Under-Cabinet and Accent Lighting Installation

Purpose

Under-cabinet and cove lighting is sold on how it looks and fails on two things nobody photographs: where the driver ended up and how far the run is from it. A driver screwed to the back of a cabinet that later gets loaded with cereal boxes is a heat problem and an access problem at once, and a constant-voltage run fed from one end goes visibly dimmer toward the far cabinet in a way the customer sees the first night and cannot unsee.

This procedure guarantees the line-to-Class-2 boundary lands inside a box that stays reachable, that every splice is in an enclosure rather than taped above a cabinet, and that the far end of the run is measured rather than eyeballed. The number that catches the dimming problem is a voltage read at the far end of the tape while it is running, and it takes about a minute.

Scope

Covers permanently installed under-cabinet luminaires, linear tape and cove or toe-kick accent lighting on 120 V supply with a Class 2 driver, in dwelling kitchens, small offices and small retail display work.

Does not cover dimmer selection, which the dimmer selection and installation SOP owns and which this procedure hands the driver's dimming protocol to. Does not cover installing a new branch circuit or a new switch location where none exists, which is a rough-in scope. Does not cover the isolation sequence, owned by the lockout tagout for branch circuit work SOP and consumed at step 2, or general luminaire replacement, which has its own standard.

Roles and responsibilities

Procedure

1. Measure the run and read the load before ordering anything. Acceptance: total run length in feet, the tape or luminaire's listed watts per foot or per unit, the total connected watts computed, and the supply point identified with its existing load. Wrong looks like a run estimated from a cabinet elevation rather than measured with a tape along the actual path, which is always longer. Stop rule: a supply point that is a required countertop receptacle outlet under NEC 210.52(C) is not consumed by permanently connected lighting; find another source. Hazard: none physical, this is measurement with everything closed; do not stand on a countertop to reach a soffit.

2. Establish an electrically safe work condition at the supply point. Acceptance: zero volts from each ungrounded conductor to every other, to the grounded conductor and to ground, with the meter proved on a known live source before and after, per NFPA 70E-2021, 120.5, under work practices at 29 CFR 1910.333(b)(2). Wrong looks like isolating one kitchen circuit when the box holds two, which kitchens routinely do. Stop rule: any reading above zero in the enclosure means a second source to find first. Hazard: 1910.333(b)(2) is written for qualified persons per 1910.332 and 1910.399, so if nobody on site meets that basis the work waits.

3. Fix the driver location before any hole is drilled. Acceptance: a location that stays accessible without removing building finish or cabinetry, is inside the ambient temperature range the driver's instructions state, is not inside a sealed or insulated cavity, and is close enough to the run to keep the feed short. Wrong looks like a driver above a soffit with no access panel, or one tucked behind a cabinet back where a dishwasher's heat rises. Stop rule: a location that fails any of those is rejected before it is convenient, because a driver is the component most likely to be replaced first. Hazard: none physical, this is a decision made with the circuit still isolated; do not test-fit a driver into a cavity you have not looked into with a light.

4. Rough the path. Acceptance: every penetration made through cabinet backs, toe kicks and finished surfaces with a listed grommet or bushing where a conductor passes through metal or a sharp edge, and every wall cavity checked before drilling. Wrong looks like a paddle bit going through a cabinet back into a concealed water line or an unmarked cable. Stop rule: an obstruction you cannot identify by scanning or by opening a small inspection hole is not drilled through. Hazard: cutting or drilling releases dust at the tech's face, and grinding or drilling masonry, tile or concrete for a cove detail releases respirable crystalline silica, so a water or vacuum dust control is used at the tool per 29 CFR 1926.1153 on construction work or 1910.1053 in general industry, and any respirator that either requires is worn only under a written program per 29 CFR 1910.134; pre-1978 painted surfaces bring EPA's RRP rule at 40 CFR 745 Subpart E.

5. Land the line side in a box. Acceptance: the 120 V supply spliced only inside a listed box with a cover, that box accessible per NEC 314.29 and covered per 314.25, the driver's line leads landed there or the driver itself mounted to it as its listing permits, box fill checked against NEC 314.16, and Class 2 conductors kept out of any enclosure or raceway containing light and power conductors per NEC 725.136. Wrong looks like a wire-connector splice taped to a cabinet top, which is a splice outside a box and a violation of NEC 300.15 on the day it is made. Stop rule: no splice is left outside an enclosure to save a box, ever. Hazard: reaching into the supply box passes any second circuit sharing it, which is why step 2 isolated the whole enclosure.

6. Mount the luminaires and run the Class 2 secondary. Acceptance: luminaires fastened to cabinet framing rather than to a thin cabinet bottom, mounted toward the front edge of the cabinet so light falls on the counter and not on the backsplash, connectors seated and locked, and polarity observed on any direct current tape. Wrong looks like a tape stuck to a cabinet bottom with its own adhesive and no mechanical retention, which releases the first summer the kitchen gets warm. Stop rule: adhesive-only mounting on a run intended to be permanent is supplemented mechanically or the product is changed. Hazard: overhead work under cabinets drops fasteners and offcuts toward the face, so eye protection stays on, and no knife cut toward the hand holding the tape.

