RF shielding issues for enclosures with holes.

[creativeoz]

This discussion covers a subject that I have sought to resolve for a couple of years. A metal box, 2" x 2" x 6" long. Two round 3/8" holes in 1 side 2" appart. (mounting holes) Application is RF. What frequencies up to 40ghz are likely to enter via these holes. My experience has shown me from VHF to 24Ghz+.
One "engineer" told me that nothing below 30Ghz would pass through these holes, based on the wavelength of aprox 3/8" for 30Ghz.
I don't accept his theory.
I know there will be varying losses over many frequencies that will pass, but I would like feeedback on the potential upper and lower frequency limits that would pass through. 2 x 3/8" holes in 0.065" panel/enclosure wall thickness. J.

 

[VEBill]

I'm not a subject matter expert in this exact subject, but I do know enough to know that his claim of "nothing" is an over-simplification.

Here's a link to a paper that provides more details and might help with more exact modeling:

http://www.umc.edu.dz/vf/images/emc/PDFs/Papers/153_112.pdf

I searched using shielding holes cutoff frequency and there are several similar hits.

 

[DanEE]

Wish the IBM paper focused a bit more on the subject of "waveguide beyond cutoff" (frequencies) which is what it is all about..

This paper addresses the subject on more detail..

http://www.formfactors.org/developer\specs\wg_overview_098.pdf

Also google the subject "waveguide beyond cutoff" in quotes.. will get lots of reading material on the subject....

I've built a couple of very inexpensive, but highly effective RF signal attenuators based on the characteristics of circular waveguide operated below cutoff frequency.

 

[creativeoz]

Thanks very very much guys & gals. I now know I am not alone!

The IBM paper gave me the specific detail I was looking for. In particular it covered the 9mm hole size (amazing), square, in their test setup. It gave me the supportive graphical data I needed to confirm my own experience of many years past.

I have found, over the years, a couple of "engineers" in positions of, acting for the prosecution, make broad (false) statements that they are rarely called upon to support with documentary evidence.

Good to see this forum of help\support, is out there in the real world.
Sincere thanks J.

 

[VEBill]

I think that the widespread misunderstanding comes from the word "cutoff". The word cutoff over-sells the shielding attenuation of such holes by implying, well... 'cutoff' (i.e. 100% or infinity dB).

It's semantics. If you're going in for the argument, get them to convert their adjectives into numbers (dB) first, then correct them if required. Have fun.

 

[creativeoz]

"Cut off" yes that was one of my issues. It's been implied to an investigation group, by an "engineer" that cut off is 100% signal loss. But as I touched on earlier. I have seen in the lab and field (as many others will have seen) that, if the RF source is strong enough.

 

[DanEE]

Agree that anyone that says cutoff is 100% loss as a general and final statement doesn't understand. There is a lot more to it. It can become effectively 100% with proper dimensions including sufficient length (the hole needs to become a duct or tube to yield sufficient waveguide cut-off attenuation); simple holes in sheet metal in many cases is not a sufficiently long enough waveguide.

"Cutoff" is a term from microwave waveguide transmission line meaning for a given geometry waveguide, RF applied below the "cutoff frequency" does not propagate in one of the TE, TM or TEM electromagnetic field modes that waveguides use to carry RF with minimum loss.

Once the waveguide is in cutoff mode, attenuation can be at a high rate over a relatively short distance. Obviously a function of frequency and geometry, but for example, an inch or so in cutoff mode at VHF/UHF frequencies can yield very high rates of attenuation. A hole in sheet metal especially at high frequency is simply too short a waveguide (it's length being the thickness of the sheet metal). The examples and formulas in the INTEL document linked above show very well the effects length in the dimensions where you begin to get very high levels of attenuation.

A good visual example of this is the cylindrical waveguide piston attenuator (Figure 9) on page 426 in the attached link.

http://books.google.com/books?id=Wr...="waveguide beyond cutoff attenuator"&f=false

For example my waveguide piston attenuators were used to drop 1/2 watt of RF down to receiver threshold levels around -132dbm with the cut-off waveguide length of about 1". The cylindrical waveguide diameter I used yielded a cut off frequency around 3Ghz and it was used to attenuate signals in the 100 to 500mhz range.

 

[VEBill]

It's still semantics (word games).

"100%" (for example: 99.6% that can be fairly rounded to "100%") is only about -24dB. So anything better than -24dB can be accurately described as "100%".

Another example, 99.99% is only -40dB. That's "100.0%" even taking into account the precise use of significant figures.

Even using the term "100%" in the field of radio and communications deserves a good slap upside the head.

I still recommend having your opponents pick a number (in dB) before you set them up for the kill.

 

[creativeoz]

Totally in agreement.

 

[zappedagain]

Is this box welded shut? If not you have slot antennas to deal with between the bolts that mount the cover. That may be much leakier than your 3/8 inch holes.

 

[Higgler]

A typical mechanical item to test dielectrics is a box with two holes (irises), called a resonator. It lets alot of RF inside the box. Actually it lets all of it in even though the irises are way below cutoff.
So a pair of holes can accept much more energy than you'd ever imagine, but that depends on what is in the box.
Best thing inside the box is salty water in a bag. Or another type of rf load material.
Holes in a box with cables coming out of them are coax lines, or essentially antennas, keeping their induced rf out needs their outerconductors connected to the box, or at least make the box with metal tubes sticking out of it where the cables come in. Add magram heatshrink to the cables (new product,Cuming Microwave, good invention).

Cutoff waveguide loss curves are a ___/----- low pass loss pattern but the very low end of the bandflattens out with. that's useful with antennas if you want to get some gain below band on horn antenna, or feed antennas.