Grounding of power line filter to Faraday cage chassis?

[goodphy]

Hello.

I've uploaded similar question before but I think it is time to make this clear.

I've seen the attached one during reading of grounding.

In the picture, the body of the power line filter (or ground of the filter) is connected to inner surface of Faraday cage and signal ground is connected to external surface of the cage. Thus if common mode noise is effectively rejected by the filter, I guess the noise must be flowing on the inner surface of the cage then...there maybe radiation of noise from this flow to inner volume of the cage.

I guess this is not good connection but could you tell me this is really reasonable connection? If not, could you please give me other way of ground connection?

 

[berkeman]

Hello.

I've uploaded similar question before but I think it is time to make this clear.

I've seen the attached one during reading of grounding.

In the picture, the body of the power line filter (or ground of the filter) is connected to inner surface of Faraday cage and signal ground is connected to external surface of the cage. Thus if common mode noise is effectively rejected by the filter, I guess the noise must be flowing on the inner surface of the cage then...there maybe radiation of noise from this flow to inner volume of the cage.

I guess this is not good connection but could you tell me this is really reasonable connection? If not, could you please give me other way of ground connection?

I don't think it matters which surface (inner/outer) of the metal box you connect to. After all, many such connections are made with bolts through the metal enclosure with lug connections, or with wires soldered into holes in the enclosure.

 

[Jeff Rosenbury]

Are you having a noise problem? If so, what can you tell us about the noise?

Is the Faraday cage intended for noise reduction or safety? (I've often seen cages near plasma bolts.)

Also, is this a medical application?

 

[the_emi_guy]

goodphy,
Your picture does not match your description.
Your description is that signal ground is connected to the external surface of the cage.
Your picture does not show anything connected to the external surface. Your equipment circuit (which I am assuming contains signal ground) looks like it is not connected to the cage at all.

If you are having an actual radiated emission issue you have come to the right place, there are lots of folks here savvy in this area, but you need to be more specific in your description.

 

[goodphy]

Are you having a noise problem? If so, what can you tell us about the noise?

Is the Faraday cage intended for noise reduction or safety? (I've often seen cages near plasma bolts.)

Also, is this a medical application?

The true noise source is gas-discharge in our High voltage lab. Thus there are not only EMI through ground line of the discharge system and ground lines but also EMI in form of radiation which is actually strong enough to sometimes shut down the computer near the system.

 

[goodphy]

goodphy,
Your picture does not match your description.
Your description is that signal ground is connected to the external surface of the cage.
Your picture does not show anything connected to the external surface. Your equipment circuit (which I am assuming contains signal ground) looks like it is not connected to the cage at all.

If you are having an actual radiated emission issue you have come to the right place, there are lots of folks here savvy in this area, but you need to be more specific in your description.

Hello.

Our main noise source is High-voltage gas discharge which gives us all king of EMI problem, strong enough to shut down the computer, It appears that there is very high strength of radiation of ~MHz to 90 MHz and ground potential is also fluctuation. The Faraday cage is actually the counter-measure for shield to radiation for sensitive equipment.

In the picture, signal ground is bottom 'GROUND' wire connected to external surface of the cage. The equipment connected to outside the cage only via live and neutral wires (current path in single-phase power line) but ground is not. Commercial power line has this connection configuration and probably to prevent common mode noise from flowing to the equipment via connected ground wire.

 

[the_emi_guy]

Still trying to understand what you have:

Is your "equipment circuit" inside of the chassis is the sensitive equipment that you are trying to protect? Or is the "equipment circuit" the high high voltage gas discharge that you are trying to contain?

Are there any other wires penetrating the chassis? The "equipment circuit" has no other connection outside of the chassis except for power? How do you monitor what it is doing?

Can you describe the chassis? How close to a Faraday cage is it? For example does it include EMI gaskets around access panels. Photos here would be great.

Your illustration shows the green ground wire entering the chassis then running some distance inside the chassis before connecting to chassis on the inside. Is this green wire also connected to the chassis right where it enters the chassis?

Is this green wire part of the power line filter assembly? If so, why is its connection to the chassis in a different place than the Y capacitors?

 

[goodphy]

Still trying to understand what you have:

Is your "equipment circuit" inside of the chassis is the sensitive equipment that you are trying to protect? Or is the "equipment circuit" the high high voltage gas discharge that you are trying to contain?

Are there any other wires penetrating the chassis? The "equipment circuit" has no other connection outside of the chassis except for power? How do you monitor what it is doing?

Can you describe the chassis? How close to a Faraday cage is it? For example does it include EMI gaskets around access panels. Photos here would be great.

Your illustration shows the green ground wire entering the chassis then running some distance inside the chassis before connecting to chassis on the inside. Is this green wire also connected to the chassis right where it enters the chassis?

