What this calculator does
A PCB trace carrying current dissipates as heat. It warms until the heat it loses balances the heat it generates: by conduction into the laminate and planes, and by convection and radiation from the board surface. The designer's question is usually the reverse: how wide must this trace be so that it rises no more than ΔT above ambient? This calculator answers that question with two models and also gives the electrical consequences:
- IPC-2221 (the classic equation), with the required width for a target current, the maximum current for a given width, and the temperature rise at a given current and width;
- an IPC-2152-based estimate, from the closed-form fit Brooks and Adam made to the IPC-2152 external-trace charts, for comparison;
- the resistance at the predicted trace temperature, the voltage drop and the power loss over the trace length.
IPC-2221
IPC-2221 (and IPC-D-275 and MIL-STD-275 before it) reproduces charts that go back to a 1956 National Bureau of Standards study. The curve fit printed in the standard is
with in amperes, in °C and the cross-section in square mils. The constant is for external layers and for internal ones. The internal value is simply the external one halved. In the original work the internal charts were not measured; they were derived by assuming that inner traces cool worse.
IPC-2152 fit
IPC-2152 (2009) replaced those charts with new measurements. It publishes charts only, and they are copyrighted. Douglas Brooks and Johannes Adam digitised the external-trace curves and found that one equation fits the 2 oz and 3 oz data very closely:
The calculator uses this equation, inverted where needed, for the "IPC-2152 fit, external" results. Their internal-trace fits have constants that vary with copper weight and width, so this calculator leaves them out. Brooks and Adam found that internal traces in the IPC-2152 data run slightly cooler than external ones, so the external fit is somewhat conservative for inner layers.
Resistance and loss
The DC resistance uses the 20 °C conductivity (58.0 MS/m for annealed copper, per IEC 60028) and the copper temperature coefficient /°C. It is evaluated at , where is the IPC-2221 rise at your current:
How to use solve-for
Required width and max current are direct outputs, so you rarely need to solve for them. Solve-for helps with the other inversions:
- Click ⇄ Solve on Trace length and target Voltage drop to find how long a run can be for a given drop budget.
- Solve Trace width for Temperature rise at I (IPC-2152 fit) to size a trace with the newer data.
- Solve Copper thickness for Max current to see what copper weight a fixed width needs.
Worked example
An external 1 oz trace ( mil m), 50 mm long, carries 2 A at 25 °C ambient with a 10 °C allowed rise.
| Quantity | Result |
|---|---|
| Required width, IPC-2221 | 0.786 mm (30.9 mil) |
| Required width, IPC-2152 fit (external) | 0.930 mm (36.6 mil) |
| Max current of a 0.75 mm trace, IPC-2221 | 1.93 A |
| Rise of a 0.75 mm trace at 2 A, IPC-2221 / IPC-2152 fit | 10.8 °C / 12.8 °C |
| Resistance at 35.8 °C, voltage drop, power | 35.08 mΩ, 70.2 mV, 140 mW |
This example teaches two things. First, for a narrow external trace the newer IPC-2152 data predicts a higher rise than IPC-2221, so IPC-2221 is not conservative everywhere. Second, switch Layer to internal and IPC-2221 asks for 2.04 mm and predicts a 52 °C rise for the 0.75 mm trace. That reflects the factor-of-two assumption, not physics: modern data shows that inner traces next to planes cool about as well as outer ones.
Accuracy and validity
| Model | Range | Notes |
|---|---|---|
| IPC-2221 equation | Up to 35 A external (17.5 A internal), widths to 400 mil, ΔT 10–100 °C, 0.5–3 oz copper (the extent of the charts) | A curve fit to old chart data that has been redrawn many times; the internal curve was derived, not measured |
| IPC-2152 fit (Brooks & Adam) | External traces, curves for roughly 5–500 mil widths | A fit to digitised charts, good to chart-reading precision. The IPC-2152 test boards had no adjacent planes |
| Resistance | DC, uniform temperature | Ignores the temperature gradient along the trace and skin effect |
The calculator warns when your inputs leave these ranges. It also warns when the trace temperature passes 105 °C or the typical 130 °C maximum operating temperature of FR-4.
Both models describe one isolated trace in still air, on a board of a particular size and material. A real board is often better than this. A solid plane 5–10 mil below a trace can lower its rise dramatically, because copper conducts heat roughly a thousand times better than FR-4. It can also be worse: many parallel traces heat each other, and boards in sealed enclosures or in vacuum lose the convective cooling. Brooks and Adam conclude that trace temperatures depend on too many variables for any chart, and that critical cases need thermal simulation.
Practical notes
- Use the finished thickness. Outer layers are usually plated up, and a "½ oz + plating" outer layer often ends up near 1.4–2 mil. Inner layers stay at foil thickness, and fabricators may etch them thinner.
- Etching narrows the trace. The copper cross-section is a trapezoid whose top is narrower than the drawn width by roughly the copper thickness. Size fine traces with margin.
- Long thin sections are the risk. A short neck-down at a pad or via has a lower current rating than the trace, but the copper around it sinks its heat. A long thin section has no such help.
- Electroplated copper conducts worse. Plated copper can have 10–20 % higher resistivity than the 58 MS/m annealed standard. Lower Copper conductivity to see the effect on drop and loss.
- Pulses heat less. These are steady-state models, and a short pulse heats a trace much less than the same DC current would. Use the RMS current for repetitive waveforms, but check the fusing limit for single high-energy events.