LED Light Bar Wire Gauge: How AWG and Voltage Drop Decide Your Harness

A 10 A auxiliary lamp circuit needs 10 AWG copper once the one-way run reaches 5 metres, because 16 AWG loses about 1.56 V over that distance, which is 11.3% of a 13.8 V supply. The lamp still lights. It lights dimmer, runs its driver harder, and lands further from its rated output than the box promised. That single comparison is the whole subject of this guide.

What follows is arithmetic, not opinion. It covers how AWG sizes are defined, what each gauge actually resists, how to compute the round-trip voltage drop on a 12 V circuit, where the 3% and 10% thresholds come from and what they do not cover, how nameplate wattage converts to the current you must plan for, why the return conductor counts as much as the supply conductor, and how the fuse follows from the gauge rather than from the lamp. Every number that is a calculation is labelled as one, and every number that comes from a published source points at that source.

Two boundaries up front. This is not a guide to what a wiring kit contains; that is covered in the LED light bar wiring kit buying guide, which sorts out relays, switches and fuse holders. Nor is it about matching a lamp to 12 V or 24 V systems; that belongs to the 12V vs 24V light bar guide. This article assumes the voltage is already correct and asks only one question: how much copper does the circuit need.

Quick answer: the gauge is a voltage-drop decision first

For a 12 V auxiliary lighting circuit, choose the conductor by voltage drop and then check it against ampacity. Not the other way round. At the currents a light bar or a pair of work lamps draws, the copper almost never melts; it simply refuses to deliver enough volts.

The short version, in five lines:

• Compute the load current at the lowest voltage the system will realistically supply, which is closer to 12.0 V than to 14.4 V.
• Measure the run from the power source to the lamp and back, because both conductors are in series with the load.
• Multiply: voltage drop equals two, times current, times one-way length, times resistance per metre.
• Hold the drop to 3% of supply for anything you care about, and treat 10% as the outer limit that only non-critical lighting should ever approach.
• Pick the fuse from the conductor, not from the lamp.

Run that sequence on a 10 A load with a 5 m one-way run and it lands on 10 AWG. Run it on the same load at 8 m and it lands on 8 AWG. Both answers are larger than the 16 AWG that a surprising number of generic harnesses ship with, which is the practical reason this subject keeps coming up in owner forums.

The rest of this guide is the derivation behind those five lines, with the numbers written out so you can check them.

Why 12 V systems are far more sensitive to wire gauge than mains circuits

The same wattage at a lower voltage means more current, and the ratio is not modest. For a fixed power, current scales inversely with voltage, so a 12 V circuit carries about 18.3 times the current of a 220 V circuit and about 19.2 times the current of a 230 V circuit.

That factor of roughly eighteen is only the first half of the problem. The second half is worse, and it is worth deriving because it explains why 12 V wiring looks so absurdly heavy next to household cable.

Voltage drop across a two-conductor run is:

V_drop = 2 × I × L × R

where I is current in amperes, L is the one-way length in metres, and R is the conductor resistance in ohms per metre. Substituting I = P / V gives:

V_drop = 2 × P × L × R / V

so the drop as a fraction of supply voltage is:

V_drop / V = 2 × P × L × R / V²

The percentage drop therefore scales with the inverse square of system voltage. Moving the same power from 230 V to 12 V multiplies the percentage loss by (230 / 12)², which is 367. Moving it from 220 V to 12 V multiplies it by (220 / 12)², which is 336. Both figures are my calculations from the equation above; the inputs are the two voltages and nothing else.

A worked illustration makes the abstraction concrete. Take 120 W and a hypothetical conductor with 0.05 Ω of round-trip resistance. At 230 V the load draws 0.522 A, the drop is 0.026 V, and that is 0.011% of supply. At 12 V the same load draws 10.00 A, the drop is 0.500 V, and that is 4.17% of supply. The same wire, the same watts, and a loss ratio of 367 to 1. Every figure here is a calculation.

This is why a 1.5 mm² household cable that happily feeds a 2 kW heater is not a template for anything on a vehicle. The physics did not change; the voltage did. When someone tells you that 16 AWG is “good for 10 A”, they are usually quoting an ampacity figure derived for mains wiring, where a volt or two of drop is irrelevant against 230 V and where the conductor does not need to preserve the load’s supply voltage.

A 3% loss at 230 V is 6.9 V, which no appliance notices; a 3% loss at 13.8 V is 0.41 V, which an LED driver notices at once. The same percentage means different things in the two worlds.

What AWG defines: diameter, area and the 39th root of 92

AWG is a logarithmic, stepped wire gauge standard used in North America since 1857 for the diameter of round, solid, non-ferrous conducting wire. Its dimensions are given in ASTM B258, and the scale is defined by two anchors: 36 AWG is 0.005 inch in diameter and 0000 AWG is 0.46 inch, a ratio of 1 to 92 across 39 steps. Each successive size therefore differs in diameter by the 39th root of 92, approximately 1.1229.

Everything else follows from that geometric progression, and three consequences matter when you are buying cable:

• Every three gauge numbers doubles the cross-sectional area, and therefore halves the resistance. Two 14 AWG conductors have roughly the same copper area as one 11 AWG conductor.
• Every ten gauge numbers multiplies area, weight and conductance by about ten. This is why the resistance of copper wire lands near 1 Ω per 1,000 ft at 10 AWG, 10 Ω per 1,000 ft at 20 AWG, and 100 Ω per 1,000 ft at 30 AWG.
• The gauge number describes the copper, not the cable. A stranded conductor’s AWG is set by its equivalent solid cross-sectional area, so a stranded cable of a given AWG has a slightly larger overall diameter than a solid one, because of the gaps between strands.