Is this green wire part of the power line filter assembly? If so, why is its connection to the chassis in a different place than the Y capacitors?

You better see the attached image. The equipment circuit is sensitive circuit to be protected from EMI generated from discharge process. This equipment is of course, has output and input connected to outside via Isolated BNC adapter but for simplicity, I've skipped to draw them.

The penetrating wires are live and neutral wires for power line and BNC cables for measurement connected in a way mentioned above.

Actually I want to know the possibility that common mode noise (rejected from the filter) flowing through inner surface of cage to signal ground case via connection 3 and 2 can be the source of radiation to inside, rather than pursuing real design of the Faraday cage.

If so, my real design will include additional shielding room for power line filter inside the cage.

 

[the_emi_guy]

Good, I am following you now.

You are right, common mode current can enter the chassis on line/neutral then take whatever path it can to get to ground, this includes the green wire or your BNC cables.

Your challenge is how to prevent this from causing noise coupling to your sensitive circuits.

First thing to be aware of is that power line filters are generally designed to limit conducted emissions below 30MHz. If you really have 90MHz fields that you need to block choose your power line filter carefully, many provide very little attenuation at 90MHz due to inductor interwinding capacitance. You may want to include clip on NiZn ferrite around line/neutral either outside the chassis or inside just before power line filter.

Second, make sure your ground wire is connected to the chassis right at point 1. If it is not you will be steering line/neutral common mode current through that interior wire (2) on its way out to ground.

 

[Jeff Rosenbury]

You might try a separate, shielded ground cable. The shield should be grounded to the cage, but the cable should go to a grounding pit as far as possible from the cages' ground. See if you can insert one loop in the ground cable with a carefully chosen ferrite ring. You should be trying to block the RF while allowing a safety ground, so choose wisely. (Ferrite has a frequency dependence and you might be able to get it to saturate at safety levels of current.) Make sure to calculate carefully for safety. Adding inductance to the ground line can be a no-no.

As for getting the power in, can you build a shielded and filtered DC bus pair? Then you could use an inverter for equipment power. Of course inverters are noisy...

Getting signals in is harder. You might look at optic-couplers. Make sure to check their data sheets.

How big are the gaps in your cage? Is it possible some harmonics are slipping through? It's something to think about.

In general your cage isn't going to block everything. An ideal cage would, but unless you build it from a superconductor... You might check what is getting through the cage with a spectrum analyzer.

 

[goodphy]

You might try a separate, shielded ground cable. The shield should be grounded to the cage, but the cable should go to a grounding pit as far as possible from the cages' ground. See if you can insert one loop in the ground cable with a carefully chosen ferrite ring. You should be trying to block the RF while allowing a safety ground, so choose wisely. (Ferrite has a frequency dependence and you might be able to get it to saturate at safety levels of current.) Make sure to calculate carefully for safety. Adding inductance to the ground line can be a no-no.

As for getting the power in, can you build a shielded and filtered DC bus pair? Then you could use an inverter for equipment power. Of course inverters are noisy...

Getting signals in is harder. You might look at optic-couplers. Make sure to check their data sheets.

How big are the gaps in your cage? Is it possible some harmonics are slipping through? It's something to think about.

In general your cage isn't going to block everything. An ideal cage would, but unless you build it from a superconductor... You might check what is getting through the cage with a spectrum analyzer.

Hello.

I'm sorry but I have to ask what shielded ground cable means? You said shield of the cable connected to the cage and central conductor is configured in what way? I need to get more details as the concept of shielded cable is wired for me since I even don't know whether shield and conductors are grounded at the same point or plane.

 

[Jeff Rosenbury]

Hello.

I'm sorry but I have to ask what shielded ground cable means? You said shield of the cable connected to the cage and central conductor is configured in what way? I need to get more details as the concept of shielded cable is wired for me since I even don't know whether shield and conductors are grounded at the same point or plane.

Run an insulated cable inside a metal conduit. Ground the conduit to the cage. Ground the cable to whatever they use for ground where you are. (Perhaps a copper spike in the soil, perhaps a salt well, maybe the metal plumbing. It varies depending on soil/code.) Just make sure the cage is physically earthed separately from the ground cable. That way the cable will only pick up signal coupled from the conduit (which theoretically should be nearly zero, since it is also a Faraday cage -- a long thin one).

The idea is to provide more electrical isolation since there's clearly not enough.

I think the inside/outside the cage thing is a bit misleading since, at low frequencies the skin depth isn't going be deep enough to stop bleed through the wire and at high frequencies the skin depth will limit the current to the outsides of the wire (and thus the inside of the cage). There should be a sweet spot frequency where this would work, but since your having problems... (I could be wrong on this.)

What material is your cage? Are the wire intersections all electrically sound? I guess what I'm really asking about is your budget. How much can you afford to get this right? Some of my suggestions are not cheap.