How to Verify an Actuator Reached Its Commanded Position

Why this matters

An actuator that runs its full travel and moves the driven element only part way commissions clean, reports normal, and cripples the system at exactly the load where nobody is watching. The feedback signal will not catch it, the controller will not catch it, and a capacity check at design load frequently will not catch it either, because a heat transfer coil delivers most of its output in the first part of the stroke. The only measurement that settles it is a physical one, taken at the driven element, across the full range plus a midpoint. This is that procedure.

1. Establish the failure position and neutralize stored energy before your hands go anywhere

A spring-return actuator holds mechanical energy whether or not it is powered. Find out which way it goes at rest, from the marking on the housing rather than from the drawing, then isolate and lock out the equipment under 29 CFR 1910.147 and let the spring drive the device fully to that rest position before you approach the linkage. Confirm arrival by sight.

For a non-spring device, the blade or stem can still move under its own weight or under system pressure once a coupling is loosened, so block the driven element mechanically before loosening anything. If the valve is on a hot or pressurized circuit, do not break the pressure boundary at all for this test - everything in this procedure is external observation. If a repair later requires opening the circuit, isolate it, depressurize it, drain it and let it cool first, because a hot water line will scald on contact well below boiling.

Skipping this step is how somebody ends up with a hand between a damper blade and its frame when control power is restored.

2. Set a fixed reference and a measurement method with a stated resolution

You need a datum that does not move: a scribed mark on the actuator housing, a bracket, a frame edge. Mark the actuator output shaft against one datum and the driven element against a separate datum. Two marks, because one mark cannot tell you whether the motion crossed the coupling.

State how you are measuring and to what resolution before you take a reading. For a linear valve stem, a depth gauge or a rule referenced to a fixed point, resolving to a hundredth of the stated stroke. For a rotary damper, an angle finder against the frame, resolving to a degree. Without a stated resolution you cannot later tell a real difference from a reading error, which is exactly the judgement step 5 depends on.

3. Find out what the feedback signal is made of, before you use it for anything

Read the actuator documentation and determine whether its feedback output is generated from the command inside the device or sensed from the output shaft. If it is generated from the command, it is a repeat of the command and it carries no information about position at all. If it is sensed, it reports the actuator's own shaft and still says nothing about what happened past the coupling.

Do this now rather than later, because a tech who believes the feedback is a measurement will stop the procedure here when the numbers agree. A sibling card covers the three quantities this distinction separates.

4. Sweep the full range under command and record both marks

Command 0 percent. Record both marks. Command 100 percent. Record both marks. Wait for the actuator to complete its travel and stop moving before each reading, using the device's stated full-stroke timing as the minimum wait rather than eyeballing it.

Do not put your hands in the sweep path to steady anything, and do not hold a gauge against a moving element. On a spring-return device you cannot remove power to take the reading, because removing power is what drives it home, so the measurement has to be taken from outside the path the linkage sweeps, referenced to your fixed datum. Any command action means the control circuit is live; if that requires being inside an enclosure carrying line voltage, de-energize and prove dead with live-dead-live against a known source under NFPA 70E-2021, 120.5 for the line-voltage side, and if the test genuinely requires energized work, that is the narrow condition 29 CFR 1910.333(a)(1) permits it under, with a meter and leads rated for the circuit's category and available fault current plus the protection the arc-flash assessment calls for.

5. Add the midpoint, in both directions

Command 50 percent from below, record. Command 50 percent from above, record. This is the step people skip and it is the one that separates two faults that look identical at the endpoints.

A device that is short on total travel but linear within its own range gives a midpoint at exactly half its measured full travel, and the same value from both directions. A slipping coupling or a worn linkage gives a midpoint that differs depending on which way you arrived, and the difference is the backlash. Both endpoints can be identical in the two cases; only the midpoint tells them apart.

6. Judge against a stated tolerance with a stated Boolean

Write the tolerance before you take the readings. A workable default, tuned to the device documentation where it gives one: full travel within 5 percent of the device's stated stroke, and the midpoint within 5 percent of half the measured full travel, and both must pass. The Boolean is load-bearing. An OR, or an average of the two, passes a device that is badly short on range but perfectly linear, which is the most common version of this fault.

7. Only now measure the controlled variable, and only at the endpoints

Measure what the system delivers with the device commanded fully closed and fully open. That catches a valve that does not seat, which is a leak-by problem the travel measurement will not reveal if the stroke reference was taken from the wrong end.