The last point is where listings get loose. “16 AWG” is a claim about copper area. Coarse strands, copper-clad aluminium instead of copper, or thick insulation that makes a cable look substantial change none of it. Aluminium has about 61% of copper’s conductivity, needs roughly 56% more cross-section for the same resistance, and behaves like copper about two gauge sizes smaller.

Six gauges from 18 to 8 AWG, in one table

The table below is the reference point for everything else in this article. Diameters and areas are computed from the ASTM B258 geometric definition; resistances are computed from the resistivity of copper at 20 °C, and then re-expressed at 75 °C and 105 °C using the 0.393% per °C temperature coefficient of the international annealed copper standard. Every figure in the resistance columns is therefore a calculation, not a catalogue quotation.

Gauge	Diameter (mm)	Area (mm²)	R at 20 °C (Ω/m)	R at 75 °C (Ω/m)	R at 105 °C (Ω/m)	R at 20 °C (Ω per 1,000 ft)
18 AWG	1.024	0.823	0.02039	0.02479	0.02720	6.214
16 AWG	1.291	1.309	0.01282	0.01559	0.01711	3.908
14 AWG	1.628	2.081	0.00806	0.00981	0.01076	2.458
12 AWG	2.053	3.309	0.00507	0.00617	0.00677	1.546
10 AWG	2.588	5.261	0.00319	0.00388	0.00425	0.972
8 AWG	3.264	8.366	0.00201	0.00244	0.00268	0.611
6 AWG	4.115	13.302	0.00126	0.00153	0.00168	0.385

Two independent checks say the columns are right. The 10 AWG figure of 0.972 Ω per 1,000 ft at 20 °C reproduces the published rule of thumb of about 1 Ω per 1,000 ft, and the diameters match a published AWG-to-metric table giving 16 AWG as 1.29 mm and 10 AWG as 2.59 mm, against computed 1.291 mm and 2.588 mm.

Read the table as ratios. Doubling the area halves the resistance, so 16 AWG to 13 AWG would halve it, and 16 AWG to 10 AWG is a factor of about 4. Each step down the gauge numbers buys roughly 26% less resistance, paid for in copper, weight, stiffness and terminal size.

Metric cable is sold by cross-section, so convert to area before comparing: 1.5 mm² sits just above 16 AWG (1.31 mm²) and 2.5 mm² sits between 14 AWG (2.08 mm²) and 13 AWG. Round towards more copper, never towards less.

Conductor resistance at 20 °C, 75 °C and 105 °C

Published resistance tables are usually quoted at 20 °C, and a light bar harness almost never runs at 20 °C. Copper’s resistance rises by about 0.393% for every degree Celsius above 20, which is the basis of the 75 °C and 105 °C columns in the previous table.

The multipliers are worth memorising because they are easy to apply: at 75 °C a copper conductor has 1.216 times its 20 °C resistance, and at 105 °C it has 1.334 times. A 16 AWG conductor measured at 0.01282 Ω/m on the bench is 0.01559 Ω/m when the harness is warm and 0.01711 Ω/m when it is hot.

This is not a theoretical nicety. A bundle of conductors taped into a loom near an engine bay sits well above ambient. Size from a 20 °C figure and you have sized for a condition the cable will not be in.

The conservative approach costs almost nothing. Compute at 75 °C, which is 1.216 times the 20 °C figure and about 22% more drop than the bench number. An answer comfortably inside the threshold at 75 °C will be comfortably inside it cold.

A “105 °C” marking on cable is the insulation’s temperature rating, not a prediction that the conductor will reach 105 °C. The two are related only through the ampacity table, which assumes a conductor hot enough to bring its insulation to rated temperature at rated current. Using the 105 °C resistance column is therefore pessimistic, not wrong.

The voltage-drop equation and its factor of two

The equation for a two-conductor DC circuit is:

V_drop = 2 × I × L × R_per_metre

The factor of two is the part that people drop, and dropping it is the single most common arithmetic error in auxiliary lighting wiring. Current leaves the source on the supply conductor and returns on the ground conductor. Both conductors have resistance. Both are in series with the lamp. Both therefore contribute to the drop the lamp sees.

The authoritative statement of this is blunt. When ABYC explains how circuit length is measured for voltage-drop tables, it specifies the length of conductor from the positive source of power connection to the device and back to the negative source of power connection, explicitly including both the positive supply conductor and the negative return conductor. The same page notes that voltage drop occurs across conductors, connections, switches, fuses, circuit breakers and any other component in the circuit, because all of them have some resistance.

Rearranged for buying decisions, the equation becomes a minimum-area calculation:

A_required = 2 × I × L × ρ / V_allowed

with ρ the resistivity of copper, 16.78 nΩ·m at 20 °C. Working in areas rather than gauges is often easier, because areas add and subtract sensibly: two parallel conductors of equal area behave like one conductor of double the area, which is why doubling up a run is a legitimate fix when a single large cable will not fit.

A unit check catches most mistakes: amperes times metres times ohms per metre gives volts. An answer in millivolts usually means resistance per kilometre; an answer at half the expected value usually means a one-way length.

