Board Level Shield Can Design And Attachment

A board level shield is a metal cover placed over a group of components to stop the radio frequency energy they generate from leaving the area and to stop external energy from reaching them. It is a mechanical part fitted to the board, and its performance depends as much on how it is attached and grounded as on the metal it is made from.

This article covers what a shield can does, how it is constructed and grounded, how openings are handled, and how removal for rework is planned.

What A Shield Can Does

A shield reduces the coupling between the circuits inside it and the outside world. It works because a metal enclosure reflects and absorbs the field, and because it provides a return path for the currents that flow on the surface of the board beneath. Its effectiveness depends on the completeness of the enclosure and on the impedance of the path from the shield to the ground of the board.

A shield is not a substitute for good layout. A circuit that radiates because its return path is broken will still radiate inside the shield, and the energy will leave through the openings, the seams and the board itself. The shield is the last measure, applied after the layout has been made as quiet as possible, and the principles behind that work are set out under EMI suppression design principles.

Fence And Lid Constructions

A two piece shield uses a fence, a wall of metal soldered or pressed onto the board around the perimeter, with a separate lid that clips onto it. The fence is attached during assembly and the lid is fitted afterwards, which means the components inside can be inspected, tested and reworked before the lid goes on. The lid can be removed later with a tool, which makes the assembly serviceable.

A one piece shield is a formed metal can with its own walls and legs, placed as a single part and soldered along the perimeter. It is cheaper and simpler to fit, and it cannot be removed without desoldering, which makes any rework inside it a difficult operation. The choice between the two is usually made on the expected number of reworks and on how the assembly is tested rather than on electrical performance, since the two behave similarly when both are well grounded.

Grounding And The Return Path

The shield has to be connected to the ground of the board along its whole perimeter, at intervals short compared with the wavelength of the highest frequency of interest. A shield that is grounded at one point only behaves as an antenna rather than as an enclosure. That is why the fence is soldered continuously, or with clips at a close pitch, and why the ground plane beneath the perimeter is continuous.

The connection also has to be a low impedance path for the current. A single via connecting the shield to an inner ground plane is not enough, because the inductance of that via rises with frequency. In practice the perimeter sits over a row of vias or over a solid ground area that is itself well connected, and the arrangement of that grounding is part of the power and ground planning described under ground routing and power trace planning.

Openings, Vents And Heat

Every opening in a shield leaks, and the leak depends on the largest dimension of the opening rather than on its area. A long thin slot is much worse than a row of small holes with the same total area, because the slot behaves as a slot antenna at a wavelength related to its length. Where an opening is needed for a connector, its length is minimised and it is divided where possible, and where holes are needed for airflow they are made small and closely spaced.

Heat is the constraint that forces many of those openings. A shield over a device that dissipates significant power traps the heat and raises the temperature of the components inside, which can exceed their rating. The remedies are to use the shield itself as a heat spreader by thermally coupling the device to it, to provide a controlled opening, or to use a shield with a thermally conductive interface to the case. Each of these reintroduces a design problem elsewhere, and the trade is settled by measurement of the temperature inside the shield in the real product.

Removal And Rework

A two piece shield is designed to be removed. The lid is lifted with a tool, and the fence stays on the board, so the board is not disturbed. The lid has to be strong enough to survive several removal and refitting cycles without losing its spring, and the fence has to be strong enough that the lid can be clipped to it repeatedly without deforming. Both are mechanical requirements that should be stated when the shield is specified.

A one piece shield is removed by desoldering, which requires a hot air tool or a dedicated removal fixture and heats the board locally. The perimeter solder has to be melted all round at the same time, or the shield lifts at one corner and tears a pad. The design measure that makes this possible is a perimeter that allows access to the solder joint and a ground connection that can be reworked. Where the shield covers a large area, the thermal mass makes removal difficult and the two piece construction is the practical choice.

Design And Assembly Rules

The layout should provide a continuous ground perimeter under the fence, with vias tied to the ground plane at a close pitch, and a clearance inside the fence for the components and for the solder fillet. The fence footprint is a keep-out on all layers except the ground, and it should be wide enough that the fence can be soldered without touching a neighbouring conductor.

The assembly sequence has to place the shield after the components and after any test that needs access to them. Where the shield is fitted before test, the test points inside it become inaccessible and the coverage is lost. Where the lid is fitted after test, the sequence has a step that must not be forgotten, because a missing lid is an electromagnetic compliance failure that no electrical test will detect. The rules that keep this kind of process step visible are part of EMI reduction through stackup and layout.

FAQ

Does a shield make a noisy circuit quiet? It reduces what leaves the area and what enters it, but a circuit that radiates because of a broken return path will continue to radiate inside the shield and to leak through the openings. Layout comes first.

How often does the shield need to be grounded? At intervals short compared with the wavelength of the highest frequency of interest, which in practice means a close pitch of vias along the perimeter rather than a few connections.

Can a shield be used as a heat sink? It can, if the device is thermally coupled to it, and this is a common reason for choosing a metal can over a plated plastic one. The coupling has to be designed, because an air gap inside the shield is an insulator.