Enclosure Aperture Leakage Calculator

What a shielded box actually leaks through its slots, seams, display cut-outs and vents — the aperture’s own shielding effectiveness, the frequency at which it goes to zero, the penalty for many apertures, and the gain a thick wall buys as a waveguide below cutoff.

Last updated 24 Sep 2026

Leakage through an aperture, seam or vent

Shape, length, depth, count, frequency -> SE

What shape is the opening?The distinction matters enormously and not for the reason most people expect. Leakage is set by the LONGEST dimension, not by the area, so a long thin slot is far worse than a round hole of the same open area. A seam between two panels, a lid gap, a card-guide slot and an unpainted joint are all slots.

Longest dimension (mm)For a slot or seam, its LENGTH — the full unbonded run between two fasteners or two bond points, not the gap width. For a round hole, the diameter. This is the single number the answer depends on.

Slot width (mm)Used only to work out the open area and to compare with an equal-area round hole. IT DOES NOT AFFECT THE LEAKAGE in this model, and that is the point: closing a 1 mm gap to 0.1 mm without shortening it buys you nothing. What shortens a seam is more fasteners, or a conductive gasket that turns one long slot into a continuous bond.

How many apertures?Apertures on the same face, close together compared with a wavelength — a vent pattern, a row of fastener gaps. The penalty is 10 log N, so a hundred holes cost 20 dB, not 40. Widely separated apertures on different faces do not add this way and each has to be considered on its own.

Wall thickness, or the depth of the opening (mm)The depth the opening tunnels through, which for a plain hole is the wall thickness. This is what turns a hole into a short waveguide below cutoff, and the effect is proportional to depth over the longest dimension — so a 10 mm deep honeycomb cell 2 mm across is worth 160 dB and a 0.5 mm hole in foil is worth nothing.

FrequencyAperture leakage gets worse at 20 dB per decade as frequency rises, until the aperture reaches a half wavelength and stops shielding altogether.

Frequency unit

The opening, drawn as a geometry: a panel seen face on with the aperture in it, and the same aperture seen edge on below so the depth is visible. Leakage is set by the LONGEST dimension — the slot's length, the hole's diameter — and the slot's width appears nowhere in the answer. Depth turns the opening into a waveguide below cutoff, which is the only thing that helps a hole. The drawing switches between a slot and a round hole with the shape selector, and the status line turns amber near the opening's half-wave resonance and red at or above it, where the opening stops shielding and becomes a slot antenna.

17.9dBExample

a 100 mm seam 0.2 mm wide in a 1.5 mm wall, one of them, at 200 MHz

Leakage is set by the longest dimension, not the area

SE_aperture = 20 log₁₀ ( λ / 2L ) dB, for L « λ
f_resonance = c / 2L    the frequency at which that goes to zero
N apertures: −10 log₁₀ N
depth: + K · (t / L) · √(1 − (f/f_c)²)    K = 8.686 · 2π / λ_c
round hole: λ_c = πd / 1.8412, so K = 31.98    slot: λ_c = 2L, so K = 27.29

L: the LONGEST dimension of the opening — a slot’s length, a hole’s diameter. The slot’s width does not appear at all, which is why narrowing a seam without shortening it achieves nothing
K: the below-cutoff attenuation constant, derived from the cutoff wavelength rather than quoted. 1.8412 is the first root of the derivative of the Bessel function J1, which sets the TE11 cutoff of a circular guide
fc: the opening’s waveguide cutoff. Below it the depth term applies and grows as the frequency falls; at it the term is zero and the opening propagates
N: the number of openings close together on the same face. Their leakage POWERS add, so the penalty is 10 log N and not 20 — a hundred holes cost 20 dB

Worked example

a 100 mm seam 0.2 mm wide in a 1.5 mm wall, one of them, at 200 MHz
At 200 MHz the wavelength is 1.499 m. The seam's longest dimension is 100 mm, so λ/2L is 7.495 and the aperture term is 20 log of that = 17.50 dB
The 0.2 mm width appears nowhere in that. Closing the gap to 0.02 mm would change nothing at all; the seam is 100 mm long either way
A slot's waveguide cutoff is c/2L = 1.499 GHz, and 200 MHz is 0.1334 of that, so the depth term is 27.29 × (1.5/100) × √(1 − 0.0178) = 0.41 dB — almost nothing, because the wall is thin compared with the seam's length
One aperture, so no N penalty. The total is 17.9 dB
For comparison: the seam's open area is 20 µm², and a round hole of that area would be 5.046 mm across and worth 43.4 dB — 25.9 dB better for exactly the same hole in the metal
And that penalty is frequency independent: it is 20 log(100 mm ÷ 5.046 mm), which is 25.9 dB at every frequency the model is valid at
The seam stops shielding altogether at 1.499 GHz, where 100 mm is half a wavelength. This page will not print a number above that