One practical note on the “L” you measure. Measure the cable route, not the straight line. A roof bar on a crew-cab pickup can need 6 to 7 m each way once the cable crosses the engine bay and runs up the A-pillar, where a tape measure suggests 3 m.

Where the 3% rule comes from, and what it does not cover

The 3% figure is an engineering practice threshold, most clearly documented in marine DC systems. ABYC E-11 establishes two maximum voltage drop categories for DC conductor sizing: a 3% maximum for panelboard main feeders, bilge blowers, electronic equipment, navigation lights and other circuits where voltage drop should be kept to a minimum, and a 10% maximum for lighting other than navigation lights and other circuits where voltage drop is not critical.

The same thresholds appear in general 12 V references. One widely used table of maximum current against wire gauge and length states plainly that conductors should not be sized with voltage drops exceeding 3%, and works the figure as 12 V × 3% = 0.36 V.

ABYC’s own worked example shows how quickly a modest resistance eats a 12 V budget. A load drawing 10 A through a circuit with 0.06 Ω of total resistance, including both conductors and all connections, drops 0.6 V. From a 12.0 V source the load receives 11.4 V. That 0.6 V is 5% of 12 V, and the same page points out that 3% of 12 V is only 0.36 V while 10% is 1.2 V.

Three things this rule is not. It is not a vehicle lighting regulation: I found no US federal or ECE provision that sets a voltage-drop limit for auxiliary lamp circuits, so anyone presenting 3% as a legal requirement is overstating it. It is not a substitute for ampacity: ABYC is explicit that conductor sizing cannot be based on voltage drop alone, that the conductor must satisfy both its required ampacity and the applicable voltage-drop limit, and that where the two indicate different sizes the larger conductor must be used. And it is not a warranty of performance: 3% is a threshold that keeps a well-behaved load well-behaved, not a guarantee that a specific lamp will hit its rated output at the end of the run.

Which threshold should you apply? A light bar used for night driving is a circuit “where voltage drop should be kept to a minimum” in ABYC’s language, so 3% is the target. An intermittent rear work lamp is closer to the “not critical” category, where 10% is defensible. What is not defensible is applying 10% to a forward-facing driving light and then wondering why it looks yellow and dim next to the same bar on a shorter harness.

For a forward-facing lamp, hold the drop to 3% of supply. At a 13.8 V supply that is 0.414 V. At a rested 12.6 V battery it is 0.378 V. At a nominal 12.0 V it is 0.360 V.

Worked example: 16 AWG at 10 A across 3, 5 and 8 metres

Here is the calculation that most often surprises people who have been told 16 AWG “handles 10 amps”.

Take a 16 AWG copper conductor at 75 °C, so R = 0.01559 Ω/m, and a supply of 13.8 V. Apply the round-trip equation across one-way lengths of 3, 5 and 8 m and currents of 5, 10, 15 and 20 A. The 3% limit at 13.8 V is 0.414 V. Every figure in the table is a calculation from the equation above.

Current	One-way length	Voltage drop (V)	Voltage at lamp (V)	Drop as % of 13.8 V
5 A	3 m	0.468	13.332	3.39%
5 A	5 m	0.780	13.020	5.65%
5 A	8 m	1.247	12.553	9.04%
10 A	3 m	0.936	12.864	6.78%
10 A	5 m	1.559	12.241	11.30%
10 A	8 m	2.495	11.305	18.08%
15 A	3 m	1.403	12.397	10.17%
15 A	5 m	2.339	11.461	16.95%
15 A	8 m	3.742	10.058	27.12%
20 A	3 m	1.871	11.929	13.56%
20 A	5 m	3.119	10.681	22.60%
20 A	8 m	4.990	8.810	36.16%

Read the 10 A row across and the answer is unambiguous. At 3 m, 16 AWG drops 0.936 V, which is 6.78% of 13.8 V. It fails the 3% target and merely approaches the 10% ceiling. At 5 m it drops 1.559 V, or 11.30%, which fails both thresholds. At 8 m it drops 2.495 V, or 18.08%, and the lamp sees 11.305 V while the alternator is charging.

Even at 5 A, 16 AWG fails the 3% test at every one of those lengths: at 3 m it drops 0.468 V, or 3.39%, which is just outside; at 5 m it is 5.65% and at 8 m it is 9.04%.

Two honest qualifications. At 20 °C the 10 A, 5 m case drops 1.282 V rather than 1.559 V, which is 9.29% instead of 11.30%; the conclusion holds, but a cold measurement reads lower than a hot one. And the percentages are of 13.8 V. With the engine off and the battery at 12.24 V, the BCI figure for about 50% state of charge, the same drop is 12.7% of supply and the lamp sees 10.68 V.

That is the real answer to “why does 16 AWG at 10 A cause problems”. It is not that the copper fails. It is that the lamp stops receiving the voltage it was designed around, and everything downstream of that is consequence.

Ampacity versus voltage drop: two different tests

Ampacity is the current a conductor can carry continuously without its insulation exceeding its rated temperature. Voltage drop is how much of the supply voltage is lost getting the current there. They are different questions with different answers, and for 12 V auxiliary lighting the voltage-drop answer is almost always the binding one.

The standard ampacity reference in North American practice is the National Electrical Code table of allowable ampacities for insulated conductors, whose copper values for the gauges in this article are reproduced below. The three columns are keyed to 60 °C, 75 °C and 90 °C conductor temperature ratings. This is a published table, not a calculation.

