Three types of sensors are most commonly
used to measure temperature in industrial
environments: thermocouples, resistance
temperature detectors (RTDs) and thermistors.
Each has its unique advantages, disadvantages
and signal conditioning requirements.
Thermocouples are inexpensive, proven sensors
and provide the widest range of temperature
measurement. TC’s generate their own
signal and do not need excitation. They are
among the most numerous sensors used.
TC’s do have some drawbacks. Substituting
one thermocouple for another one of the
same type can produce a slightly different
output voltage, forcing recalibration of the
signal conditioner for best accuracy.
TC’s may also be contaminated by the envirionment.
This contamination is frequent
enough that virtually all thermocouple signal
conditioners provide open-circuit or ‘burnout’
detection. Dataforth’s SCM5B37
thermocouple signal conditioning modules
can be configured for either upscale
or downscale burnout detection.
TC operation is based on two physical properties.
When a metal rod is heated at one end,
a small voltage (Thomson EMF) develops
between the hot and cool ends. If two dissimilar
metals are joined and heated at their
junction but not connected at the unheated
ends, a similar EMF occurs. This is called the
Peltier EMF. The magnitude and polarity of
the Peltier EMF are dependent on the temperatures
of the junctions and the combination
of the two metals involved.
The algebraic sum of the Thomson EMFs
and the Peltier EMF appear at the unjoined
ends of the metal pair. This EMF is the basis
for all thermocouple operation and is called
the Seebeck EMF. If both the joined and open
ends of the metallic pair are at the same temperature,
this EMF is zero. If the temperatures
at the open ends are equal and kept constant,
the Seebeck EMF is a direct function of
the temperature at the measurement junction
and can be used to measure that temperature.
The Seebeck EMF depends on the TC’s composition
and ranges from 10 to 80mV for
full-scale output. See Table 1 for common
thermocouples and their measurement
Table 1. Thermocouple Types
|TYPE||COMPOSITION||MEASUREMENT LIMITS (°C)|
|J||Fe vs Cu-Ni||-210 to 760|
|K||Ni-Cr vs Ni-Al||-270 to 1372|
|T||Cu vs Cu-Ni||-270 to 400|
|E||Ni-Cr vs Cu-Ni||-270 to 1000|
|R||Pt-13% Rh vs Pt||0 to 1768|
|S||Pt-10% Rh vs Pt||0 to 1768|
|B||Pt-30% Rh vs Pt-6%Rh||0 to 1820|
|C||W-5% Re vs W-26% Re||0 to 2320|
|N||Ni-14.2% Cr-1.4% Si vs Ni-4.4% Si-0.1%Mg||-270 to 1300|
TC outputs have non-linear relationships
(Figure 5) to the measured temperature. In
addition, each thermocouple type has its own
non-linear characteristic. This makes it difficult
to provide universal linearization for the
many types of TCs available. However,
Dataforth’s SCM5B47 series
linearization and has a complete selection
to match all popular TC types.
Thermocouple accuracy depends on the composition
and purity of their metals and their
fabrication. Usually a thermocouple will not
be more accurate than 0.5% to 1% of its total
measurement range. This translates to a
measurement error as large as 2°C for some
TC types. TCs almost always require amplification
and cold-junction compensation.
It is important to note that thermocouples
always indicate the difference between two
temperatures at two junctions. The measurement
junction is the one whose temperature
is of interest. The other junction is either
maintained at a reference temperature (0°C)
by physical means or this condition is simulated
electronically. It is called the reference
Physically maintaining one or more reference
junctions at a constant temperature is
not an easy or desirable solution for industrial
measurements. Instead, the reference junction
is created by bringing the measurement
thermocouple wires to the amplifier and connecting
them to a terminal block. This terminal
block is often called an iso-block (shorthand
for isothermal terminal block). Its high
thermal conductivity assures that the terminals
for the thermocouple wires are at the
same temperature. SCM5B modules use the
very predictable voltage drop of a silicon
diode to measure the terminal block temperature
and imitate a thermocouple in an ice
bath at 0°C. This entire process is called
cold-junction compensation and the circuit is
called the cold-junction compensator (CJC).
While the technique sounds complicated, it is
easy to implement electronically. Figure 6
shows a block diagram. The electronic compensation
is usually 2-3 times more accurate
than the thermocouple itself. It allows precision
instrumentation to provide accurate temperature
readings even when the ambient
(cold-junction) temperature moves through
large swings occurring in industrial applications.
RTD’s are among the most well-behaved and
precise temperature measuring devices available
for industry. An RTD is a precision
resistor with a well-defined resistance vs.
temperature curve. RTDs are classified
according to their material composition and
change in resistance versus temperature
(Alpha Value). See Table 2.
Copper, nickel, and platinum RTDs enjoy
wide-spread use, although the platinum RTD
is now almost universally specified for new
industrial installations.The platinum RTD
offers many outstanding characteristics
including high accuracy, wide measurement
range and chemical resistance to many of the
nastier atmospheres in industrial applications.
The PT1OO is the most commonly used
RTD curve and displays an ice-point resistance
of 100 ohms. It is based upon several
European standards and is supported by the
IEC as well. PT200, PT500 and PT1000
RTDs have appeared on the market recently,
but they are simply multiples of the basic
PT100 curve in their behavior. That is, a
PT500 sensor will have five times the resistance
of the PT100 at the same temperature.
The higher resistance allows use of less
material and provides a cost savings. It also
can provide a smaller sensor. If, however, the
smaller sensor is used, careful attention must
be paid to the level of excitation current used.
Too much current will cause the sensor’s
small mass to self-heat and degrade its accuracy.
