Fixing Actuator “Hunting for Signal” by Adjusting the Deadband Setting

In a modulating actuator, the control board continuously compares two quantities: the command (setpoint) coming from the PLC, controller, or DDC system, and the actual position feedback coming from the feedback potentiometer. The board drives the motor in the direction that reduces the difference — the error. When the actuator is well-behaved, the error shrinks until it falls inside the deadband, the motor stops, and the valve holds position.

Hunting is what happens when that simple loop never reaches a stable state. The classic signature is a low-frequency, roughly sinusoidal cycling of position and motor current around the setpoint:

• The stem moves toward the setpoint, then reverses before the position feedback actually matches the command.
• Cycles repeat continuously at intervals from a few seconds to a minute or two — not a one-off overshoot.
• Motor temperature climbs and cycle counters rack up even though the process setpoint has not changed.
• Valve position readback on the DCS/SCADA screen jitters around the target instead of holding a steady number.

The cost of leaving hunting untreated is measurable: a modulating actuator rated for, say, 100,000 cycles can burn through that life in under a year of continuous 30-second hunting cycles, and the process side pays in product variability. That is why the fix — usually a ten-minute configuration change — is worth doing properly the first time.

The deadband (also called the neutral zone or input sensitivity band) is a deliberately small window around the setpoint inside which the positioner is allowed to do nothing. If the command is 50.0% and the valve is sitting at 49.4% with a 1% deadband, the error of 0.6% is inside the band and the motor stays off. If the valve drifts to 48.5%, the error of 1.5% exceeds the deadband and the motor drives until the position re-enters the band.

Deadband exists precisely to prevent hunting. Every real actuator loop contains irreducible noise sources — electrical noise on the command line, wiper ripple and wear on the feedback potentiometer, backlash in the gear train, and stick-slip friction in the valve packing. If the deadband is smaller than the combined amplitude of those disturbances, the positioner sees a perpetual “error” that flips sign as the valve crosses the setpoint, and it motors back and forth forever. The actuator is hunting because it is trying to be more accurate than its own hardware and signal can be.

Two related terms are worth keeping straight. Hysteresis is the mechanical play in the drive train — the position difference between approaching a point from above versus from below — and it is a property of the hardware, not a setting. Deadband is the electrical window the control board uses to ignore small errors. If the gearbox has significant backlash (hysteresis), the deadband must be wide enough to cover it, or the positioner will oscillate as it alternately approaches the setpoint from each side. You cannot tune hysteresis away with a wider deadband alone — the correction only masks it — but in practice a modest widening is the right first step, and Section 5 explains how to verify whether the feedback side needs attention instead.

One more common source of apparent “hunting” sits outside the actuator: a noisy command. If the 4–20 mA loop itself carries ripple or the 0–10 V signal is corrupted by electromagnetic interference, the positioner is faithfully chasing a moving target. That problem is not fixed by any deadband value, which is why the very first diagnostic step below is measuring the command signal, not changing settings.

Fifteen minutes of measurement saves you from chasing the wrong adjustment. The goal is to answer one question: is the command stable, and does the feedback follow it?

1. Put the loop in manual at the controller and hold a fixed setpoint (for example 50%).
2. Measure the command at the actuator terminals with a true-RMS multimeter or a mA loop calibrator in series. A stable 4–20 mA command should read within ±0.05 mA (±0.3%) with no visible ripple on a fast-responding meter; a 0–10 VDC command should be steady to ±10 mV or better. If the signal itself is dancing, the source or the cable run is the problem — see the 4–20 mA loop diagnostics referenced below.
3. Watch the position feedback on the actuator’s readout (or the DCS position signal) for 2–3 minutes without touching anything. Note whether the valve is cycling continuously, drifting slowly, or holding steady.
4. Watch the motor current / torque readout if available. Repeated reversing at low load points to hunting; high, steady current points to mechanical binding; no motor activity while position readback jumps points to a feedback fault.

