Voltage Drop Investigation on a Long Run

Purpose

On a long run to a garage, a well, a barn or a shop, a measured voltage drop on its own tells you nothing, because a long run is supposed to drop. The number that matters is the difference between what the run measures and what its own conductor size, material, length and current say it should measure. Skip that calculation and you either condemn a sound circuit that was simply built too small, or you accept a genuine developing fault because the run is long and everyone expects it to sag.

Getting it backwards is expensive in both directions. Rewiring an outbuilding to fix a corroded splice wastes a day and a trench. Leaving that splice because the run is long leaves a resistive joint dissipating heat in a junction box nobody opens.

Scope

Covers a voltage complaint on a branch circuit or feeder supplying a detached structure or a remote load in a dwelling or small commercial occupancy, through to a finding of undersized-by-design, defective connection, or defective conductor.

Does not cover the general voltage complaint screen, the service neutral check or the recorder deployment, which the Power Quality Complaint Investigation SOP owns and which runs first. Does not cover feeder sizing for new work, load calculation for a service upgrade, or the disconnecting means and grounding design at a separate structure.

Roles and handoffs

Procedure

1. Get the four facts the calculation needs before any meter comes out. Establish conductor material, conductor size, one-way length measured rather than estimated, and the actual current the load draws. Acceptance: all four written down, with material confirmed by stripping and looking at a conductor end rather than by assumption. Wrong looks like assuming copper on a 1970s outbuilding feeder. Stop rule: any of the four unknown stops the procedure, because a calculation missing one is not conservative, it is arbitrary. Hazard: reading a conductor end means opening an enclosure, so the circuit is opened and proved dead on a known live source before and after per NFPA 70E-2021, 120.5, with work practices at 29 CFR 1910.333(b)(2), and a damp outbuilding is treated as a wet location where nobody works energized at all.

2. Compute the expected drop and write the conditions next to it. For a single-phase two-wire run, expected drop in volts is 2 times K times current times one-way length, divided by the conductor's circular mil area, where K is about 12.9 for copper and about 21.2 for aluminum at a 75 C conductor operating temperature. Acceptance: one expected drop in volts, with the material, the K used, the circular mil area from NEC Chapter 9 Table 8, and the current all on the same line. Wrong looks like carrying a copper K onto an aluminum run, which understates the expected drop by about 1.64 times and invents a defect that is not there. Stop rule: on conductors larger than about 1/0, or where power factor is well below unity, this approximation drops reactance and the calculation moves to the impedance values in NEC Chapter 9 Table 9 instead. Hazard: none at this step, it is arithmetic done at the truck.

3. Measure the actual drop at both ends at the same moment, under the same load. With the load running at the current used in step 2, read voltage at the source end and at the point of use simultaneously, either with two meters or one meter and a helper. Acceptance: both readings and the load current recorded within the same minute, with the measurement points named. Wrong looks like reading the far end with the load off, which shows almost no drop on a circuit that collapses under use. Stop rule: a load current that differs from the step 2 figure by more than a couple of amperes means step 2 is recomputed at the measured current before anything is compared. Hazard: energized readings at both ends, so use a meter and leads rated CAT III 600 V or better, seat the leads before looking at the display, and wear the arc-rated clothing and face protection your program assigns per NFPA 70E-2021, 130.5 and 130.7 at the source-end enclosure.

4. Compare measured against expected using one gate, stated once. Acceptance: measured drop within 25 percent of the calculated expected drop, evaluated on the single two-point measurement rather than on an average of several. Wrong looks like comparing the measured drop against the NEC informational-note design recommendations of 3 percent to the farthest outlet and 5 percent combined, and calling the circuit defective on that basis; those are design targets for new work and they say nothing about whether an existing run has a fault. Stop rule: measured inside the gate ends the fault investigation, and the finding is undersized by design. Hazard: none at this step, it is a comparison.

5. Convert the excess into resistance so the finding has a physical size. Where the gate fails, subtract expected from measured and divide by the measured current to get the unintended resistance in the run. Acceptance: one resistance figure in ohms, with the subtraction and the division written out. Wrong looks like reporting only that the drop is high, which gives the next tech nothing to hunt for. Stop rule: an excess that would dissipate serious heat at the operating current is a developing fault rather than a performance complaint, and the circuit stays off until it is found. Hazard: none at this step, but the number changes the urgency of everything after it.

6. Bound the excess by halving the run, not by replacing terminations. Measure at an accessible midpoint under the same load, then at the quarter point on whichever half carries the excess. Acceptance: a span or a device pair whose drop is out of proportion to its length. Wrong looks like re-terminating every splice along the run hoping to catch it. Stop rule: where a single connection carries most of the excess, stop measuring and open it dead, because the power dissipated at a resistive joint is the voltage across it times the current through it. Hazard: opening buried or outdoor boxes means water plus electricity, so the circuit is dead and proved dead before a cover comes off, and a box holding standing water is drained and the enclosure assessed for replacement rather than reclosed.