A slot against a round hole of the same open area

Identical open area, 15.9 decibels apart, at every frequency. This is the whole reason ventilation is drilled rather than louvred and seams are fastened rather than merely closed. A useful corollary: splitting one long slot into several short ones is a large win, because the aperture term improves by 20 log of the length reduction while the N-aperture penalty only costs 10 log N.
Opening	Longest dimension	Open area	SE at 100 MHz
Slot 50 mm × 1 mm	50 mm	50 µm²	29.5 dB
Round hole, same area	7.98 mm	50 µm²	45.5 dB
Difference	—	none	15.9 dB

What depth buys, at frequencies well below cutoff

Below cutoff a hole is an evanescent waveguide and its attenuation is 32.0 dB per unit of depth over diameter for a round hole, 27.3 dB per unit of depth over length for a slot. Both constants are derived from the cutoff wavelength: πd/1.8412 for the circular TE11 mode and 2L for the rectangular TE10 mode. The effect is enormous and it is the only reason a ventilated enclosure can be shielded at all — but it vanishes as the frequency approaches cutoff, and a thin wall gives essentially none of it.
Opening	Depth ÷ dimension	Waveguide attenuation
3 mm hole in 1 mm sheet	0.33	10.7 dB
3 mm hole in a 6 mm boss	2.0	64.0 dB
2 mm honeycomb cell, 10 mm deep	5.0	159.9 dB
100 mm seam in a 1.5 mm wall	0.015	0.41 dB
10 mm slot in a 10 mm wall	1.0	27.3 dB

Half-wave resonance: the frequency at which an opening stops shielding

At c/2L the aperture term is exactly zero: the opening is a half-wave slot and radiates as efficiently as a resonant antenna. The third column is a third of that, which is where the small-aperture approximation starts to flatter. If your frequency of interest is above it, the honest answer is that this model cannot tell you and a measurement or a full-wave solve can.
Longest dimension	f = c / 2L	And the model is only trustworthy below
5 mm	29.98 GHz	9.993 GHz
10 mm	14.99 GHz	4.997 GHz
25 mm	5.996 GHz	1.999 GHz
50 mm	2.998 GHz	999.3 MHz
100 mm	1.499 GHz	499.7 MHz
200 mm	749.5 MHz	249.8 MHz
500 mm	299.8 MHz	99.93 MHz

The truth a wall-thickness calculation cannot tell you

A shielded enclosure is not limited by its walls. Half a millimetre of aluminium is worth over a hundred decibels against a plane wave anywhere in the HF band, and no real box comes close to that, because real boxes have lids, seams, connectors, display windows, ventilation and cables. This page is about the openings.

Length, not area. The leakage through a small opening goes as the ratio of the wavelength to twice its longest dimension. The width does not appear. A 100 mm seam 0.1 mm wide leaks exactly as much as a 100 mm seam 2 mm wide, and the instinct to close gaps rather than shorten them is the most common and most expensive mistake in enclosure design. The fix for a long seam is more fastener points, or a conductive gasket that makes the joint continuous. The fix for a display window is a mesh or a conductive coating, which replaces one long opening with many short ones — and that trade is a win, because splitting a slot improves the aperture term by 20 log of the length reduction while costing only 10 log N for the extra openings.

Depth is the one thing that helps a hole. Below its cutoff frequency a hole is a waveguide that does not propagate, and the field decays exponentially down it. The attenuation is about 32 decibels for every unit of depth over diameter in a round hole, and about 27 for a slot, which is why a honeycomb vent panel with 2 mm cells 10 mm deep gives over 150 dB while a 2 mm hole in foil gives nothing. Both constants come straight from the cutoff wavelength — πd/1.8412 for a circular guide, 2L for a rectangular one — and both collapse as the frequency approaches cutoff.

