 |
Web Hosting by Netfirms | Free Domain Names by Netfirms |
|
|
|
By Fred Schraff, P.E.
Adapted from an article that appeared in the June 1996 edition of
Measurements & Control
Temperature measurement is an important part of scientific experiments, research and development, and industrial processes. The thermocouple provides a simple and efficient means of measuring temperature because it produces a voltage which is a function of temperature. This voltage can be read using an analog to digital converter (or any voltmeter), and the temperature can be inferred from consulting standard tables.
A thermocouple works because there is voltage drop across dissimilar metals which are placed in contact. This voltage is a function of temperature. In principle, a thermocouple can be made from almost any two metals. In practice, several thermocouple types have become standard because of desirable qualities such as linearity of the voltage drop as a function of temperature and large voltage to temperature ratio. Some common thermocouple types are designated as J, K, T, E, N28, N14, S, R, and B. Figure 1 shows a type T thermocouple. The thermocouple leads are joined by welding or soldering. (Thermocouple wires are commonly twisted together, but this is not recommended and can lead to inaccuracies.)
One might think that making an accurate temperature measurement is as simple as connecting the thermocouple wires across the terminals of a voltmeter, measuring the voltage and looking up the corresponding temperature in a table. To demonstrate the errors introduced in this procedure, I placed the junction of a type T thermocouple in boiling water (known to be at 100°C) and read the voltage across the leads. The reading was 3.634 mV, which corresponds to 86.1°C. Errors this large are intolerable in most applications.
This temperature error arises because the connection of the thermocouple leads to the voltmeter constitutes two additional thermoelectric junctions that subtract voltage from the signal being measured.
This problem can be remedied using the arrangement shown in Figure 2. One thermocouple junction is held in an ice bath at 0°C. This is called the reference junction. The other thermocouple junction is the temperature probe. If the probe is at 0°C, then there is no thermoelectric voltage across the leads because the thermoelectric voltages created by each junction cancel each other out. Thermocouple tables usually assume that a reference junction is held at 0°C.
The configuration of Figure 2 avoids the problem of additional thermoelectric voltages being generated at the instrument as long as the terminals are at the same temperature. Thermoelectric voltages are still generated at the junction of the wire and terminal, but now the voltage drop generated at each terminal is the same, and so is cancelled by the other.
Using the setup shown in Figure 2 with the probe junction in boiling water, I measured 4.511 mV across the thermocouple using IOtech's DaqBook/216 and DBKl9 thermocouple card. Looking up the corresponding temperature in a table gives 104.4°C. This is still a few degrees off, but it is a much better measurement than without the reference junction.
Most of the remaining error arises from a small offset introduced in the DaqBook's internal electronics. (All electronic instruments have some offset.) We can obtain a much more accurate temperature reading by measuring this offset and subtracting it from the voltage across the thermocouple before conversion to temperature. The offset voltage was measured across a shorted channel of the DBKl9 thermocouple card and found to be 0.245 mV. This gives an adjusted thermocouple voltage of 4.266 mV, which corresponds to 99.7°C.
The difference between the 99.7°C and the expected 100.0°C can arise from a number of sources. Fluctuations in the barometric pressure or impurities in the water can change the boiling point of water. Furthermore, imperfections in the thermocouple wire can cause smaIl deviations in the thermoelectric voltage from the published tables. Finally, a small temperature difference between the terminals connecting the thermocouple to the DaqBook will be reflected in the temperature measurement.
Considering these factors, an error of 0.3°C is better than the expected accuracy and acceptable in most applications. However, the need to maintain an ice bath and separate thermocouple for every thermocouple being used is not acceptable. The ice bath can be eliminated by knowing the temperature of the reference junction because the opposing thermoelectric voltage generated by the non-zero temperature of the reference junction can simply be added to the voltage reading.
The law of intermediate metals provides the final step in simplifying the thermocouple arrangement. In this case, the law of intermediate metals says that the constantan-copper-terminal portion of the setup in Figure 3A is electrically identical to the constantan-terminal portion in Figure 3B, as long as the constantan-copper and copper-terminal junctions in Figure 3A are all at the same temperature.
Consequently, by measuring the temperature of the terminals, we have
effectively measured the temperature of the reference junction, and we
can obtain an accurate temperature reading without actually having a reference
junction. A single thermocouple on the terminal block eliminates the need
for a reference junction on each thermocouple channel. This technique is
called cold-junction compensation.