If the matrix points to the deadband, the feedback, or the limit switches, proceed to the corresponding fix below. If it points to the command signal, step away from the actuator entirely and work the loop: shield and route the cable away from VFDs and power feeders, bond shields at one end only, and check for ground loops. Loop testing and EMI fixes for a valve that ignores or misreads its command are covered in the 4–20 mA diagnostic guide referenced above.

Where the deadband is configured depends on the actuator generation:

• Smart / digital actuators expose the deadband as a parameter in the local menu or over Modbus/BACnet (often labeled Deadband, Input sensitivity, or Neutral zone, expressed in % of span). Write down the current value before changing anything.
• Analog positioners with DIP switches or trimmers set the deadband (and often a companion input damping/filter) with a switch or a small potentiometer. The nameplate or the manual tells you the per-step value; if not, measure it by stepping the command and finding the smallest change that produces motion.

Use this step-by-step procedure — always change in small increments and test at a steady setpoint:

1. Record the factory/current deadband. Typical defaults are 0.5–1.5% of span. Write it down so you can return to it.
2. Set the deadband to 1% of span if the loop is hunting at its current (usually smaller) setting. On actuators whose minimum step is a fixed value, choose the next step above the current one.
3. Command 50% and hold it for two minutes. Count the number of motor reversals. One settling correction or none = good. Continued cycling = go to step 4.
4. Increase the deadband in 0.5% steps (1.0% → 1.5% → 2.0%) and repeat the two-minute test each time. Stop at the smallest value that fully stops the cycling.
5. Verify accuracy is still acceptable. Command 25%, 50%, and 75%. The settled position should sit within the deadband of the target. If the value that stops hunting also leaves an unacceptable offset, the loop needs a fix beyond deadband alone — see Sections 5 and 6.
6. Re-test at both travel ends and after the actuator has warmed up, because backlash and friction change with temperature.

If your actuator’s positioner also offers an input filter or damping time parameter, a small filter (tens of milliseconds) can smooth high-frequency signal ripple without sacrificing response speed. This is often a better tool than a very wide deadband when the command is electrically noisy but otherwise healthy.

The feedback potentiometer is the actuator’s “position sense.” The control board converts the wiper voltage into a 0–100% position reading and compares it with the command. Two distinct feedback faults cause hunting-like behavior:

• Zero/span drift — the pot’s electrical travel no longer matches the mechanical travel (it was disturbed, the coupling slipped, or the actuator was rebuilt). The board then believes the valve is at 50% when it is really at 48%, and keeps correcting against a phantom error.
• Wiper noise / dead spots — a worn or contaminated track makes the position reading jump or momentarily drop out as the wiper crosses the worn zone. The control board interprets each jump as a position error and reacts, producing irregular cycling that a deadband change never fully cures.

A full-stroke calibration re-matches the electrical and mechanical travel, and it is a required step after any repair that touches the drive train, after replacing the actuator on the valve, or when the symptoms in row two of the diagnostic matrix appear:

1. Disconnect the command so the actuator cannot be driven by the loop during calibration (or set the controller output to manual and hold it).
2. Drive the actuator to the mechanical closed (0%) stop using the local hand/auto controls. Confirm the valve disc actually seats — do not trust the readout yet.
3. Set the zero point. In the calibration menu select Zero / 0% / Closed and confirm. On trimmer-based positioners, adjust the zero trim until the position readout shows 0.0% (or the feedback voltage matches the datasheet value, typically 0 V for 0–10 V feedback spans).
4. Drive fully open (100%) to the mechanical stop, confirm the disc is against the stop, then set Span / 100% / Open and confirm (or trim until the readout shows 100.0%).
5. Stroke the actuator 0% → 100% → 0% once slowly, then repeat the zero and span settings one more time — the second pass compensates for gear backlash and settles the pot at its true endpoints.
6. Check linearity mid-travel: command 25/50/75% and confirm the readout tracks within the actuator’s stated accuracy (typically ±1%). Large mid-range deviations point to a worn track or a slipped coupling rather than calibration.
7. Log the feedback resistance or voltage at each point so the next calibration has a baseline.

