Power line EMI filter + feedthrough filter for a Faraday cage

In summary, cheaper feedthrough filters go down to 100mhz while more expensive ones can go up to 1mhz. The combination of a cheaper feedthrough filter and a regular power line filter should be effective at attenuating EMI between 100khz and 5ghz.
  • #1
27
4
Dear forumers,

I need a 50hz 250VAC power line filter for an EMI shielding faraday cage. The filter should work between 100khz and 20ghz, attenuation should be >=60dB between 1mhz - 5ghz.

Regular power line EMI filters like this
EMI-Filter-Power-Line-Filter-Mains-Filter-EMI-EMC.jpg

don't seem to go that high (only up to a few hundreds of mhz or so).

I see that there are professional feedthrough filters
HPR-Installation-Example-300x279.jpg

that do the filtering alone but they are quite expensive.

So I am thinking about using a cheaper feedthrough filter or capacitor (one that starts to filter above 10mhz) in combination with the regular one.

The regular one's spice model would look like the one in the attachment.Has someone any experience with this combination or any counter-proposal against this?

Regards,
Akos
 

Attachments

  • Screenshot_20180929_210342.png
    Screenshot_20180929_210342.png
    26.2 KB · Views: 574
  • EMI-Filter-Power-Line-Filter-Mains-Filter-EMI-EMC.jpg
    EMI-Filter-Power-Line-Filter-Mains-Filter-EMI-EMC.jpg
    9.8 KB · Views: 1,673
  • HPR-Installation-Example-300x279.jpg
    HPR-Installation-Example-300x279.jpg
    8.2 KB · Views: 1,407
Engineering news on Phys.org
  • #2
If you want 60dB up to 5GHz, you need to pay for the expensive filter. And what are you using for your Faraday Cage? How big is it? What else are you trying to feed through the walls?
 
  • #3
The cage's dimensions will be ~2m x 1m x 1m (not a rectangular but an irregular one). The cage is for personal use.
I've been reading datasheets e.g. https://eu.mouser.com/datasheet/2/382/resin-sealed-bolt-in-filters-372229.pdf
and these show that high frequency insertion loss is independent of capacitance, so in general the bigger the capacitance, the greater the low frequency attenuation. However, small capacitance (1000pf for example) filters are much cheaper.
I would also like to lead an usb and a cat-5 cable through the metal walls, these aren't that important.

So the small capacitance feedthrough filter would work as a high frequency filter (>10mhz) while the regular one would go from 100khz-100mhz or so.
 
  • #4
Filtering at 100MHz is very different from filtering at 5GHz. Are you familiar with why this is true? :smile:
 
  • #5
No I don't think I'm familiar with it (I'm googling it). EMI shielding papers tell that any power & data cable that goes thru the shielding must be filtered. All I know that there are parasitic properties of inductors, capacitors that makes these filters problematic to work at microwaves. So for microwaves, very small values make a big difference. Thanks for the eye opening but I don't get it. Why I couldn't insert a feedthrough filter to filter microwaves, then a regular power line filter to filter radio frequencies?
 
  • #6
flowwolf said:
All I know that there are parasitic properties of inductors, capacitors that makes these filters problematic to work at microwaves.
Correct.
flowwolf said:
Why I couldn't insert a feedthrough filter to filter microwaves, then a regular power line filter to filter radio frequencies?
You may be able to series-connect filters for different parts of the stop-band that you want, but there are lots of other issues to make the shielding work. Seams in your Faraday Cage welding/soldering can be a problem, for example. And if you are not careful about your filtering, you can impair the communication between the inside and outside of the cage.

Can you say more about your application? Have you looked at commercially-available shielding solutions? I use shielded enclosures in my EE R&D work weekly, and even though I've built some of them on my own, the best solution is almost always to buy one from the available vendors (who also offer options for filtered feedthroughs)...

http://www.vicommtech.com/data/imag..._equipments/302016/manual/shielding_box_6.jpg

shielding_box_6.jpg
 

Attachments

  • shielding_box_6.jpg
    shielding_box_6.jpg
    12.5 KB · Views: 1,018
  • Like
Likes dlgoff and flowwolf
  • #7
Thanks for the answers berkeman, I'll try the series connection then, I have no idea how much it can impair the communication.
I know about the other issues (seams, gaps and honeycombs for air ventilation). I've been reading EMI shielding papers like this http://www.dtic.mil/dtic/tr/fulltext/u2/a257412.pdf for a while.
I've been checking commercial products as well, but they're out of budget in general.

