Chassis to Ground - connect it or not?

[shaiko]

Suppose we have an electronic device with a metal chassis and a PCB mounted in it.
What are the prons and cons of shorting the PCB's ground planes to the chassis of the metel shell?

Please don't say "EMI immunity"...be as elaborate as you can.

 

[barry]

EMI immunity...and ground loops.

That is SUCH a broad question. You need to do some reading. There are a LOT of variables at play here. Where are your power supplies? How are they connected to ground? How are they connected to the chassis? Where? What else is in the system? Is the PCB in question sensitive to noise? Is it a noise generator? How is power connected to the board?

 

[HTA]

The chassis should be connected to a protrective ground(earth in many power stystems). If there is an isolating transformer in the power supply it should be better not to connect the PCB to ground. The problem with connecting the ground in a undefined way is that you introduce ground loops which could cause a lot of problems.

Enjoy your design work!

 

[shaiko]

I'm designing a PCB mounted inside a metal enclosure. This metal enclosure is mechanically fastened to a moving platform.
This moving platform feeds my PCB with power and other IO's on a single connector.

1. Should I use an isolated power supply? If yes, please explain why.
2. Should I short circuit the PCB's ground to the chassis of the metal box enclosure? What will happen if I don't?

 

[HTA]

An isolated power supply has several advantages. It isolates the subsystem from the main supply and avoids ground loop in distributed power supplies. If you have a human operating with your moving platform it make sense to connect the enclosure to a protective ground and have an isolated power supply. Since the metal box is shielding the PCB against external EMI it should be connected to ground, or better to shield ground of the supply cable(if available). Is the input power the AC line voltage, with or without a protective ground?

Enjoy your design work!

 

[FvM]

I think, the problem can't be discussed without knowing the instrument kind (e.g. RF, audio, precision measurement) and it's external signal connections.

In some applications direct connection of low voltage power and signal ground to enclosure and protective earth is pretty standard, in some it's typically avoided.

Low voltage power supplies are always isolated from mains. Even in an instrument with protective earth connection of the enclosure, a low voltage power supply would implement reinforced insulation and not strictly require earthing of the low voltage ground.

 

[shaiko]

HTA,

You say:

connect the enclosure to a protective ground and have an isolated power supply. Since the metal box is shielding the PCB

I thought the metal case shields the PCB because of it being a conductive enclosure...a Faraday's Cage should act like such regardless of the PCB's ground being shorted to the cage or not

The input voltage from the moving platform to the PCB is 28VDC (not AC).

 

[FvM]

a Faraday's Cage should act like such regardless of the PCB's ground being shorted to the cage or not

HTA or others didn't tell different.

Presumed you have a perfect Faraday cage. Then the interesting question is, how do you keep external interferences from propagating through external power or signal connections, perforating the shield.

 

[shaiko]

FVM,

a Faraday's Cage should act like such regardless of the PCB's ground being shorted to the cage or not.

Other then the cage's weak spots that you mentioned, it provides an overall good blockage agaist surrounding radiation...
So why not always keep our PCB ground floating relative to the chassis (as a general design rule)?
I'm trying to make sense of galvanically isolating the power input with an isolated DC/DC only to have it later violated with a PCB ground to chassis short!
please explain.

 

[FvM]

Design the I/O filtering and decide to which potential you want to "discharge" injected interference currents. Most engineers end up with a low impedance capacitive coupling to case ground.

Filtering is usually necessary for isolated supply inputs as well, because the coupling capacitance is sufficiently high to couple external interferences.

 

[shaiko]

And what may happen if I simply leave my PCB floating?
Whithout connectring it to the chassis...

 

[FvM]

You didn't consider any filter element against case ground, so the common mode voltage carried by the supply lines will be "copied" to the PCB ground, even across an "isolating" DC/DC converter. The PCB ground becomes hot against the shield, or viewn from the PCB side, the board is surrounded by a hot shield which injects interferences to any critical node through the case-to-board capacitance. In return, through the same coupling mechanism. PCB bound EMI will be transmitted to the outside, conducted as common mode interfernce by the supply cable.

 

[shaiko]

Very well.
And what about the complete opposite?
If we do want the chassis and the PCB's ground to have an electrical link - why not make it a strong high conductivity link? (i.e connect both planes with many vias all accross the PCB)

 

[FvM]

If we do want the chassis and the PCB's ground to have an electrical link - why not make it a strong high conductivity link?

That's the usual way for RF instruments. But as previously discussed, ground loops may be a problem in some applications, suggesting a capacitive link between circuit ground and PE. The best way has to be chosen per application.

 

[marce]

I would suggest purchasing and reading Henry Ott, 'Electromagnetic compatibility Engineering'.

His WEB site:
http://www.hottconsultants.com/