How do you choose between different temperature controllers?

A temperature controller can be customized to meet the specific needs of a thermal control system.

Figure 1. Hysteresis provides an area of no action to prevent chatter or rapid on/off switching.
Temperature control utilizes a feedback loop that begins with a system's measured temperature. A temperature sensor - typically a thermocouple, RTD or thermistor - measures a process's real-time temperature and feeds the reading back to a controller. The controller compares the measured temperature to the setpoint temperature and actuates devices like heaters or valves to bring the temperature to the desired setpoint. The three most common methods of control are on/off, proportional and proportional-integral-derivative (PID).

On/Off Control. On/off is a simple, economical control method. For heating applications, the heater being controlled is either completely on (if the process temperature is below the setpoint) or completely off (if the process temperature is at or above the setpoint). On/off control does not account for how far the actual temperature is from the setpoint. Because the process temperature must cross the setpoint for the output to change, frequent cycling between on and off is common. Rapid switching is known as chatter.

Temperature controls utilize a feedback loop to control process temperature.
To help eliminate chatter, on/off controllers have a buffer zone or hysteresis (also referred to as dead band), which is an area of no action between on and off. Usually preconfigured into the controller, hysteresis is expressed as a percentage of the temperature range or a number of degrees (figure 1). Even with the hysteresis, on/off control experiences temperature overshoot and un-dershoot due to system inertia. Be-cause temperature never stabilizes at setpoint, on/off control is best suited for noncritical applications that experience slow temperature changes.

Proportional Control. When a system requires a higher level of accuracy or experiences frequent load changes, a proportional temperature controller should be selected. Multiple types of proportional control exist. Depending on the application, some types are more appropriate than others.

Figure 2. Manual reset eliminates droop and brings the process temperature to setpoint.
Time-proportioning control is appropriate when the device being actuated can only be completely on or completely off. Time-proportioning control prevents repeated cycling above and below setpoint by allowing the user to set a proportional band and cycle time. The proportional band usually is expressed as a percentage of the full temperature scale or a number of degrees. If the proportional band is too narrow, oscillation will occur around the setpoint. If the band is too wide, a longer stabilization time will result.

In heating applications employing time-proportioning control, the heater is on as long as the measured temperature is below the proportional band. Once the temperature enters the band, the heater is cycled on and off. As the process temperature approaches the setpoint, the heater is switched on for decreasing amounts of time and off for increasing amounts of time. At the midpoint of the proportional band, cycle time for both on and off is 50%. Timed cycling between on and off within the proportional band helps prevent setpoint overshoot.

True proportional control employs the same proportional band as time-proportioning control, but this method utilizes linear outputs such as current or voltage to vary the degree of on or off rather than varying the amount of time. For example, a solenoid valve that ac-cepts a 4 to 20 mA signal allows the controller to proportionally vary current supplied to the valve based on the de-viation from setpoint once the process temperature enters the proportional band. This situation allows the valve to be 0 to 100% on, or open, instead of completely on (open) or off (closed).

The primary problem with both time-proportioning and true proportional control is that stabilization usually occurs slightly above or below the actual setpoint. This phenomenon is known as offset or droop (figure 2). On many controllers, a manual reset is available to return the process temperature to setpoint. This manual adjustment must be made by an operator and may prove to be inconvenient. Proportional-integral (PI) control addresses this problem by incorporating an automatic reset to compensate for droop before it occurs (figure 3).

Figure 3. Proportional-integral control eliminates offset.
With PI control, deviation from setpoint is integrated over a set time period and added to the proportional signal to shift the proportional band toward the offset, returning the process temperature to setpoint. Reset only can occur if the process temperature is within the proportional band. This prevents large undershoots or overshoots during startup or load changes. PI control is suitable for systems with stable setpoints and varying loads, such as environmental changes. PI control eliminates offset, but requires a long stabilization time. Proportional-derivative (PD) control measures the derivative (rate) of deviation from setpoint and adds it to the proportional signal, shifting the proportional band to minimize overshoot (figure 4). PD control can be used to minimize initial overshoot and correct large system disturbances that require a prompt response such as batch introduction. PD control is the fastest system stabilizer, but it does not account for offset or droop.

Figure 4. Proportional-derivative control decreases initial overshoot, but offset remains.
For systems that frequently change loads and setpoints, a controller combining proportional-integral-derivative (PID) control will provide precise control. PID control anticipates necessary corrective actions using a linear combination of the deviation from setpoint, its integral (reset) and its derivative (rate) to modify the output function. PID control helps eliminate oscillation around the setpoint (figure 5) but requires tuning PID parameters. Tuning can be done manually, but it requires hours of trial and error and is not recommended for those who are not trained. Autotuning is a viable alternative to manual tuning. This feature allows the controller to calculate the control parameters based on a few on/off cycles. Some controllers have adaptive tuning, where PID parameters are monitored and corrected continuously to optimize system stability and eliminate oscillation around setpoint.

Fuzzy logic PID control is a development in machine intelligence that utilizes true or false facts composed of a series of ones and zeros, giving the controller a range of response possibilities. Fuzzy logic allows the controller to create a temporary setpoint that continuously changes as the process temperature approaches setpoint. This essentially eliminates overshoot during startup or load changes. Fuzzy logic is too complex to be discussed in detail in this article, but numerous books on the topic are available.

Figure 5. PID control provides the combined benefits of reduced overshoot (PD) and the elimination of offset (PI).

Consider Your Options

As important as the control action being used, additional features and options help customize a controller to meet the specific needs of a thermal control system. When selecting a controller, remember to consider external factors and performance features including:

  • Number of sensor inputs, their type and placement.
  • Number of outputs and types (mechanical relay, solid-state relay, voltage pulse, current or voltage).
  • Output action (direct or reverse acting).
  • Ramp and soak profiling.
  • High/low alarms.
  • Analog retransmission.
  • Digital communication.
  • Power rating.
  • Agency approvals.
  • Operator lockout.
  • Housing size and type.

    By taking advantage of select features, a controller can be configured to meet the needs of your specific application.