Gauge	60 °C (A)	75 °C (A)	90 °C (A)	Overcurrent limit noted in table footnote (A)
18 AWG	not listed	not listed	14	not stated
16 AWG	not listed	not listed	18	not stated
14 AWG	20	20	25	15
12 AWG	25	25	30	20
10 AWG	30	35	40	30
8 AWG	40	50	55	not stated
6 AWG	55	65	75	not stated

Now compare that with the previous section. 16 AWG at 90 °C is listed at 18 A. A 10 A load is therefore thermally unremarkable on 16 AWG; it is about 56% of the tabulated figure. Yet the same 10 A load over 5 m loses 11.3% of its supply voltage. Both statements are true at once, and the one that matters for a lamp is the second.

This is why “16 AWG is rated for 18 amps” is a misleading sentence in a product listing. It is not false as an ampacity claim under the table’s conditions. It is simply an answer to a question nobody asked. The question that matters is whether the lamp receives adequate voltage, and ampacity says nothing about that.

There is also a caveat about the table itself that makes the ampacity figure less useful here. It is titled for conductors in a raceway, and its values assume a 30 °C ambient with no more than three current-carrying conductors. An automotive loom is a bundle of many conductors routed through hot regions, so real ampacity there is lower than the tabulated figure, not higher. For 18 and 16 AWG the footnote states no overcurrent limit at all.

The resolution is a two-test rule, and it is the one ABYC states for DC systems generally: size for ampacity, size for voltage drop, and where the two disagree use the larger conductor. On a 12 V lamp circuit voltage drop usually wins by a wide margin; when it does not, take the larger.

Minimum gauge lookup at 3% and at 10%

If you would rather read a table than do the algebra at the counter, the two tables below are the same calculation inverted. For each combination of current and one-way length they give the smallest gauge that holds the drop inside the threshold, computed at 75 °C with a supply of 13.8 V. The 3% budget is 0.414 V and the 10% budget is 1.380 V. Cells reading “>6” mean that even 6 AWG does not meet the threshold at that combination, which for a 12 V circuit is a signal to shorten the run, raise the system voltage, or accept a lower standard.

Current	3 m	5 m	8 m	10 m	12 m
5 A	14 AWG	12 AWG	10 AWG	10 AWG	8 AWG
10 A	12 AWG	10 AWG	8 AWG	6 AWG	6 AWG
15 A	10 AWG	8 AWG	6 AWG	>6	>6
20 A	8 AWG	6 AWG	>6	>6	>6
25 A	8 AWG	6 AWG	>6	>6	>6
30 A	6 AWG	>6	>6	>6	>6

The second table relaxes the target to 10%, which is the ABYC figure for lighting other than navigation lights. It is included because some readers genuinely are wiring an intermittent rear work lamp rather than a driving light, and because seeing both tables side by side shows how much copper the stricter standard actually costs.

Current	3 m	5 m	8 m	10 m	12 m
5 A	18 AWG	18 AWG	16 AWG	14 AWG	14 AWG
10 A	16 AWG	14 AWG	12 AWG	12 AWG	10 AWG
15 A	14 AWG	12 AWG	10 AWG	10 AWG	8 AWG
20 A	14 AWG	12 AWG	10 AWG	8 AWG	8 AWG
25 A	12 AWG	10 AWG	8 AWG	8 AWG	6 AWG
30 A	12 AWG	10 AWG	8 AWG	6 AWG	6 AWG

The gap between the two tables at 15 A and 5 m is three gauge sizes: 8 AWG at 3%, 12 AWG at 10%. That is the cost of a stricter standard, and it is the reason a single “recommended wire gauge” chart without a stated threshold is not useful.

Check the 3% table against independent published data and it holds up. A widely used 12 V chart, which states the 3% basis and measures length as the total run out and back, gives 10 AWG for 10 A over 30 ft of total length, and 12 AWG for 10 A over 20 ft. Thirty feet is 9.1 m of round trip, or about 4.6 m one way, and twenty feet is 6.1 m of round trip, or about 3.0 m one way. My table gives 10 AWG at 5 m and 12 AWG at 3 m for 10 A. The two sources agree at both checkpoints, which is what you would expect if both are doing the same arithmetic.

These tables cannot tell you your harness temperature or your alternator output. They are computed at 75 °C and 13.8 V. At 14.4 V the percentages fall by about 4%; at a 12.2 V battery they rise. Treat them as the design case, not a measurement.

Nameplate watts, LED watts and the current to plan for

Every gauge decision starts with a current, and almost every product page gives a wattage. The conversion is where the first real uncertainty enters, because the voltage you divide by is not fixed.

Nominal lamp power	Current at 12.0 V	Current at 13.8 V	Current at 14.4 V
60 W	5.00 A	4.35 A	4.17 A
120 W	10.00 A	8.70 A	8.33 A
180 W	15.00 A	13.04 A	12.50 A
240 W	20.00 A	17.39 A	16.67 A
300 W	25.00 A	21.74 A	20.83 A
360 W	30.00 A	26.09 A	25.00 A

All values are calculations from I = P / V. The spread is about 15% between the 12.0 V and 13.8 V columns, and that spread propagates directly into your gauge choice: a 240 W bar is 20.00 A at 12.0 V and 17.39 A at 13.8 V. Looked up in the 3% table at 3 m, 20 A needs 8 AWG while 17.4 A lands on 10 AWG: a 15% difference in current moves the answer by two gauge sizes.

