ESD Protection at Board and Enclosure Level

Static discharge is a fast, high-current event that lasts nanoseconds and arrives through paths the designer did not draw. It enters through a connector, a keypad, a seam in the enclosure or a metal part, and it leaves through whatever connection to earth happens to exist. Everything between those two points is exposed.

Good ESD protection is therefore designed at two levels at once: the enclosure, which decides where the current can enter, and the board, which decides where it flows once it is inside. This article covers both, and the layout measures that make a protection device actually work.

Two Levels of Protection: Enclosure and Board

The enclosure is the first line. A fully sealed insulating housing with no exposed conductors gives the discharge nowhere convenient to enter, and the board can then be designed without heroic measures. The difficulty is that most products do have exposed metal: connectors, switches, screws, decorative plated parts and the metal chassis itself.

The board is the second line, and its job is to redirect what gets in. The two mechanisms are blocking and diverting. Blocking means keeping the current out of the sensitive area with distance, barriers and insulated surfaces. Diverting means giving it an easy path to the main ground before it reaches anything that matters. Every practical measure belongs to one of those two categories.

How an ESD Current Actually Reaches a Circuit

The current has to have both an entry point and an exit. If a product is not connected to earth, the discharge still flows, coupling through the parasitic capacitance between the board and its surroundings, which is why the problem exists even on battery-powered devices with no earth connection.

The entry points are the interfaces that a user can touch and the openings in the housing. Once inside, the current flows toward the reference of the system, and it will take any low-impedance route, including one that passes through a signal line, a ground trace shared with a sensitive circuit, or the power distribution network. That is why the analysis starts with a list of the interfaces and the openings rather than with the components.

Enclosure Measures: Conductive Parts and Grounding

Conductive materials attract and accumulate charge, so a design that is expected to pass a severe ESD requirement should avoid exposed metal where it can. Where a metal part is necessary, it should have a defined and evenly distributed ground connection, with a spring contact or a metal finger preferred over conductive foam or fabric tape, which have higher and less predictable impedance.

The spacing rule is simple and worth enforcing: an exposed metal part should be kept at a distance from components and traces, with a few millimetres as a working minimum. Where a part cannot be grounded, its exposure should be reduced by making it thick, recessing it, or covering it. Where a component sits behind a seam or an opening, a physical barrier such as a high-temperature tape or an insulating shield keeps the discharge from reaching it directly.

Board-Level Measures: Plane Area and Layer Count

The board ability to absorb a discharge depends on how much low-impedance copper it has. A larger ground area neutralises more charge before a damaging voltage develops locally, which is one reason a board with a full ground layer behaves very differently from one without. On a design too small for a complete ground layer, at least one continuous ground layer should be provided, connected well to the battery or supply return.

Where the board is very small, its ability to absorb charge is limited and the enclosure has to do more of the work. This is a useful way to frame a design review: if the board is small and the enclosure has exposed metal, the two weaknesses compound, and the protection has to be added at the interfaces rather than assumed from the plane.

Placing a TVS Diode So It Works

A transient voltage suppressor only protects what it can reach before the disturbance does. The rule is that the signal should arrive at the suppressor first and reach the protected device afterwards, which means the device belongs right at the connector with the trace routed through it. A suppressor placed elsewhere on the same net, connected by a length of trace, has enough lead inductance to let the transient past before it conducts.

The ground connection of the suppressor matters as much as its signal connection. It should have the shortest possible path to the main ground, with several vias if it has to change layers, sized according to the via design rules used on the rest of the board, because the impedance of that path is what determines the clamping voltage the device actually delivers. Combining a suppressor with a high-frequency bypass capacitor between the supply and ground of the protected device further limits the voltage difference the device sees.

Protecting the Reset and Clock Circuits

A reset pin is one of the most common victims, because a short disturbance there stops the processor rather than merely corrupting a reading. The protection is a series resistor of the order of one to ten kilohms close to the pin, combined with a small capacitor from the pin to the adjacent ground, typically one to ten nanofarads. The resistor limits the injected current and the capacitor absorbs the charge.

The layout around the reset circuit has its own rules. The connection should be short rather than a long loop, and it should not run near the board edge or alongside an interface that a user can touch. The same treatment applies to a clock or oscillator circuit, which should be kept compact, surrounded by ground, and well away from any connector that could inject a discharge close to it. These measures also improve behaviour during radiated immunity testing, because a layout that is sensitive to a fast transient is usually sensitive to a radio-frequency field as well.

Software Measures and Their Limits

Software can recover from some disturbance, but it cannot prevent the disturbance. The measures that work are those that assume a transient is short: reading a control input several times and accepting a value only when the readings agree, discarding the highest and lowest samples in an analogue sequence, and using a watchdog to restart a processor that has stopped responding.

The limits are worth knowing. A filter on a signal widens the disturbance as well as attenuating it, so a bead and capacitor combination can turn one disturbance into a longer one that a confirmation routine cannot reject. Where a system must respond quickly, the software defence has to be chosen with that in mind. The most serious failure, a processor that locks up completely, is often a layout problem rather than a software one, and the protection of traces near the board edge is part of the structural answer.

Reviewing a Design for ESD

The review should start with the interfaces, because that is where the current enters. Every connector, switch, keypad and exposed metal part gets a protection device, and every protection device is checked for both its placement at the entry point and its ground connection. A device that is present but poorly connected is worse than no device at all, because it gives false confidence.

The second pass covers the routing: no important signal such as a clock or reset running along the board edge, no unprotected line reaching a controller pin directly, and unused board area filled with ground connected to the main ground. The principles used to suppress EMI in layout and the general rule of keeping the loop area small complete the picture, and together they cover the great majority of failures seen in practice.

FAQ

Is a TVS diode needed on every connector? On every connector that a user can touch, and on every line that leaves the enclosure. Internal board-to-board connectors that are never exposed are a lesser risk, though the line may still need protection if it runs a long distance.

Why does a protection device sometimes not help? Placement. If the trace reaches the protected device before it reaches the suppressor, or if the suppressor ground connection is long and inductive, the transient passes anyway. The device has to be at the entry point with a short ground return.

Can a metal enclosure solve the whole problem? It changes the problem rather than removing it. A metal housing shields the board but also provides a discharge path, so the grounding of the housing and the spacing between it and the board become the critical parameters.