A smart positioner has to learn the physical limits of the valve it drives before it can position accurately: exactly where the valve is fully closed and exactly where it is fully open. On most modern digital positioners you do not set those endpoints by hand; you launch an auto-calibration routine, often called autostroke or autocal, and the positioner drives the valve to both stops itself, measuring the travel. Once it knows the endpoints it maps its 4-20 mA input onto that stroke. Calibrating a positioner well means getting the feedback linkage set up correctly first, running that routine, verifying the result against the input, and choosing the right characterization curve for how the loop should behave.
Calibrate a Valve Positioner in one line: To calibrate a valve positioner, first mount and set up the feedback linkage so the positioner senses travel over the correct range, then run the auto-calibration or autostroke routine, which drives the valve fully closed and fully open so the positioner learns its travel endpoints. Afterward, verify the calibrated stroke by commanding known percentages through the 4-20 mA input and confirming matching physical travel. If autocal fails, set the zero and span manually at the closed and open positions, and select the flow characterization curve, linear or equal-percentage, the loop requires.
Before any calibration runs, the positioner has to be able to sense travel correctly, and that depends entirely on the feedback linkage. For a linear actuator the feedback arm or magnet must track stem movement across the valve's rated travel without binding or running out of range at either end. For a rotary actuator the positioner's shaft or non-contact sensor must be aligned so the full quarter turn or the actuator's rotation lands within the sensor's usable arc. Getting this mechanical setup right is the step people rush and regret: if the feedback runs off the end of its range at full travel, no calibration routine can produce an honest stroke, because the positioner cannot see where the valve really is.
With feedback set up, you launch the auto-calibration routine from the positioner's local interface or a handheld communicator. The positioner takes control of the valve and drives it to one stop, records the feedback reading there, drives it to the other stop, and records that too. From those two measured endpoints it establishes what feedback value corresponds to fully closed and fully open, and it also learns characteristics like how much air the actuator needs and how fast it responds, which it uses to tune its own control. The whole point of autostroke is that the positioner discovers the real travel of this specific valve rather than assuming a nominal value, so its subsequent positioning is referenced to physical reality.
After the routine finishes you verify rather than assume it worked. Command the valve through the 4-20 mA input to 0, 50, and 100 percent and confirm the physical travel matches, using the valve's travel scale. A successful autocal should put the valve hard on its seat at the closed command and at full open at the top command, with the midpoint landing near half travel adjusted for whatever characterization is selected. If the endpoints are right and the midpoint tracks, the calibration is good; if the valve stops short of its seat or the travel is compressed, the feedback setup or the routine needs another look before you trust the loop.
Once the positioner knows the physical travel, the characterization curve decides how it maps input signal to that travel. A linear characterization moves the valve travel in direct proportion to the input, so 50 percent input gives 50 percent stroke. An equal-percentage characterization deliberately bends that relationship so equal changes in signal produce equal percentage changes in flow, which compensates for valve and system behavior and is common where the installed flow characteristic needs shaping. Many smart positioners can apply this curve electronically, replacing the mechanical cam that older positioners used. Choosing the wrong curve does not break the calibration, but it makes the loop control differently than the process engineer intended, so the selection has to match the loop's design.
Rotary and linear actuators differ enough in setup that it is worth calling out. Linear actuators move a sliding stem, and the feedback typically comes from an arm following that stem, so the concern is arm geometry and travel range. Rotary actuators turn a shaft, and the feedback comes from the shaft angle, so the concern is angular alignment and making sure the sensed rotation matches the valve's open and closed positions. In both cases the autostroke routine is conceptually the same, drive to both stops and learn the endpoints, but the mechanical checks before it differ, and a feedback element aligned for one geometry will not behave correctly on the other.
Autocal does not always succeed, and knowing the manual fallback keeps you working when it fails. Common causes of a failed autostroke are insufficient or unstable air supply, a feedback element out of range, a valve that will not reach a hard stop, or excessive friction that confuses the routine. When autocal cannot complete, you set the endpoints by hand: drive the valve fully closed and capture that as the zero, drive it fully open and capture that as the span, then confirm the two points hold. Manual zero and span give you a working calibration while you chase down whatever kept the automatic routine from finishing, whether that is an air-supply problem or a mechanical one.
A freshly calibrated positioner is accurate on the day you calibrate it, but valves live in a world of rising packing friction, air-supply changes, and mechanical wear that slowly pull the assembly away from that calibration. The value of a smart positioner is that it knows its own learned travel and can report how hard it is working to hold position, and those internal diagnostics are exactly what a monitoring system can carry away from the valve. Rather than discovering a decalibrated positioner during an upset, you can watch its behavior between calibrations.
When positioner diagnostics and valve position are trended in a cloud SCADA platform such as Merobix, the signs that a positioner needs recalibration show up as data rather than surprises. A rising drive signal needed to hold the same position, a growing gap between commanded and actual travel, or endpoints that no longer reach the seat all appear in the trend. That lets a technician decide from evidence when to rerun autostroke, targeting the valves whose diagnostics have started to wander instead of recalibrating on a blanket schedule.
The monitoring record also verifies the recalibration itself. After you rerun autostroke on a valve that had been positioning poorly, the trend should show commanded and actual position snap back into agreement and the drive effort settle down. Because the platform kept the history from before the work, you have a clean before-and-after that confirms the calibration actually restored the valve rather than just clearing an alarm, which is the kind of proof a one-time field check on its own cannot provide.
Autostroke drives the valve to both physical stops so the positioner can measure and learn where the valve is fully closed and fully open. From those two endpoints it maps its 4-20 mA input onto the real travel and often tunes its own control by learning the actuator's air demand and response speed. The routine matters because it references the positioner to the actual valve rather than to a nominal travel figure.
The usual causes are an unstable or insufficient instrument air supply, a feedback element that is out of alignment or running off the end of its range, a valve that never reaches a firm stop, or friction high enough to confuse the routine. Fixing the mechanical setup or the air problem often lets it complete. When it still will not finish, you fall back to setting the zero at fully closed and the span at fully open manually.
Not usually, because the characterization curve only changes how the learned travel is mapped to the input signal, not the travel endpoints themselves. Once autostroke has established the physical closed and open positions, switching between linear and equal-percentage reshapes the input-to-travel relationship without needing another autocal. You should still verify a few points through the 4-20 mA input afterward to confirm the valve moves as the selected curve intends.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.