It has been some six years since I wrote about the inaccuracies and instabilities caused by misuse of thermocouple extension cable on temperature control systems. In 1989, the International Electrotechnical Commission (IEC) published IEC 584-3. This worldwide standard could replace the jungle of different cable color codes and do away with the consequent problems. Yet even now, some 18 years on, the standard is virtually invisible in North America.
Two examples illustrate the ongoing failures caused by color confusion.
Case 1: Furnace Burnout.A newly wired furnace suffered burnout of all six of its silicon carbide elements at a cost of some $9,000. The controllers were in good order and showed normal working temperature, though the furnace was clearly much hotter. The type R thermocouples were good and located properly, but somehow the controllers were being deceived.
Cause.The extension cables were Type J and should have been Type R. The red was connected as positive and white as negative; that is, the wires crossed at both ends. Red as positive sounds logical, being a common worldwide convention, but not on North American thermocouple wires. This is an understandable but costly error.
Likewise, it was an elusive problem, because for a while after the furnace came up to temperature, there was no problem. But, as the thermocouple head heated up in relation to the controller terminals, the Type J extension cable injected a large negative signal into the controller. This made it read low by as much as five times the thermocouple head-to-controller temperature difference, causing it to severely overheat the furnace. Knowledge of extension cables and their color codes could have prevented this damage.
Case 2: Injection Molding.An injection molding shop was making an excessive number of rejected parts. The controller showed the desired temperature, so it was not a suspect.
A technician took a two-hour flight and found the thermocouple extension cables to be of the correct alloy but reversed at both ends. The same mistake as Case 1; not so severe or destructive but still costly.
You have verified that you are using the right thermocouple for the temperature and atmosphere. It is located where it sees the work temperature and calibrated correctly. Your controller is accurate, calibrated for the thermocouple in use and is tuned for good control stability. Yet the process appears to overheat gradually over the first few hours -- or minutes -- after startup. This is more likely on an imported or newly installed process or one where the thermocouple has been rewired or worked on. The wire colors could be unusual. The wires could be crossed at both ends. A single crossover would be obvious, sending the indication downscale as the temperature builds up. Your fast action is called for here.
How the Controller Receives Signals from the Process.The circuit in figure 1 shows the signal route from the hot tip of the thermocouple to the temperature controller. (Celcius is used in this example but the same principle could be demonstrated in Fahrenheit.)
There are three contributions to the millivolt signal that the controller receives and acts upon, totaling 200°C (360°F) worth in this case.
- Those generated between the hot junction and head of the thermocouple. In this case, this accounts for 180°C (324°F), being proportional to temperature difference.
- Those generated by the extension cable between the thermocouple head and the controller. At this time, it is 0°C (0°F) because there is no temperature difference yet.
- Those generated by the controller, representing its own (ambient) temperature, which is 20°C (36°F) in this example. It has to contribute this so as not to be short by an amount equal to its own ambient temperature. This is called cold junction compensation.
After startup, the outside of the process begins to warm up. Suppose the temperature of the thermocouple head rises from 20°C, where it began, to 40°C (72°F). The thermocouple now delivers 160°C (288°F) of signal, and the correct extension cable will generate the 20°C, representing the 20°C end-to-end difference. The controller will still see 200°C (360°F).
Trap 1. You Use Copper Instead of Thermocouple Extension Cable.The copper fails to generate the 20°C of signal that the proper extension cable contributes in the above example, and the controller faces the prospect of seeing 180°C (324°F) instead of the true 200°C (360°F). However, it will get to work as soon as any shortfall appears and will already be turning the process up towards 220°C (396°F) in order to be satisfied and reading 200°C (360°F). This will raise no suspicion because the controller will be indicating the desired temperature -- 200°C -- all the time that the temperature is slowly climbing above 200°C.
Ignoring thermocouple non-linearity, the temperature will settle out too high by an amount about equal to the head-to-controller temperature difference.
I’ll have more on traps next month.