I could say more about the application only in private.
 
  • #8
flowwolf said:
I know about the other issues (seams, gaps and honeycombs for air ventilation).
What are you planning on using for door seams and any other access port seams? Something like strips of EMI spring fingers or gaskets?

https://www.tech-etch.com/shield/images-shield/finger-stock-becu-gasket-profiles.jpg

finger-stock-becu-gasket-profiles.jpg
 

Attachments

  • finger-stock-becu-gasket-profiles.jpg
    finger-stock-becu-gasket-profiles.jpg
    32.7 KB · Views: 978
  • #10
For access port seams, it depends on what it is. Power line filters can be screwed to the enclosure, but it'll likely be something custom in which case I will screw it as densely as possible.
I'm using screws with washers to join the sheets (galvanized metal sheets).

For doors, I'm not sure. There was a paper that showed EMI spring fingers are preferable over gaskets.
This is for windows but I'm planning to implement something like this (by using a thicker metal door):

990810r.gif


For usb and cat-5 ethernet cables the situation is more problematic. The cat-5 cable isn't really important, but
it seems that only usb data line filters with optical cable (thru a waveguide) should be employed:
7895-DVI-USB-Data-line-filter-main-image.png


Some webpages suggest that filtering the Vcc and GND wires improves the situation but the data lines simply cannot be filtered.
(I'm thinking on this one too.)

For the power line filter, cheaper feedthrough filters like this one https://eu.mouser.com/datasheet/2/415/4106-000-32394.pdf
go down to 100mhz. On the contrary, filters that go down to 1mhz are ~50 times more expensive.
There are many manufacturers, but it also depends on what I can buy here..
 

Attachments

  • 990810r.gif
    990810r.gif
    38 KB · Views: 1,106
  • 7895-DVI-USB-Data-line-filter-main-image.png
    7895-DVI-USB-Data-line-filter-main-image.png
    145.5 KB · Views: 1,001
  • #11
It may sound funny, meat grinder plates seem to be a cheap alternative for ventilation instead of honeycomb panels.
A grinder plate with 3mm holes and with a thickness of 15mm has a cutoff frequency of ~60ghz and works well up to 20ghz, attenuation is "160dB - 20*lg(n)" which is still >= 80dB.

51YOMqDmAUL._SY463_.jpg
 

Attachments

  • 51YOMqDmAUL._SY463_.jpg
    51YOMqDmAUL._SY463_.jpg
    22.5 KB · Views: 895
  • Like
Likes sophiecentaur, Tom.G and dlgoff
  • #12
People usually have unreasonable expectations of their home made Faraday enclosures. RF is insidious stuff and will leak in all over the place. If you don’t have deep pockets then your performance is bound not to be stunning.
The first thing to do is to specify what level of interference you actually need - seriously. You may not in fact need the ultimate screened box.
You may actually need no in/out leads at all if you use batteries and measure and record everything in a well sealed box. That would be one extreme design philosophy.
You May get away with just power cables in and optical links for signals in and out.
Anything less will be a compromise but that may not matter.
 
  • Like
Likes Paul Colby, flowwolf, CWatters and 1 other person
  • #13
sophiecentaur said:
RF is insidious stuff and will leak in all over the place.

+1

I've done time in a Faraday cage hundreds of meters down a salt mine. Some of the most frustrating days of my career. You can spend a fortune and follow all the good practice and theory as much as you like and still struggle to keep it in/out. Then someone comes along with a bit of kit thrown together in 10 mins that should leak all over the place and yet it doesn't. Then your boss wants to know why you need another £10K and two more weeks while they don't.
 
  • #14
I don’t believe your visitor’s claims unless the performance of their kit relates to some luckily chosen frequencies.
Bosses are like that, anyway!
 
