How-to
Fixing voltage drop on a long 277 V lighting feeder
Short answer
The drop budget on this feeder limits the loop to 0.40 Ω — 0.267 Ω per conductor — so size up the copper or split the load across the three phases of the wye system, which relaxes the limit sixfold. Raise the source only after the two-endpoint check, because a fixed boost that fixes full-load voltage also overvoltages the panel at light load.
On this page
- How much voltage drop is allowed on this feeder?
- What loop resistance does the drop budget allow?
- Does this run need copper well above ampacity size?
- Should I split the load across the three phases?
- Do I need a 480/277 V transformer at the source?
- Can transformer taps replace bigger wire?
- Can a buck-boost transformer go at the source end?
- How do I verify the fix and close it out?
- Questions asked
Fix this with conductor resistance, not transformation. The lighting voltage already exists: on a 480 wye system each phase-to-neutral circuit is 480 V ÷ 1.732 = 277.1 V, so nothing has to convert anything. The drop budget at full lighting load limits the loop to 0.40 Ω, which with equal conductors is 0.20 Ω each — 0.267 Ω on the per-1000-ft basis used in the tables. Splitting the load across the three phase-to-neutral circuits relaxes the limit to 1.60 Ω. Raising the source trades a full-load sag for a no-load rise, so boosting is the last move.
How much voltage drop is allowed on this feeder? #
Plan for 8.31 V, which is 3% of the phase-to-neutral voltage and matches the feeder figure that NFPA 70, the National Electrical Code, 2023 edition, recommends in its informational notes alongside 3% for branch circuits and 5% combined. The run from the electrical room to the lighting panel is a feeder; the luminaires' own circuits are branch circuits with their own share of the budget. Taking the full 3%: 0.03 × 277.1 V = 8.31 V, and the target used through this article sits just inside it. NFPA 70 marks these figures as informational notes — recommendations rather than mandatory rules — but specifications and inspectors enforce them routinely, so confirm the application with your authority having jurisdiction and the equipment documentation.
What loop resistance does the drop budget allow? #
The loop — phase conductor out, neutral back — may total no more than 0.40 Ω, which with equal phase and neutral conductors is 0.20 Ω each, or 0.267 Ω per 1000 ft at operating temperature. Worked example, stated inputs: single-phase load at the lighting voltage, 20 A, one-way length 750 ft, equal copper phase and neutral conductors.
- Allowed loop resistance: allowed drop ÷ current = 8 V ÷ 20 A = 0.40 Ω
- Each conductor: 0.40 Ω ÷ 2 = 0.20 Ω over the run
- Per-1000-ft basis: 0.20 Ω scaled up by 1000 ÷ 750 = 0.267 Ω
The current returns through the neutral, so the calculation uses the whole loop, not one conductor. Use AC resistance at conductor operating temperature — NFPA 70, Chapter 9, Table 9 lists it at 75 °C by raceway type — not room-temperature DC resistance. For loads with poor power factor or harmonic current, run the final check with resistance and reactance against measured current; straight I × R is a screening estimate, and the final selection gets checked against the governing code and the connected equipment's data.
Does this run need copper well above ampacity size? #
Yes — on a run this long voltage drop, not ampacity, sets the conductor size, and you take the smallest copper whose 75 °C AC resistance in NFPA 70 Chapter 9, Table 9 is at or below the 0.267 Ω per-1000-ft figure for your raceway. Expect a conductor several sizes above anything this load requires for ampacity alone. Steel conduit adds reactance, so where it applies use the effective-impedance column for your raceway rather than resistance alone. Check that the panel and breaker terminal markings accept the conductor size and insulation temperature you land on. Larger copper also cuts standing loss: at the limit the two conductors dissipate 2 × (20 A)² × 0.20 Ω = 160 W, and every size up reduces that for the life of the feeder.
Should I split the load across the three phases? #
Yes, and do it before any boosting: the load split evenly across the three phase-to-neutral circuits is 6.67 A per phase. Worked example, balanced case, same budget:
- Phase current: load ÷ 3 = 6.67 A per phase-to-neutral circuit
- Allowed conductor resistance: budget ÷ 6.67 A = 1.20 Ω each
- On the per-1000-ft basis: 1.60 Ω against 0.267 Ω
That is six times the single-circuit allowance, because the current falls to a third and a balanced three-phase feeder no longer doubles the drop through a shared return. Steps:
- Inventory the fixture load circuit by circuit rather than trusting the panel total.
- Assign about one third of the fixtures to each phase.
- Measure each phase current and each phase-to-neutral voltage under normal operation.
- Recalculate drop from the current each conductor actually carries.
The neutral needs its own check: electronic drivers draw third-harmonic current that adds in the neutral instead of cancelling, so measure the neutral and size the grounded conductor for the current you read.
Do I need a 480/277 V transformer at the source? #
No — a separate 480/277 V transformer buys nothing, because the voltage already exists: on the wye system each phase-to-neutral circuit is the lighting voltage while phase-to-phase is 480 V. The plant's supply transformer — a 4160 V-class primary in this installation — already delivers the lighting voltage, so the problem is drop, not transformation. A two-winding unit added in front of the panel brings no-load loss, load loss, and nameplate impedance that worsens regulation and makes inductive-load starting harder; read %Z from the nameplate rather than assuming a value. Decision: no additional transformer.
