Basics of EM/RF Shielding - Thin-walled enclosures

[CurtainCall]

Hello,

TL;DR: If someone could please confirm that this is right, (more details and thought process below) attenuation for thin-walled enclosures can be approximated by:

Attenuation [dB] = 20 log ( (?/2a)*(r/a) ) - 20 log (n[sup]1/2[/sup]);
Where:
? = wavelength
a = largest cross-sectional dimention of the aperture
r = distance of frequency source to aperture
n = number of similar apertures with edge-to-edge distances of less than ?/2
d = depth of the aperture (material thickness)

NB: s < ?/2 > a > d to be valid

Thank you,

- Kirk


My boss has asked me to put together a design reference for EM/RF sheilding for internal and external support. Unfortuately being a mechanical/aerospace engineer this is a little out of my league. If anyone would mind going over and critiquing what I have learned so far I would appriciate it.

EM/RF sheilding is used to isolate the component to prohibit interference from the environment to the component and vise-versa.

To isolate the component a conductive structure is required to "balence" internal and external charges.

A perfectly sealed Faraday cage would be the ideal solution, but is not always practical or desired.

The two main sources of leaks/losses are seams and apertures.

NB: Due to the similarity between apertures as "holes" and antenna apertures, for simplicity I will refer to "holes" from now on.

Seams are caused by the manufacturing process and can be treated with gaskets or specialized fabrication techniques such as double flanging to maximize contact area between the conductive materials.

Screws and other fasteners can be used to create discontinuties in seams to reduce losses.

Untreated seams can be considered as narrow, lengthy holes.

Holes in enclosures are required for cooling/ventalation, external access (power/data transfer), etc.

The exact geometrical shape of the hole is irrelevant to the losses occured (to a point).

Though different shapes can have unrelated benificial effects, ie honeycomb structures for airflow.

The largest of the cross-sectional (CS) dimensions of a hole is important.

If the holes have significant depth, ie the material is thick, or are coupled to tubes; the holes can be assumed to act as waveguides.

A waveguide will attenuate the signal of any signal below a certain cutt-off frequency.

The cutt-off frequency is approximately the frequency at which the largest CS dimension is equal to a half wavelength.

For waveguide-like holes a simple approximation of signal attenuation is:

Attenuation = C * (d/a) * sqrt( (1-(f/fc)[sup]2[/sup]) ) [ignore][dB][/ignore];
Where:
C = waveguide coefficient; Generally 30, rectangular hole 27, circular hole 32
d = depth of hole (material thickness or length of coupled tube)
a = largest CS dimension
f = frequency of the field
fc = cutt-off frequency

NB: this equation is only valid for waveguide like holes, ie d>>a

For thin-walled enclosures:

Wavelengths smaller than twice the largest CS dimension (2a) will not be attenuated.

Wavelengths equal to 2a will have 0dB shielding, inferring the cutt-off frequency to be:

fc = C/2a;
Where:
C = propogation velocity of an EM wave (ie speed of light)

For Wavelengths greater than 2a attenuation can be approximated via:

Attenuation [ignore][dB][/ignore] = 20 log(?/2d);
Where:
? = wavelength

NB: This is only valid for ?/2 > a > d
NB: Maximum attenuation is the attenuation of a solid barrier with no holes, aka faraday cage construction
NB: To neglect the effects of signal noice within the enclosure the component must be atleast distance, d, away from the enclousure

If you cannot place the component far enough away, then you can approximate the effect of signal noise:
fc = (C/2a)*(r/a);
Where:
r = distance to the hole

and

Attenuation [ignore][dB][/ignore] = 20 log(fc/f) = 20 log( (?/2a)*(r/a) )

If there are multiple similar holes and spaced closely together, ie s < ?/2, where s is the spacing between holes. The attenuation can be approximated by:

Attenuation [ignore][dB][/ignore] = 20 log(?/2) - 20 log(n[sup]1/2[/sup]);
Where:
n = number of holes
s < ?/2 > a > d
s = edge to edge hole spacing

Now there is a whole lot more to go through with: different arrays of holes, absorbtion loss of waveguides, thick and thin conductive coatings, etc. However, I just wanted to make sure I was on the right track, before I start learning something inconsistant.