The safe convention is to divide by the lowest voltage the system will realistically present while the lamp is in use, which is not 12.0 V in most cases and certainly not 14.4 V. The BCI state-of-charge table gives 12.65 V for a fully charged starter battery at rest, 12.45 V at 75%, 12.24 V at 50% and 12.06 V at 25%, all at 26 °C after a 24-hour rest. A vehicle at rest is somewhere in that band, and parasitic loads put it into what Battery University calls a quasi-closed-circuit condition.

Charging voltages come from the other end of the range. Lead-acid charge-voltage limits run from 2.30 V to 2.45 V per cell, and the three-stage CCCV regime is commonly expressed as limiting the charge to 2.40 V per cell, which is 14.40 V across six cells, then dropping to a float of about 2.30 V per cell, which is 13.8 V across six cells. Recommended float for flooded batteries is 2.25 V to 2.27 V per cell. Those figures are the origin of the 13.8 V and 14.4 V numbers used throughout this article, and they describe a charging source, not a guarantee about what your lamp terminal sees.

The second uncertainty is what the nameplate wattage refers to. A lamp’s input power includes driver losses; its LED power is the power that reaches the emitters. They are not the same number, and a listing that says “240 W” rarely states which one it means. I could not find a general convention that suppliers follow, so treat the nameplate figure as an input-power claim and, where the difference matters, plan for the higher current. If the true input power is higher than the nameplate, your gauge will be marginal in exactly the way this article is trying to prevent.

A third uncertainty is measurement condition. A wattage quoted without a voltage is incomplete. If the supplier measured at 13.8 V, the current at 12.0 V will be about 15% higher for the same power. Ask for the voltage, and if the answer is vague, use 12.0 V.

Constant-power loads: current rises as voltage falls

An incandescent lamp behaves helpfully when voltage sags: its current falls roughly with the square root of the voltage, so a marginal conductor self-limits. An LED lamp with a regulated driver does the opposite. Holding string current constant holds output power roughly constant, so input current rises as input voltage falls.

Supply voltage at the lamp	Current drawn by a 120 W constant-power load
14.4 V	8.33 A
13.8 V	8.70 A
12.6 V	9.52 A
12.0 V	10.00 A
11.0 V	10.91 A

These are calculations from I = P / V under the stated constant-power assumption. Real drivers are not perfect constant-power sources, and near the bottom of their input range they typically drop out of regulation or shut down, but over the normal operating band the trend holds: a 15% fall in voltage produces about an 18% rise in current.

The consequence for conductor sizing is that the loop does not correct itself. If the conductor is marginal at 13.8 V, it becomes worse as the voltage falls, because the current rises at the same time as the drop per ampere stays fixed. The two effects multiply rather than cancel. A circuit that drops 11.3% at 10.00 A will drop more than 11.3% once the current has risen to compensate.

There is a second-order effect that is worth knowing about even though it is small. That extra current flows through the same conductor and dissipates I²R in it, so the conductor runs hotter, so its resistance rises, so the drop rises again. For a correctly sized circuit this converges harmlessly. For a badly undersized one it converges on a hot harness.

The practical instruction is simply to compute the current at the low end of the voltage range and the drop at the high end of the temperature range. Those two choices are both conservative and they cost you at most one gauge size.

Neither the 3% nor the 10% threshold is a law, and neither is enforced by any vehicle lighting regulation I could find. What they are is a defensible way to make the same decision twice and get the same answer, which is more than can be said for choosing a gauge by eye. If you want to see what the symptom looks like when a circuit is already marginal, the guide to a light bar dimming at idle walks the same fault from the measurement side, and the beam distance guide explains why a voltage-starved lamp loses reach out of proportion to its measured lumens.

The return conductor is part of the calculation

Every figure above assumes two conductors of equal resistance. That assumption fails the moment a circuit is grounded to the vehicle body instead of run back to the battery, and it fails in a way a meter on the positive post will not show.

ABYC’s definition of circuit length is explicit that both the positive supply conductor and the negative return conductor are included in the total. General DC cable-sizing guidance says the same thing in one sentence: the ground cable is as much a part of a circuit as the positive cable, and it must be sized the same (Logistics Operational Guide, cable sizes and wiring). That source also names undersizing relative to the load as the most common installation error in DC systems.

A chassis return is not a copper conductor of the same length. The path runs through the lug, the fastener, the panel it is bolted to, the spot welds or seam adhesive joining that panel to the next, and eventually the block. Each interface is a resistance, and several were never designed to carry current.

The rule that follows is short. For any lighting circuit above about 5 A or longer than about 3 m, run a dedicated return conductor of the same gauge back to a common ground terminal, not a self-tapping screw into sheet metal. It costs one more run of cable and removes a whole category of intermittent faults.

What a corroded ground adds, in volts

I found no published figure for the resistance of a corroded vehicle body ground, so the table below is not a measurement. It is a sensitivity calculation: the extra voltage a given interface resistance costs at a given current.

Interface resistance	Extra drop at 10 A	At 15 A	At 20 A	At 30 A
0.02 Ω	0.20 V	0.30 V	0.40 V	0.60 V
0.05 Ω	0.50 V	0.75 V	1.00 V	1.50 V
0.10 Ω	1.00 V	1.50 V	2.00 V	3.00 V
0.20 Ω	2.00 V	3.00 V	4.00 V	6.00 V
0.50 Ω	5.00 V	7.50 V	10.00 V	15.00 V

The point is the comparison with the earlier worked example. Ten amps through 16 AWG over 5 m cost 1.559 V; ten amps through a single 0.10 Ω ground costs 1.00 V before the supply conductor has contributed anything. A circuit with good cable and a bad ground is a bad circuit.

