ESD

[phenol]

Suppose we have a board with copper pour ground plane on top and bottom layers and an external electromechanical device controlled by the board. The device has a metal shell accessible to users and therefore a subject of electrostatic discharges. It will be touching a sheet metal spring soldered to the board to dissipate the charge. Now the question is what would the best interface between the spring pad and board ground be... direct connection, an etched pattern spark gap, a string of current limiting resistors in order to reduce ground bounce, capacitors,...?
the esd pulse is air discharge 15kV

[madires]

Is that device powered by batteries, a wall wart or mains?

[phenol]

it's an automotive device, so go figure-truck battery, alternator...

[madires]

Is the metal shell mounted on the truck and electrically connected to the chassis?

[phenol]

no, the metal shell is mounted in said automotive device, the exterior of which is mostly made of plastic. The device is then connected to truck's grid via power supply cables with local emi filtering, ferrite beads, caps, that sort of things.
i am not being specific about the nature of this device on purpose , but it belongs to one of the largest automotive electronics manufacturer in Europe.

[madires]

And they don't know how deal with that setup?  I don't got any idea of automotive standards, but I would go for the string of resistors and/or ground the metal shell to the chassis via a dedicated wire.

[phenol]

Their idea is a single 0805 200 ohm resistor from spring pad to pcb ground. Im not entirely sure who knows what there, but when i shoot the sucker with the esd gun, there is a spark clearing the small gap between spring pad and pcb ground regardless of the resistor there. Long ground wires is a no no. ESD pulse has sub-nanosecond rise time. Equivalent circuit is 150pf via 330ohm resistor, so 10s of amps of discharge current that decays in less than 100ns or so, but causes all sorts of damage during fraction of that time

[madires]

AFAIK, 10kV easily cross a distance of 1cm in air. That means you would have to create an area with sufficient clearance and put several resistors in series. Though, the 0805 could be a spark gap on purpose.

[phenol]

this is what i would have done, along with plenty of keepout area under the spring pad on adjacent layers. I've seen others do it with a string of say 3 10-ohm resistors in parallel with 100p-1n-100n caps.

[phenol]

in this case it's not HBM, but IEC61000-4-2, so peak current could be as much as 6 times higher for equivalent voltages