  • #15
Educate me. I'm Faraday cage ignorant. Aren't all these problems frequency dependent? Is it correct to call the museum image below a Faraday Cage?

hqdefault.jpg
 

Attachments

  • hqdefault.jpg
    hqdefault.jpg
    10.4 KB · Views: 993
  • Like
Likes flowwolf
  • #16
anorlunda said:
Educate me. I'm Faraday cage ignorant. Aren't all these problems frequency dependent? Is it correct to call the museum image below a Faraday Cage?
If it does the job at the required frequency then it is a FC. But you couldn’t sell it for wideband general lab use.
 
  • Like
Likes anorlunda
  • #18
sophiecentaur said:
RF is insidious stuff and will leak in all over the place

Sounds reasonable, most documents emphasize this all the time. And it's something I've experienced myself (and I take it seriously).
Someone said in a yt video if joints allow water to flow in, that will allow EM waves to leak in.

Technically, it's an enclosed EMI shielding box not just a Faraday cage, I mean it's solid (a cage that is made of wires is still a FC AFAIU) but I don't know too much about polarization.

CWatters said:
I've done time in a Faraday cage hundreds of meters down a salt mine.
That must have really been frustrating, but what could have been the problem in following theory?I don't exactly understand the propagation of EM waves btw. In theory, a waveguide can keep out EM waves below a specific frequency. EM waves are not longitudinal waves but how is it possible that particles can't go thru such a waveguide (let's assume a circular waveguide)?
 
Last edited:
  • #19
flowwolf said:
Someone said in a yt video if joints allow water to flow in, that will allow EM waves to leak in.
Not necessarily. A microwave oven is not waterproof, right?

And the opposite statement is also not necessarily true -- it's easy to make metal-to-metal waterproof seams that are not RF tight...
flowwolf said:
I don't exactly understand the propagation of EM waves btw
This may help: http://www.rfcafe.com/references/electrical/waveguide.htm

:smile:
 
Last edited:
  • Like
Likes dlgoff
  • #21
berkeman said:
Not necessarily. A microwave oven is not waterproof, right?

OK, but that is just for a specific frequency, cell phones still work in an oven because the frequency is not the same.
The youtube video was about an EMP-proof shielding box (where the frequency is unknown).

There are optical usb cables for 100$. But what if the equipment still transmits adverse frequencies thru it?

The following product is claimed to remove all the interference by reconstructing the USB signal, so no adverse signal will be found on the output:

https://www.djmelectronics.com/usb-emi-rfi-filter.html
 
  • #22
anorlunda said:
Is it correct to call the museum image below a Faraday Cage?
It works well enough to protect the occupant from what is basically a DC risk. Inside it will be full of relatively low level RF shash at all frequencies but that is irrelevant in this application. The Faraday suits that workers on live HV cables use need to be of closer weave than that cage because there is significant RF power when the flash-over occurs. See after 1m in this video (it goes on a bit and is very dramatic).
Edit: It could also be to do with the fact that the suit is very close to the body. I would bet that a 'bird cage' could be more wide spaced.
 
  • Like
Likes flowwolf
  • #23
Interesting video.

btw. When mounting a waveguide or a honeycomb vent onto the shielding enclosure, does it matter which side I mount it?
 
  • #24
flowwolf said:
btw. When mounting a waveguide or a honeycomb vent onto the shielding enclosure, does it matter which side I mount it?
The wall of a shielded room will need to be a double wall. The walls are not electrically bonded except at the door, and the power filter feed-through. A honeycomb vent will be mounted inside the wall cavity, but attached to only the inner wall. The outer wall will also need a honeycomb or mesh that does not contact the inner honeycomb.

A guide will have some length and be bonded to one wall at each end. The power filter box should be built as a long box with partitions and removable covers. Half of those covers will be accessible from outside the outer wall, the other end from inside the inner wall. Each box section will have another filter stage, separated by a partition with feed-through capacitors. The filters should not all be simple LC or RF choke filters, there must also be resistive snubbers, AC capacitor coupled, to absorb the reflected noise, or you will end up with all sorts of resonances.
 
  • Like
Likes flowwolf
  • #25
I built my first Faraday cage while I was still at school. That taught me the futility of trying to reach perfection. I can honestly say I have not needed a screened room since then.

There is a screened room in the Engineering Department at the local University that I can use if I ever need one. The availability of a screened room makes it possible to realize that a screened room is very rarely actually needed. When it is needed, the required specifications will be known, and will certainly not require attenuation of the band from DC to Daylight to below the thermal noise floor in your analyser.

