Lighting Control and Occupancy Sensor Commissioning

Purpose

Leave a control system the occupants will not defeat. The sensor that gets taped over, jumpered out, or called in as a warranty failure is almost never a dead sensor. It is a working sensor still on its factory time delay and sensitivity, dropping the lights on a person sitting still at a desk, in a room whose occupancy pattern nobody looked at before the device went up.

It also guarantees that a control never switches something it must not switch. A sensor placed on a circuit that also feeds an emergency head will turn off egress lighting on a timer, and it will pass every test anyone thinks to run, because the fixture works perfectly right up to the moment it matters.

Scope

Covers installation, replacement and commissioning of wall-box and ceiling-mounted occupancy and vacancy sensors, daylight sensors, and time-switch or panel-level lighting controls in small commercial interiors: offices, retail, restaurants, and the common areas of multifamily buildings.

Does not cover networked or addressable systems with a commissioning agent and a submitted sequence of operations, which belong to that agent; the state acceptance-test forms where a jurisdiction requires a certified acceptance technician, which the California Title 24 Recessed Lighting Airtight SOP touches for that state; or fixture selection and layout.

Roles and handoffs

Procedure

1. Read the room before choosing where anything mounts. Note the space type, where people sit, how still they sit, what blocks line of sight, and where diffusers blow. Acceptance: a per-space note recording occupancy pattern, obstruction, and the sensor technology chosen against it, with the reason. Wrong looks like passive infrared specified for a room where people work seated behind a partition, since infrared detection depends on motion crossing its field and a seated typist barely crosses anything. Stop rule: a space whose required control function is unclear from the permit documents goes back to the estimator before a hole is cut. Hazard: none at this step, it is an observation made from the floor.

2. Prove the switch box dead, then confirm what conductors are actually in it. Open the branch circuit, lock and tag under 29 CFR 1910.333(b)(2), and prove dead on a known live source before and after per NFPA 70E-2021, 120.5. Then identify a grounded conductor at the box, which NEC 404.2(C), in the edition your jurisdiction has adopted, requires at most switch locations with stated exceptions. Acceptance: zero volts at the box on both proofs, and a grounded conductor present or the applicable exception identified in writing. Wrong looks like discovering there is no neutral after the old switch is off the wall and the customer is standing there. Stop rule: no neutral and no applicable exception means the device selection changes or the branch is re-pulled; it is never solved by borrowing a neutral from another circuit. Hazard: this is qualified-person work under 29 CFR 1910.332 and 1910.399, and a switch loop can be fed from a circuit the panel schedule does not name, which is exactly what the before-and-after proof catches.

3. Check the sensor against the load, not just the room. Confirm the device's rating covers the connected load type and that the load exceeds its stated minimum. Acceptance: connected load inside the marked minimum and maximum for that load type, read off the device label and recorded. Wrong looks like a switch rated for a magnetic ballast load holding on a small LED driver bank below its minimum, which produces a fixture that glows faintly when it should be off. Stop rule: a load outside the rated band stops the install and the device is exchanged rather than tried. Hazard: none at this step beyond the step 2 isolation, which stays locked.

4. Mount and aim for coverage, then prove the coverage by walking it. Set the position away from diffusers and out of the shadow of beams and shelving, then use the device's walk-test mode. Acceptance: detection at every point of a grid walk covering the full space, including the two most remote corners and the seated positions, with detections and misses recorded. Wrong looks like a walk done straight toward the sensor, which is the one direction infrared detects worst and so flatters the result. Stop rule: any miss inside the covered space moves the aim or adds a device; the delay is never lengthened to hide a coverage gap. Hazard: cutting or drilling a ceiling of unknown vintage can disturb texture, tile or insulation that may contain asbestos, so unidentified ceiling material is not cut, drilled or scraped until the owner has it identified, because disturbing it is an inhalation route no lighting job justifies.

5. Set the time delay and the sensitivity per space type, not per factory default. Our shop starts at 20 minutes in private offices and conference rooms, 15 minutes in open office areas, and 5 minutes in restrooms, closets and storage, then tunes from complaints and from the walk-test result. Acceptance: the delay and sensitivity written on the settings sheet for every space, matching what is set on the device. Wrong looks like a room left on the factory default because the dip switches were behind the lens. Stop rule: a delay set beyond what the adopted energy code permits for that space type fails the inspection, so the code limit governs and the shop default yields to it. Hazard: none at this step, it is a setting made at the device with the circuit still isolated.

