What an Actuator Does and How Far It Actually Went
Why this matters
Three numbers get called "position" on a service call: what the controller commanded, what the feedback signal says, and where the driven thing physically is. They are three separate quantities produced by three separate mechanisms, and almost every ticket written as "actuator failed" is one of them being trusted as if it were the other two. The command costs nothing to read and proves nothing. The feedback is often a copy of the command wearing a different name. Only the third one moves air or water, and it is the only one nobody measures.
The call
An indoor air handler with an economizer. Complaint: on mild mornings the space overcools late and the unit runs mechanical cooling when it should be riding on outside air.
At the controller: outside air 55, return air 75, mixed air 68, economizer command 100 percent outside air, actuator feedback 100 percent. The screen is internally consistent. Everything agrees. Nothing agrees with the building.
The number that broke the story
Mixed air is a mass-weighted blend of the two streams, so the fraction of outside air can be solved from the three temperatures: return minus mixed, divided by return minus outside air. That is 75 minus 68, over 75 minus 55, which is 7 over 20, or 35 percent.
Check it forward: 0.35 times 55 is 19.25, and 0.65 times 75 is 48.75, summing to 68.0. The arithmetic closes.
That solution assumes both streams have similar specific heat, which they do at these conditions, and that the mixed-air reading is taken past the point of full mixing. A single-point sensor in a stratified mixed-air plenum reads whichever stream washes over it and produces a fraction that is not a blend of anything - which is why an averaging element strung across the plenum face is standard on economizer control, and why the first thing checked on this unit was that it had one. It did. A sibling card covers what sensor location does and does not buy you.
So: commanded 100 percent, reported 100 percent, delivering 35 percent. Two of the three numbers were wrong, and both of them were on the screen.
Why the feedback agreed with the command
The actuator's feedback signal did not come from a position sensor on its output shaft. It was generated inside the same device that received the command, from the command, on the assumption that the motor ran for the commanded duration and therefore arrived.
A feedback signal produced inside the device that received the command is a repeat of the command, not a measurement. It confirms the actuator believes it went. It cannot detect a slipped coupling, a jammed blade, a linkage that hit an obstruction, or a stroke that was set wrong at commissioning, because none of those events are visible from inside the actuator housing.
Some actuators do carry a genuine position sensor referenced to the output shaft, and those will catch a stalled motor or a stroke that never completed. Even those cannot see past the coupling: they report where the actuator's own shaft is, which is not where the damper blade is. There is no feedback signal anywhere in this class of device that reports the position of the driven element, unless somebody deliberately installed a separate end switch or position transmitter on the blade itself.
Establishing where it actually went
This part is physical and it happens with the equipment shut down.
Before opening anything. Isolate the unit, lock and tag it under 29 CFR 1910.147, and confirm the blower wheel has stopped by sight, because a belt-drive wheel coasts long after the contactor drops out. A spring-return actuator holds mechanical energy whether or not it is powered: with control power removed, let the spring drive the device fully to its rest position and confirm it has arrived before your hands go near the linkage, and keep them out of the plane the blade and its links sweep through even after that. If the actuator is non-spring-return, the blade can still fall under its own weight or under duct pressure when the coupling is loosened, so block it before you loosen anything.
Mark two things, not one. Put a mark on the actuator's output shaft against a fixed reference on its housing, and a separate mark on the damper blade shaft against a fixed reference on the frame. One mark tells you the actuator moved. Two marks tell you whether the movement got across the coupling.
Sweep the full range and measure both. Command 0 percent, record both marks. Command 100 percent, record both marks.
On this unit the actuator output shaft swept the full 90 degrees between the two commands. The damper blade shaft swept 32 degrees. The set screw on the coupling had backed off enough to slip under load, so the actuator ran its full travel into its own internal stops every time while the blade came along for part of the ride.
What the two numbers do not tell you
The blade travelled 32 of 90 degrees, which is 36 percent of its rotation, and the mixing solution said 35 percent outside air. Those two numbers landing close together is a coincidence of this particular damper and this particular system, and reading a relationship into it is the mistake this section exists to prevent.
Blade angle is not proportional to flow. The installed characteristic of a damper depends on its authority, meaning the ratio of the damper's own pressure drop when fully open to the total pressure drop of the path it controls. At low authority a parallel-blade damper passes well over half its full flow at a third of its travel, because the rest of the system, not the damper, is setting the flow. At high authority the same damper at the same angle passes far less. Opposed-blade construction changes the curve again.
The practical rule that follows: measure the controlled variable, not the angle. The angle tells you whether the actuator's motion reached the blade, which is a mechanical question. The controlled variable tells you what the system is doing, which is the question the customer asked. You need both, and neither substitutes for the other.
The three quantities, and what each one is good for
| Quantity | How you get it | What it proves | What it cannot show |
|---|---|---|---|
| Command | Read it on the controller | The control logic decided to move | Nothing about the device |
| Feedback signal | Read it on the controller or at the actuator | The actuator believes it arrived, or, if genuinely sensed, that its own shaft arrived | Anything past the actuator's output shaft |
| Physical position | Mark and measure at the driven element, powered down | Where the blade or stem actually is | What that position does to flow |
| Controlled variable | Measure the process quantity the loop exists to control | What the system is actually delivering | Which link caused a shortfall |
Work down that table on any "actuator failed" call and the diagnosis falls out of which row first stops agreeing with the row above it.
Where this changes
On a two-position device, the failure surface is much smaller and end switches carry most of the load. A device with genuine end switches on the driven element is reporting physical position, and the marking exercise is only needed when the switches themselves are suspect or when they are mounted on the actuator rather than on the blade.
On a valve rather than a damper, the same three quantities apply but the controlled variable is harder to read, because a heat transfer coil delivers most of its capacity in the first part of the stroke. That nonlinearity hides a partial-stroke problem at design load and exposes it at part load, which is why a valve with a mis-set stroke commissions clean and gets worse as the season mildens. A sibling card covers verifying a commanded position properly.
Where the actuator drives something that serves a protective function, a stuck or partial position is not a performance problem to be measured at leisure. Establish the protective function's state first, and do not command the device through its range to test it while the process it protects is running.
How to verify you got this right
After correcting the coupling, re-run the same evidence in the same order, because a repair that only satisfies one of the three quantities is how this fault comes back.
Re-sweep both marks. Both shafts must move the same angle across a 0 to 100 percent command, within the play you can see by eye. If the blade shaft still falls short, the stroke is set wrong rather than the coupling being loose, and that is a different adjustment.
Re-solve the mixing fraction at a commanded 100 percent. With outside air and return air re-measured at the time of the test, the fraction should land near 100 percent, not near the commanded number on the screen. Two readings of the same three temperatures, ten minutes apart, guard against catching a transient.
Do not accept feedback as confirmation. It agreed before the repair and it will agree after. Nothing about the repair changed what that signal is made of.
References
- 29 CFR 1910.147 - control of hazardous energy, for isolation, lockout and the release of stored mechanical energy in a spring-return actuator before working on a linkage
- Manufacturer documentation for the actuator's stroke, feedback type (echoed command versus sensed position) and coupling torque, and for the damper's flow characteristic
- Trade-standard practice for economizer mixed-air sensing and for damper authority in system design
- See related: How to Verify an Actuator Reached Its Commanded Position; The Failure Position and Why It Was Chosen; Where a Sensor Is Reporting From and Why It Matters