ONLINE EXCLUSIVE: What are Your Thermocouples Really Telling You?
How TCs OperateTo understand how thermocouples operate, we first must understand what a TC does. A TC is like a small battery that produces a voltage that increases with temperature. In 1821, Thomas Seebeck discovered that when two dissimilar metals are joined they produced a voltage. This “Seebeck voltage” is useful because it changes with temperature and is repeatable. Because of this repeatability, we are able to develop accurate tables for various dissimilar metal combinations that make up thermocouples. The American National Standards Institute (ANSI) has established standard designations for some of these metal combinations, and the National Institute for Standards and Technology (NIST) has published tables of temperature and voltage. Other combinations are universally used but do not have the “official status,” and even more pairs have been reported in literature but are not being widely used.
Accuracy vs. RepeatabilityNext, it is important to understand the difference between accuracy and repeatability. Accuracy is the ability to “hit a target,” while repeatability is the ability to hit the same place each time, even if it is not the targeted point. Let’s say that a TC placed in boiling water gives a voltage that corresponds to 98.0?C on the NIST tables. It can be said that that TC is accurate to within 2.0?C. If the same TC is used to measure the same boiling water 100 times and always reads 98.0?C, the TC is repeatable without any range. However, if the readings vary between 97.8 and 98.3, it is repeatable within 0.5?C.
The first source of TC error is an accuracy issue that comes from the wire itself. This is mainly due to contamination of the metal. The standard type K TC has a maximum error of 2.2?C or .75% of the reading, whichever is largest. This means that with a reading of 200?C, the “real temperature” could be as high as 202.2 or as low as 197.8 for a particular wire set. This problem gets worse at higher temperatures. For example, at 850?C, the reading could be as high as 856.4 or as low as 743.6?C, while at 1200?C it could be 1209 or 1191?C.
In a recent experiment, six type K TCs were made from a single roll of wire and run together in a furnace at 850?C. Care was taken to ensure that the bi-metal junctions were all at the same temperature and that the readout device was in good condition. The repeatability of over 200 readings was within ±0.8?C with a standard deviation of 0.29.
The results of this experiment prove that all thermocouples have variability, regardless of how carefully they are made or installed. Manufacturers should therefore take that variability into account when evaluating their operating temperature.
Setting Special LimitsIn many cases standard error limits of accuracy are good enough, but what happens if you need to be more precise? An easy answer is to specify TCs with “special limits of error.” These special limits cut the error in half, though they typically add 15% to 20% to the cost. Some TC suppliers include a wire deviation statement with each shipment.
But while the special limit wire increases accuracy, it can still cause a problem. At 200?C, the maximum range of “special limits wire” is 201.1 to 198.9, while at 1200?C it is 1204.8 to 1195.2. Figure 1 shows the error bands (also known as “The Error Cone”) for type K TCs. Table 1 shows the limits of error for some of the other TC types.
If you want absolute accuracy or accuracy over a wide range, you will need a TC that is certified and can be traced to the National Bureau of Standards at multiple temperatures. Certification can cost between $75 and $100 for the first temperature and another $20 to $30 for each additional point. Certification or “special limits” reports do nothing to alter the reading of the TC. They simply document the offset at the measured temperatures.
Mutual CalibrationIf you want to compare the readings from two different TCs but are not interested in obtaining an accurate temperature, you can calibrate them to each other. Mutual calibration is done by making one TC a “de-facto standard” and comparing it with the other (or others) at various temperatures. This can be accomplished by running the TCs next to each other in the process, or in a special oven, and noting the difference at various temperatures. Assuming that the TCs see the same temperatures, the difference can then be used to establish “fudge factors” for equalization. The fudge factors are used to equalize the data from TCs in the same way that deviations are used to calibrate certified TCs.
Using a certified TC as the standard in the mutual calibration procedure allows you to use the fudge factors to obtain accurate and comparative temperatures. By numerically adding (or subtracting) the results from the individual TCs and the certified TC, you can determine the offset for each TC. This method detracts slightly from the accuracy of the second-generation TCs but allows you to retain the certified TC for subsequent checks. The retention of a standard is good practice whether a certified TC or the mutual calibration method is used. However, even the standard or certified TC will drift over multiple runs.
Age MattersThe age, number of cycles, and atmosphere a TC has seen can also affect its accuracy and repeatability. The error of a type K TC begins to change after just a few uses at high temperatures in hydrogen. This change is somewhat less in air but still exists.
To prove this theory, five of the six TCs that were previously used to test repeatability were used to profile an air furnace. The remaining TC was withheld as a reference. After 10 trials at 850?C, the five TCs used in the furnace read over 3.5?C higher than the one that was withheld.
It is interesting to note that the new range of the five TCs in this trial was 2.0?C, which is nearly 2 1/2 times the original range. Over longer periods of time, the insulation, wire and bead can degrade and give false readings. Poor insulation can allow a short to form, and this short adds additional bi-metal junctions (little batteries) to the circuit that change the overall voltage and result in false readings. The amount of time that you can reliably use a TC depends on your process parameters and accuracy requirements. Sometimes a TC will last for years, while other times it is good for only a few runs.
Connection ErrorsAnother source of error is the extension wire or other bi-metal junctions that become part of the circuit. Other bi-metal junctions are formed when a TC wire is connected to junction blocks or equipment. In some applications, the TC can be connected directly to a meter that compensates for internal junctions. However, when this type of connection is impractical, an extension wire is needed, and the use of extension wire adds more junctions and more potential for error.
This does not mean that all added junctions are bad. Known “reference junctions” are used to establish accurate thermocouple tables. The National Bureau of Standards (NBS) uses a 0?C ice point junction as the reference for its tables.
Equipment/Installation MistakesThe last source of error is due to equipment and installation mistakes. If you use the incorrect extension wire or connector, the TC reading will suffer. In fact, using any of the wrong components will add additional bi-metal junctions to the circuit and give false readings.
Using the wrong type of TC can also have disastrous results. For example, if a type K junction is used with equipment that was calibrated for type J, the results of a 10 millivolt (mV) TC will be read as 186?C when it is really 246?C. At 40 mV, the reading would be 714?C when it is actually 968?C. Mistakes like this will most certainly mess up a process and, in some cases, destroy equipment.
Table 2 shows temperature that corresponds to the mV output for a few types of TCs. By cross-referencing the temperatures in the table, you can see the results of mixing the wrong TC and equipment.
Controlling Your ProcessWhen evaluating your firing temperature, don’t forget to consider the variables that can affect the accuracy of your thermocouples. The metal combination used in a thermocouple; its limits of error; whether it has been calibrated; its age, number of cycles, and the atmosphere in which it is used; external junctions; and equipment and installation mistakes can all affect temperature readings. If obtaining a precise temperature reading is important to your operation, make sure you are using the right kind of thermocouples and that they are installed correctly with the minimum number of external connections. By understanding how thermocouples operate and where errors are likely to occur, you can gain better control over your firing or sintering process.
For Further Reading
- The Temperature Handbook, Omega Engineering, 2000.
- Temperature Measurement Handbook, Vol. XII, Thermocouples, RTDs and Accessories, NANMAC Corp.