When controller output reverses direction, a control valve assembly may show no observable response until the signal crosses a finite input range. During that interval, valve travel, flow, or the measured process variable can remain unchanged even though the controller is requesting a correction.
This behaviour is control valve deadband, but the measured value depends on which input and output are being compared. A deadband based on stem travel is not automatically the same as the deadband seen in flow or process response under load.
The symptom does not identify the responsible component. Diagnosis starts by comparing controller output, actual valve travel, and the process variable around a controlled direction reversal, then checking the valve, actuator, positioner, air supply, and mechanical connections as one assembly.
What control valve deadband means
Control valve deadband is the range through which an input signal can change after reversing direction without producing an observable change in the selected output. It is normally expressed as a percentage of the input span.
The input for a complete control valve assembly is usually the controller output. The output under observation may be stem or shaft travel, flow rate, or the measured process variable. These outputs are not interchangeable, so a deadband value is incomplete unless the test identifies which one was measured.
After the input reverses, the signal must cross the no-response range before movement or process correction resumes. The deadband can differ between increasing and decreasing signals, which means both directions and the reporting method matter.
A bench test that compares input signal with actuator travel describes only that test arrangement. It does not by itself establish how the installed valve changes flow under pressure load or how the process variable responds in the operating loop.
Deadband vs hysteresis, stiction and dead time
Calling every delayed or jumpy valve response deadband hides the evidence needed to select the next check. The four terms describe different relationships, although more than one can appear in the same control valve assembly.
| Term | Observed pattern | Possible mechanism | Useful check |
|---|---|---|---|
| Deadband | Input reverses and changes through a finite range before the selected output responds | Backlash, friction, shaft wind-up, linkage clearance, or positioner and relay behaviour | Compare input and output around controlled reversals in both directions |
| Hysteresis | The same input corresponds to different outputs on increasing and decreasing paths | Friction, magnetic effects, elastic deformation, or mechanical play | Plot complete increasing and decreasing input-output curves |
| Stiction | Valve travel remains stuck while force builds, then moves in a jump | Static friction in packing, seals, guides, bearings, or contacting trim | Look for a repeatable stick-slip signature in signal, pressure, and travel trends |
| Dead time | Output begins responding only after an elapsed time delay | Signal processing, pneumatic volume, tubing restriction, transport delay, or measurement lag | Measure time from the input change to the first verified output response |
The terms can overlap in service. Packing friction may create a stick-slip pattern while also increasing the signal change required after reversal. Backlash can contribute to both deadband and a difference between increasing and decreasing travel curves. The test variables, direction, load, resolution, and calculation method should therefore accompany every reported value.
If the main symptom is a time delay rather than a reversal-dependent no-response range, the checks in the control valve response time guide provide a more suitable starting point.
How deadband affects control-loop performance
After the controller reverses its output, it must cross the deadband before the valve delivers a corrective change. The process can continue moving away from setpoint during this interval, especially when disturbances are frequent or the required corrections are small.
The controller may continue increasing its demand because the measured process variable has not responded. Once the valve finally moves, the accumulated correction can be larger than the process needs. The controller then reverses again, and the same no-response range may contribute to sustained cycling.
The severity depends on loop dynamics, valve authority, process gain, disturbance size, and whether the reported output is travel, flow, or the process variable. A travel-based deadband that appears small on a bench can have a more visible process effect where the installed flow characteristic has high gain. The reverse can also occur where a travel change produces little flow change.
Deadband is therefore not only a valve accuracy number. Its practical consequence is determined by the complete path from controller output through valve movement and flow to the process measurement.
What causes deadband in a control valve assembly
The responsible no-response range can be distributed across the valve, actuator, positioner, feedback mechanism, and connecting parts. Each cause below is a diagnostic branch, not a root-cause conclusion from the symptom alone.
Sliding-stem friction and packing
Packing, stem seals, guides, and contacting trim resist the start of stem movement. When static friction exceeds running friction, actuator force can build without travel and then release as a jump. Packing condition, adjustment, stem surface condition, temperature, and process deposits can all change this behaviour.
Friction-related deadband should be checked under representative load. A low-load bench stroke may not reproduce the force balance present when differential pressure acts on the plug.
Rotary backlash and shaft wind-up
Clearance in splines, keys, pins, couplings, gears, or linkages can absorb actuator motion after a direction reversal before the closure member changes position. Elastic twist in shafts and connections can create a similar load-dependent effect.
The position measured at the actuator may then differ from the actual disc, ball, or plug position. The location of the travel feedback device must be recorded before its signal is treated as proof of closure-member movement.
Actuator and positioner effects
An actuator with inadequate force or torque margin may struggle to overcome friction or process load near particular travel positions. Spring setting, diaphragm or piston condition, actuator sizing, and available supply pressure can change the response after reversal.
A positioner can reduce position error only when its sensing, pneumatic, and mechanical paths are working correctly. Feedback-link clearance, calibration, relay dead zone, restricted pneumatic passages, or inappropriate settings can add to the measured deadband. The valve positioner guide explains how these elements interact with the actuator and control signal.
Wear, load and instrument-air conditions
Wear can enlarge clearances in pins, bearings, linkages, and drive connections. Process deposits, damaged guides, changing seat load, or an uneven torque profile can make the no-response range position-dependent rather than constant across the stroke.
Low or unstable instrument-air pressure, restricted tubing, contamination, and regulator or filter problems may imitate or amplify valve-response faults. Air pressure trends and actuator chamber pressures help separate supply behaviour from mechanical resistance.