Where this model stops. These are small-aperture approximations. They assume the opening is much smaller than a wavelength, that the surrounding sheet is a good conductor bonded all round, that nothing is coupled to the opening from inside, and that the enclosure has no cavity resonance nearby. Every one of those assumptions fails somewhere. The page prints the frequency at which the longest dimension reaches half a wavelength — where the model says zero and the reality is an efficient slot antenna — and a frequency a third of that, above which it is already flattering. It will not print numbers above resonance, because extrapolating a small-aperture formula into the microwave region is how a leaky enclosure comes to look sealed on paper.

What is still missing after all that. Coupling. The formulas above describe how well an opening blocks a field that happens to arrive at it; they say nothing about how strongly whatever is inside the box is driving it. A switching node routed along the inside of a seam is a completely different problem from the same seam with a ground plane between it and everything else, and the difference is tens of decibels. Enclosure shielding is as much about internal layout as it is about the metalwork.

And cables beat all of it. An enclosure with perfect seams still has cables leaving it, and a cable shield terminated with a pigtail carries interference straight through the wall as if it were not there. That is the transfer impedance page, and it is usually the first thing to fix. The solid-wall page gives the upper bound this page erodes, and the RE102 margin page is where the requirement lives.

Frequently asked questions

Does making a seam narrower help?

No — not in this model and not much in practice. Leakage is set by the longest dimension of the opening, which for a seam is its length between bond points. Closing a 1 mm gap to 0.1 mm leaves a 100 mm slot either way. What helps is shortening it: more fasteners, a conductive gasket, a continuous weld. Halving the length between bonds buys 6 dB.

Why is a round hole so much better than a slot of the same area?

Because the answer depends on the longest dimension and not on the area. A 50 × 1 mm slot and a 8.0 mm round hole pass exactly the same open area, and the slot is 16 dB worse at every frequency — 20 log of the ratio of the two dimensions. This is why ventilation panels are drilled with many small holes rather than cut as a few long louvres.

How much does wall thickness help?

Through the waveguide-below-cutoff effect, and only in proportion to depth over the longest dimension. A 3 mm hole in 1 mm sheet gets about 11 dB from depth; the same hole through a 6 mm boss gets about 64 dB. For a long seam the ratio is tiny and the effect is negligible. And it all disappears as the frequency approaches the opening’s cutoff.

What happens above the half-wave frequency?

The opening stops being an aperture and becomes an antenna. At λ = 2L the model gives exactly zero, and above it the real structure has resonances that a small-aperture approximation cannot describe at all. This page clamps the aperture term at zero there and says so, rather than printing a negative number that would look like a calculation.

Do a hundred small holes really only cost 20 dB?

10 log N, yes — provided they are on the same face and close together compared with a wavelength, so their leakage powers add. That is why a perforated panel works: a hundred 3 mm holes give up 20 dB against one 3 mm hole, but each of them is worth far more than one large opening of the same total area. Holes spread over different faces of the enclosure do not combine like this and should be assessed separately.

Can I add this to the wall’s shielding effectiveness?

No. They are not in series in that sense — the enclosure leaks by whichever route is worse, so the aperture number effectively replaces the wall number wherever it is lower. In almost every real enclosure the aperture is lower by a wide margin, which is the point of this page.

Related calculators

References

Ott HW. Electromagnetic Compatibility Engineering. Wiley, 2009. Chapter 6 Shielding, sections 6.10 Apertures, 6.10.1 Multiple Apertures, 6.10.2 Seams and 6.11 Waveguide Below Cutoff. Section numbers and titles verified against the author’s own published detailed contents listing.
Schelkunoff SA. Electromagnetic Waves. D. Van Nostrand Company, New York, 1943. The waveguide cutoff wavelengths from which the depth-attenuation constants on this page are derived — πd/1.8412 for the circular TE11 mode and 2L for the rectangular TE10 mode.
IEEE Std 299-2006, IEEE Standard Method for Measuring the Effectiveness of Electromagnetic Shielding Enclosures. The measurement these estimates would be compared against. Copyrighted; named for the method only.
MIL-STD-461G, paragraph 5.18 RE102, radiated emissions, electric field and paragraph 5.21 RS103, radiated susceptibility, electric field. Aperture leakage works in both directions and the same opening limits both requirements.