A control valve loop check proves that a command leaving the control system actually moves the valve to where it should be and reports back honestly. It is one thing to trim a positioner on the bench; it is another to confirm that 50 percent from the DCS produces half travel on the real actuator with the real air supply, and that the position feedback returning to the control system agrees. A proper loop check walks the valve through its full range in both directions, confirms the fail action on loss of signal, and measures how much lost motion the assembly carries. Done before startup, it catches reversed action, poor seating, and sticky travel while they are still cheap to fix.
Loop-Check a Control Valve in one line: To loop-check a control valve, command it from the control system in steps of 0, 25, 50, 75, and 100 percent and confirm the positioner drives the actuator to matching physical stroke at each step, with the position feedback returning the same value to the DCS. Verify the valve seats fully at 0 percent, opens fully at 100 percent, moves in the correct direction, and goes to its designed fail position when the signal or air is removed. Measuring the difference between the up and down passes gives you the deadband and hysteresis of the assembly.
The heart of the loop check is stepping the valve through its range from the same place the process will command it, the control system output, not a hand-held source jumpered at the positioner. You send 0 percent and confirm the valve is fully closed, then 25, 50, 75, and 100 percent, pausing at each step to read the actual physical travel against the valve's travel scale or a dial indicator on the stem. The point is to prove the whole chain: the analog output card, the field wiring, the positioner, the I/P, the air supply, and the actuator all cooperating to put the valve where the command said. A valve that reaches only 90 percent travel at a 100 percent command has an air, linkage, or calibration problem that a bench check might have hidden.
At every step you also confirm the position feedback that travels back to the control system. Most modern valves report position either as a second 4-20 mA signal from a transmitter or digitally over HART or a fieldbus, and the loop check is where you prove that feedback matches reality. Command 50 percent, confirm the stem is at half travel, and confirm the DCS shows the valve reporting 50 percent open. A feedback signal that disagrees with the actual stem position will mislead operators and any logic that watches valve position, so catching a miscalibrated or reversed feedback here is a real save.
Direction and end conditions deserve explicit attention because they are the errors that cause the worst surprises. Confirm the action is not reversed: increasing command should move the valve the intended way, opening an air-to-open valve and closing an air-to-close one, and a reversed positioner will drive it exactly backward. At the extremes, confirm the valve seats hard at its closed command rather than stopping a few percent open, since a valve that will not fully close cannot shut off flow, and confirm it reaches full travel at the open command rather than stalling short. These end checks are where seating problems and travel-stop misadjustments reveal themselves.
Fail action is a safety property, and a loop check is the right time to prove it because you can watch the valve do it. Every control valve has a designed position it must take when its signal or its air supply is lost, fail-closed, fail-open, or fail-in-place, chosen so the process ends up safe. To verify it, remove the command signal or bleed the instrument air and watch the valve move to its designed position under spring force. A fail-closed valve that does not fully close on air loss, or a valve that drifts the wrong way, is a finding that matters far beyond control quality, and the loop check is often the only time anyone deliberately takes the air away to confirm it.
Deadband and hysteresis describe the lost motion in the assembly, and you measure them by comparing the up pass to the down pass. Deadband is how much the command must change before the valve begins to move at all after a reversal of direction; you see it as a command increment that produces no stem movement right after you reverse. Hysteresis shows up as a difference in stem position between the ascending and descending passes at the same command, for example the valve sitting at 48 percent travel on the way up and 52 percent on the way down when both are commanded to 50. Both come from packing friction, linkage slop, and actuator lost motion, and a large value means the valve cannot hold a precise position, which shows up as sluggish or hunting control.
Reading these numbers turns a pass-fail check into a diagnosis. A small, symmetric hysteresis is normal and expected from packing friction. A large deadband right around a reversal usually points to worn linkage or a loose feedback connection, and a growing hysteresis over successive checks tracks packing that is tightening as it ages or a positioner losing its tune. Recording the deadband and hysteresis at commissioning gives you the baseline that later checks compare against, so a valve that has begun to stick is caught as a change from its own history rather than only when control visibly suffers.
A loop check is a snapshot, but the properties it measures keep changing in service, which is why the position feedback it verifies matters long after commissioning. Once the feedback is proven honest, the control system has a continuous, trustworthy picture of where every valve actually is, not just where it was told to go. That distinction is the whole value of position feedback: a valve commanded to 50 percent but sitting at 40 because it is sticking is invisible without feedback and obvious with it. The loop check is what earns the right to trust that signal.
When valve command and valve position are both trended in a cloud monitoring system such as Merobix, the deadband and hysteresis you measured once at commissioning become something you can watch drift over time. A growing gap between command and feedback, or an increasing spread between where the valve lands opening versus closing, is packing friction rising or linkage wearing, and it appears in the trend as a slow change long before the valve fails to control. That lets maintenance be scheduled from evidence rather than from a fixed interval or an outright failure.
The same visibility protects the fail-action assumption that a loop check verifies once. If a valve is supposed to drive closed on trip and the position feedback is trended, an actual demand event leaves a record showing whether the valve reached its safe position and how quickly. Reviewing those records confirms the safety behavior held in the real event, not just on the day of the commissioning check, which is the kind of after-the-fact verification that manual loop checks alone can never provide.
The standard set is 0, 25, 50, 75, and 100 percent, run both ascending and descending. The five points span the full travel and the middle points reveal any nonlinearity, while running both directions exposes deadband and hysteresis. Command each step from the control system output rather than a jumper at the positioner so the whole loop, including wiring and the analog output card, is included in the check.
Remove the command signal or bleed off the instrument air and watch the valve move under spring force to its designed position, whether that is fully closed, fully open, or in place. Confirm it reaches that position completely and in the correct direction. A loop check is often the only planned opportunity to deliberately take the air away and prove the fail action, so it should not be skipped even when everything else looks correct.
Both come from lost motion in the assembly. Deadband, the command change needed before the valve starts moving after a reversal, and hysteresis, the position difference between the up and down passes at the same command, arise from packing friction, worn or loose linkage, and actuator lost motion. A small symmetric amount is normal from packing, but a large or growing value means the valve cannot hold position precisely and points to wear or a loose feedback connection.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.