Do not try to verify position from a midpoint capacity reading. A heat transfer coil's output is strongly nonlinear in flow, and a control valve's installed characteristic depends on its authority, meaning the ratio of the valve's own full-open pressure drop to the total drop of the circuit it controls. At low authority an equal-percentage plug behaves nearly quick-opening and most of the coil's capacity arrives in the first part of the stroke, so a half-open valve can deliver most of full output. A midpoint capacity reading in that condition tells you nothing about where the stem is.

The worked check

A modulating valve on a hydronic heating coil. Nameplate valve stroke is stated on the device label; call it S and measure everything as a fraction of it, with a depth gauge resolving to 0.01 S. Tolerance set in advance per step 6.

Ascending sweep:

Descending sweep returned 0.72, then 0.35, then 0.00.

Read the linearity first. On the ascending sweep, 0.36 divided by 0.72 is exactly 0.50, so the command-to-travel relationship is linear within the range the device actually covers. The descending midpoint of 0.35 differs from the ascending 0.36 by 0.01 S, which is one count at the gauge's stated resolution. That is at the limit of what the measurement can distinguish, so it is not evidence of backlash; it is not evidence of the absence of backlash either, and if backlash mattered here the honest next move is a finer gauge, not a confident sentence.

Read the range second. Full travel is 0.72 S against a stated stroke of 1.00 S, which is 28 percent short. That fails the 5 percent full-travel tolerance outright. The midpoint at 0.36 against half of the measured 0.72, which is 0.36, deviates by zero and passes its own test. Under the AND stated in step 6, the device fails. Under an OR or an average, it would have passed, which is why the Boolean was written down first.

Read the closed end third. The stem reached 0.00 against a fully-seated reference at 0 percent command, so it does close. Had the range been mis-set the other way, the same 0.72 S of travel could have run from 0.28 to 1.00, leaving the valve permanently off its seat with a leak-by nobody would find with a travel measurement alone.

What this costs the system, stated carefully. How much capacity 28 percent of missing lift removes depends entirely on the trim characteristic and the installed authority, and cannot be read off the travel numbers. At low authority almost none of it shows at design load, which is precisely why the unit commissioned clean. What is certain from the numbers alone is the loop gain: the controller's full 0 to 100 percent command now moves the stem across 0.72 of its stroke instead of 1.00, so each percent of command produces 28 percent less stem motion than the design assumed. The loop is slower to correct than it was tuned to be, and its top end is capped at whatever 0.72 S passes. That reads in the field as a system that holds fine in mild weather and cannot reach setpoint on the coldest morning.

The fix is a stroke or range adjustment at the actuator, not a new valve and not a new actuator, and it is confirmed by re-running steps 4 through 6.

How to verify you got this right

Re-run the full sweep after any adjustment, both directions, and record it. An adjustment made and not re-measured is a guess. The full travel must now land within the stated 5 percent of S, and the midpoint within 5 percent of half of that.

Confirm both marks still move together. If the driven element's travel came into tolerance because the coupling was re-clamped, verify at 0, 50 and 100 that the two shafts are still in step. A coupling tightened without a torque reference tends to slip again under spring load, and the spring load is the one you never see in normal operation.

Leave the numbers on the equipment. Write the measured full travel and the midpoint on a tag or in the equipment record with the date. The next tech's fastest route to this fault is a prior reading to compare against, and without one they will repeat this entire procedure to learn what you already knew.

If a protective device was tripping before the adjustment, do not close the ticket on the travel fix alone. Establish what condition reached that device and confirm it no longer does, on measurement. A limit that opened because the process genuinely reached its trip point was working, and a repair that merely stops it from opening reaches the same end state as jumpering it.

References

• 29 CFR 1910.147 - control of hazardous energy, covering isolation and the release of stored mechanical energy in a spring-return actuator before work on a linkage
• 29 CFR 1910.333(a)(1) - live parts to be de-energized before work, and the narrow conditions permitting energized troubleshooting
• NFPA 70E-2021, 120.5 - process for establishing and verifying an electrically safe work condition
• Manufacturer documentation for the actuator's stated stroke, full-stroke timing, feedback type and range adjustment, and for the valve's trim characteristic
• See related: What an Actuator Does and How Far It Actually Went; The Failure Position and Why It Was Chosen