EMI Shielding on PCBs: Cans and Grounding

What Shielding Can and Cannot Do

A shield can is a metal enclosure soldered or clipped over a circuit to keep electromagnetic energy in or out. It works by reflecting and absorbing fields, and it can only do so if it is connected to a reference with a low impedance and if it encloses the circuit completely. A can with a poor ground connection behaves as an antenna; a can with a slot larger than a fraction of a wavelength leaks like an open window. Shielding is therefore a grounding problem as much as a mechanical one, and the design effort belongs on the connection between the can and the board rather than on the metal itself.

The Shield Can Footprint

The footprint is a fence of pads or a continuous ring on which the can sits, with the wall soldered or clipped along its length. A continuous ring gives the best electrical performance and the simplest analysis, since there are no discrete gaps, but it consumes board area and makes the courtyard large. A fence of discrete pads relies on the can’s own fingers bridging between pads and produces a series of small apertures, which is usually acceptable below a few gigahertz if the pad spacing is small compared with the wavelength. The pad spacing is therefore the design parameter that matters, and it should be chosen from the highest frequency the circuit generates or is sensitive to rather than from the mechanical convenience of the can.

Ground Via Fences

A single row of stitching vias along the can footprint connects the ring to a ground plane beneath it, and that is the difference between a shield that works and one that merely looks like a shield. Without the vias, the ring is a piece of copper on the surface with an uncontrolled path to ground, and the shield current has to find its own way, which produces resonances and unpredictable performance. The vias should be spaced closely enough to present a low impedance at the highest frequency of interest, which in practice means a few millimetres rather than a centimetre, and they should be placed along the entire perimeter including the corners.

Stitching and Plane Integrity

The plane under the shield has to be continuous, because a split in the reference plane beneath a shield creates a slot that radiates. Where the design has separate analog and digital grounds, the shield usually sits over one of them and the boundary has to be planned so that the shield ring does not cross the split. Where the plane is fragmented by the antipads of the many vias in a dense area, the shield’s vias may not have a good plane to connect to, and a second ground layer stitched to the first is often the practical answer. The check is to look at the plane under the shield outline and confirm that it is solid metal.

Openings and Cable Penetrations

Every opening in a shield is an aperture, and an aperture radiates in proportion to its size relative to the wavelength. A slot that is a small fraction of the wavelength is largely harmless; a slot that approaches a quarter wavelength becomes an efficient antenna. The practical implication is that a long, narrow opening, such as the gap between a can and the board at a connector, is worse than a small round hole, because the relevant dimension is the longest one. Cables that leave a shield are a direct path for the noise, and the usual treatment is to filter the signals at the boundary or to use a shielded connector whose shell is bonded to the can.

Keep-Outs Under a Can

The area under a can belongs to the shield, so it should contain no test points that must be probed, no components that are expected to be tuned at the factory, and no keep-out zone that a stencil cannot reach. In practice, tuning is often needed, which is why so many radio designs leave a small removable access cover or place the tuning components at the edge of the shield. The can also has to clear the tallest component inside it, and the height has to be confirmed before the footprint is fixed, because a can that touches a component or that cannot be fitted is a redesign. Where the can must be reworkable, a socket or a clip system is used instead of solder, at the cost of a slightly less reliable electrical contact.

Conductive Gaskets and Board-Level Shielding

Where a can is fixed rather than clipped, a conductive gasket or a conductive adhesive can provide the contact between the can wall and the board. Where the shield has to be applied to a large area or a complex shape, a sprayed conductive coating or a metal foil with a conductive adhesive can be used, though these are more common on the enclosure than on the board. At board level, the alternative to a can is to move the noisy element or to improve the layout, which is often cheaper and more reliable than adding metal.

Design Checks

Check that the shield ring has stitching vias around the whole perimeter, that the plane beneath is continuous, that the pad spacing suits the frequency, and that no signal crosses the ring except through a filter. Confirm the components inside fit under the can, that the tuning access works, and that the courtyard does not collide with adjacent parts. Finally, plan the shield early: a can added at the end of the design is usually a can that does not fit.

FAQ

Does a shield can always fix an EMC problem? No. It works only if it is well grounded and encloses the circuit, and it does nothing for noise that leaves through a cable.

What spacing should the shield pads have? Close compared with the wavelength of the highest frequency of interest, usually a few millimetres rather than centimetres.

Why do I need vias around the shield ring? To connect the ring to the reference plane with a low impedance. Without them the shield is not a shield.

What is the worst kind of opening? A long narrow slot, because the relevant dimension is its length. A slot near a quarter wavelength radiates efficiently.

Should test points be inside the shield? Only if they can be probed after fitting, otherwise provide access or move them outside.

Conclusion

A shield can is a grounding structure, so the design effort belongs on the ring, the vias, the plane beneath it and the apertures. Plan the footprint early, fence it with vias on a spacing that suits the frequency, keep the reference plane solid, and account for every opening. Shielding interacts with PCB design and layout, the fabrication of the footprint and the vias is covered under PCB manufacturing, and the assembly of the can belongs to SMT PCB assembly. Radio boards with shields are normally developed through prototype PCB assembly in 2026.