Electronic – Faraday cage – reducing EMI

The following forum post might be interesting for you: Mathematical derivation of the Faraday cage from the Maxwell Equations.

Especially post #4 from Astronuc (please have look in the link above for the complete citation):

Well trivially, it's Gauss's law. Inside the hollow conductor there is not charge, so the enclosed charge is zero, so the electric field is zero everywhere.

Now more directly, consider the most trivial case of the center of a hollow sphere, with 'uniform' charge on the surface. For each charge, there is an equal charge diametrically opposed, and therefore at the center the electrical fields (vectors) are equal and opposite, so they cancel.

Now, consider any point, off-center. One cannot apply the opposite point charge, but rather one must consider opposing surfaces, \$dA\$, which would have charges \$σ_1dA_1\$ and \$σ_2dA_2\$. Now think if two cones with vertices touching (and having same solid angles) and colinear (parallel) axes, with heights \$r_1\$ and \$r_2\$. The \$E\$ from one is just \$\dfrac{σ_1dA_1}{r_1^2}\$ and the other is \$\dfrac{σ_2dA_2}{r_2^2}\$, but realize that \$dA_i\$ is proportional to \$r_i^2dΩ_i\$, where \$dΩ\$ is the solid angle enveloped by cones and subtended by \$dA_i\$.

So \$E_i\$ is proportional to \$\dfrac{1}{r_i^2}\$, and \$dA_i\$ is proportional to \$r_i^2\$, and the term cancel which then leaves equal charges (\$σdΩ\$) opposing each other, and therefore the electric fields cancel, i.e. \$\vec{E}=0\$.

First off, get Electromagnetic Compatibility Engineering by Henry W. Ott and read most of it. Secondly there are no hard/fast rules when it comes to EMC engineering, it's more of an art, and the best way to answer questions is to build and test (you'll need some equipment like an RF generator or a spectrum analyzer, it also helps to have access to a lab with the equipment needed to carry out emission tests, but you don't have to.)

Need I faraday cage between chasis plane and ground plane ? Will It be helpfull to reduce any EMI ?

It really depends on what tests you need to clear, most countries require something like FCC class A (or B) or EN55022. The difference is usually where the product goes (industrial or residential). See what the regulations are for the areas of the product you want to sell in (or get a consultant). Most oscillators and microprocessors emit some kind of radiation and need some kind of shielding to pass the radiation tests. Usually a metal enclosure that covers most of the PCB is sufficient in the products I build.

Source: https://powermateusa.com/wp-content/tech.html

How can I draw traces that must be connect any point where is out of shield? Is below figure proper? Can I draw traces just between the shield pins/pads. Won't it break the shielding? Or I have to use vias and draw trace on bottom layer?

It depends on what you want to shield against for normal everyday environments, shielding isn't necessary for digital devices. Very sensitive analog devices (uV or nV level amplifiers and sensors) will need shielding. The first thing is to know what frequencies the electronics are susceptible to. If these are digital pins, it's unlikely that regular boring radiation will cause 100's of mV needed to flip a bit on a 3.3V digital channel. If you are in a higher radiation environment, then you could always put a small cap to ground (in the nF or pF range) or even make a small low pass filter depending the frequency needs of your digital channels. A differential channel like RS485 will not be as susceptible to noise due to it's differential nature.

Which components I must cover? My covering groups that I explained above is proper? Or I must cover all components in 1 shield. Or any other combinations ?

Again this depends on if you are trying to block radiation or prevent radiation from escaping your device. If you are trying to block radiation from leaving a board, usually a shield over the sources of radiation is sufficient (microprocessors and clocks). If your are trying to block it, then a much more in depth analysis (which requires a book to explain) will be needed.

Where shield pads must be connect to ? Directly to the ground or chasis or anywhere?

The best place to connect the pads of the shield are for the current generated on the shield to return back to the source (the path of least inductance). Minimize the inductance between the shield and the source of the RF. Usually this is PCB ground, but doesn't have to be. There are some situations where this is not possible.