Dataforth’s SCM5B34 and SCM5B35
signal conditioning modules specifically use
0.25mA excitation current to eliminate this
The D1OO curve is a commonly used US
RTD curve and is also supported by a
Japanese standard. For many years, the
SAMA (Scientific Apparatus Manufacturer’s
Association) curve was popular used in the
US, but it has since fallen into disuse. The
PT1OO curve now dominates, finding use in
85-95% of the industrial applications.
RTDs can be used in several configurations
that reflect the cost factors and the degree of
accuracy desired. Figures 7,8 and 9 show
SCM5B34 and SCM5B35 modules in 2-
wire, 4-wire, and 3-wire connections respectively.
The 2-wire configuration is used when
signal lines are short, highest accuracy is not
required, and lowest cost of installation is of
paramount importance. Because RTD’s must
be excited by a current, signal-line resistance
will appear in the apparent resistance of the
RTD. If the resistance is high, or if the temperature
coeffecient of the line resistance is
high, errors will occur, giving a false temperature
The most accurate connection method is to
excite the RTD by using two power leads to
carry the excitation current, and using two
additional wires to sense the voltage at the
RTD directly. If high impedance circuits are
used to measure the voltage on the second
pair of wires, no appreciable current will
flow through them and the voltage measured
will only be that at the RTD. This setup,
shown in Figure 8, is used in almost all laboratory
or other high-accuracy situations. The
SCM5B35 is specifically designed for this
Table 2. RTD Types
|DESIGNATION ||MATERIAL ||ICE POINT RESISTANCE ||ALPHA VALUE|
|PT100 ||Platinum ||100Ω ||0.00385|
|PT200 ||Platinum ||200Ω ||0.00385|
|PT500 ||Platinum ||500Ω ||0.00385|
|PT1000 ||Platinum ||1000Ω ||0.00385|
|D100 ||Platinum ||100Ω ||0.003916|
|SAMA ||Platinum ||98.129Ω ||0.003923|
|NI120 ||Nickel ||120Ω ||0.00672|
|CU10 ||Copper ||10Ω ||0.004274|
A reasonable compromise between 2-wire
and 4-wire connections is the three-wire connection.
It offers high accuracy and a lower
cost of wiring. This connection is the one
most frequently used in factory or plant
instrument systems. Figure 9 gives the
details. By connecting the wires as shown,
equal currents flow from balanced current
sources in the SCM5B34 through the ground
wire, and back through the top and bottom
RTD connections. If the line resistances are
equal, the voltage drops from the amplifier
inputs to the top and bottom of the RTD will
be equal and the error voltages added to each
line will also be equal. The two line voltage
drops cancel, and the differential input to the
amplifier is the actual voltage at the RTD.
The voltage drop in the common (ground)
wire will be twice that in the other two lines
and will add to both amplifier inputs equally.
Thus it appears as a common mode voltage to
the amplifier inputs. One may question the
basic assumption that the line resistances are
equal; using twisted three-wire cable helps
make this assumption correct.
One of the few drawbacks to the use of
RTD’s is their nonlinear change in resistance
with temperature. Figure 10 shows the nonlinearity
of a PTIOO RTD measuring from
0°C to 450°C: it is almost 2% of span.
Dataforth SCM5B34 and SCM5B35 modules
provide hardware non-linearity correction
to 0.015% typical.
The RTD has become the preferred sensor for
temperature measurement when accuracy,
repeatability and interchangability are
required. The PT100 RTD has a uniform
non-linear resistance curve, is resistant to
most harsh environments, and is particularly
robust for industrial measurements.
Thermistors are relatively inexpensive
devices exhibiting very large changes in
resistance for small changes in temperature
(typically 4-6%/°C). For example, a typical
thermistor may have a nominal resistance of
30kΩ at 25°C, but have a resistance of 2.5
kΩ at 85°C. This large change in resistance
makes line resistance to the thermistor a very
small source of error and use of the thermistor
can avoid the three-and four-wire connections
common with the RTD.
In the past, thermistors have had very poor
interchangeability. That is, an instrument calibrated
for one thermistor would require
major recalibration when a new thermistor
was substituted. Also, the relationship
between temperature and resistance could
vary significantly from lot to lot and even
between sensors from the same manufacturing
lot. Low-cost thermistors still retain these
Today, manufacturers can supply thermistors
with vastly improved interchangeability. This
fact has given the thermistor a boost as a serious
process temperature sensor. This interchangeability
has its limitations, however.
The temperature span over which interchangeability
exists lies between 50°C and
100°C of the thermistor’s total measurement
range. As an example, some manufacturers
can achieve 0.1°C interchangeability from
0°C to 70°C or from 120°C to 180°C.
Thermistors are also somewhat limited in
their absolute temperature range. Realistic
limits lie between -100°C and 450°C. This
can eliminate them from consideration for
some industrial measurements, but they still
remain ideal for low-cost sensing within their
Thermistors are very nonlinear devices.
Figure 11 shows the behavior of a typical
thermistor. Thermistors can, however, be
used in or purchased as part of a resistor network
whose output is highly linear over most
of the useful temperature measurement range
of the device. Figure 12 shows a resistorthermistor
network whose behavior is quite
linear over a modest range as shown in
Figure 13. Note that the network resistance
now approaches that of some RTDs, and the
three- and four-wire connections used with
RTDs may be necessary for best accuracy.
Dataforth offers several signal conditioning
modules for thermistors. Because of the large
variety of thermistors available, consult
Dataforth for details.
One of the big advantages of thermistors is
their small size. This gives them perhaps the
best thermal response time of almost any
temperature sensor. Some can react in milliseconds
to large temperature variations.
The down side of this small mass is self-heating
of the sensor from the excitation current.
This heating can pose a significant source of
error. Since thermistors are typically highresistance
devices, large currents will produce
large self-heating errors. Pay close
attention to the manufacturer’s recommendations
in this area.