6. Restore power and verify all four control functions, including the one that must not be controlled. Close the breaker from the hinge side rather than square in front of the door, then test turn-on, hold with an occupant seated and still, turn-off after the set delay, and manual override. Then confirm every emergency and egress-fed fixture on that circuit stays lit when the sensor times out. Acceptance: lights on within the device's stated response, holding continuously through a seated test at least as long as the set delay, off within the delay plus its tolerance after vacancy, override working, and the emergency head still lit with the controlled load off. Wrong looks like a test performed by walking in and out, which never checks the seated hold or the emergency separation. Stop rule: an emergency-fed fixture that goes dark on timeout is a wiring defect corrected before handover, never solved by disabling the sensor. Hazard: re-energizing is where a conductor pinched behind the device announces itself, so approach the panel door from the side and keep the working space clear.

7. Run the functional test the adopted energy code asks for, and record who ran it. IECC Section C405 and ASHRAE 90.1 Section 9, in whichever your state has adopted and in that edition, require lighting controls to be functionally tested, and some editions and states require a party other than the installer to perform it. Acceptance: a completed record naming the tester, the date, each space, and the result for each required control function. Wrong looks like the installing tech signing their own test where the adopted edition does not allow it, which the inspector rejects. Stop rule: an unclear requirement is resolved with the AHJ before the inspection is called, not during it. Hazard: none at this step, it is paperwork completed at a desk.

8. Brief the occupants and leave the override where they can find it. Show whoever works in the space how to force lights on and off and what the delay is. Acceptance: a named person who can demonstrate the override, and the settings sheet left with the manager. Wrong looks like a crew leaving without telling anyone, which is how a sensor gets taped over in week two. Stop rule: no available occupant means the sheet goes to the manager with a note to brief them. Hazard: none at this step, it is a conversation held in a lit space.

The record this produces

One settings sheet per space: space name, sensor type and technology, mounting position, the walk-test result as detection points achieved out of points walked, the delay in minutes, the sensitivity setting, the connected load against the device's rated band, and the step 6 result for each of the four control functions including the emergency separation check. Plus one functional test record per project naming the tester.

The manager keeps the sheet, because the first complaint call is answered by reading it rather than by dispatching a truck. The office keeps it because a room whose delay was raised twice is a coverage problem being managed as a timing problem, and that is only visible across visits.

Worked pass: four spaces in a leased office suite

Step 1: two private offices with seated desk work, one open area of six workstations, one restroom. Dual-technology sensors chosen for the offices and the open area on the seated-occupancy reason; a passive infrared wall-box vacancy sensor for the restroom.

Step 2: breaker 12 locked and tagged, proved dead on a known live source before and after, zero volts at each box. Grounded conductors present at three of the four boxes. The restroom switch is an old switch loop with no neutral, so under step 2's stop rule the device selection changes to a model listed for that condition rather than a neutral being borrowed from the adjacent lighting circuit.

Step 3: the open-area load is 8 LED troffers, inside the device's marked band for electronic loads. Recorded and passed.

Step 4: walk test on a grid. Offices detect at 8 of 8 points each, restroom at 4 of 4. The open area detects at 11 of 12; the miss is the far corner workstation, shadowed by a structural beam. Under step 4's stop rule the delay is not touched. A second sensor is added at the far end and the walk repeats at 12 of 12.

Step 5: 20 minutes set in each private office, 15 in the open area, 5 in the restroom, all written on the sheet and confirmed against the adopted energy code's limits for those space types.

Step 6 is where the pass fails. Turn-on, seated hold and override pass in all four spaces: the open area holds continuously through a 20-minute seated test against a 15-minute delay, and turns off 16 minutes after vacancy. Then the emergency separation check: with the open-area load off, the emergency-fed troffer by the corridor door goes dark with it, because it is on the same switched conductor. Step 6's stop rule fires, the sensor is not disabled, and that fixture is re-landed ahead of the control. Retest: controlled load off, emergency troffer lit.

Step 7: the second tech runs and signs the functional test on all four spaces, since the installing tech's signature is not accepted under the edition this permit was pulled under.

Step 8: the office manager demonstrates the override, keeps the settings sheet, and is told the restroom goes dark five minutes after the last movement, which is the one number people call about.

References

• NEC 404.2(C), in the edition your jurisdiction has adopted, for the grounded conductor required at switch locations and its exceptions
• IECC Section C405 and ASHRAE 90.1 Section 9, in whichever your state has adopted and in that edition, for automatic shutoff, occupant sensor requirements and the functional testing of lighting controls; confirm with the AHJ who is permitted to sign the test
• 29 CFR 1910.333(b)(2) with 1910.332 and 1910.399, and NFPA 70E-2021, 120.5, for the isolation and live-dead-live sequence at step 2
• The specific device's installation instructions, which govern minimum and maximum connected load, walk-test mode and coverage pattern
• See related: Emergency and Egress Lighting Annual Test SOP, Dimmer Selection and Installation SOP, California Title 24 Recessed Lighting Airtight SOP