Introduction

When making temperature measurements using a standalone digital multimeter (DMM) or a DMM integrated into a data acquisition system, common errors will occur. We focus here on helping users eliminate potential measurement errors and to achieve the greatest accuracy with a DMM.

A temperature transducer measurement is typically either a voltage or resistance measurement converted to an equivalent temperature by software conversion routines. Therefore, the types of errors that are typically associated with voltage and resistance measurements (such as loading errors, ground-loop errors, etc.) must also be taken into account for temperature measurements. (For more information on the errors associated with voltage and resistance measurements, refer to Sources of Measurement Errors in Digital Multimeters.)

The largest source of error in temperature measurements is generally the transducer itself. Descriptions of the three popular types of temperature transducers and their advantages and disadvantages are discussed.

Types of Temperature Transducers

Specific measurement requirements will help to determine which temperature transducer type to use. Each has an associated temperature range, accuracy, and cost. The table, below, summarizes some typical specifications for three popular transducers, information that is useful when selecting the transducer for a particular application. The transducer manufacturers can provide exact specifications for a particular transducer.

Parameter	Thermocouple	RTD	Thermistor
Temperature Range	-210ºC to 1820ºC	-200ºC to 850ºC	-80ºC to 150ºC
Measurement Type	Voltage	2- or 4-Wire Resistance	2- or 4-Wire Resistance
Transducer Sensitivity	6 µV/ºC to 60 µV/ºC	= Ro x 0.004ºC	= 400 W /ºC
Probe Accuracy	0.5ºC to 5ºC	0.01ºC to 0.1ºC	0.1ºC to 1ºC
Cost (U.S. Dollars)	$1 / foot	$20 to $100 each	$10 to $100 each
Durability	Rugged	Fragile	Fragile

RTD Measurements

An RTD (resistive temperature detector) is constructed of a metal (typically platinum) that changes resistance with a temperature change in a precise, and known, way. An RTD has the highest stability of the temperature transducers mentioned, and its output is also very linear. This makes an RTD a good choice for high-accuracy, long-term measurements. The resistance of an RTD is nominal at 0° C and is referred to as R_o.

RTDs can be measured using a 2-wire or 4-wire resistance measurement method. The 4-wire resistance method (with offset compensation) provides the most accurate way to measure small resistances. Note, however, that connection lead resistance is automatically removed using the 4-wire method.

Thermistor Measurements

A thermistor is constructed of material whose resistance changes non-linearly with changes in temperature. Thermistors have a higher sensitivity than thermocouples or RTDs. This makes the thermistor a good choice when measuring very small changes in temperature. However, thermistors are very non-linear, especially at high temperatures, and function best below 100° C. Because of their high resistance, thermistors can be measured using a 2-wire resistance measurement method.

Thermocouple Measurements

A thermocouple converts temperature to an associated voltage level. When two wires composed of dissimilar metals are joined, a voltage is generated which is a function of the junction temperature and the types of metals in the thermocouple wire. Since the temperature characteristics of many dissimilar metals are well known, a conversion from the voltage generated to the temperature of the junction can be made. For example, a voltage measurement of a T-type thermocouple (made of copper and constantan wire) might look like this:

Fig. 1

Notice, however, that the connections made between the thermocouple wire and the DMM make a second, unwanted thermocouple where the constantan (C) lead connects to the DMM's copper (Cu) input terminal. The voltage generated by this second thermocouple affects the voltage measurement of the T-type thermocouple.

If the temperature of the thermocouple created at J2 (the LO input terminal) is known, the temperature of the T-type thermocouple can be calculated. One way to do this is to connect two T-type thermocouples together to create only copper-to-copper connections at the DMM's input terminals, and to hold the second thermocouple at a known temperature.

An ice bath can be used to create a known reference temperature (0° C) (see Fig. 2.) Once the reference temperature and thermocouple type are known, the temperature of the measurement thermocouple can be calculated.

Fig. 2

The T-type thermocouple is a unique case since one of the conductors (copper) is the same metal as the DMM's input terminals. If another type of thermocouple is used, two additional thermocouples are created. For example, take a look at the connections (see Fig. 3) with a J-type thermocouple (iron and constantan):

Fig. 3

Two additional thermocouples have been created where the iron (Fe) lead connects to the DMM's copper (Cu) input terminals. Since these two junctions will generate opposing voltages, their effect will be to cancel each other. However, if the input terminals are not at the same temperature, a measurement error will result. To make a more accurate measurement, extend the copper test leads of the DMM closer to the measurement, and hold the connections to the thermocouple at the same temperature (see Fig. 4.)

