What is the fusing current formula for a PCB trace?
Onderdonk’s equation gives I = A × √(log10((Tm − Ta) / (234 + Ta) + 1) / (33 t)), with A in circular mils. A 10 mil wide, 1 oz copper trace is estimated to reach its melting point in 100 ms at about 8.2 A, and in 1 s at about 2.6 A. This is a melt estimate with no margin, not a current rating, and it ignores cooling, so real traces on FR-4 can take 1.5 to 6 times more before they open.
Fusing current formula: Onderdonk and Preece equations
Onderdonk’s equation solves the heating of copper by a constant current, including the rise of resistance with temperature, and assumes no heat escapes.
I = A × √( log10((Tm − Ta) ÷ (234 + Ta) + 1) ÷ (33 × t) )
- I
- current in A
- A
- copper cross-section in circular mils
- t
- time in s
- Tm
- final temperature in °C (1083 °C for melting copper)
- Ta
- starting copper temperature in °C
For a trace, the area in circular mils is the width times the thickness (both in mils) times 4/π. Starting from 20 °C and going to 1083 °C the bracket is a constant, 0.0217, so the time for a given current is:
Preece’s law for copper has no time term. It was fitted to wires in free air, with d the diameter in inches. For a trace, use the diameter of a round wire of the same area, which gives the second form with A in square inches:
I = 10244 × d^1.5 I = 12277 × A^0.75
Worked examples
Example 1: a 10 mil signal trace in a 100 ms fault
1 oz copper is 1.38 mil thick, so A = 10 × 1.38 × 4/π = 17.5 circular mils. With t = 0.1 s and a 20 °C start, I = 17.5 × √(0.0217 ÷ 0.1) = 8.2 A. For a 1 s fault the equation gives about 2.6 A, but treat that as a lower bound: by 1 s the board is pulling heat away, and Brooks and Adam’s simulated 1 oz traces took at least twice the Onderdonk current to fuse.
Example 2: how long a 100 mil trace lasts at 15 A
A = 100 × 1.38 × 4/π = 175 circular mils, so t = 0.0217 × (175 ÷ 15)² = 3 s. Preece’s estimate for the same cross-section is 16 A. The two agree in size, but at this time the board is already pulling heat away, so a real trace lasts longer. An IPC-2152 based calculation for 15 A on this trace, which includes cooling, gives about 180 °C.
Example 3: a trace that has to survive 20 A for 5 s
A = 20 × √(5 ÷ 0.0217) = 304 circular mils. In 1 oz copper that is 173 mil (4.4 mm) wide, with no margin. Derate it for a real design: this is the width that just reaches the melting point after 5 s.
PCB trace fusing current by width and copper weight
Onderdonk estimate for 1 s, starting at 20 °C. These are lower bounds: Brooks and Adam’s simulated traces on FR-4 fused at 1.63 to 3.87 times the equation at 1 s. Multiply by 3.16 for 100 ms, or by 0.45 for 5 s.
| Width | 0.5 oz | 1 oz | 2 oz | 3 oz |
|---|
| 6 mil (0.15 mm) | 775 mA | 1.55 A | 3.1 A | 4.65 A |
| 10 mil (0.25 mm) | 1.29 A | 2.58 A | 5.16 A | 7.75 A |
| 20 mil (0.51 mm) | 2.58 A | 5.16 A | 10.3 A | 15.5 A |
| 50 mil (1.3 mm) | 6.46 A | 12.9 A | 25.8 A | 38.7 A |
| 100 mil (2.5 mm) | 12.9 A | 25.8 A | 51.6 A | 77.5 A |
| 200 mil (5.1 mm) | 25.8 A | 51.6 A | 103 A | 155 A |
Wire fusing current by AWG
Onderdonk at three durations, and Preece. For wire in air, the no-cooling assumption only holds for short times, about 0.1 s for a 0.2 mm wire and about 1 s for a 2 mm wire.
| AWG | Diameter | Area | 10 ms | 100 ms | 1 s | Preece |
|---|
| 30 | 0.255 mm | 0.0509 mm² | 148 A | 46.8 A | 14.8 A | 10.3 A |
| 28 | 0.321 mm | 0.081 mm² | 235 A | 74.4 A | 23.5 A | 14.6 A |
| 26 | 0.405 mm | 0.129 mm² | 374 A | 118 A | 37.4 A | 20.6 A |
| 24 | 0.511 mm | 0.205 mm² | 595 A | 188 A | 59.5 A | 29.2 A |
| 22 | 0.644 mm | 0.326 mm² | 945 A | 299 A | 94.5 A | 41.3 A |
| 20 | 0.812 mm | 0.518 mm² | 1.5 kA | 475 A | 150 A | 58.5 A |
| 18 | 1.02 mm | 0.823 mm² | 2.39 kA | 756 A | 239 A | 82.9 A |
| 16 | 1.29 mm | 1.31 mm² | 3.8 kA | 1.2 kA | 380 A | 117 A |
| 14 | 1.63 mm | 2.08 mm² | 6.04 kA | 1.91 kA | 604 A | 166 A |
| 12 | 2.05 mm | 3.31 mm² | 9.61 kA | 3.04 kA | 961 A | 235 A |
| 10 | 2.59 mm | 5.26 mm² | 15.3 kA | 4.83 kA | 1.53 kA | 333 A |
What goes wrong in practice
Use these numbers to check whether a short fault can open a trace before the upstream fuse or current limit acts, by comparing the time here with the protection’s clearing time or its I²t. Do not use them to set a current rating: that comes from a temperature-rise method such as IPC-2152, and the trace width calculator linked below.