After recalibration, repeat the two-minute hold test from Section 4. Many actuators that “needed” a 3% deadband settle happily at 1% once the feedback actually tells the truth.

Mechanical limit switches (cam-operated micro switches) exist to cut power at the ends of travel so the actuator never drives the valve past its seat or into the bonnet. They have no role in mid-stroke proportional control — so if your actuator hunts at 50%, the limit switches are almost certainly innocent. But two limit-switch-related faults do look like hunting, and both occur at the travel ends:

• Early tripping: a cam is adjusted too tight, so the switch trips and kills power before the valve reaches the commanded end position. In proportional mode the board still sees an error, re-energizes the motor, and the valve repeatedly approaches the end, trips, and relaxes — an end-of-travel cycling that a deadband increase will never fix.
• Contact chatter: the cam roller rests right at the switch’s actuation point, so vibration makes the contact bounce on and off, producing rapid power blips and erratic position behavior at the ends of stroke.

Re-setting the limit cams is a mechanical adjustment, not an electronic one:

1. Confirm the symptom location. The fault must reproduce only within the last few degrees of travel. If it happens mid-stroke, stop — this section is not your fix.
2. Drive to the affected end and loosen the cam (the cams are normally accessible under the terminal cover or through the cam window; loosen the retaining screw, never remove it completely).
3. Set the closed/open cam with a small margin: the switch should actuate when the valve disc is against the seat (closed) or at the mechanical stop (open), with roughly 2–5% of travel of overrun margin so normal seating force does not chatter the contact. Use the valve manufacturer’s seat-stop setting, not the actuator’s full electrical travel, as the reference.
4. Tighten and verify with a multimeter: across the switch’s common/NO terminals, confirm it changes state crisply at the desired position and does not bounce as you wiggle the linkage by hand.
5. Re-run the full-stroke calibration from Section 5, because changing the cams changes the mechanical endpoints the positioner must learn.

While the covers are off, do the two-minute mechanical sanity check: with power off, move the handwheel or manual override through a full stroke and feel for roughness, and check the coupling and stem connection for play. Backlash here feeds directly into the hunting loop, and it is cheaper to find by hand than by watching the motor hunt. A valve that binds or sticks will defeat any tuning — that is a valve binding and over-torque problem (see Section 1) rather than a tuning fault, and it is diagnosed separately.

Hunting rarely comes out of nowhere. It appears after a repair, after the valve is replaced or repacked, after the loop wiring is rerouted, or after years of pot wear. A small preventive routine keeps it away:

• Document the tuned deadband per actuator in the CMMS or loop folder, along with the date and the positioner firmware version. When a “tuning” appears to drift, you will know instantly whether someone changed a parameter.
• Schedule a feedback check at every turnaround: read pot resistance or feedback voltage at 0/50/100% and compare with the baseline from the last calibration. A pot nearing the end of life announces itself in the mid-travel readings long before it starts hunting.
• Protect the signal path: use shielded twisted pair, grounded at one end, and keep command wiring out of trays shared with VFD output cables. Most intermittent hunting that returns after a “successful” fix is electrical noise, not the actuator.
• Watch cycle counts on smart actuators. A sudden jump in cycle count at a constant process setpoint is the earliest possible hunting alarm — investigate while the wear is still negligible.
• Re-tune after any mechanical intervention: repacking a valve, replacing the gearbox, or re-mounting the actuator changes friction and backlash, and the old deadband may no longer be appropriate.

If you are commissioning new actuation rather than rescuing an existing loop, choosing the right hardware up front removes most of these problems: pick a modulating actuator whose positioner exposes a real deadband parameter (rather than a fixed sensitivity), and follow proper installation practice from the start. Our step-by-step guide to installing electric valve actuators covers mounting, wiring, and the commissioning sequence that gives a clean baseline for tuning.

Frequently Asked Questions