designer’s notebook
The Art of Stifling Interference on a Printed Circuit Board Assembly
A printed circuit is an antenna for transmitting and receiving energy.
A raging fraternity party with thumping house music can be annoying as the morning hours approach. Noise suppression ordinances to the rescue! The partiers have two choices: quiet down or get shut down.

In that sense, the fraternity party is like building an electronic circuit. If our machines make too much “noise” in any part of the spectrum, it’s game over.

Just like kids can stop trampling everyone’s lawn and come inside, shut the doors, windows, shutters and even the fireplace flue, we can also contain unwanted spectral emissions. Left unchecked, a printed circuit is an antenna for transmitting and receiving energy from within and outside the board.

Even a well-designed PCB has compromises. Our goal is to be ready to react to spurious emissions that take us beyond the allowable threshold. We start with a power budget, and a noise floor, and our electronics must comply with those design criteria while meeting regulatory requirements.

As we bring different functions aboard, the problems in the near field multiply. We think of high-speed transmission lines as the focus of our EMI abatement efforts. They are seen as the cause of the problem, owing to their signal rise times or a harmonic of that momentary event.

When the root cause is identified, the solution often comes in the form of an additional filter of some kind. Sometimes, that’s not enough, and a more extensive re-spin of the board is needed. Just as good fences make good neighbors, so does metallic shielding around various circuits on the PCB.

Large crowd of participants
FIGURE 1. More participants mean more noise.
Generational shrinkage of a WiFi radio over the product lifecycle
FIGURE 2. Generational shrinkage of a WiFi radio over the product lifecycle. They are much smaller these days.
Coupling: That trace right there is really noisy. Pass it on! If nothing else, a frame around a circuit helps define it as a group. Set off by the shield footprint, the part is already on its way to a happy state of affairs. The shield footprint is typically secured to inner ground planes by a ring of vias along the perimeter. That’s all part of the 3-D Faraday cage that suppresses crosstalk within the board itself.

Coupling is a nonlinear thing. The amount of coupling one element will receive from another is a function of the length and width of the gap between the two elements, as well as the magnitude and frequency of the signals. With all that factored in, doubling the gap will square the amount of isolation. That’s a bargain, but it works both ways, of course. Cut the gap in half and the coupling is squared. Yikes!

An interactive heatmap to show us the price of rotating an inductor or adding an extra decoupling cap would be nice. Right now, that is an outside function known as SI/PI. Signal integrity/power integrity is the branch of engineering that simulates circuits and saves us from critical mistakes. As the PCB becomes more advanced, SI/PI simulations become much more important.

Coexistence: Making each constituent equally uncomfortable. The only way to get space for something is to take it from something else. This fact of nature is in play for most power distribution networks. We often have just enough copper to meet the overall power requirements and then only if it is apportioned exactly (FIGURE 2).

Getting the balance right is most important when power is limited. Mobile consumer electronics walk a tightrope of high performance with low power consumption. Inevitably, the battery grows to crowd the electronics together. When it comes to the final form factor, the shields will cover nearly all the components as a substitute for the space they had on the breadboard.

After all circuits are fully integrated and playing nice together, it’s time to consider the world of other devices. Bottling up EMI is helped by the metal enclosures, plus an awareness of the noise sources and the circuit elements most affected by noise.

Aggressors: The party animals of the electronics world. We can depend on some circuits to be a house party. One of the characteristics of a noisy circuit element is that they never have a rest state. A crystal will hum along nonstop. A switch mode power supply turns on and off at a certain rate by design. These two players are not usually switching their states all that fast. It’s those harmonics that come into play. Clock traces do the same thing and may not need the help from harmonics, but they get it.

The strategy here is to create a jail that keeps the spurious signals bottled up. For crystals, I like to isolate them with a moat in the copper pour that helps contain their emissions (FIGURE 3). Not only is the copper pulled well back, no ground vias are nearby to introduce the crystal to the inner planes.