Real traces fuse at their weakest spot, so a narrow neck, a pad transition or an etched-thin section sets the result, not the nominal width. Heat flows into the board quickly: across six simulated traces, Brooks and Adam found the fusing current was 1.51 to 3.46 times the Onderdonk figure at 0.5 s and 2.2 to 6.3 times at 5 s, with narrower traces gaining the most. A plane 12 mil under the trace extended the fusing time by 33 % to over 100 %. Outer layers usually gain plating, so use the finished thickness. The equation also stops at the onset of melting, without the heat of fusion, and it assumes constant DC. For a pulse that is not rectangular, compare its ∫i²dt with the I²t the calculator shows.
A trace that has fused has failed destructively. The heat can damage the laminate and neighboring parts even where it looks intact, so the board should come out of service rather than be repaired.
Glossary
- Fusing current
- The current that brings a conductor to its melting point in a stated time.
- Circular mil
- The area of a circle one mil in diameter. A round wire’s area in circular mils is its diameter in mils, squared; one square mil is 1.273 circular mils.
- Adiabatic
- With no heat exchanged with the surroundings. Onderdonk’s equation assumes the conductor keeps all its heat.
- I²t
- Current squared times time, a measure of the heating energy of a pulse. Fuses are rated by it, and the Onderdonk model gives a fixed I²t for a given conductor area.
- Heat of fusion
- The extra energy needed to turn solid copper at its melting point into liquid. The equation stops before this step.
- Copper weight
- Foil thickness given as ounces of copper per square foot. 1 oz is about 35 µm, or 1.38 mil.
Frequently asked questions
What is the fusing current of a PCB trace?
The current that heats the copper to its melting point (1083 °C in the published equation) in a given time. It depends on the cross-section and on the time: a 10 mil, 1 oz trace is estimated at 8.2 A for 100 ms and 2.6 A for 1 s.
What is the Onderdonk equation?
The Onderdonk formula is I = A × √(log10((Tm − Ta) / (234 + Ta) + 1) / (33 t)), with I in amps, A in circular mils, t in seconds and temperatures in °C. It is the analytic solution for a copper conductor heated by a constant current with no cooling at all. The earliest known reference to it is a 1928 General Electric Review article.
What is the Preece equation?
I = 10244 × d^1.5 for copper, with d in inches. W. H. Preece fitted it to experiments on wires in air in the 1880s. It has no time term, so it is a single rough figure for a sustained overcurrent.
How accurate are these equations for traces on a PCB?
Only roughly, and they err low. Brooks and Adam report board cooling taking over in under 1.5 s: their simulated traces on FR-4 fused at 1.63 to 3.87 times the Onderdonk current at 1 s and 2.2 to 6.3 times at 5 s. They still treat it as a usable estimate out to 4 or 5 s, and Adam advises reading the result to one significant figure.
Can a PCB trace be used as a fuse?
Only as a last-resort, one-time failure. The current needed to open a trace is uncertain by the factor above, and the event can damage the laminate and nearby parts, so the board should be replaced. Where a fuse is really needed, use a fuse component with a rated I²t.
Why is the fusing current so much higher than the trace width calculator gives?
They answer different questions. A trace width calculator sets a safe continuous current for a small temperature rise, often 10 to 20 °C. Fusing current is the point of destruction, with no margin, and the laminate is damaged well before the copper melts.
Does ambient temperature change the fusing current?
A little. The equation uses the starting copper temperature, so a trace already running at 85 °C fuses at 92.8 % of the current of one starting at 20 °C.
How do you calculate the fusing current of a copper wire?
Use the same Onderdonk equation with the wire’s area in circular mils, which is its diameter in mils squared. To use this as a wire fusing current calculator, choose an AWG gauge or a diameter: 20 AWG is estimated at 480 A for 100 ms. In air the no-cooling assumption only holds for short times, about 0.1 s for a 0.2 mm wire and about 1 s for a 2 mm wire.
Sources
- Brooks and Adam, PCB Fusing Currents, PCD&F, September 2015
- Brooks and Adam, Who Were Preece and Onderdonk?, PCD&F, July 2015
- Brooks, Onderdonk’s and Preece’s Equations: How Do They Compare?, PCD&F, June 2023
- Adam, TRM White Paper 10: The Adiabatic Wire (derivation and limits)
- NBS Handbook 100, Copper Wire Tables (AWG sizes)
Design the whole board, not one number
ProtoFlow is free AI PCB design software: describe a circuit and get a schematic with real parts, PCB layout with an autorouter, SPICE simulation and Gerber files. Native KiCad, Altium and EAGLE import and export. Free on macOS, Windows and Linux.