Selecting Sensors to Probe Cold Temperatures
Choosing a temperature-sensing technology for measuring lower temperatures has different considerations than higher temperature applications. How should you make your choice?
An engineer trying to select the best temperature-sensing technology generally has two choices: a thermocouple (TC) or a resistance temperature detector (RTD). There are many resources to help with such a selection — provided the application involves moving into higher temperature ranges — because most situations involve some sort of heat source.
But, what if the application is going the other direction? Do the same guidelines apply when moving into areas below room temperature or below the freezing point of water? Looking at each of these technologies, individually, with reference to temperatures below 32°F (0°C) presents a different picture than going to the hotter side.
Using Thermocouples for Low Temperature Applications
The basic thermocouple is the oldest electronic temperature-measuring approach and is still the most widely used. It can be used to measure cold temperatures, but its inherent characteristics should be kept in mind.
If a piece of wire is hotter at one end than the other, heat will flow from hot to cold. Electricity also will flow if it has some means to complete the circuit. (A temperature gradient also produces an electromotive force.) The voltage created is proportional to the temperature differential, but different metals can have different characteristics. As a result, some create a higher voltage for the same temperature differential. If two wires with different characteristics are connected, the junction put into a heat source, and if the other ends are connected to a voltmeter, the side with the higher voltage constant will create a higher voltage (figure 1a).
This approach works, but it only shows the temperature difference between the two ends of the wires. It does not provide a specific temperature reading by itself. High school physics classes demonstrate how a thermocouple works by inserting a second junction into a second measuring point (figure 1b). Such experiments usually use ice water as the second measuring point because it is at a more-or-less constant temperature of 32°F (0°C).
It is possible to determine the temperature at the hot end (the sensing junction) because the known temperature at the reference junction is provided by the ice water, hence the term, reference junction. This reference junction value can be any temperature so long as it is known.
Because having glasses of ice water in most locations is not practical, a second temperature sensor is used in industrial applications. This sensor can be an RTD or thermistor to provide the reference point.
Now, returning to our physics class, imagine inserting the sensing junction into a block of dry ice so it is now colder than the reference junction in the ice water (figure 1c). What happens? The voltage polarity is reversed because the reference junction is now the warmest spot, reversing the heat and electrical flow. A look at a typical thermocouple table will show that voltages go negative at 32°F (0°C). This is not an electrical characteristic so much as a convention relating back to the traditional ice water reference junction.
The greater the temperature difference between the sensing junction and reference junction, the higher the voltage generated. As a practical matter, the worst operating point for a thermocouple is when the two junctions are close to the same temperature. This is because the voltage will be zero, or nearly so. Automation systems do not like situations where a value is zero because it is difficult to differentiate between a correct value and an open circuit. The system will need to have the ability to verify continuity in such a case.
Moreover, when the temperature difference is relatively low, the voltage is very small and subject to problems from electromagnetic interference (EMI) inducing voltage in the wires. For example, a technician triggering a walkie-talkie near a thermocouple line can create a voltage spike because it acts as an antenna.
These problems should not eliminate the possibility of using a thermocouple as a low temperature sensor, provided that the concerns have been addressed appropriately. Selecting the most appropriate thermocouple type is critical. Some, such as Type B, simply are not suitable for temperatures below 32°F (0°C). Types E, K, N and T have ranges down to -454°F (-270°C), with Type T especially popular for cryogenic applications.
Once below -148°F (-100°C), most thermocouples begin to lose linearity (figure 2). This is not necessarily a problem because it is a known characteristic and can be corrected. Not all temperature transmitters or controllers are set up to work at the low end, however, so users must make sure that any device used in these applications has the required capabilities.
Using Resistance Temperature Detectors for Low Temperature Applications
RTDs have replaced thermocouples in many applications because they provide a direct temperature reading without the need for a reference junction. The technology takes advantage of the characteristic of some materials to change electrical resistance in response to changes in temperature. Nickel, copper and platinum all have this characteristic, but platinum has the widest range, so it is used the most. While RTDs do not have the high temperature capabilities of some thermocouple types, they are comfortable working in subzero applications.
The most common type of RTD is the 100 Ω platinum, which is linear over its useful range, changing by 0.39 Ω per °C. Its reference resistance is 100 Ω at 32°F (0°C), so the theoretical lowest limit is -428°F (-256°C), where its resistance reaches zero; however, it does not make sense to run to the very limit.
When moving down into the sub -148°F (-100°C) world, it is not the type of sensor as much as the construction that becomes critical. Industrial RTDs can have the actual bit of platinum configured as thin wire wrapped in a coil, or deposited as a thin-film element on a board (figure 3). The coiled-wire types, where the wire is either wrapped around a mandrel or encased in a ceramic insulator, can usually withstand -400°F (-240°C). The thin-film configuration — with its multiple layers —tends to bottom out at -58°F (-50°C).
RTDs have accuracy classes, specifying the tolerance range for each. The sophistication of the application will determine which type needs to be deployed. While RTDs have a high level of linearity, this characteristic still breaks down somewhat below -148°F (-100°C). Because the loss of linearity makes them harder to characterize, higher grade RTDs do not claim as low a range as lower classes. A Class AA sensor will not be recommended below -58°F (-50°C); by contrast, Class A and Class B sensors bottom out at -148°F (-100°C) and -320°F (-196°C), respectively (figure 4), but accuracy will be sacrificed somewhat at the lower levels of this range. The -320°F (-196°C) value is based upon the boiling point of liquid nitrogen, which is often used for low temperature calibration. Even though some RTDs can operate below -320°F (-196°C), such applications are rare, and a user can only hope for Class B accuracy at best.
Working in Cold Process Environments
Whether working with a thermocouple or RTD, the signal coming from the sensor is going to be low in magnitude and thus not very robust. This leaves it vulnerable to degradation if sent over a long distance. For thermocouples, any mismatch with extension wires, corrosion or simply poor terminations will be detrimental to accuracy. If the input channel at the automation system receiving the signal cannot handle the low value, the value sent will be unusable.
RTDs are somewhat easier to work with, but at reduced temperatures, the resistance becomes low. This makes it harder to measure accurately when other installation-related interference comes into play.
Installing a temperature transmitter as close as possible to the sensor can reduce these problems. The transmitter can convert the signal from a few millivolts to a more robust 4 to 20 mA current loop or a digital signal such as Foundation Fieldbus, either capable of being transmitted over long distances without the need for special cabling. Wireless transmitters can use a protocol such as WirelessHART, allowing new devices to be deployed without cabling.
Sophisticated transmitters (figure 5) also can perform diagnostic routines capable of verifying sensor performance, and they can send this information to the automation system using HART, Foundation Fieldbus, Profibus or wireless protocol. Any increase in signal noise or sensor-loop resistance caused by wiring degradation or sensor deterioration can be detected before an outright failure. If there is a break, diagnostics can spot the problem immediately and provide a warning to operators.
In conclusion, working with low temperature applications — with the exception of extreme cryogenic applications, which are not covered in this article — does not require specialized sensors. The ability to provide a reliable reading across a wide range can help in applications such as specialized heat treatment of tool steels where the process can involve both low temperatures to hold the workpieces around -238°F (-150°C), followed by a long ramping over several days to 752°F (400°C). All this often happens in the same oven, so having devices able to perform across the entire temperature range is advantageous.
Such applications can perform with great accuracy and reliability, provided a user takes appropriate care in device selection and installation.