7. Restore power and measure the electrical acceptance. Acceptance: driver loading at or below the maximum its instructions state, with a shop default of 80 percent of rated output where the instructions give no derating; voltage measured at the far end of the run while it is on, with the total loss from the driver's output terminals to that far end at or under 3 percent of nominal as a shop default; and no visible difference in brightness end to end. Wrong looks like a run that measures fine at the driver and reads several volts lower at the last foot, which is the drop the customer sees as dimming. Stop rule: a run over the loss gate is corrected by feeding it from both ends, splitting it into two runs, or upsizing the feed, and it is not handed over dimming. Hazard: this step puts power back with a tech's hands near a metal counter edge, so the deadfront is on before the breaker closes, the body is to the hinge side at the panel, and nobody is holding an unterminated conductor when it energizes.

8. Verify dimming, heat and appearance, then close out. Acceptance: the dimming protocol matched to the driver per the dimmer selection SOP and swept with no flicker or dropout; the driver's case and the tape's channel checked by hand after thirty minutes and found warm rather than hot to hold; and no visible hot spot, socket shadow or light spill onto the backsplash at eye level. Wrong looks like a driver too hot to keep a hand on, which means it is loaded past its rating or sitting in a cavity with no air. Stop rule: a driver that hot is de-energized and relocated or resized, not left with a note. Hazard: touching an energized driver's case is done with the back of the hand and only on a listed, grounded assembly, never by reaching blindly into a cavity.

The record this produces

One record per run: total run length, watts per foot and total connected watts, driver model with rated output and the loading percentage computed, driver location described well enough to find it, feed conductor size and one-way length, the measured voltage at the driver's output and at the far end with the loss as a percentage, the dimming protocol used, and any correction made at step 7.

The driver location and the loading percentage are the fields with the long life. A driver is the first thing to fail in these systems and the hardest to find without a note; a run recorded at 73 percent of driver output tells the next tech that a customer's request to add another cabinet's worth is a new driver, not a splice. The measured far-end loss is what settles a future dimming complaint without re-measuring.

Worked pass: kitchen under-cabinet tape, 16 ft of run, one driver

Step 1 measures 16 ft of actual run at 4.4 W per foot, so 70.4 W connected, drawing 70.4 divided by 24, or 2.93 A at 24 V. The supply is taken from a switched lighting circuit, not from a required countertop receptacle.

Step 2 proves the supply box dead, meter checked at 121 V on a known live kitchen receptacle before and after. Step 3 places the driver inside the upper cabinet over the range hood run, on the cabinet side wall, reachable by opening one door and outside any sealed cavity.

Step 4 drills three cabinet-back penetrations with grommets, using a vacuum at the bit. Step 5 lands the 120 V supply in a listed box with a cover behind the cabinet's access opening, and the Class 2 conductors leave that box separately from the power conductors.

Step 6 mounts the tape in an aluminium channel screwed to the cabinet bottoms at the front edge, with the adhesive backing supplemented by the channel's own fasteners.

Step 7 fails. The driver is rated 96 W, so 70.4 W is 73 percent of it, which clears the 80 percent default. But the feed is 18 AWG, 12 ft one way, so 24 ft of conductor at the 7.77 ohm per 1000 ft NEC Chapter 9 Table 8 gives for solid uncoated copper at its 75 C basis is 0.19 ohm, and at 2.93 A that is 0.55 V. Measured, the driver's output terminals read 24.0 V, the tape's first foot reads 23.4 V, and the far end reads 22.2 V. From the driver's terminals to the far end is 24.0 minus 22.2, or 1.8 V, which is 7.5 percent of 24 V against the 3 percent gate of 0.72 V. The stop rule fires.

Two corrections are made because two terms are involved. The run is re-fed from both ends, which halves each segment and so quarters the tape's own drop, taking 1.2 V to 0.3 V, with the new worst point at the middle of the run rather than at its end. The feed is upsized to 16 AWG, where 24 ft at 4.89 ohm per 1000 ft is 0.12 ohm, and at 2.93 A that is 0.34 V. Total loss becomes 0.34 plus 0.30, or 0.64 V, which is 2.7 percent of 24 V and clears the gate. Re-measured, the midpoint reads 23.4 V against 24.0 V at the driver, and no brightness difference is visible along the run.

Step 8 sweeps the dimmer with no flicker, and after thirty minutes the driver case is warm and holdable.

References

• NEC (NFPA 70) Article 725 for Class 2 circuits including 725.136, 300.15 for boxes at splices, 314.25 and 314.29 for covers and accessibility, 314.16 for box fill, and 210.52(C) for required countertop receptacle outlets, in the edition your jurisdiction has adopted
• NEC (NFPA 70) Chapter 9 Table 8 for conductor resistance, and 110.3(B) for installation per listing instructions
• 29 CFR 1926.1153 for respirable crystalline silica in construction and 29 CFR 1910.1053 in general industry, with 29 CFR 1910.134 for any respirator use
• 29 CFR 1910.333(b)(2), with qualified-person definitions at 1910.332 and 1910.399; NFPA 70E-2021, 120.5 for the proving sequence
• See related: the dimmer selection and installation SOP, the light fixture replacement standard SOP, the lockout tagout for branch circuit work SOP