At this same point, of course, as you intend it to do.

What I meant by "quiet" is as follows:
Sometimes you can see the ground connections of TVS, running to the chassis ground, not being routed properly, means not being routed by the help of a local ground plane with shortest connections to the chassis, but by rather inductive copper traces. Then, right at the grounds of TVS you will observe rather high voltage drops (refering to the chassis ground) during ESD events. So, such a point would not be a good location to have the signal ground being connected to, because it will see relevant voltage drops caused by ESD currents. Means, such a location would not be "quiet".

You should connect the signal ground of 4-layer board directly to the above chassis ground, not to any copper trace also running to chassis ground, to minimize all these voltage drops.


Then, the ground lines of cables can be connected directly to the chassis ground, as you stated it here:


You got it: The ground star point arrangement at the location you mentioned will provide the best results.

But don't take this "star point" phrase too literaly. Nothing is gained, if you feed each ground connection of TVS via a separate and snaky copper trace to the central ground "star point". This would result in great disappointement. All these ground return currents of TVS must reach the star point via a ground plane. And if too many parts are to be connected to the star point, that the star point no longer looks like a "point", then use a larger ground star point. Means, the connection between enclosure and the chassis ground plane of your 4-layer board is allowed to be provided by a larger area, just large enough so that each TVS's ground reaches the chassis ground via a ground plane (means width of connection is larger than length).

An interesting question arises from the above: What is a "quiet" location for ESD events? Or, Where do we want all this ESD stuff being shunted to?
Well, it's clear that we want to keep all these annoying ESD effects outside of our device. So, there's only one location where we can the ESD currents shunt to, namely the metal enclosure itself. By connecting all the grounds of TVS and all the grounds of optional low pass filters to the chassis ground we can obtain optimum results. Especially if we combine these filtering and ESD suppressing measures with the concept of Faraday cage, which helps us to keep nearly everything outside of an (entirely) closed metal enclosure! ESD events must always be shunted to the metal enclosure, well away from internal electronics.

Even better results can be obtained, if shielded cables are used, with the shield connected to chassis ground at both ends, forming a Faraday cage arrangement, where the "enclosure" appears to be extended over all the cables. Then, it's very difficult for ESD events to inject their unsane energy into the internal electronics.

Finally, a very nice example of how to use a chassis plane for the TVS and how to use common mode filtering for the signals fed from the chassis plane to the signal plane can be seen here:

http://www.8052.com/forum/read.phtml?id=52259

Take note, how many different grounds are distinguished...

Kai