for self-heating and selecting a device with an inad-
equate temperature range are common mistakes made when using
thermistors. However, there are a few ways to reduce self-heating.
The typical test current is a constant DC current. Using a pulsed DC
current helps reduce self-heating effects, because the current is ap-
plied for a short portion of the measurement cycle, thereby reducing
the overall developed power in the thermistor.

Another common mistake is measuring temperature outside of the

range of the thermistor. Always consider the range of temperature to
be measured. Remember, the temperature range of the thermistor is
about a few hundred degrees C. Some manufacturers even caution
that extended exposure even below operating limits will cause the
thermistor to drift out of tolerance.

 June 2007 How to Select the Right temperature Sensor

Figure 2: A schematic diagram showing a copper-to-copper connection. Figure 3: Four-wire resistance measurement setup.

Specifications are subject to change without notice.

All Keithley trademarks and trade names are the property of Keithley Instruments, Inc.
All other trademarks and trade names are the property of their respective companies.

A G R E A T E R M E A S U R E O F C O N F I D E N C E

Keithley Instruments, Inc. ■ 28775 Auror a Road ■ Cleveland, Ohio 44139-1891 ■ 440-248-0400 ■ Fax: 440-248-6168 ■ 1-888-KEITHLEY ■ www.keithley.com

Belgium china finland fr ance germany india

Sint-Pieters-Leeuw Beijing Espoo Saint-Aubin Germering Bangalore
Ph: 02-363 00 40 Ph: 8610 -82255010 Ph: 09-88171661 Ph: 01-64 53 20 20 Ph: 089-84 93 07-40 Ph: 080 22 12 80-27/28/29
Fax: 02-363 00 64 Fax: 8610 -82255018 Fax: 09-88171662 Fax: 01-60 -11-77-26 Fax: 089-84 93 07-34 Fax: 080 22 12 80 05
w w w.keithley.nl w w w.keithley.com.cn w w w.keithley.com w w w.keithley.fr w w w.keithley.de w w w.keithley.com

italy japan korea Malaysia netherlands singapore

Milano Tokyo Seoul Kuala Lumpur Gorinchem Singapore
Ph: 02-553842.1 Ph: 81-3-5733-7555 Ph: 82-2-574-7778 Ph: 60 -3-4041- 0899 Ph: 0183-63 53 33 Ph: 65-6747-9077
Fax: 02-55384228 Fax: 81-3-5733-7556 Fax: 82-2-574-7838 Fax: 60 -3-4042- 0899 Fax: 0183-63 08 21 Fax: 65-6747-2991
w w w.keithley.it w w w.keithley.jp w w w.keithley.co.kr w w w.keithley.com w w w.keithley.nl w w w.keithley.com.sg

sweden Switzerland taiwan UNITED KINGDOM

Solna Zürich Hsinchu Theale
Ph: 08-50 90 46 00 Ph: 044-821 94 44 Ph: 886-3-572-9077 Ph: 0118-929 75 00
Fax: 08-655 26 10 Fax: 41-44-820 30 81 Fax: 886-3-572-9031 Fax: 0118-929 75 19
w w w.keithley.com w w w.keithley.ch w w w.keithley.com.t w w w w.keithley.co.uk

© Copyright 2007 Keithley Instruments, Inc. Printed in the U.S.A. No. 2816 0207

How to Select the Right temperature Sensor June 2007