Page 457 - The Mechatronics Handbook

0066_Frame_C19 Page 79 Wednesday, January 9, 2002 5:27 PM

Law 1. A third metal introduced in the circuit with both ends of the third metal at an isothermal point
does not affect the thermally induced voltage of the original pair.

There are two important implications associated with this law. The first means that the nature of the
electrical contact between the wires at the junction is not critical, and that the thermocouple itself can
be made up of two wires that are soldered, brazed, welded, or swaged together. In all these cases, a third
metal is present at the junction whether it is the filler in soldered and brazed connections, the intermediate
alloy produced by welding dissimilar metals, or the metal swage holding the ends of the wire together.
This does not mean that there are not other concerns involved with how the junction is formed. It will
obviously not do to solder wires together and then use the junction above the melting point of the solder.
Most commercially prepared thermocouples are welded for that reason. This law also allows for a metallic
item whose temperature is being measured to serve as the actual junction by attaching the thermocouple
leads directly to it. This might be done to avoid the time needed to transfer energy between the object
and an independent thermocouple assembly. The second implication of this law allows a measuring
circuit made of conductors other than those used in the thermocouple to be inserted in the circuit as
long as both connections between the measuring circuit and the two thermocouple wires are at the same
temperature.
Law 2. The temperatures along the wires do not affect the thermally induced voltage characteristic of the
temperature of the two junctions.
This means that the thermocouple leads can be conveniently routed through various temperature regions,
and that only the temperatures at the junction and the monitoring location are important in determining
the voltage.

Law 3. Each metal has its own voltage gradient for a given temperature gradient independent of the wire
used to monitor that voltage.
This means that each type of metal can be calibrated against a standard and that the calibration is valid
for each type of thermocouple that can be made from this wire.

Thus far the discussion has been limited to open circuits because this avoids the complications of
2
energy being carried away from a hot junction by the electrons or the I R losses in the conductors that
both produce heat and reduce the measured voltage. Using high-impedance amplified voltmeters or
differential voltmeters having infinite impedance when balanced, allows the practical open circuit voltage
to be measured and eliminates these sources of error. However, since the wires used in thermocouples
are usually metals with reasonable heat conduction, energy may be inadvertently removed from the
measured system by simple heat conduction along the wires.
To actually use a thermocouple the temperature at one junction must be known and some sort of
calibration table or polynomial curve fit must be used to convert the measured voltage to temperature
at the junction. There are published tables and polynomials for common pairs of metals used in thermo-
couples referenced to the ice point of water (Croarkin et al., 1993). These are based upon an average
alloy but the alloy actually purchased cannot be precisely the same as the one represented by the table.
This unavoidable variation of alloy content and application of standard tables is the major source of
error in thermocouple readings. Despite the best efforts of the manufacturers, this variation can lead to
as much as 2% error in the reported voltage for a given temperature measurement. Individual spools of
wire or assembled thermocouples can be calibrated to minimize this source of error.
The temperature of the reference junction must be known to determine the temperature of the
measuring junction from the measured voltage. If the temperature of one of the reference junctions is
not the same as the reference temperature of the calibration table, the voltage associated with the known
temperature of the reference junction can be algebraically added to the measured voltage to determine
the voltage that would have been measured if the reference junction were, in fact, at the defined temperature
of the table. This is not as difficult as it first appears. If the reference junction is kept in an ice bath, no
correction is needed when using the normal calibration tables. To make this easy, there are commercial

©2002 CRC Press LLC