Why Doesn't a Grounded Chassis Create Ground Loops?

[691175002]

A lot of industrial automation electronics is very insistent that every device should have exactly one unique path to ground, but they are generally also connected to ground through the chassis or din rail.  Furthermore some devices like VFDs use shielded cables that are connected to ground at both ends, providing yet another path to ground.

I understand why ground loops should be avoided in a general sense, but I'm really confused as to why the "rules" seem to be ignored so often without explanation.

Are ground loops only a problem when they get large, or does sheet metal cause fewer problems because of its shape?

[ruffy91]

If done right chassis ground and signal ground are indeed separate. ESD suppressors and connector shells are connected to the chassis, while signal ground is brought back to the chassis ground in only one point.
See here  https://www.emcfastpass.com/avoid-the-most-common-failures-with-these-7-essential-emc-design-rules/ or here  http://www.hottconsultants.com/questions/chassis_to_circuit_ground_connection.html for reference.

[Housedad]

If done right chassis ground and signal ground are indeed separate. ESD suppressors and connector shells are connected to the chassis, while signal ground is brought back to the chassis ground in only one point.
See here  https://www.emcfastpass.com/avoid-the-most-common-failures-with-these-7-essential-emc-design-rules/ or here  http://www.hottconsultants.com/questions/chassis_to_circuit_ground_connection.html for reference.

That can change too about where signal grounds go.  In the Hitachi WJ200 series VFD, it requires not only signal and control grounds to come back to the VFD connection panel, but also the shields for the cables.  The VFD handles the shunting of grounds to a separate ground screw for circuit ground, while the mounting is a direct chassis bond.  (it is mounted via the heatsink).  I spent days pulling my hair out trying to learn about NEC required grounding and bonding for control circuitry in a cabinet when I designed a unique VFD control for my milling machine.  Green wire everywhere!!

[Kalin]

There is a big difference between grounding and bonding. You bond all of the non current carrying parts of electrical equipment together and give it a guaranteed low impedance path to your system grounding conductor.

The kind of ground that you would use in a signal cable is more of a shield to prevent interference and serves more as a way to sink induced voltage in the shield of the cable.

The only cable I have pulled for VFD's has been teck90 or in conduit and the only connections for the sheath and bare conductor have been bonding. The purpose of that is not signal integrity but one of safety. If you are careful to follow the rules about bonding throughout an installation there should be no issues. The main problem comes from having connections between distinct system groundings or improper grounding when creating a neutral in a distribution that can result in potential difference between the grounding conductors where two systems interconnect. If you follow the code you will have nothing to worry about and one day you will understand how it all fits together and makes sense :-)

Sent from my SM-G930W8 using Tapatalk

[RJFreeman]

Ah Ground loops, the joy they bring (NOT!).
So just to throw a couple of ideas into the mix, since I do not know the specifics:

Ground loop noise is a result of current flow through a loop, where the current is usually induced by a magnetic field (there can be instances where the current is from say a ground return on a power supply, or even decoupling capacitors, shunting supply noise to ground, this is when we look at star earthing things) .
As we are usually dealing with signal voltages, the ground loop noise between two ends of a cable (for example) is I x R (Current times resistance, when talking about noise from RF sources, than this would be current times impedance).

So, to reduce this noise, we want to reduce either I or R.
If we break the loop, we reduce I to zero, so no voltage drop between different parts of the signal ground.
But, the Gotcha here is, that much like an open transformer winding we end up with entire voltage across the open gap. Of course if this is in the power supply, or back where we don't have any signal paths, then that would not be a problem.

The other alternative is to reduce R as much as possible, so this is where we connect a heavy duty Bus bar between the grounds on two pieces of interconnected equipment

[T3sl4co1l]

They do, that's the whole point.

The point is to keep the currents outside of the signals, by wrapping the signals in contiguous shielding.  As long as the currents stay on the outside of those shields, the signals inside don't know the difference!

Or vice versa, because everything in linear E&M is reciprocal: the high voltage, intensely noisy wires coming out of an unfiltered VFD must be plumbed inside shielded conduit, otherwise those high frequency currents get into everything else!   

Tim

[Electro Detective]

Isn't it easier to invest in some Audiophool cabling and job done?! 

[David Hess]

If the chassis is used for multiple ground (common) points, then it does result in ground loops.  In some cases this is acceptable and in other cases, differential signal paths or other techniques are used to reduce common mode noise.

Tektronix did this in their 7000 series oscilloscopes but since the signal paths were all differential, it was not much of a problem.  But in the areas where single ended signaling was used, it is possible to see interference from things like the sweep gate and power supply ripple.