In my legitimate surveillance activities I monitored many targets who needed a screened room, but did not know it. I am beginning to think that only criminals and government cipher machines actually need screened rooms.
 
  • Like
Likes berkeman
  • #26
Baluncore said:
I am beginning to think that only criminals and government cipher machines actually need screened rooms.
We rent time at a local lab's shielded anechoic chamber for our FCC-type radiated EMI testing and certification. It is used both for radiated emission compliance, as well as EN 61000-4-3 RF Radiated Immunity testing.
 
  • #27
Baluncore said:
The wall of a shielded room will need to be a double wall.

Is that because of "the futility of trying to reach perfection"? Again, what is the problem by following theory? Papers show that double walls don't double the shielding effectiveness.

Baluncore said:
When it is needed, the required specifications will be known, and will certainly not require attenuation of the band from DC to Daylight

Good point actually, and you can see that I told berkeman that I can answer it in private only.

One needs such a shielding enclosure when the adverse frequency isn't really known in advance.
Criminals require a shielding enclosure if (and only if) they use some kind of electromagnetic weapon against their target (but it is widely accepted that is a conspiracy theory isn't it?). So the priority would be to see who have such weapons and then who builds shielding.
 
  • #28
I accept that there are situations where EMI testing and certification actually specify that a screened anechoic environment be used. It is only the certification process that requires the controlled environment. Later, the instrument and the certificate work quite well outside.

It is commercially very sensible to only rent a certified instrumented screened room when required by certification standards. The premium instruments used in screened rooms are expensive to buy and maintain, partly because they must be repeatedly certified as meeting specifications, which takes time and requires several other certified instruments to complete the task.
 
  • Like
Likes berkeman
  • #29
flowwolf said:
Papers show that double walls don't double the shielding effectiveness.
Can you reference those papers.
Two separated walls, having a separate medium in the middle, with the inherrent two mismatched impedance interfaces, will usually outperform one wall having twice the thickness.

flowwolf said:
Criminals require a shielding enclosure if (and only if) they use some kind of electromagnetic weapon against their target (but it is widely accepted that is a conspiracy theory isn't it?). So the priority would be to see who have such weapons and then who builds shielding.
No. The use of RF equipment may radiate energy and information that can be easily followed by law enforcement. Wrong-doers get caught because they are naïve and so do not have sufficient shielding. They foolishly transmit using codes or scramblers that draw attention to themselves.
 
  • #30
Baluncore said:
Can you reference those papers.

I remember reading that in more than one document, those showed that when doing precisely, the 2nd layer improves SE by some (30 or so) dB, but it will be less than double.

"LAMINATED AND NESTED SHIELDS Occasionally, it is assumed that the SE of enclosures can be doubled by doubling the walls or by adding shields contained within the enclosure, Fig.9. It will be demonstrated that the resulting SE is less than double."
https://www.ipen.br/biblioteca/cd/ieee/1999/Proceed/00350.pdf

Also here: http://www.jpier.org/PIERL/pierl01/06.07110706.pdf

Baluncore said:
No. The use of RF equipment may radiate energy and information that can be easily followed by law enforcement.
It depends on what type of criminals we are talking about. Some of them keep up with times. I agree that a full-blown energy weapon would burn things like metal objects, or radar frequencies would bounce off metal but there are other ways to commit crimes undetected (implants etc..)

btw. Wouldn't a two-layered wall cause resonances and worsen SE at specific frequencies?
 
  • #31
flowwolf said:
I remember reading that in more than one document, those showed that when doing precisely, the 2nd layer improves SE by some (30 or so) dB, but it will be less than double.
Double the energy is a change of 10*Log(2) = 3dB. I think you were expecting doubling the dB which is squaring the SE?

According to your ref; Kistenmacher and Schwab, 1996. IEEE. Page 351.
My statement is shown to be true for conductive non-magnetic materials, for Cu see fig. 12.
The effectiveness increases with frequency. Note the plot there assumes the same total mass of material.

My statement is false for magnetic resistive screen materials. See fig 13. Which is what you expect from ferrite slabs that are designed to allow EM propagation between the magnetic grains.

