Some ESD protection questions (spark gaps, earth/ground connection issues)

Envision a complete circuit between the ESD source and your product: This must include a complete return path. Often, this involves the protective 'earth' or grounding system. Your goal is to divert as much of the ESD energy as possible away from your circuits, both to protect circuit elements and to prevent disruption of equipment operation. If you insulate your product completely, there will be no ESD to the equipment, but this is impractical in cases where conductive parts must be exposed, particularly if those conductive parts cannot be connected to earth ground. The best defense, then, is to provide deliberate adjacent paths to ground, for example, the grounded 'ring' around the perimeter of a circuit board. If possible, design the product so that grounded conductive parts extend out in such a way that any ESD conducts to ground rather than into your circuit.

For any ungrounded paths into your circuit that may still take an ESD hit, you have these basic defenses to consider: a) series resistance b) series inductance c) shunt resistance and shunt breakdown devices such as zeners, transils, MOVs, spark gaps d) shunt capacitance

For high speed signals, it may be impossible to incorporate any appreciable amount of series resistance or inductance or shunt capacitance without degrading the signal, so you're left with the low- capacitance tranzorb solution. The common terminal of the tranzorb(s) should be connected to earth ground with the lowest possible impedance. If this doesn't do the job, reconsider how you might prevent direct ESD to signal nodes: shielding, shrouded connectors, etc.

Keep in mind that the high currents and fast risetimes associated with ESD will develop high potentials across stray inductances in the return path, AND there may be significant magnetic field coupling of ESD currents to adjacent circuitry. Therefore, loop area of the complete ESD current path should be considered.

Circuit board spark gaps will produce widely varying results depending on humidity, dust buildup, and other factors, and I think your estimate of 1 to 2 kV for 8 mils is probably a bit low for most conditions. Spark gap tubes have the advantage over semiconductor devices of maintaining relatively shunt voltages even for very high current levels, so they are often used to absorb more powerful transients such as those resulting from indirect lightning strikes.

In Summary:

1) Design packaging so that product is either totally insulated from ESD or any discharges go preferentially to grounded structures, shunting all ESD current away from internal circuits. 2) Series and and shunt any remaining filter nodes that must take a direct ESD hit.

As you get farther into this subject, you'll also discover the difference between insulating systems that dissipate vs collect static charge and between deliberate low-impedance discharge of ESD versus dissipating or bleeding off charge to prevent high amplitude ESD. Paul Mathews

Surely USB has a grounded shell and provision that the ground connection mates first, then the power connections, then the signal? There's no real threat of high potentials/spark/currents there, unless your shield melts in a lightning strike. If you're canny about the other I/O ports, they can be pretty robust, too, and NOT because of TVS diodes or other addons.

Spark gaps (commercially available ones, look like disk capacitors but with a little saw-cut slot) will take a lot of current and don't often fail, BUT they're intended for rather high breakdown thresholds. Unless you expect high currents at high voltages (like an isolated power supply that floats at a few hundred volts from ground), the spark gap is going to be too insensitive, won't protect delicate circuitry well. Low-pressure gas has better breakdown thresholds, and a Ne lamp can be a very effective clamp if you can tolerate the (about 100V) turnon threshold.

Grounded-track-surround is a variant of a guard ring. Multiple guard rings connected to successive potentials is the hallmark of HV circuitry.

Guard rings are effective if you expect tracking (conduction through foreign substances after your PC boards get dirty), and I'd expect to see them used a LOT where a circuit board is expected to have hundreds of volts per cm kinds of gradients. USB and ' I/O ports' isn't really in that range, usually. With multiple guard rings, connected with spark gaps, you can make your printed circuit board REALLY intimidating, even without international warning symbols. Techs will quake every time they open the box...

Typical ESD testing is at low event energy, with voltages (2 kV) that can be handled with minor precautions. But, any controlled, safe environment for such testing IS a laboratory, by definition. If you don't want to 'go out' to a laboratory, you build one for yourself.

You might be able to find some useful information on this page:

Chris