PCB Trace Width and Current Calculator
Give it the current, allowed temperature rise, copper weight and layer, and it inverts the common IPC-2221 chart fit for the minimum trace width, then adds the resistance, voltage drop and heating of that run. Calculated in the browser; nothing is uploaded.
Result
Worked examples
Every row is produced by our Python reference implementation using the same constants and thresholds the page recalculates with; decimals go only as far as you can hand-check.
| Case | Inputs | Minimum trace width mil | Minimum trace width mm | Required cross-section area mil² | Copper thickness (from the selected oz) mil | Resistance of that length Ω | Voltage drop at this current mV | Trace dissipation W |
|---|---|---|---|---|---|---|---|---|
| Supply entry at 2 A, 10 °C rise, 1 oz outer layer, 20 mm run | 2 A · 10 °C · 1 oz · outer · 20 mm | 30.8 | 0.781 | 42.4 | 1.378 | 0.0126 | 25.2 | 0.0503 |
| Same 2 A / 10 °C with 2 oz copper: the area requirement is unchanged, so the width halves | 2 A · 10 °C · 2 oz · outer · 20 mm | 15.4 | 0.391 | 42.4 | 2.756 | 0.0126 | 25.2 | 0.0503 |
| Same 2 A / 10 °C / 1 oz moved to an internal layer: k halves, watch the width | 2 A · 10 °C · 1 oz · internal · 20 mm | 80.0 | 2.033 | 110.3 | 1.378 | 0.0048 | 9.7 | 0.0193 |
Where the number comes from
- I: steady DC current (A)
- ΔT: allowed temperature rise (°C, above room temperature)
- A: trace cross-section area (mil²)
- k: 0.048 (outer layer) / 0.024 (internal layer)
- t: copper thickness, 1 oz = 1.378 mil = 35 µm
The fit reproduces the published IPC-2221 current capacity chart; it is not the standard text. All constants are listed below, please check them.
Constants: k=0.048 (outer layer) / 0.024 (internal layer) · b=0.44 · c=0.725 · 1 oz=1.378 mil=35 µm · 1 mil=0.0254 mm=25.4 µm · rho(Cu, 20 °C)=0.0172 ohm·mm²/m
What this result cannot be used to decide
This is a steady-DC temperature-rise floor. It ignores impedance and differential-pair requirements, pulsed and motor-start currents, the current gain from solder-mask openings, and copper lot / board thickness tolerances. Before mass production, critical nets go back to the IPC-2152 charts, a thermal simulation or a measured temperature rise on a real board.
Frequently asked questions
Which formula does this use?
The common fit for the IPC-2221 current capacity chart, I = k·ΔT^0.44·A^0.725, with k = 0.048 for outer and 0.024 for internal layers and A in mil². It reproduces the shape of the chart in the standard; it is an empirical fit, not the standard text, and it is not a conformity claim for any board.
Can I use the width it returns directly?
Only as a "no narrower than" floor. It constrains temperature rise, nothing else: impedance control, differential-pair spacing, neighbour heating in dense layouts, pulsed current and inrush all need their own simulation or measurement.
Why does the internal layer need so much more width?
The fit halves k for internal layers: the copper is buried in dielectric and heat can only conduct sideways. For the same current and rise the required area is about 2.6× the outer layer — the exponent 0.725 turns a 2× coefficient difference into 2^(1/0.725) ≈ 2.62×.
Does trace length change the width conclusion?
No. The temperature-rise model only sees cross-section and cooling conditions; length determines resistance, drop and total dissipation. That is why length is a separate input on this page and only feeds the resistance rows.
What about pulsed current, like a stalled servo?
This is a steady-DC assumption. For pulsed loads work it out from duty cycle and thermal time constant, or measure the temperature rise on the real board. The page will not invent a pulse derating coefficient for you — that needs an actual waveform.
Why does 0.5 A come out only a few mil wide — is that usable?
The formula is simply optimistic in the small-cross-section region: the IPC-2221 chart was fitted from bulk-copper trends, and later IPC-2152 measurements show narrow traces carrying less. We reproduce the original fit without that correction, so at low currents it still prints a very small width — but fab minimum width, impedance rules and habit push real designs to 10-15 mil anyway, and the page says so whenever the result falls under 8 mil.
Related on this site
Data and formulas last checked：2026-11-02