2.5.11 Thermocouples

164 Measurement

produce a consistent emf change. Therefore, there must be individual thermocouple calibration tables for each type of thermocouple.

If you look at a thermocouple calibration table, you will see that it has a reference junction at 0°C. This reference point is used because of an interesting complication that arises when a thermocouple is hooked up to a voltmeter. To explain this phenomenon, first look at one specific type of thermocouple, a type T (cop- per-constantan design). Also, assume that the wires in the voltmeter are all copper. Once the thermocouple is hooked to the voltmeter, we end up having a total of three junctions (see Fig. 2.31):

Junction 1. The original thermocouple junction of copper and constantan.

Junction 2. The connection of the constantan thermocouple wire to the copper wire of the volt meter.

Junction 3. The connection of the copper thermocouple wire to the copper wire of the volt meter.

We are hoping to find the temperature of Junction 1, but it seems we now have two more junctions with which to be concerned. Fortunately, Junction 3 can be ignored because this connection is of similar, not dissimilar, metals. Therefore, no emf will be produced at this junction. Junction 2 presents a problem because it will produce an emf, but of an unknown temperature; this means that we have two unknowns in the same circuit, and we are unable to differentiate between the two. Any voltmeter reading taken at this point will be proportional to the temperature difference between the first and third junction.

If we take Junction 2 and place it in a known temperature, we can eliminate one of the two unknowns. This procedure is done by placing Junction 2 (now known as the reference junction) in an ice bath of 0°C (known as the reference temperature). Because the temperature (voltmeter) reading is based on an ice-bath reference temperature, the recording temperature is referenced to 0°C (see Fig. 2.32).

Junction 1

Junction 2 (Reference temperature)

IC6 bath

Fig. 2.32 A thermocouple using an ice bath for a reference temperature.This illustration is from The Temperature Handbook©1989 by Omega Engineering, Inc. All rights reserved. Reproduced with the permission of Omega Engineering, Inc., Stamford, CT 06907

References for voltmeter/thermocouple readings (found in such books as the CRC Handbook) will usually indicate that they are referenced to 0°C. Because Junctions 3 and 4 are of similar metals, they have no effect on the emf.

This example is limited because it is representative only of thermocouples with copper leads. In all other thermocouples, there are likely to be four dissimilar metal junctions and therefore up to four Seebeck effects. However, by extending the copper wires from the voltmeter and attaching them to an isothermal block, placing the same type of wire from one of the thermocouple leads to the part between the icebath and the isothermal block, and then attaching the thermocouple wires to the isothermal block (to eliminate thermal differences at these two junctions), it is possible to cancel out all but the desired Seebeck effect (see Fig. 2.33).

It is obvious that the use of an ice bath is inconvenient and impractical. Maintaining a constant 0.0°C temperature can be difficult, and there always is the possibility of the icebath tipping over. Fortunately there are two convenient techniques to circumvent the need for an actual ice bath for thermocouple measurements.

One approach requires measuring the equivalent voltage of the ice-bath reference junction and having a computer compensate for the equivalent effect. This technique, called software compensation, is the most robust and easiest to use. However, depending on the equipment used, the lag time involved for the compensation may be unacceptable.

Alternatively, it is possible to have hardware compensation by inserting an electric current (typically from a battery) to provide a voltage which electronically offsets the reference junction. This approach is called an electronic ice-point reference. This method has the simplicity of an ice bath, but is far more convenient to

Fig. 2.33 An iron-constantan thermocouple using an isothermal block and an ice bath. This illustration is from The Temperature Handbook©! 989 by Omega Engineering, Inc. All rights reserved. Reproduced with the permission of Omega Engineering, Inc., Stamford, CT 06907

Ironically, the sophistication of modern controllers typically makes most of the concerns about how temperature is determined irrelevant. The greatest concern for the user is to select the right type and size of thermocouple for the specific job and environment, then to select the proper controller for that type of thermocouple.

Although there are many overlapping temperature ranges among various thermocouple types, all types of thermocouples perform better for some jobs than others. For example, Type K is significantly less expensive than Type R, although both thermocouples can read into the 1000°C range. This similarity would lead one to believe that either choice is adequate, however Type K is preferred because of its lower cost. If all you need are occasional readings of such temperatures, you could get by with Type K. However, if you expect to do repeated and constant cycling up to temperatures in the 1000°C range, Type K would soon fail and would need to be replaced. Unfortunately, you are not likely to notice failure developing until something has obviously gone wrong. Because the potential costs of failure are likely to be greater than the additional cost of the more expensive thermocouple, it may be more economical to obtain the more expensive type R at the outset.

Sample size should also be considered when selecting thermocouple size. If a large thermocouple (with a high heat capacitance) is used on a small sample (with a low heat capacitance), the sample's temperature could be changed. As stated before, the act of taking an object's temperature can change its temperature.