How to detect and measure control valve deadband
A deadband test must begin by naming the input and output. Controller output versus verified valve travel measures a different relationship from controller output versus flow or process variable.
Start with synchronised operating trends
Review time-aligned controller output, positioner demand, verified valve travel, process variable, and relevant pressure signals. The useful part of the trend is the period immediately before and after a direction reversal.
A repeatable failure to return after reversal can indicate deadband when controller output changes while verified valve travel remains stationary. If valve travel changes but flow or the process variable does not, the investigation should include installed valve gain, pressure conditions, process dynamics, and measurement resolution.
Use a controlled direction-reversal test
The public ISA closed-loop troubleshooting method uses controlled output changes and trend comparison to help distinguish process, tuning, and final-control-element behaviour. Site operating limits and safety procedures determine whether such a test is permitted.
- Choose a stable operating period and confirm that the test is authorised for the process, valve, and fail action.
- Record the current controller output, verified valve travel, process variable, supply pressure, and relevant process pressures.
- Move the input in one direction until the selected output has changed and settled sufficiently for the test resolution.
- Reverse the input in small controlled increments, allowing enough settling time after each step.
- Identify the first repeatable change in the selected output and calculate the input change from the reversal point.
- Repeat the test in the opposite direction and at other relevant travel or load conditions when the authorised procedure allows it.
The reported result should include signal span, reversal direction, step size, settling criterion, measurement resolution, valve position, process load, and calculation method. Noise, filtering, backlash in the feedback device, or coarse travel indication can otherwise be mistaken for valve deadband.
Separate bench travel from installed response
A bench test is useful for checking command-to-travel behaviour under defined supply and loading conditions. It does not reproduce every process force, pressure drop, piping interaction, or flow response present in service.
Research on installed control-valve deadband measurement illustrates why process data may be needed when the required output is flow rather than stem position. Acceptance records should state whether the result came from a bench stroke, installed travel test, flowing test, or loop-response assessment.
Protect the process and preserve the evidence
Do not place a loop in manual, stroke a valve, change packing load, bypass an interlock, or disturb a pressurised assembly unless the site procedure specifically authorises the action and controls stored energy and process consequences. Where an online test is unsuitable, collect operating trends and reproduce the relevant assembly conditions during an authorised offline test.
How to reduce deadband without masking the cause
The corrective action should follow the measured fault branch. Retuning the controller before verifying the final control element may change the cycle without removing the no-response range.
| Suspected branch | Evidence to confirm | Corrective direction |
|---|---|---|
| Linkage or drivetrain clearance | Lost motion across pins, keys, couplings, gears, or shaft connections after reversal | Inspect connections and restore specified fit, alignment, and tolerances |
| Packing or seal friction | Actuator force or pressure builds before a sudden travel jump | Review packing condition, adjustment, materials, stem finish, and sealing requirements |
| Actuator force or torque margin | Response deteriorates at particular pressure loads or travel positions | Verify actuator sizing, spring setting, supply pressure, load assumptions, and fail action |
| Positioner or pneumatic path | Demand, output pressure, and travel trends reveal calibration, feedback, relay, tubing, or supply irregularities | Check air quality, regulator, tubing, calibration, feedback parts, and manufacturer settings |
| Valve internal condition | Travel resistance varies with position or inspection finds deposits, guide damage, or contacting trim | Clean, repair, or replace affected parts under the approved maintenance procedure |
| Installed flow or process response | Valve travel responds correctly but flow or process variable remains insensitive | Review pressure drop, valve authority, installed characteristic, sizing, trim, and process measurement |
A positioner replacement is not a universal remedy. It may improve command-to-position control when the actuator and valve are mechanically capable, but it cannot remove worn drivetrain clearance or repair an undersized actuator.
Packing adjustment also requires care. Reducing packing load without checking the packing design, pressure, temperature, emissions requirement, and leakage risk can exchange a movement problem for a sealing problem.
Once the assembly responds repeatably, controller tuning can be reviewed against the corrected installed process. The evidence sequence in the control valve troubleshooting guide can help organise the remaining mechanical, pneumatic, and process checks.
Any adjustment, dismantling, or live movement must follow the authorised isolation, stored-energy, and functional-test procedure for the site.
What deadband should be specified for a new control valve package
A deadband percentage without test conditions is not a complete acceptance requirement. The specification must identify what is driven, what is measured, how reversal is applied, and whether the assembled valve is tested under load.
The Emerson Control Valve Handbook gives manufacturer guidance of 1% or less, with performance near 0.25% presented as preferable for reducing process variability. These figures are useful as a benchmark, not as a universal acceptance rule for every valve type, service, project, or test arrangement.
A project specification should define:
- The input variable, signal range, and direction of each test.
- The output variable, such as stem travel, shaft position, flow, or process response.
- The complete valve, actuator, positioner, feedback, and accessory configuration.
- The travel position, pressure load, air supply, temperature, and other relevant test conditions.
- The step size, settling criterion, measurement resolution, calculation method, and required repeatability.
- The acceptance limit and the test record to be supplied with the assembled package.
The specified limit should follow the process-control requirement and applicable project documents. It should also distinguish a factory command-to-travel test from an installed test intended to demonstrate flow or loop response.
For a control valve package review, provide the signal, travel range, actuator and positioner configuration, differential pressure, supply conditions, test method, and required acceptance record. Those data allow the deadband requirement to be assessed against the complete assembly instead of a context-free percentage.