Can a PCB Carry 100 A? PCB Current Capacity in Copper

Most boards never see more than a couple of amperes. Consumer electronics typically run below 2 A continuously, and even a dense industrial board rarely exceeds 5 A or 10 A on any one net. Then a motor drive or a power conversion stage arrives with a continuous current of 80 A, and once transients and design margin are added, the requirement becomes 100 A or more. At that point the ordinary approach of widening the trace stops working, and the question becomes a structural one.

Where the Limit Comes From

The current capacity of a copper conductor is set by its resistance and by how much heat it can shed. Resistance depends on the resistivity of copper, the length of the path and the cross-sectional area, and since the resistivity is fixed, the only variables available to the designer are length and area. In a board, the area comes from the trace width multiplied by the copper thickness, and copper thickness is specified in ounces: one ounce of copper is about 35 micrometres, two ounces is about 70 micrometres, and heavier foils scale the same way. The immediate conclusion is that a high-current path should be as short and as wide as possible, and that thicker copper helps in direct proportion.

The Three Engineering Variables

In practice, engineers describe the PCB current capacity of a board in terms of three numbers: copper thickness, permitted temperature rise and conductor width. A common reference point is that a 100 mil (2.5 mm) wide trace on 1 oz copper, allowed to rise 10 degrees Celsius, carries about 4.5 A. Two features of that figure are worth noting. The first is the temperature rise, which is a design choice: allowing a larger rise increases the current the same trace can carry, at the cost of a hotter board and a shorter service life for everything nearby. The second is that the relationship between width and current is not linear. As the conductor gets wider, the additional current gained per millimetre falls, because the outer parts of a wide conductor are less effective at transferring heat. Any calculation should therefore be based on a curve or a table rather than on a simple ratio.

Working Through 100 A

Applying those variables to 100 A makes the scale of the problem clear. A single 1 oz trace would need an impractical width, so the practical route is heavier copper and a much wider conductor. Four ounce copper, a trace width of around 15 mm, the same conductor duplicated on both sides of the board, and additional heat sinking to hold the temperature rise down is a realistic combination. Even then, the resulting conductor is a copper area rather than a trace, and the layout should be planned around it from the beginning, because a 15 mm wide path consumes routing space on every layer it passes through. Where the current is even larger or the space is tighter, the right answer is usually to stop treating the board as the conductor.

Parallel Layers and Thermal Vias

Using several layers in parallel is the standard way to increase the cross-section without widening the board. Two identical conductors on the top and bottom layers, joined at both ends, halve the resistance, and inner layers can be added where the stackup allows. The layers must be joined properly at every transition, which means a sufficient number of vias rather than a symbolic connection. These vias also act as thermal paths, moving heat from the conductors into the inner planes and out to the surface, and their count should be calculated from the current and from the heat that has to leave the conductor. Where the current path turns or narrows, the effective width at that point governs the result, so a wide trace that necks down at a via field or a component pad is limited by the neck. The general relationship between width, thickness and current is set out in this article on trace width and current calculation.

<img src="https://www.gopcba.com/wp-content/uploads/2024/09/Wire_Harness_Assembly.webp" alt="bus bar connection carrying high current on a power board” />

Bus Bars and Terminals

The alternative to carrying the current in copper foil is to carry it in hardware. Terminal posts, surface mount studs and PCB terminal blocks rated for the current can be mounted on the board, and the high-current connection can be made with a cable and a crimped lug instead of a trace. This approach also removes the heat from the board, since the cable carries it away, and it makes the connection serviceable. A custom bus bar goes further: a formed copper bar bolted to the board or to the chassis is the usual industrial solution for very high currents, and it is what transformers, switchgear and server racks use. The bus bar has a large cross-section, is air-cooled on all sides, and can be bolted at both ends with a defined contact area, which makes the joint predictable. Its dimensions rather than the board”s copper then become the limiting factor.

Connections, Joints and Assembly

A high-current path is only as good as its weakest joint. Every connector in the path has a current rating and a contact resistance, and a connector chosen for signal use will fail long before the copper does. Solder joints in a heavy copper area need more heat to form, because the copper conducts the heat away from the joint, so the assembly process has to be adjusted for the board rather than the other way round. Where a bolted joint is used, the contact area, the surface finish and the torque all affect the resistance, and the joint should be specified rather than improvised. Thermal relief, which is normally desirable to make soldering easier, is the wrong choice on a high-current pad, where a solid connection is needed for both current and heat.

Verifying the Design

A high-current board should be measured rather than assumed. The two useful measurements are temperature and voltage drop. A thermal image taken at full load shows whether the heat is where the calculation predicted and whether any joint is running hotter than its surroundings. A four-wire resistance measurement across the path, or a simple voltage drop at a known current, gives the actual resistance to compare against the calculated value. Both measurements should be taken at the maximum expected ambient temperature, because a joint that is acceptable at 25 degrees Celsius may not be acceptable inside an enclosure at 60. Where the layout allows, the design should also be tested at the worst-case transient, since the thermal behaviour over milliseconds is different from the steady state.

Design Recommendations

For a 100 A path, the sequence that works is to decide first whether the current belongs in copper foil at all. If it does, use heavy copper, duplicate the conductor across layers, keep the path short and free of necks, plan the space for it before other routing, and provide enough thermal vias to move the heat into the planes. If it does not, use terminals and cables or a bus bar, and let the board carry the control and sensing signals instead. Whichever route is chosen, the stackup has to support it, and the options for copper weight and layer arrangement are described in this article on layer stackup from one to eight layers. Where the current is distributed across a large area of copper, the choice between a solid and a meshed fill also affects the result, as described in this article on copper flooding, mesh or solid.

FAQ

Can any ordinary board carry 100 A? No. It requires heavy copper, a very wide conductor, several layers in parallel and additional cooling, or a move to terminals, cables or a bus bar.

Does a wider trace always carry proportionally more current? No. The gain per millimetre falls as the conductor widens, because the outer regions are less effective at dissipating heat, so tables or curves should be used rather than a simple ratio.

Why is thermal relief wrong for a high-current pad? Because a relief reduces the copper cross-section at the pad, adding resistance exactly where the current density is highest and restricting the path for heat to leave the joint.