Your reference, the Bahadorzadeh, Moghaddasi and Attari, 2008, paper, appears to be discussing cascaded apertures in the wall of a cavity. Figure 2 shows two walls give about a 12dB advantage over one. That is significantly better than a factor of two and supports my suggestion that the same mass distributed amongst several walls is better than one thick wall. The paper goes on to show increased wall separation increases screening.

flowwolf said:
btw. Wouldn't a two-layered wall cause resonances and worsen SE at specific frequencies?
Resonant cavities need coupling to be part of the circuit. A 100mm thick timber stud wall supporting and separating two conductive sheets will not support a significant resonance. Galvanised steel is cheaper than copper and can be easily soldered along the seams.
 
  • Like
Likes flowwolf
  • #32
Baluncore said:
Double the energy is a change of 10*Log(2) = 3dB. I think you were expecting doubling the dB which is squaring the SE?

Yes, for power 10*Log(2) = 3dB is the double.

Let's say that an enclosure with a single wall has an SE of 60dB. It seems logic that a nested enclosure (a double walled one) have an SE of 120dB, which is the doubling of dB (that is what I was expecting yes).

We can agree that double walls (the nested ones) improve SE anyway. What I wanted to ask that when done correctly, do I really need nested enclosures?

Baluncore said:
Resonant cavities need coupling to be part of the circuit. A 100mm thick timber stud wall supporting and separating two conductive sheets will not support a significant resonance.

What about something smaller like 20mm? Are there any rules on this, or do I have to do simulations to know the resonances? I didn't know that resonant cavities need coupling (you mean electrical connection?).

Baluncore said:
Galvanised steel is cheaper than copper and can be easily soldered along the seams.

+1 Yes, galvanised steel seems to be solderable.
 
  • #33
flowwolf said:
It seems logic that a nested enclosure (a double walled one) have an SE of 120dB, which is the doubling of dB (that is what I was expecting yes).
When you cascade two attenuators you square the attenuation, unfortunately that is not directly applicable to double walls. Is it better to use two conductive walls of half the thickness or one thick wall? Probably more walls, but it involves understanding SE as a function of wall thickness, which is frequency dependent.
A conductive wall has external circulating RF currents due to the external world on the outer surface, with internal circulating RF currents due to your activities on the inner surface. Those currents induced in one face only diffuse very slowly (100 m/s) through the material due to back emf, hence skin effect. You need to keep those surface currents separate which in reality comes down to the quality of the aperture seals, which provide the only short path around the wall.

That means your door and vent seals must be very reliable, because any break in conductivity concentrates surface currents and allows them to wrap around the edge of the conductive screening sheet. Keep apertures away from external edges where circulating currents are naturally higher, so the door should be in a face. It must have metallic wipers that eliminate the conductivity gap when closed. Any imperfect seal becomes a slot antenna or an aperture coupling inside to outside.

With double walls the door will also have two insulated conductive surfaces, each with it's own wipers. The only ground connection between the double walls will be the conduit used for the power supply which is the middle part of the cascaded filter box.

There is a construction requirement. A timber stud wall, floor and roof will all be self supporting. That suggests independent inner and outer conductive walls can be attached. Fold the material around the edges so your conducting joints are on less stressed flat faces. Avoid TIG or MIG welding, use plumbers solder along all the seams.

flowwolf said:
What about something smaller like 20mm?
You are talking waveguides with cutoff frequencies. But you must inject the energy and then couple it out again so avoid apertures.
 
  • #34
ETS-LINDGREN. Double Electrically Isolated RF Enclosures for Industrial, Communication, and Research and Development Applications.
http://www.ferret.com.au/ODIN/PDF/WebsiteCustomers/78.pdf
 
  • Like
Likes berkeman and flowwolf
  • #35
This is a great link, thanks. I like the way the panels joined, you're right, this method is excellent.

flowwolf said:
What about something smaller like 20mm?

I meant, what if the space is smaller than 100mm between the walls. 20mm would be a bit small but something like 40-50mm? Is there any rule?
 

Suggested for: Power line EMI filter + feedthrough filter for a Faraday cage

Replies
8
Views
1K
Replies
39
Views
2K
Replies
14
Views
784
Replies
3
Views
2K
Replies
2
Views
737
Replies
15
Views
4K
Replies
15
Views
995
Back
Top