Fig. 4

This circuit will give accurate temperature measurements. However, it is not very convenient to make two thermocouple connections and keep all connections at a known temperature. The Law of Intermediate Metals eliminates the need for the extra connection. This empirical law states that a third metal (iron, Fe, in this example) inserted between two dissimilar metals will have no effect upon the output voltage, provided the junctions formed are at the same temperature. Removing the reference thermocouple (see Fig. 5) makes the connections much easier.

Fig. 5

This circuit is the best solution for accurate thermocouple connections.

In some measurement situations, however, it would be nice to remove the need for an ice bath (or any other fixed external reference). To do this, an isothermal block can be used to make the connections (see Fig. 6.) An isothermal block is an electrical insulator, but a good heat conductor. The additional thermocouples created at J1 and J2 are now held at the same temperature by the isothermal block when, once known, allows accurate temperature measurements to be made. A temperature sensor is mounted to the isothermal block to measure its temperature.

Fig. 6

Sources of Errors in Thermocouple Measurements

Following are descriptions of the common sources of errors associated with thermocouple measurements.

Reference Junction Error
A thermocouple is typically formed by welding or soldering two wires together to make the junction. Soldering introduces a third metal into the junction. Provided that both sides of the thermocouple are at the same temperature, the third metal has little effect.

Commercial thermocouples are welded using a capacitive-discharge technique. This technique is used to prevent overheating of the thermocouple wire near the junction, and to prevent the diffusion of the welding gas and atmosphere into the thermocouple wire.

A poor weld or bad solder connection can also cause errors in a thermocouple measurement. Open thermocouple junctions can be detected by checking the resistance of the thermocouple. A resistance measurement of more than 5 kW typically indicates a defective thermocouple.

The HP 34970A Data Acquisition/Switch Unit from the Hewlett-Packard Company offers a built-in, automatic thermocouple check feature. By enabling this feature, the instrument measures the channel resistance after each thermocouple measurement to ensure a proper connection. For more information on the HP 34970A, access the HP Basic Instruments web site at www.hp.com/go/bi.

HP 34970A Data Acquisition/Switch Unit

Diffusion Error
Diffusion in a thermocouple wire is the process of changing the type of alloy along the wire itself. Atmospheric particles can actually diffuse into the metal. These changes in the wire alloy introduce small voltage changes in the measurement. Diffusion is caused by exposure to high temperatures along the wire, or by physical stress to the wire such as stretch or vibration.

Temperature errors due to diffusion are hard to detect since the thermocouple will still respond to temperature changes and give nearly correct results. The diffusion effects are usually detected as a drift in the temperature measurements.

Replacing a thermocouple that exhibits a diffusion error may not always correct the problem. The extension wire and connections are all subject to the effects of diffusion. Examine the entire measurement path for signs of temperature extremes or physical stress. If possible, keep the temperature gradient across the extension wire to a minimum.

Shunt Impedance
High temperatures or corrosive atmospheres can result in degradation of the insulation used for thermocouple wire and extension. These breakdowns appear as a resistance in parallel with the thermocouple junction and are especially apparent in systems using a small gauge wire where the series resistance of the wire is high.

Shielding
Shielding reduces the effect of common-mode noise on a thermocouple measurement, noise that can be generated by sources such as power lines and electrical motors. The noise is coupled to the unshielded thermocouple wires through distributed capacitance. As the induced current flows to ground through the DMM, voltage errors are generated along the distributed resistance of the thermocouple wire. Adding a shield to the thermocouple wire will shunt the common-mode noise to ground and preserve the measurement (see Fig. 7.)

Fig. 7

Common-mode noise can dramatically affect the DMM. A typical thermocouple output is a few millivolts and a few millivolts of common-mode noise can overload the input to the DMM.

Conclusion

We have given you here an overview of the popular types of temperature transducers and described the advantages and disadvantages of each. In addition, we have described the common sources of errors associated with thermocouple measurements. When making temperature measurements, keep in mind that there are errors associated with the temperature transducer itself as well as errors related to wiring connections and measurement environment.

Biographical Information

David Heintz has been a technical writer with HP’s Electronic Measurements Division in Loveland, Colorado, since 1984. He has written customer documentation for numerous test and measurement instruments, and his work has been recognized by the Society for Technical Communication.

Scott Stever is an R&D section manager with HP’s Electronic Measurements Division in Loveland, Colorado. He has been with HP since 1979, and has held numerous positions within R&D.

		Click here to get your listing up.