Thanks for taking the time to go through this rather lengthy brain dump.

Attached you will find a brief PDF I was using as one reference. If anyone can recommend something better I'd appreciate it.

- Kirk

 

[biff44]

Well, that is a pretty valiant effort for someone who is not a microwave engineer. Sadly, for you, there is no simple equation that predicts the leakage. It is more complicated than you could begin to imagine, and probably would need the exact dimensions and the use of a $85K emag analysis program to solve accurately.

You are right, if it is a "solid box" with a hole in it, and the hole is smaller diameter than the waveguide cuttoff frequency, AND the hole has a little depth to it, THAT can be predicted. However, if you have a grid of such holes, forming a leaky array, that is much harder to predict the performance of.

And how would you even define a "poorly formed" seam? Even a very small crack of the right length can radiate energy very well. You would have to know the exact location of wall-to-wall contact points to know the precise geometry.

You will either need to analyze a specific structure with an EMAG field solver program, or make up some test samples and test them with calibrated antennas.

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[VEBill]

Also, any RF that leaks into, or orginates within, the chamber will (unless the chamber is lined with absorbing cones or similar) bounce around inside creating multipath, standing waves, and - potentially - endless confusion.

Are you planning to buy (or build) an E3 chamber? If you're thinking of building one from scratch, then start with the door (because that's the tricky bit).

 

[CurtainCall]

Thanks for the quick replies biff44 & VE1BLL, both of you are very active on the forums and I feel very confident with your answers.

I know this was far out of my league, but we're a small firm and my boss probably just wants something/anything on paper so that we have a place to start.

Our products are usually contained in simple, small (up to 4x6x12") anodized Al enclosures, usually with internal PCB guides and a sliding top plate for access (oh god the EM geometrical nightmare). We're not shooting for anything resembling sealed, we just want to know what frequencies we can expect to "leak" through so we can advise our customers of potential interference issues while we try to improve.

Unfortunately you are both right. With the effects of standing waves, multipath, and related harmonics, I'd have to plug in our model to a EM field solver to get any real answer but I don't think our company has the resources for something like that right now. A E3 chamber might be a good middle ground (I assume that's just a perfect faraday cage to use with calibrated antenna).

Looks like all those aerodynamic courses are really paying off

Thanks for your help, I'll try to come up with something general, but I'll mention that we should build a chamber to test the products that we have and recommend for the future that we invest in something like Solidworks' ElectroMagneticWorks (or another program you recommend) if we want something more detailed, verifiable, and otherwise actually engineered.

- Kirk

 

[CurtainCall]

Yep, nevermind - did not realize what an E3 chamber (reverberation) was - we definately don't have the resources (read someone who knows what their doing) for that.

But I can probably build a decent faraday cage and as was suggested probe the working enclosure to get decent results during component operation.

- Kirk

 

[VEBill]

Sorry, I misunderstood your first post (small chassis versus walk-in chamber).

Before worrying about radiated emissions, have you already dealt with conducted emissions? Typically there is a lot of work to do with cleaning up the leakage from the I/O before you'd reach the level where the radiated emissions from the seams and gaps of a metal chassis would be worth mentioning.

 

[CurtainCall]

Hello VE1BLL,

I should have been more descriptive of our problem, most of our products are focused around wireless transceivers/receivers/transmitters with varing applications, almost all of them are battery powered and if I understand right, conducted emissions are coupled to the power cord, since our products have no central power distribution I would assume I could neglect conducted emissions.

But a very good point to keep in mind regardless, thank you

- Kirk

 

[Higgler]

Take a look at waveguide cutoff

http://en.wikipedia.org/wiki/Cutoff_frequency

That's usually the first line of estimating how much leaks into a hole or more likely between two screws you've used to connect a cover to a frame.

Just knowing that 1/2 wavelength is the cutoff in waveguide is a good starting point.

I've done alot of isolating antennas, which is similar to your leakage problem. The solution all depends on the level of protection needed. Many screws is the simplest, to gaskets and absorber and metallic paint as the optimum.

One note to scare you. We use resonators to measure material. It's a perfect box 0.459" cube, with two tiny 0.020" holes at opposing ends. Energy can flow into one hole and out the other hole with virtually no loss when the box dimension is half wavelength. Hence at higher frequencies, alot gets into a box between two screws holding a cover to a frame. Even 0.001 gap lets alot of energy into a box between screws.

