Three-position floating control is how you steer a motorized valve or damper when all you have is three commands: raise, hold, and lower. There is no analog positioner telling the actuator to go to 63 percent - instead the controller pulses a raise or lower contact to run the motor open or closed, and holds both contacts off to let it stop and stay wherever it landed. The output floats: it stays put until told to move, which is where the name comes from. This page explains three-position floating control as a specific output type, covering the neutral zone, pulse duration, and how a SCADA or PLC system can infer position without any feedback.
Three-Position Control in one line: Three-position floating control drives a bidirectional motorized actuator using three discrete commands - raise (drive open), lower (drive closed), and hold (both off) - instead of an analog position signal. The controller pulses the raise or lower contact when the error exceeds a neutral zone, and holds when the error is inside it, so the actuator floats at its last position. It needs no positioner and, in its basic form, no position feedback.
A three-position floating output has exactly three states, driven by two contacts. Energize the raise contact and the actuator motor runs in the open direction for as long as the contact is held. Energize the lower contact and it runs closed. Energize neither and the motor stops, holding position by its own gearing or brake. The controller never commands an absolute position; it only commands motion, direction, and duration, and the physical integration of that motion by the motor is what sets where the valve ends up.
This is why it is called floating. Between commands the output floats freely at whatever position the last move left it, neither drifting nor being actively held to a target. It is a natural fit for slow-moving final elements - large dampers, motor-operated valves, louvers - where an analog positioner would be expensive or impractical and where the process itself changes slowly enough that stepwise motion is perfectly adequate. The actuator is, in effect, an integrator that the controller pokes forward and backward.
It helps to distinguish three-position floating control from the broader idea of floating control in general. Floating control is any control mode whose output changes at a rate rather than to an absolute value, and it includes single-speed and proportional-speed variants. The three-position form is the specific, very common implementation using two discrete raise/lower contacts and a dead center, and it is what people usually mean on plant floors when they say a valve is under floating control.
The heart of a three-position controller is its neutral zone, also called the deadband or dead center. This is a band of error around setpoint within which the controller issues no command at all and simply holds. Without it, the controller would chatter endlessly, pulsing the motor open and closed around setpoint and wearing out the actuator, because there is no such thing as being exactly on target. The neutral zone has to be wide enough to swallow measurement noise and the actuator's own overshoot, but narrow enough that the loop still controls tightly.
When the error leaves the neutral zone, the controller drives the motor in the correcting direction. How it does so distinguishes crude from refined implementations. The simplest scheme runs the motor continuously until the error re-enters the band, which is easy but tends to overshoot. More refined schemes pulse the output - a short burst of raise or lower, then a pause to let the process respond, then reassess - and they make the pulse duration proportional to how far outside the band the error sits. Bigger error, longer pulses; small error just past the edge, short taps. This pulse-width behavior is how a floating controller approximates proportional action with a purely on/off actuator.
Tuning a three-position loop is therefore a matter of setting the neutral zone, the pulse and pause timing, and the actuator's full-stroke travel time so that the loop settles without hunting. Too narrow a neutral zone or too aggressive a pulse and the valve hunts back and forth; too wide or too timid and the loop is sluggish and leaves a large standing error. Because the actuator travel time is usually fixed by hardware, most of the tuning lives in the neutral zone and the pulse timing.
A basic three-position floating loop often runs with no position feedback at all - no limit-switch potentiometer, no 4 to 20 milliamp position transmitter. The controller does not know where the valve is; it only knows the process variable and whether it needs to raise or lower. This keeps wiring and cost down, which is a large part of the appeal, but it means position is something you infer rather than measure. The usual method is a simulated position: the system integrates the run time in each direction against the known full-stroke travel time to estimate a position, resetting the estimate at the ends when a limit switch or a fully-run timer says the actuator has hit a stop.
That inferred position is exactly the kind of value a SCADA layer likes to compute and historize. A cloud SCADA platform such as Merobix can log the raise and lower commands, accumulate estimated position from run time, and present a floating valve as a percentage on a dashboard even though the field device never reports one. Trending that estimate alongside the process variable lets a remote operator see whether the valve is drifting toward a limit or hunting, and it makes an otherwise opaque floating loop legible from far away.
The catch is drift. Because the estimate is dead reckoning from run time, it accumulates error - gear backlash, a slipping clutch, a stroke time that changed with temperature - and without periodic recalibration against a real limit the displayed position slowly parts ways with reality. Good practice is to re-reference the estimate every time the actuator drives fully to an end stop, and to alarm when the commanded and inferred behavior stop agreeing. On remote unmanned sites, a floating valve that is quietly stuck can keep receiving pulses while going nowhere, so having the SCADA system watch for a valve that is being commanded to move but whose process effect never changes is a cheap and valuable safeguard.
The neutral zone, or dead center, is a band of error around setpoint in which the controller issues neither a raise nor a lower command and simply holds the actuator. It prevents the motor from chattering open and closed around setpoint, since the process is never exactly on target. It must be wide enough to absorb noise and overshoot but narrow enough to keep control tight.
It controls on the process variable alone, pulsing raise or lower when the error leaves the neutral zone, and lets the motor integrate that motion into a position. To display or use position, the system infers it by dead reckoning - integrating run time in each direction against the known full-stroke travel time - and re-references the estimate whenever the actuator hits an end stop. No positioner or feedback transmitter is required for the loop to function.
Floating control is the broad category of control whose output moves at a rate rather than to an absolute value, and it includes single-speed and proportional-speed variants. Three-position floating control is the specific, common implementation using two discrete raise/lower contacts plus a hold state and a neutral zone. When people describe a motorized valve as floating on a plant, they usually mean this three-position form.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.