Can transformer taps replace bigger wire? #
Only if both endpoints pass: a tap change on the supply transformer moves the secondary in steps — typically 2.5%, worth 6.9 V on the phase-to-neutral circuit — and raises no-load voltage by the same amount it lifts full-load voltage. Typical nameplates offer two 2.5% raises and two 2.5% lowers; the nameplate and its tap diagram control. One raise step is 6.9 V, short of what this feeder drops. Scenario check, illustrative inputs: a fixed boost of the full shortfall puts 285 V on an unloaded panel, which is 285 × 1.732 = 494 V line-to-line. If the upstream secondary also runs 5% high at light load — a condition to measure, not assume — the stack reaches about 299 V line-to-neutral, or 518 V line-to-line. Both endpoints must sit inside the marked ratings of the panel, breakers, luminaires, and drivers. Decision: taps trim a feeder that is nearly right; they do not rescue one that is undersized.
Can a buck-boost transformer go at the source end? #
Yes — a buck-boost unit is not restricted to the load end; you can install it in the electrical room at the source, provided the connection diagram shipped for that exact unit matches your phase arrangement, neutral, grounding, current, and boost ratio. Installed there it boosts the whole feeder at one fixed ratio, so the no-load endpoint moves exactly as it does with a tap. As a connected autotransformer it provides no isolation, which this service does not need. It is also small: the boost portion is only 160 VA (the boosted volts times the load current) against the 5.5 kVA load, so the unit and its overcurrent protection follow the manufacturer's sizing and connection data for that exact model. Decision: allowed at the source end, chosen only after the conductor and balancing branches fail, and only when both endpoints pass.
How do I verify the fix and close it out? #
Close it with simultaneous source-and-remote readings at full load and at light load, true-RMS instruments, and both endpoints inside the marked ratings of the panel, breakers, drivers, and luminaires.
- De-energize, lock out, and test for absence of voltage before opening any panel or terminal box; NFPA 70E, 2024 edition sets the safe-work practices and protective equipment for the live readings this procedure needs.
- Record all three line-to-line and all three line-to-neutral voltages at the source under maximum lighting load, and again at minimum load.
- Measure each phase current and the neutral current with a true-RMS clamp meter.
- Rebalance separable lighting circuits across the three phases and repeat the readings.
- Compute each conductor's drop from measured current times its Chapter 9, Table 9 impedance, including terminations and splices.
- If a boost still seems necessary, calculate the minimum full-load and maximum no-load remote voltage for each available tap or buck-boost connection before energizing anything.
- Accept the arrangement only when the remote panel holds within 8.31 V of the source at full load and every no-load reading stays inside each device's marked ratings.
The decisions, by case:
- Single circuit kept: copper at or below the 0.267 Ω per-1000-ft limit per conductor at 75 °C, terminals verified.
- Load split three-phase: conductors at or below the 1.60 Ω relaxed limit, neutral sized from the measured harmonic current.
- Boost unavoidable: the one setting whose two endpoints, including measured upstream rise, fit inside every marked rating — otherwise, more copper.
Questions asked
Can aluminum conductors meet the same 0.267 Ω limit?
The limit does not change with material: the per-conductor resistance figure at operating temperature stands for the single-circuit case. Aluminum conducts roughly 61% as well as copper for the same cross-section, so an equal-size aluminum conductor reads about 1.6 times the resistance and you step up sizes until the listed AC resistance meets the limit. Where aluminum is used, terminals and lugs must be listed and installed for it.
Does the conductor also have to pass an ampacity check?
Yes, and it is a separate calculation from voltage drop. Lighting counts as a continuous load, so NFPA 70, 2023 edition sizes the feeder conductor at 125% of the load: 25 A minimum ampacity before temperature or fill corrections, unless the overcurrent device and assembly are rated at 100% of their ampere rating. On a run this long the drop requirement lands on a much larger conductor anyway, so verify ampacity last, not first.
Is 285 V line-to-neutral a problem for 600 V equipment?
285 V line-to-neutral is 494 V line-to-line (285 × 1.732), under the 600 V maximum system voltage this gear is typically marked with — but read the marking on each panel, breaker, and driver, because the marking and listing control, not the percentage. The number that actually bites is the light-load endpoint, where the fixed boost stacks on any upstream rise.
How much standing loss does the three-phase split save?
With the same conductors, single-phase loss is 2 × I² × R and the balanced three-phase loss is 3 × (I ÷ 3)² × R — one sixth as much. At this feeder's numbers, 160 W falls to about 27 W over the run, saved every hour the lights are on.
Why do the verification readings need a true-RMS meter?
Electronic lighting drivers draw pulsed current rich in harmonics, and average-responding meters calibrated for sine waves misread it — both the current and any distorted voltage. A true-RMS clamp meter and true-RMS voltmeter report the heating-equivalent value the conductor actually carries. Measure the source and remote ends at the same time, and log one full-load window and one light-load window.