 

[CurtainCall]

Ah I see, thank you Higgler.

I knew high frequency's would be a problem, but I didn't realize a thousanth of an inch gap would be so detrimental. We usually use a combination of gaskets and screws, though we don't use EMI gaskets yet...(shakes head)...

It seems to be the general concensus that reducing the apperture dimensions (while avoiding array losses) as well as controlling for assembly based leaks is the best way to start. Followed up with a simple faraday-cage/probe analysis to examine what frequencies we still need to address.

As an aside, even though I sound pretty ignorant, I can assure you all of our products meet emission regulations, but that happens only after outsourced testing. We're trying to reduce costs and add more value to the customer by engineering ahead of time.

Everyone who has commented has really helped, not only me, but our customers.

Thank you

- Kirk

 

[CurtainCall]

EDIT: going over my original post, I found a couple of typo’s:
“For Wavelengths greater than 2a attenuation can be approximated via:

Attenuation [dB] = 20 log(?/2d);
Where:
? = wavelength”

Should read:

Attenuation [dB] = 20 log(?/2a);

And
“If there are multiple similar holes and spaced closely together, ie s < ?/2, where s is the spacing between holes. The attenuation can be approximated by:

Attenuation [dB] = 20 log(?/2) - 20 log(n[sup]1/2[/sup]);
Where:
n = number of holes
s < ?/2 > a > d
s = edge to edge hole spacing”

Should read:

Attenuation [dB] = 20 log(?/2a) - 20 log(n[sup]1/2[/sup]);

I apologize for any confusion.
- Kirk

 

[Higgler]

If your box inner components contain metal only, it's more difficult to RF seal the box from outside RF energy. If you have RF absorber inside, your metal enclosure actually seals better.

Might sound odd, but it's true.

If you ever want to see how much a cable leaks to another cable, locating them next to each other can show -110 dB coupling leakage. Placing the cables inside a metal box can give you -55 dB. I did this measurement to see how much two 50 ohm loads at the end of a pair of cables couple to each other. With the energy that leaked radially contained in the metal box, coupling between the 50 ohm loads jumped up.

Hence, I'd suggest some lossy material inside each box to help improve your effective shielding. All RF amplifiers and high isolation RF swithes add absorber inside their boxes for similar purposes.

 

[CurtainCall]

Very Interesting Higgler,

I'll have to look into that, I don't know if we currently use RF absorbers in all our products, the one's I've been dealing with don't but I'll look and see if maybe some of the more top-shelf products use something like that.

I'm guessing that the absorption by the material inside reduces the interference between the internal and external signals (reflections etc), which in turn reduces noise?

- Kirk

 

[Higgler]

Yes, it will absorb the signal that leaks in, but more importantly, dampen any half wave resonance inside your box.

When a multiple of a half wave length is set up inside your box, it's like opening up your entire box at the specific 1/2 wavelength or N*half wave. A huge amount of RF can get in during a half wave resonance, even with a tiny hole in your cover at higher frequencies.

To check the results of whether absorber can help, without the time and cost of buying just the right absorber. Just put a wet sponge/paper inside and remeasure your RF leakage.
Dense foam absorber is often 5 dB per inch loss at 2.5 gHz and water is 10 dB per inch loss at that frequency. If the wet towel/sponge gives some results, change to salt water to see if you get a larger improvement (much more loss at lower frequencies with salt).
It helps alot to have absorber inside a box, like putting it in a waveguide. Don't use "surface wave absorption" info, thats attenuation in a wave sitting outside a box typically.

You can put these wet towels or sponges in a plastic bag, same RF results and you can protect your equipent. A bag of water would work too.

Foam absorber (Google LS-30) is common at low frequencies (under 2 gHz), Magram is common at higher frequencies (2-18 ghz)

good luck,

 

[CurtainCall]

Good tip about the plastic bag, Higgler! I cocked my head when you first mentioned a wet sponge, and then did a double take with the salt water comment

Thank's I'll try that our after the break when we finally get back to work after all that turkey.

Thanks everyone who's lent me their expertise,

Happy Holidays!

- Kirk