ABYC’s voltage drop page lists corrosion among the causes of excessive resistance, alongside conductors that are too small, long runs, and loose or poor-quality connections, and notes that excessive resistance also produces heat. That is the mechanism behind the brown, slightly melted crimp on an older truck.

The confirming test is a loaded drop measurement, not a dead resistance check. ABYC notes that voltage-drop testing under load is a practical tool for confirming that DC equipment is receiving the voltage it needs. Put the meter across the joint, switch the load on, and read the difference.

Fuses protect the conductor, not the lamp

The common misconception is that the fuse is there for the lamp. It is not. The fuse opens the circuit before the conductor overheats, so its rating is bounded above by the conductor and only loosely related to the lamp.

Two published statements frame the rule. The 2005 National Electrical Code ampacity table carries a footnote stating that, unless specifically permitted elsewhere, overcurrent protection shall not exceed 15 A for 14 AWG, 20 A for 12 AWG and 30 A for 10 AWG copper (Elliott Electric Supply, Table 310-16). And general 12 V design guidance warns that failure to use an adequate conductor size may result in a fire, and instructs that wires always be secured with fuses (Engineering ToolBox). Both say the same thing: the protection device and the LED light bar wire gauge are a matched pair, and neither is chosen in isolation.

Gauge	NEC 90 °C ampacity (A)	Overcurrent limit in table footnote (A)	Fuse band used in practice (engineering judgement)
16 AWG	18	not stated	10 A or lower
14 AWG	25	15	10 to 15 A
12 AWG	30	20	15 to 20 A
10 AWG	40	30	20 to 30 A
8 AWG	55	not stated	30 to 50 A
6 AWG	75	not stated	45 to 70 A

The third column is quoted from the published table; the fourth is my judgement, derived from the ampacity column and constrained where the footnote gives a figure. For 18 and 16 AWG the 2005 table states no overcurrent limit at all, which means a small-gauge circuit has to be protected conservatively rather than by a number lifted from a chart.

One condition a gauge table cannot express also applies. The fuse must be rated for the circuit voltage and for its environment, so mount and seal the holder and keep the run from battery to fuse short, so the unprotected length is minimal.

The lamp sets the lower bound on the fuse, not the upper one. A 10 A load wants a fuse comfortably above 10 A to survive inrush and below the conductor ampacity to do its job. On a marginal conductor that window can close entirely, which is the real symptom of an undersized harness: the fuse that protects the wire is too small to run the lamp.

The order of operations: current, gauge, fuse, relay

The LED light bar wire gauge decision is one step in a fixed sequence, and doing the steps out of order produces harnesses that are over-fused and under-sized at the same time.

• Establish the load current at 12.0 V, not at the charging voltage.
• Measure the one-way cable route and double it.
• Choose the smallest gauge that holds the drop inside 3% at working conductor temperature.
• Choose a fuse at or below that gauge’s ampacity and above the load current.
• Choose a relay and switch rated for the load with margin, mounted where they can be serviced.
• Choose terminals that accept the chosen gauge, because 6 AWG does not fit a ring terminal made for 12 AWG.

Three cases show how the answers move. All figures are calculations at 75 °C, 13.8 V, with a 3% budget of 0.414 V.

Case	Load at 12.0 V	One-way run	Drop with chosen gauge	Chosen gauge	Fuse band
Single 60 W pod (Clone Pro 7 inch)	5.00 A	2.5 m	0.390 V (2.82%) with 16 AWG	16 AWG	10 A
Single 120 W bar (Jungle Series 40 inch)	10.00 A	4.0 m	0.310 V (2.25%) with 10 AWG	10 AWG	15 to 20 A
Single 240 W bar	20.00 A	4.0 m	0.390 V (2.83%) with 8 AWG	8 AWG	30 to 50 A

The first row is the corrective to the rest of this article. A small pod on a short run genuinely works on 16 AWG, and telling every reader to buy 8 AWG for everything would be as unhelpful as telling them all to buy 16 AWG. What matters is that the answer came out of arithmetic rather than habit.

The third row is the case that catches people. A 240 W bar needs 8 AWG at 4 m, which most owners would not think to buy for a light bar, and a fuse in the 30 to 50 A band, which is larger than what many universal harnesses carry. If the harness in the box has 16 AWG and a 15 A fuse, it was never adequate for that bar.

Parallel lamps and the trunk conductor

Two lamps in parallel draw twice the current, so the trunk that feeds both must be sized for the total while each branch is sized for its own load. Cable-sizing guidance makes exactly this point about lighting circuits sharing supply and ground conductors: where several lights are fed from a common pair, the supply conductor must be sized for the total load of all of them.

Configuration	Total current at 12.0 V	Trunk for 3 m one-way	For 5 m	For 8 m
2 × 120 W	20.00 A	8 AWG	6 AWG	above 6 AWG
3 × 120 W	30.00 A	6 AWG	above 6 AWG	above 6 AWG
4 × 120 W	40.00 A	6 AWG	above 6 AWG	above 6 AWG

The 4 × 120 W case at 3 m passes on 6 AWG with a computed drop of 0.367 V, or 2.66%. At 5 m the same array needs more copper than 6 AWG offers: 6 AWG over 5 m drops 0.612 V, or 4.43%, outside the 3% budget. The correct answer there is a shorter trunk with a relay and a local distribution point, or a higher system voltage, not a cable that will not pass through the grommet.

