ESD Protection Placement And Routing Rules

Electrostatic discharge protection is designed as a circuit and then destroyed by its layout. A protection device that is placed a centimetre from the connector it is meant to defend will allow the transient to travel through the board, couple into neighbouring traces and reach the integrated circuit before the device has time to clamp. The device does its job only if the pulse is given a short, low inductance path to ground that does not pass through the rest of the design.

This article covers what the protection has to achieve, where the device belongs, how the return path is arranged, and how the result is verified.

What The Protection Has To Do

The device has to limit the voltage that reaches the sensitive pin to a value the pin can survive, and it has to do so within the rise time of the pulse. That second requirement is the one that layout controls. A clamping diode has a response time of a nanosecond or less, but the trace that connects it to the connector has inductance, and the voltage across that inductance adds directly to the clamping voltage seen by the circuit behind it.

The energy also has to go somewhere. A transient that arrives on a signal line is diverted to a reference, and the reference has to be able to accept the current without rising in voltage. If the return path is a narrow trace shared with the circuit ground, the current will raise the local ground potential and inject noise into every device that shares it, which is how a successful clamp becomes an functional failure.

Where The Device Belongs

The device goes between the connector and everything else, at the point where the signal enters the board. It should be placed within a few millimetres of the connector pin, on the same side of the board as the connector where possible, with the connection from the pin to the device made as short and as wide as the layout allows. Nothing else should be connected to that segment except the protection device.

The signal then leaves the protected node and travels to the circuit. That order is the whole principle: the pulse must meet the device before it meets anything else, including the trace that runs to the receiver. If a stub branches off the node before the device, the pulse will travel down the stub and couple into whatever is at the end of it, so the protected node should be a short tee rather than a long run with a device hanging off the middle.

The Return Path

The current path is a loop, and the loop area is what determines the inductance. The device connection to its reference should be short, wide and direct, and it should go to a ground plane rather than to a thin trace. Where the ground plane is on another layer, the device should be connected to it through several vias placed immediately at its ground pad, because a single via adds inductance that appears in series with the clamp.

The ground reference itself must be solid under the connector. In a design with a split ground, the protection device and the connector belong to the same reference, and that reference has to be the one the protected circuit uses, or the current will find a path between the two planes through the circuit. The general principles of plane arrangement are described under ground routing and power trace planning.

Routing Around The Device

The trace that carries the signal from the connector to the device should be as wide as practical, because a wide trace has lower inductance and also reduces the voltage drop while the clamp is conducting. The trace inductance between the connector pin and the device is what converts a fast edge into a voltage, so it is the length of that segment rather than the width of the rest of the net that decides the result. After the device the trace can return to its normal width, since the residual pulse is small, but it should not run parallel to an unprotected trace for any distance, or it will couple into it.

Where several protected lines enter through the same connector, the devices should be grouped so that they share a short ground connection but each line keeps its own short path. Sharing the ground return between devices is good practice; sharing the signal path is not, because a transient on one line would then develop a voltage that appears on the others.

Connector, Chassis And Board Edge

A connector that is accessible to a user is the usual entry point for a discharge, and the layout at that point is a compromise between the signal quality and the protection. Signals are usually routed to the inside of the board before the protection device, and the chassis connection is made at the connector shield, with the shield tied to chassis ground. The treatment of the board edge and the way the shield is handled are described under ESD and PCB edge traces.

The same arrangement has to work for radiated immunity as well as for direct discharge, because a cable that acts as an antenna brings energy into the same node. The separation of the noisy interface are from the sensitive circuit, and the filtering that supports it, are covered under EMI suppression design principles.

Verification

The design is verified by injection. A discharge is applied to the connector pins, to the case and to any accessible metal, at the levels the product is required to survive, and the equipment is monitored during and after the test. A functional reset, a latched output or a permanent failure is a design defect, and the location of the fault usually points to the path the current took.

A second check is a current probe or a near field measurement around the protection device, which shows whether the current is flowing where the layout intended. If the return current is found on a signal trace rather than on the plane, the layout has a defect that the functional test may not have exposed, and it will appear later as an intermittent failure in the field.

Numbers Worth Knowing

Rule of thumb figures make the trade-offs visible. A trace of one millimetre in free space has an inductance of roughly one nanohenry, and a capacitor discharge can rise in a fraction of a nanosecond, so a millimetre of extra length is not academic. At a discharge current of 30 amperes with a rise time of one nanosecond, a path of 10 nanohenries develops about 300 volts across it, which is far more than the clamp voltage of the device and more than most integrated circuits will survive.

The same arithmetic explains why the layout and not the device datasheet sets the performance. Two boards with identical parts can differ by hundreds of volts in the voltage that reaches the pin, and the difference is entirely due to the length and the loop area of the connections around the device. A layout review that checks those connections before the board is made is the cheapest protection measure available.

FAQ

Can a single protection device serve several lines? It can, for lines that share a reference and never carry a discharge at the same time, but the ground connection then has to carry the sum of the currents and must be designed for twice the current of the worst single line.

Does the device have to be on the top side? It should be on the same side as the connector whenever possible, so that the via between them does not add inductance. Placing it on the opposite side adds a via pair that can double the effective clamping voltage.

Is a series resistor useful? It is, in combination with the device, because it limits the current that reaches the pin and forms a divider with the clamp. It is placed after the protection device, between it and the circuit.