Loading…
Size a trace from its current, copper weight and allowed temperature rise, then read the voltage drop, the hot resistance and the fault current. IPC-2221 and the IPC-2152 universal chart are both on the page, with the thermal environment exposed.
Direct answer
A 3 A trace on 1 oz copper with a 10 °C rise is about 1.8 mm wide by the IPC-2152 universal chart. The IPC-2221 fits bracket it: about 1.4 mm on an outer layer and 3.7 mm on an inner layer, because the legacy fit gives an inner layer roughly half the current. Width is half the answer: a 50 mm route at 3 A drops roughly 30 mV, and a low-voltage rail cares as much about that drop as about temperature. Check the drop, the connector and the fault current.
Calculator
Enter the current, the copper weight and the rise you can tolerate. The panel solves the cross-section, then reports the resistance at the hot copper temperature, the voltage drop, the power loss and the one-second melt current. The thermal section lets you add a board thickness, a nearby plane, airflow and altitude. A fabrication minimum rounds a solved width up, so the answer is a trace the etcher can hold.
Model and validity range. IPC-2221 Brooks-Adam chart fit and the IPC-2152 Figure 5-2 fit, with temperature-dependent copper resistivity, an Onderdonk melt current, a 12% internal-layer rise premium and first-order thermal corrections for board thickness, plane proximity, airflow, altitude and copper weight. The legacy fit is optimistic on an outer layer and conservative on an inner layer; the IPC-2152 chart sits between them. Charts span 2 to 100 °C of rise, 0.5 to 3 oz copper and 30 A; outside that range the result is an extrapolation.
Direction
Current and copper
Peak current. A duty cycle is applied as RMS for heating.
Rounds a solved width up to the fab's floor. Blank means no rule.
Sets the hot resistance and fusing current; the width target is a rise above ambient.
Resistance and fault
Model and thermal environment
Applied only while the environment corrections are on.
0 means no nearby plane. The plane terms need a distance above zero.
Only used when a plane distance is entered and corrections are on.
Trace current capacity
Continuous current, hot resistance, voltage drop and fault current for one cross-section.
1.840 mm for 3.00 A
1.840 mm for 3.00 A
IPC-2152 universal chart, with the thermal corrections shown below.
72.5 mil for the entered current and rise.
Continuous current at the width and rise in the model.
Hot copper at 35.0 °C.
99.3 square mil.
At 35.0 °C over the entered length.
0.86% of the supply.
Copper dissipation over the entered length.
Melt current for 1.000 s.
RMS current divided by the copper area.
IPC-2221 and IPC-2152 side by side
| Metric | IPC-2221 | IPC-2152 |
|---|---|---|
| Required width | 1.375 mm | 1.840 mm |
| Ampacity at width | 3.00 A | 3.00 A |
| Rise at current | 10.0 °C | 10.0 °C |
| Copper area | 0.0478 mm² | 0.0640 mm² |
The IPC-2152 number includes the layer term and the environment terms. Turning the environment corrections off strips the board, plane, airflow, altitude and copper terms but keeps the layer term, so the internal-layer premium stays visible.
Thermal correction factors
The product is 0.862, so the chart rise is multiplied by that factor.
What this does not cover
Free, no account. An engineer reviews these inputs before the fixed fee is issued.
Results are engineering estimates, not design verification. Confirm them against the fabrication drawing, a TDR coupon or a 3-D extraction before release. Read the tool terms.
A wrong number, a rejected in-range value or a missing model? Send it to the engineer.
Every result above traces to an equation below. Read the equations and their validity limits.
Equations
Each block gives the equation the calculator evaluates, every symbol with its unit, and the range over which the model holds, with what it cannot see stated alongside. The button in the corner copies the LaTeX source.
The historical chart fit: a current raises a copper cross-section by an amount that depends on the rise to the 0.44 power and the area to the 0.725 power.
The equation is the Brooks and Adam fit to the charts in IPC-2221 Figure 6-4. The two exponents are not physical constants; they are the slopes of straight lines on a log-log plot of measured current against area.
The constant k carries the layer assumption: 0.048 for an outer layer exposed to air and 0.024 for an inner layer surrounded by laminate. That factor of two is the source of most of the disagreement with IPC-2152.
Model limits
The fit is a curve through the published charts, not a field solve. It holds for the chart's 10 to 100 degree rise and 0.5 to 3 oz copper; outside that band it extrapolates and can understate a thin trace.
The IPC-2152 universal chart as a two-coefficient curve fit, with an exponent that changes with the allowed temperature rise.
The chart's lines are straight on log-log axes, so each temperature rise has its own multiplier and slope. The fit interpolates both coefficients between 2 and 100 degrees.