That alternative solves many cases. Run one heavy feed from the battery to a fused distribution point near the lamps, then short branch runs from there. The total current travels over a shorter heavy cable, each branch is short enough to size modestly, and the long control wire from the dashboard carries milliamps. Browse the LED light bars category with that layout in mind rather than assuming every bar takes the same harness.

Long runs, trailer circuits and rear-mounted lamps

Length compounds with current rather than adding to it. Ten amps over 12 m needs 6 AWG to stay inside 3%, at a computed 0.367 V or 2.66%. The same 10 A over 5 m needed only 10 AWG. Slightly more than double the length cost four gauge sizes.

Rear work lamps and trailer lighting are where this bites. A lamp at the tailgate of a long-bed truck may be 6 to 7 m from the battery along the cable route, and a trailer adds a connector interface, a second cable length and another set of terminals, each of which is a resistance and each of which is a corrosion site.

One design is bad and two are sound. The bad one runs thin cable the whole way and accepts the drop, giving a lamp that is bright at fast idle and dull at tickover. Either size the cable for the full run, or mount the relay and fuse near the battery, run a heavy feed to a rear distribution point and switch the relay with a thin control wire. The second costs less copper and routes more easily.

Connectors and terminals: the resistance nobody publishes

Every connector is a resistance, and no manufacturer publishes a single number you can add to a drop calculation with confidence, because the value depends on the crimp, the plating, the current, the age and the environment. I found no authoritative per-connector figure that applies across products, so this article does not invent one.

What is documented is how to make the interface as good as it can be. DC cable-sizing guidance is direct: cable lug connections should be securely crimped to the wire termination with a band, and not soldered in place. A proper crimp is a cold weld between conductor and barrel, made with a tool that deforms the barrel by a controlled amount. A soldered joint wicks solder into the strands, creates a stiff point where vibration concentrates, and can leave flux residue behind.

Three habits remove most connector faults. Match the terminal to the gauge so the barrel is filled by strands rather than air; match the ring to the stud so contact is the full face of the ring; seal the transition with adhesive-lined heat-shrink so moisture cannot corrode the strands from inside, which is the failure that looks like a dead lamp and is a broken wire.

Too small and too large: the actual consequences

Both errors are real and they fail differently. Undersizing is a performance and safety problem; oversizing is a cost, fit and serviceability problem.

What you observe	Likely gauge-related cause	Check that confirms it
Lamp dimmer than the same model on another vehicle	Drop too high, usually an undersized supply or return conductor	Measure volts at the lamp terminals with the load on
Lamp brightens sharply with engine speed	Supply voltage rising while the drop stays proportional, so the lamp is marginal at low voltage	Log lamp-terminal volts at idle and at 2,000 rpm
Harness warm or insulation soft near the lamp	Conductor undersized for continuous current, or a poor crimp concentrating heat	Thermal check after 20 minutes, then re-crimp
Fuse blows intermittently on a circuit that runs	Undersized conductor forces a fuse small enough to nuisance-blow, or a short exists	Fit the correct gauge, then re-evaluate the fuse
Voltage fine unloaded, collapses under load	High-resistance joint or corroded ground rather than gauge	Loaded drop measurement across each joint
Cable will not fit terminal, grommet or conduit	Gauge chosen without checking the mechanical envelope	Measure barrel and bore before ordering cable

The LED light bar wire gauge question is usually framed as a trade between thin enough to route and thick enough to work. The mechanical constraint is real but it is checked once at purchase, while the electrical constraint operates every time the lamp is switched on.

Oversizing deserves a fair hearing. General DC guidance is unambiguous that a marginally oversized cable carries no performance penalty, and on a 12 V circuit the extra copper buys margin against a hot harness, a corroded joint and a partly discharged battery. Where it costs you is mechanically: 6 AWG is stiff, has a large bend radius, will not pass a factory grommet, and needs terminals that can take it. Size to 3% at 75 °C, then go up one gauge if the route is long, hot or hard to redo.

A buying checklist for harnesses and loose cable

Nothing forces a supplier to tell you what you need, so this list is ordered by how often the omission causes a return.

• The gauge is stated as a number and a standard, for example 12 AWG, not as “heavy duty”.
• The conductor is stated as copper. Copper-clad aluminium of the same AWG is not equivalent, because aluminium has about 61% of copper’s conductivity and needs roughly 56% more area for the same resistance (copper conductor properties).
• Both supply and return conductors are the same gauge, or the harness states that it expects a chassis return.
• The insulation temperature rating is stated and suits the route; a 105 °C product is not an 80 °C one.
• The fuse rating is stated and is consistent with the gauge rather than with the marketing wattage.
• The relay rating is stated as continuous current, not peak.
• The length is the routed length, not the straight-line length.
• The terminal sizes are stated, so the ring fits the stud and the barrel accepts the conductor.
• The connector sealing is stated, because an unsealed connector under a hood is a maintenance item.
• The harness reaches a proper ground terminal, not just the nearest panel.

The LED light bar wire gauge on the label is the most useful line on the page and the one most often missing. A supplier who will not state it in AWG has not designed the harness, which is worth knowing before you pay.

When to stop and call an auto electrician

This article covers a single auxiliary circuit fed from the battery through a fuse and a relay. Several situations sit outside that scope, where the right action is to stop rather than extrapolate.