Layout of a switch mode power supply (SMPS) will feature an inductor on the output pin. The shape that connects the output pin(s) to the inductor is the power supply’s megaphone. That connection must be short and only wide enough for full power transfer. Aim for a rectangular placement of each power supply sub-section. Voltage conditioning can go near the main supply, but individual power supplies should be located nearer their respective loads.

The repeated focus on clean power is because a good power distribution network will help cover for some shenanigans on the routing portion of the board. It’s just a fact that some chips require a few compromises to get through the fan-out.

Isolating a crystal’s ground pins from the surrounding copper pour
FIGURE 3. Isolating a crystal’s ground pins from the surrounding copper pour. (I rotated the circuit along with the route-keep-out after this screen grab.)
Off-the-shelf shielding solutions with pre-cut “mouse holes.”
FIGURE 4. Off-the-shelf shielding solutions with pre-cut “mouse holes.” The crossmembers stiffen the shield walls and provide a central pickup location for automated assembly. The lids can be perforated for airflow. (Source: Laird)
Victims: Some circuits are designed to listen, and that’s what they do. If you are already fighting EMI/EMC issues, it pays to take a good hard look at the power distribution, as well as the grounding scheme. Sensors and other similar devices are effective at picking up unwanted interference. In their case, the shield isn’t so much a jail as it is a penthouse where the listening device of whatever type can do its thing in peace.

Receiver circuits are more susceptible than the transmitter side. The higher magnitude of the TX signal raises the noise floor. Most of the traces on a board can be classified as one or the other, with I/O lines being a little of both. Locate the sensitive receive circuits as far as practical from the noisy inductors and other culprits. An elegant layout has these attributes and provisions to increase isolation, should the need arise.

The most common shielding type is the two-piece type that has walls and a lid that presses on over the walls (FIGURE 4.) This allows field service once the shield is installed, while the single-piece shield must be unsoldered from the board for work under the hood. Another way to use a single piece shield is to solder down little clips that hold the shield in place. Is that still a one-piece shield? The companies that make those also make a few standard-size covers to plug and play.

Of course, a two-piece will cost a little more. A four-sided shield with 90° bends keeps costs down. A six-sided L-shape or chamfered edge is not so bad, but costs will rise for polygonal versions. Multiple pockets within a single shield are possible. These are popular with analog circuits where multiple radios are in use.

Materials are generally stainless steel, with aluminum or nickel-silver alloys as the base. The solderable parts will be tinned, with other plating possible. Placement considerations are driven by fitting as much into as little space as possible. Watch the component headroom under the shield, including any webbing and overhang.

Mouse holes: To bury or not to bury the transmission lines; that is the question. Traces can route in or out of the shield on the surface layer through a series of slots located along the base of the wall. Walls with a continuous footprint require internal routing or cutting out a so-called mouse hole in the bottom of the wall where a trace can pass under.

Sometimes the available space demands a special shape for the shield, even though we plan for these things. When generating a bespoke footprint, follow some of the examples by leaving the pattern of mouse holes so that same shield could potentially be used for other purposes or accommodate future growth.

The mouse hole should be at least as high as the dielectric thickness to the reference plane. The width will consider the trace geometry but also the wavelength in some circumstances. Guidelines will range from a quarter of a wavelength to a twentieth, depending on isolation goals.

Holes in the top of the shield are not just for appearances or weight savings, though they accomplish that. The holes in the top work in concert with the lower slots for convective airflow. Another common feature is a few pegs built into the walls that require sizable plated through-holes in the board.

Think about clustering aggressors apart from victims during the floor planning and keep the geometry conducive to being corralled by a Faraday cage. Use the schematic diagram to document which enclosure each component will live in. The earlier you start thinking about isolating the good citizens from the frat boys, the better the outcome.

John Burkhert Jr. headshot
John Burkhert Jr.
is a career PCB designer experienced in military, telecom, consumer hardware and, lately, the automotive industry. Originally, he was an RF specialist but is compelled to flip the bit now and then to fill the need for high-speed digital design. He enjoys playing bass and racing bikes when he’s not writing about or performing PCB layout. His column is produced by Cadence Design Systems and runs monthly.