Author : Bill Earlie
14 November 2009
Platinum resistance thermometers are much more accurate than thermocouples but they have some drawbacks, notably a more limited temperature range, a higher thermal mass and usually are more expensive.
These resistance thermometers have a linear and repeatable resistance against temperature: The two common types in use are Pt100 which has a resistance of 100 ohms at 0°C and Pt25 which has a resistance of 25 ohms at 0°C.
Platinum is used because it has a stable temperature coefficient and being a noble metal is not very susceptible to contamination. Pt100 (PRT) is the most commonly used and has a temperature coefficient of (alpha) = 0.00385 (European standard) which corresponds to an average resistance change, over the temperature range 0 to 100°C, of 0.385 ohms per °C.
Both the absolute resistance value and the change in resistance per °C are both relatively small and give rise to measurement problems, especially when the resistance of the connection leads are taken into consideration.
There are other standards also in use – for instance, the U.S. standard for pt100 has an alpha of 0.00392.
Types of measurement
When measuring the resistance of a Pt100 a test current is forced through the component and the test meter measures the voltage at its terminals. The meter then calculates and displays the resulting resistance; this is known as a two-wire measurement.
It should be noted that the meter measures the voltage at its terminals and not across the component.
As a result of this, the voltage drop across the connection leads is also included in theresistance calculation. Good quality test leads will have a resistance of approximately 0.02ohms per meter. In addition to the resistance of the leads, the resistance of the lead connection will also be included in the measurement and this can be as high as, or even higher in value, than the leads themselves.
The two-wire measurement is not recommended.
A three-wire connection is quite common in industrial applications and will eliminate most of the effect of the lead resistance on the measured value.
Care must be taken to ensure that all three wires are of equal resistance but this is almost impossible to achieve in practice. The three-wire method will not deliver the same degree of accuracy as a true four wire system, but is better than two wires.
Four wire measurements are the most accurate configuration. Two-wires are used to pass a constant current through the Pt100 and the volt drop across the unit is then measured. The impedance of the voltage measurement circuit is high and as a consequence only a very small current flows in the potential circuit, which for practical purposes can be ignored. The result is that the measurement lead resistance can also be ignored.
When measuring PRTs the measurement current used by most temperature indicators is either DC or low frequency AC. If AC is used, then care in selecting a non inductive sensor is essential as the measurement will be the impedance of the sensor rather than its true DC resistance.
There may also be some differences in the temperature measurement between sensors from different manufacturers, as their construction technique may differ, resulting in slightly different impedance values. This AC measurement does, however, eliminate any thermal emf errors that may arise.
Using DC current measurement, the true resistance value is measured and used to calculate the corresponding temperature. In this instance impedance errors are not a problem, but errors due to thermal emf must be considered.
The best method of countering any thermal emf is to measure the sensor resistances with current flowing in one direction, then reverse the current and take a second measurement.
The average of these two measurements is the true resistance without any thermal emf. This is often called the switched DC method and is selectable on the Cropico thermometers.
To obtain the best measurement results, the resistance of the Pt100 sensor must be measured with a high degree of accuracy. A temperature change of 1°C will correspond to a resistance change of 0.385 ohms so to obtain a measuring accuracy of 0.01°C (10mK) the resistance must be measured to ±0.0385 ohms.
For example: for a temperature of 100°C the resistance value will be 138.5 ohms. To measure this with an accuracy of ±0.01°C, this resistance value must be measured to ±0.0385 ohms, which is equal to ±0.028%.
If a current of 1 mA is used as the measuring current to measure 138.5 ohms (100°C), then a voltage of 138.5mV will need to be measured to ±138.5 microVolts, and to measure the temperature change of 0.01°C, a change of 3.85 microVolts must be measured.
So it’s clear that a small error in the voltage sensing measurement can create wide scale temperature measurement errors.
Self-heating
To measure the resistance of the temperature detector, a current must be passed through the device — typically a current 1 mA to 5 mA. A source current of 1 mA flowing through the 100 ohm resistance will generate 100 microWatts of heat. If the sensor element is unable to dissipate this heat it will indicate an artificially high temperature.
