Floating control, also called three-position control, is a control strategy built around a device that has three commands: raise, lower, and hold. It is the natural match for a motor-operated valve or damper that is driven open or closed by a reversible motor and simply stays put wherever it was left. Field techs meet it constantly on MOVs, louvers, and slow bulk processes, yet it is a genuinely different animal from analog PID. This guide explains how floating control works, why it is inherently integrating, and where it belongs.
Floating Control in one line: Floating control (three-position control) is a mode in which the controller issues one of three commands - raise, lower, or hold - to a reversible motor-driven final element that remains at its last position when told to hold; because the output position is the time integral of the raise and lower pulses rather than a direct analog value, floating control is inherently integrating and is used with motor-operated valves, dampers, and slow processes.
A floating controller does not send a position; it sends a direction. When the measurement drifts above setpoint by more than a neutral zone, the controller energizes the lower (close) contact and the valve motor runs closed until the command stops. When the measurement falls below the neutral zone, it energizes the raise (open) contact and the motor runs open. When the measurement is inside the neutral zone, both contacts are off and the valve floats - it holds its exact current position because a motor-operated valve has no spring return and stays where the motor left it. That is the origin of the name: between corrections the output floats freely at its last spot.
The neutral zone (a form of deadband) is essential, exactly as the differential gap is in on/off control, because it stops the motor from constantly hunting back and forth around setpoint and burning out from endless reversals. But the crucial difference from on/off control is that the final element is not slammed to a limit - it is nudged incrementally to any intermediate position, so floating control can hold a process at a modulated valve opening rather than cycling between fully open and fully closed. It gets there in small steps rather than one analog move.
The defining mathematical property of floating control is that its output is an integral. Each raise or lower pulse moves the valve a little, and those moves accumulate: the valve position at any moment is the running sum of how long the motor has spent opening minus how long it has spent closing. The controller never commands an absolute position - it only commands a rate of change. This built-in integration is why floating control can, like reset action, eliminate steady-state offset and settle the process on setpoint, without any explicit integral term being programmed. The integration lives in the actuator itself.
That same integrating nature is also its main hazard. Because the valve keeps moving as long as a command is present, floating control matches well with slow, self-regulating processes where the loop has time to react before the valve travels too far. On a fast process the valve overshoots badly, because the motor is still stroking when the measurement has already crossed back through setpoint. Tuning a floating loop is largely about the neutral zone width and the valve stroke speed rather than gain and reset numbers, which is why it feels unfamiliar to technicians used to setting PID constants.
Floating control appears most often on motor-operated valves and HVAC-style dampers, and on slow bulk processes such as large tank blending, big surge-drum levels, and building pressure or temperature control where response is measured in minutes. In these cases the reversible motor and its raise/lower contacts are cheap, robust, and forgiving, and the process is slow enough that incremental floating action holds it comfortably. Many electric-actuated block-to-throttle valves in gas plants and pipelines are wired this way, driven from a PLC's open and close digital outputs rather than an analog output card.
For an operator watching from a cloud SCADA platform, a floating loop presents differently than an analog one. Instead of a single analog output percentage, the controller exposes open and close command states and, ideally, a position feedback from the actuator. Merobix trends both the raise/lower commands and the reported valve position, so an engineer can see a damper creeping open pulse by pulse and confirm it is tracking the process. Because Merobix logs the discrete commands and the feedback together, it also exposes the classic floating-control fault - commands firing but position not changing, meaning the motor, limit switch, or gearbox has failed - from a browser rather than a truck roll.
PID sends an absolute analog output position calculated from gain, reset, and rate. Floating control sends only raise, lower, or hold commands to a reversible motor that holds its last position, so the output position accumulates from the pulses. Floating control is inherently integrating and suits slow processes with motor-operated valves, while PID modulates directly and suits faster loops.
Because the valve position is the running sum of the raise and lower commands over time - the controller commands a rate of movement, not a position, so position is the time integral of those commands. This built-in integration lets floating control remove steady-state offset without a separate integral term, but it also causes overshoot on fast processes.
The neutral zone is a band around setpoint in which the controller issues no raise or lower command and the valve floats in place. It works like a deadband, preventing the motor from constantly reversing and wearing out when the measurement sits near setpoint. Too narrow a zone causes hunting and motor wear; too wide a zone loosens control.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.