A closer look at RTD technology

An RTD is constructed by winding a length of pure platinum (or other metal) wire on a small ceramic tube, inserted at the point of measurement. The winding is done tightly, but in a format that causes no physical stress under temperature variations. The thickness and length of wire determines the total resistance. The standard is 100 ? at 0°C, though a more sensitive RTD may be up to 1000 ? or more, to get a larger signal.

When an RTD is connected to a signal-conditioner or controller, it is connected as one arm of a bridge, through which an excitation current passes to develop a voltage signal that varies with temperature. The maximum excitation current is determined by the self-heating within the RTD, which will cause measurement inaccuracy; this limits the maximum signal for a required measurement temperature range-i.e., simply increasing the excitation does not yield a greater signal. To produce a higher-level signal for indication, recording, alarms, or controls, a separate signal conditioner is needed. Some PLC and PC I/O cards allow direct RTD connection with integrated signal conditioning.

Please note that RTD connection leads themselves cause errors due to the resistance of the wires and the change in that resistance, especially when very long lead-lengths are used. This is minimized or eliminated through the use of three-wire and four-wire RTD connections (Fig. 1).

The change in resistance with temperature for all RTDs is nonlinear. For accurate conversion to temperature, a good signal conditioner or two-wire transmitter includes some form of linearizing, usually a first-approximation analog feedback technique or, if the instrument is digital, a segmented look-up table that converts the input voltage reading directly to temperature. A good signal conditioner, controller, or RTD I/O card can provide temperature measurement accuracy to a fraction of one degree over a fairly wide span of measurement, comparable with the accuracy that can be attained with a good platinum RTD.

Whatever the material and construction, an RTD is still relatively bulky, and therefore cannot measure temperature at a point. On the other hand, this technology provides an excellent means of measuring the average temperature of a surface by spreading the resistance wire over the surface.

The use of a platinum RTD may be preferred when a temperature measurement accuracy of better than 1°F or 1°C is required. However, for contact measurements of temperature above about 800°C (the maximum temperature at which platinum RTDs can be used), a thermocouple is the primary choice.

Thermocouple temperature measurement

Since the ambient temperature changes in the field or control room where the measuring instrument is located (typically the cold junction), the actual millivolt signal generated by the T/C fluctuates accordingly, causing a "cold-junction error." To offset this error, the temperature at the instrument terminals is measured and a compensating signal, the cold-junction compensation (CJC), is inserted (Fig. 2). The overall measurement accuracy depends on the accuracy of cold-junction compensation with varying ambient temperature over the normal instrument operating temperature range.

Problems arise because typical electronic instruments (for example, signal conditioners, recorders, displays, and controllers) cannot really measure the actual cold-junction temperature; they simply try to get as close as physically possible-but not close enough to eliminate inaccuracies of at least a few degrees. The use of terminal-blocks and plug-in connectors typically exacerbates the problem-because the actual thermocouple cold junction becomes more physically separated from the cold junction sensor. The cumulative effect of ambient temperature and other variations may be five to ten degrees, which can be significant.

Waiting, warm-up prolong calibration procedures

Conversion to direct temperature readout

All T/Cs have somewhat nonlinear characteristics (i.e., in terms of millivolt output relative to temperature) and a direct readout of temperature will provide accuracy only within about 20-30°C, depending on the temperature span. To read temperature over the complete span of the thermocouple, the measuring instrument (Fig. 3) requires linearization. With a conventional analog instrument, the transfer characteristic is linearized in perhaps five to ten segments, providing improvements to within a few degrees. Digital signal conditioners use a look-up table that can provide direct digital readings with an accuracy of a fraction of 1 °F or °C. Multipoint I/O cards for PLCs or PCs typically condition and linearize all inputs together, and are usually designed for just one type of T/C. General-purpose signal conditioners are programmed, either manually (by a DIP switch) or via smart PC software, to allow connection to virtually any type of T/C.

Isolation

For multipoint measurement, most PLCs and dataloggers do not provide isolation between one sensor input and another (they typically use just one common power supply). Therefore, separate four-way isolating signal conditioners (input/out put/ power/ ground) can save a lot of trouble by eliminating the possibility of future, inadvertent grounding problems.

Which should you use-RTD or thermocouple?

One solution to cold-junction error

In many applications, traceable calibrations of equipment are required at least annually, involving perhaps several thousand instruments. The calibration of thermocouple-input instrumentation can be quite time-consuming if each instrument requires several minutes to warm up or stabilize. This highlights the importance of thermocouple-input design that provides good accuracy with minimal warm-up time after the system is turned on, as well as good repeat accuracy for extended periods.

Instant Accuracy™ cold junction temperature measurement (Fig. 4), offered by Action Instruments, provides a way to minimize thermocouple measurement disadvantages. This technique uses two junction temperature sensors mounted linearly down a focal plane of the instrument printed circuit board, aimed at the terminals. By knowing or characterizing the thermal conductivity of this focal plane and measuring the differential temperature between the two sensors, extremely accurate cold junction temperature measurement is achieved.

As shown in Fig. 2, standard cold-junction compensation operates with one measurement-CJC-A, while Instant Accuracy operates with two-CJC-A and CJC-B, and calculates the temperature at the actual cold-junction.

Here's how it works. You cannot measure the exact temperature of the terminals, but you can predict what that measurement is by measuring two points along the path of the temperature variation and calculating the temperature of the third point-the cold junction. It's almost like the cold junction is not measurable-on the other side of a wall-but the measurement is calculated by extrapolation of the other two measurements. Of course, the calculations may be different for different physical geometry, but in any instrument with fixed geometry, the calculation remains the same and the overall thermocouple measurement accuracy is improved considerably.