Control Engineering

Link This | Email this | Blog This | Comments (1)

How does an RTD work?
September 8, 2008

A resistance temperature detector (RTD) exploits the fact that the resistance of most metals increases with increasing temperature. To first order, the resistance of an RTD is given by 

R = α(T-T0) + R0,

where T₀ is a baseline temperature, R₀ is the RTD’s resistance at that temperature, and α is the resistance temperature coefficient for the RTD metal. The most common type RTD is made from platinum with a baseline resistance of 100 Ω at 0 ºC. With this arrangement, the RTD has a resistance change of 0.3729 Ω/ºC.

Resistance temperature coefficient for various metals

Material	Element/Alloy	α (/°C)
Nickel	Element	0.005866
Iron	Element	0.005671
Molybdenum	Element	0.004579
Tungsten	Element	0.004403
Aluminum	Element	0.004308
Copper	Element	0.004041
Silver	Element	0.003819
Platinum	Element	0.003729
Gold	Element	0.003715
Zinc	Element	0.003847
Steel*	Alloy	0.003
Nichrome	Alloy	0.00017
Nichrome V	Alloy	0.00013
Manganin	Alloy	+/- 0.000015
Constantan	Alloy	-0.000074

* = Steel alloy at 99.5 percent iron, 0.5 percent carbon

Physically, an RTD consists of a long, thin conductor, which may be an actual wire or a thin-film deposition on an insulator, wrapped into a compact package. The conductor is long enough and thin enough to have significant resistance compared to the lead wires connecting it to the measurement circuit.

Beside relatively minor nonlinearity (down about 4 orders of magnitude for platinum), RTD errors can arise from meter loading, self heating, lead resistance, Seebeck effect, and contamination (corrosion).

Contamination and corrosion affect the RTD element’s overall resistance. Corrosion, for example, converts metal from the RTD conductor from its pure form to a more complex compound, such as oxide. Generally these compounds have higher resistivity than the pure metal. As corrosion eats into the wire or film, the conductor’s cross sectional area is reduced, and its resistance goes up independent of temperature. RTD manufacturers minimize corrosion problems by using a noble metal (usually platinum) to make the RTD element.

Contamination can reduce RTD resistance by adding somewhat conductive material to the outside, providing an alternate conduction path. The contaminating material will have a different resistance temperature coefficient, and might also be highly nonlinear, further disturbing the measurement. RTD manufacturers fight contamination by passivating the RTD surface with an electrically insulating barrier layer, such as glass. 

Reasonable design can make the Seebeck effect negligible. Having the lead wires made of different material than the RTD element itself can cause the device to act like a thermocouple. Essentially, lead material (such as copper) has a junction with one end of the (likely platinum) RTD conductor, and another junction at the other end. If the two junctions are at different temperatures, a thermoelectric electromotive force (EMF) will arise and throw the measurements off.

The solution is to maintain good thermal contact (but not electrical contact) between the two junctions. Keeping them as close as possible to the same temperature minimizes Seebeck-effect errors.

Lead resistance introduces two potential problems. To zero order, any lead resistance represents an error signal. Furthermore, if the leads are made of material different from the RTD element, they will have a different temperature coefficient, so the combined coefficient will not be correct. In addition, the temperatures of the leads may not be the same as that of the RTD, leading to additional errors.

Self heating causes what is perhaps the most pernicious error because it cannot be wholly eliminated. It appears because the only practical way to measure a resistance is to pass a current through it and measure the Ohm’s Law voltage drop across it. Passing a current through any resistance element, however, dissipates electric energy as heat. That heat raises the element’s temperature. Since temperature is what we’re trying to measure, the act of making a measurement modifies what we’re trying to measure, and so introduces an error.

All one can do is maximize thermal contact with the environment (which temperature we ultimately want to know), and minimize the current passing through the RTD. Together these strategems minimize the difference between the ambient temperature and the RTD’s temperature.

Meter loading arises from the fact that some current has to be shunted around the RTD through the voltmeter to make a measurement. This used to be a serious problem when most voltage measurements were made using D’Arsonval analog meter movements. 

Meter loading is no longer a significant problem, as long as you pay attention to it. Digital volt meters and data acquisition analog inputs start with a high impedance (often tens of megohms) amplification stage. This high impedance couples with the relatively low RTD output signal to depress meter loading to the nanoampere range. It is important, however, to assure that the meter really does have a high input impedance. 

Simple circuit for making temperature measurements with an RTD using a milliamp current source for excitation, and a high impedance voltmeter for the measurement.

These requirements constrain RTDs to a specific physical plan. The RTD element itself is relatively small, with fine platinum wires supported by a non-conducting structure able to conduct heat rapidly between the platinum and the outside environment. The platinum wire ends are physically close together where they connect to robust copper leads that have low electrical resistance. The whole is built into a rod-shaped structure with the RTD element on one end and the signal-wire terminations on the other.

Two terminations lead to a constant, low-milliampere current source providing excitation. Separate leads go to a sensitive millivolt meter. The more sensitive the meter, the lower the excitation current needs to be. Low excitation current leads to low self heating. To avoid loading the circuit, use a high impedance voltmeter.

RTDs have several advantages for control applications. They are far more linear than thermistors, and can operate reliably at higher temperatures (up to about 850 ºC). RTDs also do not have the cold-junction compensation issues that plague thermocouple measurements, and are more accurate. A standard 100 Ω platinum RTD carrying a 1 mA excitation current will produce a 100 mV output signal that varies by 0.3%/ºC, while dissipating only 100 μW as heat. If the metering circuit has a 10 MΩ input impedance, the meter load is only 10 nA. That keeps the meter-loading error at the 10 ppm level.

For more information about measuring temperatures with RTDs, download the following articles from Control Engineering:
Temperature tutorial: Thermocouple vs. RTD vs. thermistor
Control Engineering's Weekly News Tech Tips of the Week

Posted by Charlie Masi on September 8, 2008 | Comments (1)