This effect can be minimised by using a large sensor element that is in good thermal contact with its measurement environment, while allowing sufficient time for the temperature to stabilise.
An alternative is to use a short measurement pulse of current thus minimising the heating effect. A good thermometer may be configured to measure with either a continuous or a short current pulse ensuring that the best possible measurement is made.
Design and construction considerations
While platinum detectors are very stable over time, the design and manufacturing process can adversely affect these properties. During manufacture the detectors need to be heat treated to homogenise the crystal structure and remove any oxides that may have formed.
The sensor needs to be supported in a stress-free manner and the finished assembly handled without causing any impact shocks or vibration. Cycling the sensor between a high and low temperature will also increase errors. A typical drift rate for a Pt100 detector is 0.05°C per year. A high quality detector will exhibit lower drift of approximately 0.005°C to 0.01°C providing the detector is not mechanically stressed and the temperature range is limited.
The Pt100 detectors are normally constructed by bifilar winding the platinum wire onto a small bobbin. Although the detector assemblies can be quite small in size they still have a thermal mass which takes time to warm up and reach thermal equilibrium, and consequently they have a longer response time than thermocouples. The detectors are usually housed in a stainless steel sheath, which again increases the response time.
When measuring temperatures, the immersion depth is also important as heat will be conducted up the stem of the sensor giving rise to errors. The manufacturer should be consulted regarding the minimum immersion depth.
Pt100 detector can also be constructed on a flat substrate this reduces the size and can be more suitable for some applications.
Measurement errors and methods
The main sources of measurement errors are the use of two-wire and three-wire sensors, thermal emf in non-switch DC measurement systems, inductive sensors in AC measurement systems, self-heating of the sensor due to the measurement current flowing through the detector winding and insufficient stabilisation time.
The measurement errors can be minimised and eliminated by choosing a good quality precision thermometer. The accuracy can be further improved by sensor calibration and the Callendar van Dusen coefficients produced from this calibration entered into the thermometer thus modifying the standard calibration curve to fit the detector characteristics.
While the platinum thermometer is one of the most linear temperature detectors it is still necessary to linearise the measured signal.
According to the IEC standard IEC751 the non linearity can be expressed as
Rt = R0[1 + At + Bt2 + C(t - 100)t3]
Where C is only applicable when t is equal to, or less than 0° C. The standard coefficients for A, B, and C are stated in the IEC standard but may also be calculated for each individual sensor by measuring its resistance values against set temperature standards.
The Callendar van Dusen method for determining these coefficients is commonly used and based on measuring four known temperatures:
* R0 at T0 = 0°C - the triple point of water * R100 at t100 = 100°C - the boiling point of water * Rh at th = high temperature (e.g. the freezing point of zinc 419.53°C) * R1 at t1 = a low temperature (e.g. the boiling point of oxygen -182.96°C).
Using the Callendar van Dusen method, the coefficients will be calculated for you by the laboratory calibrating your sensor, so it is not necessary to describe the calculations here.
For the highest accuracy, special glass-sheathed standard PRTs, usually of 25 ohms at 0 °C, are calibrated at the fixed points of the International Temperature Scale 1990. ITS-90 specifies equations to relate the resistance to temperature and, using these uncertainties can be achieved of 0.001 °C or better. Standard PRTs can be used from temperatures as low as - 259 °C up to 660 ºC, or even 962 ºC, with some increase in uncertainty and loss of reproducibility.
Cropico precision temperature indicators are able to measure with both Pt100 and Pt25 PRT sensors and form part of a comprehensive range of precision measurement instruments available from Cropico Ltd, part of the Seaward Group. More at www.cropico.co.uk
Author: Bill Earlie, Cropico
Caption for last diagram
Colour codes for Pt100: Two, three and four-wire extension leads as per IEC 60751. Note that in practice the industrial grade sensor will have current and potential leads R1, R2 and R3, R4 connected at the same point on the sensor and therefore interchangeable when connecting to the measuring device.
Print this page | E-mail this page
This isn't a paywall. It's a Freewall. We don't want to get in the way of what you came here for, so this will only take a few seconds.
Register Now