At 10 A and 20 degrees of rise it returns 513 square mils against the chart's 500, which is inside 3% over the chart's span. That is the number the panel shows when the thermal corrections are turned off.
Model limits
The fit describes the universal chart: a 70 mil polyimide board with 3 oz copper in still air. The environment corrections on the page are first-order scalings around that curve, not a full lookup of every IPC-2152 figure.
Copper resistance grows with temperature, so the voltage drop must be evaluated at the hot trace temperature rather than at the bench value.
Copper's resistivity rises about 0.393% per degree at room temperature. A trace that runs 30 degrees above ambient is roughly 12% more resistive than the same trace at 20 degrees.
The length is the routed length, not the straight-line distance, and the area is the finished copper cross-section. On a low-voltage rail this term, not the temperature rise, is often what forces a wider trace.
Model limits
The equation is for a rectangular copper bar. It ignores the return path, via transitions and the skin effect, which matter above a few megahertz and in a narrow plane.
Each environment term scales the chart's rise, and the product is what turns the universal chart into an estimate for a specific board.
The panel computes a factor for board thickness, plane proximity, airflow, altitude and copper weight. A factor below one is a cooler trace, and a factor above one is a hotter one.
The factors are deliberately mild. They are there to show the direction and rough magnitude of the environment, not to replace a thermal simulation or a measurement on the real enclosure.
Model limits
Each factor is a first-order scaling of the heat path, not a lookup from the standard. A sealed enclosure, a potted assembly or a dense copper field needs a thermal simulation.
The adiabatic melt current for a copper conductor over a short time, the number a fuse or a fault has to be compared with.
Onderdonk assumes the conductor stores all the heat until it reaches the melting point. That makes the current depend on the inverse square root of the event duration.
A trace is converted to circular mils because that is the unit the relation is written in. The result is a fault-checking number, not a continuous rating; the continuous answer is the ampacity model above.
Model limits
Onderdonk is an adiabatic model. It overstates the current a trace survives for long times and understates how much the surrounding copper helps during a short pulse.
Sequence
A trace width is not a free variable. The copper weight is fixed by the stackup, the allowed rise is fixed by the enclosure and the laminate, and the current is fixed by the load. The only thing the layout can choose is width, and the model exists to stop that choice being a guess.
Work through the numbers in this order:
The current calculator and the Stackup Planner share the copper weight and layer assumptions. Build the stack first, then size the power traces on it.
Method
IPC-2221 is the historical chart fit. It gives an inner layer half the current of an outer layer because the laminate was treated as an insulator. That assumption is conservative for a board with a plane nearby and optimistic for a board without one.
IPC-2152 was built from a broader test program and is chart-based. The universal chart used here describes a polyimide board 70 mil thick with 3 oz copper in still air and does not split internal from external conductors. The panel therefore adds a 12% rise premium for an inner trace, far below IPC-2221's half-current rule, and exposes the other corrections that move the chart toward a real board: board thickness, a nearby plane, airflow, altitude and copper weight. Those corrections are engineering scalings around the published curve, not a re-digitisation of every IPC-2152 figure, and the page says so.
The practical rule is to compare both. If the two widths are close, the geometry is insensitive to the model. If they are far apart, the board is in a regime where the environment, not the equation, decides.
Corrections
The universal chart is one board in one still-air box. Four things move a real trace away from it, and the panel exposes each one as a named factor instead of hiding it in a fit.
A power plane is also the return path. Check the plane continuity under the trace in the Impedance Calculator and the cavity behaviour in the PDN Impedance before the current path is frozen.
Consequence
The temperature model answers how hot the copper gets. The resistance model answers what the copper does to the circuit. They use the same cross-section but different consequences, and both belong on the drawing.
Resistance is evaluated at the hot trace temperature, not at 20 °C. A 30 °C rise adds about 12% to the copper resistance, so a rail that was marginal on the bench becomes a rail that browns out in a hot enclosure. The voltage drop is that resistance times the peak current, and the percentage of the supply is the number a digital rail actually has to hold.
The fusing current is an Onderdonk adiabatic estimate. It assumes the conductor loses no heat while it reaches the melting point, which is the right model for a fault measured in milliseconds and the wrong model for a continuous rating. Read it as the upper bound of a fault event, never as an ampacity.
For a high-current input, a copper pour or a bus bar is usually the right answer, and our high-speed PCB layout service practice sizes the planes and the via field around it. Published PCB layout pricing prices the layout once the geometry is fixed.
FAQ
Send the stackup, the interface list and the power and return-path geometry inputs. An engineer re-solves them against the real layout before the fixed fee is issued.