• Continuous load above about 30 A, where alternator capacity, charge balance and routing become system questions.
• A smart charging system, battery management system or DC-DC converter between battery and loads, where supply voltage is not a simple function of engine speed.
• Any circuit that must interact with a body control module, a CAN bus, or a factory module that monitors lamp current.
• Dual-battery, auxiliary-battery or split-charge systems, where return paths and grounding points are part of a designed system.
• Existing heat damage, melted insulation, discoloured copper or the smell of hot plastic, which is a fire-risk condition rather than a sizing exercise.
• A cable that must pass through a structural panel, near fuel or brake lines, or through a region exposed to road debris.
• An installation that has to satisfy an inspection, a scrutineer or an insurer.

None of these is a reason to guess harder. Each is a reason to have someone who can measure the system look at it.

Frequently asked questions

What LED light bar wire gauge do I need for a 10 A load?

At 5 m one-way, 10 AWG holds the drop to 2.81% of 13.8 V while 16 AWG reaches 11.30%, so 10 AWG. At 3 m, 12 AWG is enough; at 8 m, 8 AWG. The gauge depends on length as much as current.

Is 16 AWG enough for a light bar?

For a small pod drawing about 5 A on a run under about 3 m it can hold the drop near 2.8% and is workable. For a 120 W bar drawing 10 A, 16 AWG drops 11.30% over 5 m, outside both thresholds. The gauge is not the problem; the combination is.

Why does my light bar look dim when the fuse is correct?

A correct fuse means the conductor is protected, not that it delivers adequate voltage. Measure volts at the lamp terminals with the lamp on. If it is more than about 3% below battery voltage, the loss is in conductors, joints or the return path.

Does the ground wire need to be the same size as the positive wire?

Yes. The return conductor carries the same current and sits in series, so it contributes the same drop for the same resistance. DC guidance states the ground cable must be sized the same, and circuit length for drop purposes includes both conductors.

What voltage should I use to convert watts to amps?

The lowest voltage the system will present while the lamp runs, nearer 12.0 V than 13.8 V. A 120 W lamp is 10.00 A at 12.0 V and 8.70 A at 13.8 V, a 15% difference that carries straight into the gauge choice.

Is 3% voltage drop a legal requirement?

No. It is an engineering practice threshold from marine DC standards and general 12 V guidance, with 10% allowed for non-critical lighting. I found no US federal or ECE regulation setting a voltage-drop limit for auxiliary lamp circuits. Treat 3% as a defensible design target.

Do I need a relay if I use thick cable?

The relay question is about the switch, not the cable. A dashboard switch rated for a few amps cannot carry 20 A continuously. Use a relay for any load above the switch rating, mount it near the battery or distribution point, and let the switch carry only coil current.

Editorial note

Written and fact-checked by the SAEC Lighting Engineering Team, Foshan. This article covers conductor sizing and voltage drop for 12 V auxiliary lighting circuits only; it does not address lamp photometry, road legality or vehicle-specific fitment. Every resistance, drop and current figure labelled as a calculation was computed from the equations and material constants cited below, at the stated temperatures and voltages.

Last updated: 2026-10-20

Sources

• American Boat and Yacht Council, “What Is Voltage Drop?”, published 08/18/2026. https://abycinc.org/what-is-voltage-drop/
• The Engineering ToolBox, “12 Volt – Amps vs. Wire Gauge: Choosing the Right Wire Size”. https://www.engineeringtoolbox.com/amps-wire-gauge-d_730.html
• Elliott Electric Supply, “Allowable Ampacities Insulated Conductors (in a Raceway) Table 310-16 2005 N.E.C.”. https://www.elliottelectric.com/StaticPages/ElectricalReferences/ElectricalTables/Allowable_Ampacities.html
• Wikipedia, “American wire gauge” (ASTM B258 dimensional definition). https://en.wikipedia.org/wiki/American_wire_gauge
• Wikipedia, “Copper conductor” (resistivity 16.78 nΩ·m at 20 °C; IACS 0.393%/°C; aluminium conductivity). https://en.wikipedia.org/wiki/Copper_conductor
• Battery University, “BU-903: How to Measure State-of-charge” (BCI state-of-charge voltage table). https://www.batteryuniversity.com/article/bu-903-how-to-measure-state-of-charge
• Battery University, “BU-403: Charging Lead Acid” (2.40 V/cell = 14.40 V; float 2.30 V/cell = 13.8 V). https://www.batteryuniversity.com/article/bu-403-charging-lead-acid
• Logistics Operational Guide, “Cable Sizes and Wiring” (ground conductor sizing, AWG to mm² conversion, crimped terminations). https://log.logcluster.org/en/cable-sizes-and-wiring
• SAEC, “LED Light Bar Wiring Kit Buying Guide: Relays, Fuses and Switches”. https://saec-lights.com/buying-guides/led-light-bar-wiring-kit-buying-guide/
• SAEC, “12V vs. 24V LED Light Bars: How to Match Voltage, Current and Vehicle Electrical Systems”. https://saec-lights.com/buying-guides/12v-vs-24v-led-light-bars-guide/
• SAEC, “LED Light Bar Dims at Idle: Charging, Wiring and Load Checks”. https://saec-lights.com/troubleshooting/led-light-bar-dims-at-idle-charging-wiring-load-checks/
• SAEC, “Light Bar Beam Distance: How Candela and Lux Decide What You Can Actually See”. https://saec-lights.com/buying-guides/light-bar-beam-distance-candela-lux/