7. Repair the bounded defect, or quote the run where the finding is design. A corroded splice or a damaged length is repaired in an accessible listed enclosure with connectors listed for the conductor material and, on a copper-to-aluminum joint, listed for that combination. Acceptance: the correct listed connector used, and the enclosure restored to its wet-location rating where it is outdoors. Wrong looks like a standard twist-on connector on an aluminum splice in a damp box, which rebuilds the same fault. Stop rule: conductors damaged in more than one place mean the run is replaced rather than patched. Hazard: stripping and re-terminating throws fragments and cuts, so eye protection and gloves stay on, and charred insulation is treated as a burned-conductor condition rather than reused.

8. Restore, then re-prove against the same expected value. Energize, run the same load at the same current, repeat the two-point measurement. Acceptance: measured drop back inside 25 percent of the step 2 expected drop, at the same current and the same two points. Wrong looks like accepting an improvement without recomputing the gate. Stop rule: still outside the gate means a second excess remains and step 6 resumes. Hazard: this puts energy back into a structure the customer uses, so before the breaker closes confirm the equipment grounding path is continuous to the far structure, confirm the required disconnecting means there operates, and confirm every ground-fault device serving that structure trips and resets on its test button, because the run you disturbed carries the fault path those devices depend on.

The record this produces

One investigation record per run: the four step 1 facts with the length stated as measured; the step 2 expected drop with its K, circular mil area and current on the same line; the step 3 pair of readings with the measurement points and the current; the step 4 comparison against the 25 percent gate; the step 5 excess resistance with its arithmetic; the bounded span; the repair with the connector type; and the step 8 re-measured pair.

The office reads the design-or-defect call to know whether it is booking a repair or quoting a trench. The customer reads the calculated drop next to the measured one, the only way to explain that a long run is allowed to sag and this one sagged twice as much as it should. The next tech reads the conductor material, the fact most likely to be assumed wrongly next visit.

Worked pass: 1979 detached garage, heater underperforms and lights dim when it runs

Step 1: the conductor end is exposed with the circuit dead and proved dead, and it is copper. Size is 10 AWG. The route is walked and measured at 180 ft one way rather than the 150 ft the customer remembered. The heater draws a measured 14 A.

Step 2: circular mil area for 10 AWG is 10,380 from NEC Chapter 9 Table 8, and K is 12.9 for copper at a 75 C conductor operating temperature. Expected drop is 2 times 12.9 times 14 times 180, divided by 10,380. The numerator is 25.8 times 14, or 361.2, times 180, or 65,016, and 65,016 divided by 10,380 is 6.3 V. Against the 121.0 V measured at the source end that is 5.2 percent, which is already past the 3 percent branch-circuit design recommendation, so this run was built small before anything went wrong with it.

Step 3: with the heater at 14 A, the source end reads 121.0 V and the garage receptacle reads 108.0 V, both within the same minute. Measured drop is 121.0 minus 108.0, or 13.0 V.

Step 4 fails and takes its stop rule. Measured 13.0 V against expected 6.3 V is 13.0 divided by 6.3, or 2.06 times expected, which is 106 percent over and far outside the 25 percent gate. This is not a design finding.

Step 5: the excess is 13.0 minus 6.3, or 6.7 V, and 6.7 divided by 14 A is 0.48 ohm of unintended resistance somewhere in the run. At 14 A that joint is dissipating on the order of 6.7 times 14, or about 94 W, into whatever encloses it, so the circuit stays off.

Step 6: a midpoint junction box in a damp crawl under the breezeway is opened dead. The drop from the source to that box is 2.4 V and from that box to the garage is 10.6 V, and 2.4 plus 10.6 is 13.0 V, which matches the two-point figure. The far half carries the excess, and the first splice inside that box is a twist-on connector with green corrosion up both conductors.

Step 7: the splice is remade in a listed wet-location enclosure with a connector listed for the conductors and the environment, and the box is replaced because it held standing water.

Step 8: same heater, same 14 A, same two points. Source end 121.0 V, garage receptacle 114.5 V, so measured drop is 6.5 V. Against the 6.3 V expected that is 6.5 divided by 6.3, or 3.2 percent over, well inside the 25 percent gate. Grounding continuity to the garage confirmed, the garage disconnecting means operated, and the receptacle's ground-fault device trips and resets on its test button. The remaining 5.2 percent design drop is written up as a separate conductor-upsizing quote rather than sold as the repair.

References

• NEC Chapter 9 Table 8 for conductor circular mil areas, Table 9 for the impedance values used where the approximation in step 2 does not apply, and the Article 210 and 215 informational notes for the 3 percent and 5 percent design recommendations, in the edition your authority having jurisdiction has adopted
• NEC Article 225, same edition basis, for the disconnecting means at a separate structure verified at step 8, and Article 250 for the equipment grounding path to that structure
• 29 CFR 1910.333(b)(2) for work practices, with NFPA 70E-2021, 120.5 for live-dead-live and 130.5 and 130.7 for the arc-flash risk assessment and PPE
• See related: Power Quality Complaint Investigation SOP, Subpanel Installation and Feeder Sizing SOP