Shielding Effectiveness Calculator

Absorption, reflection and the multiple-reflection correction for a solid conductive wall, with the wave impedance taken from the exact dipole fields rather than a three-case approximation — so the page shows you what most pages hide: a thin conductive shield is nearly transparent to a close low-frequency magnetic field, which is why mu-metal exists.

Last updated 24 Sep 2026

Shielding effectiveness of a solid wall

Material, thickness, frequency, field type -> SE

Shield materialConductivity is relative to annealed copper at 5.8 × 10⁷ S/m. These are nominal figures for nominal alloys in a nominal condition and they move: a magnetic alloy’s permeability depends on the field strength, on temperature, and on whether the part has been bent, welded or dropped since it was annealed. Use the figures from your part’s datasheet; typical values vary widely between manufacturers.

Conductivity relative to copperσ / 5.8 × 10⁷ S/m. Only used when the material above is “something else”.

Relative permeability1 for any non-magnetic metal. For a magnetic alloy this is the value at the field strength and frequency you are working at, which is not the headline number on the datasheet: mu-metal’s permeability collapses as it approaches saturation and falls steadily with frequency above a few kHz.

Wall thickness (mm)The conducting thickness, not the panel thickness. A painted or anodised surface contributes nothing; a vacuum-deposited or plated conductive layer is often 5 to 50 µm, which changes the answer completely because absorption is proportional to thickness.

FrequencyThe frequency of the interference you are trying to keep out or in. Shielding effectiveness is strongly frequency dependent — the chart below sweeps it.

Frequency unit

What kind of field?This decides the wave impedance, and the wave impedance decides the reflection loss, which is usually most of the answer. Close to a current loop the wave impedance is far below 377 Ω and reflection loss nearly vanishes; close to a voltage node it is far above 377 Ω and reflection loss is enormous. Beyond λ/2π both converge on 377 Ω and the choice stops mattering.

Distance from the source to the shield (m)Used only to work out the wave impedance. Compare it with λ/2π, printed below: inside that radius you are in the source’s near field and the electric and magnetic cases are wildly different; outside it they are the same.

A conducting wall in cross-section, drawn as a geometry. The incident wave meets the first surface and most of it is reflected there (R); what enters is attenuated as it crosses the metal (A); what reaches the far surface partly reflects back, crosses again and leaks out in a decaying series, which is the multiple-reflection correction (B) and is NEGATIVE. The three terms add in decibels. The wall's thickness in skin depths decides whether B matters: it is large below about half a skin depth and vanishes above two. Every value is live, including the wave impedance, which is what the reflection term depends on and is the reason a close magnetic source is the hard case.

32.7dBExample

a 1 mm aluminium wall, with a magnetic source 30 cm away at 1 kHz

Schelkunoff: SE = A + R + B

δ = 1 / √(π f μ σ)    A = 8.686 · t / δ dB
Z_s = (1 + j)√(π f μ / σ)    R = 20 log₁₀ [ |Z_w + Z_s|² / (4 |Z_w||Z_s|) ]
B = 20 log₁₀ | 1 − K e^−2t/δ e^−j2t/δ |    K = [(Z_s − Z_w)/(Z_s + Z_w)]²
Z_w = η₀|P|/|Q| (electric source) η₀|Q|/|P| (magnetic source) η₀ (plane wave)
P = 1 + 1/(jβr) − 1/(βr)²    Q = 1 + 1/(jβr)

A: absorption loss. Proportional to thickness and to the square root of frequency, conductivity and permeability. It is the only term that does not care what kind of field it is
R: reflection loss, from the mismatch between the wave impedance and the metal’s own surface impedance. Usually the largest term at low frequency against an electric or plane wave, and nearly zero against a close magnetic source
B: the multiple-reflection correction, and it is NEGATIVE. It matters when the wall is thin in skin depths, which is when A is small, and it vanishes once A passes about 15 dB. A page that leaves it out overstates a thin shield
Zw: the wave impedance where the shield is. Taken here from the exact broadside fields of an ideal dipole, which reduce to η0/βr close to an electric source, η0βr close to a magnetic one and η0 beyond λ/2π — with no arbitrary switch between the three

Worked example

a 1 mm aluminium wall, with a magnetic source 30 cm away at 1 kHz
Aluminium is 0.61 of copper's conductivity and non-magnetic, so σ is 35.38 MS/m and μ is μ₀. The skin depth at 1 kHz is 1/√(πfμσ) = 2.676 mm
The wall is 0.374 skin depths thick, so absorption is 8.686 × that = 3.25 dB
At 1 kHz λ/2π is 159 Mm, so 30 cm is deep in the near field. A magnetic source there has a wave impedance of 2.369 mΩ — milliohms, not 377 ohms
The metal's own surface impedance is 14.94 µΩ. The two are within a factor of a few hundred of each other, so the mismatch that produces reflection loss is weak and R is only 32.04 dB
Because the wall is under two skin depths thick the multiple-reflection correction bites: B = -2.63 dB, and it is subtracted
SE = 3.25 + 32.04 -2.63 = 32.7 dB, a field ratio of 42.92  to one
The same wall against a PLANE WAVE at the same frequency would give 136.5 dB, and against an electric source at the same distance 240.5 dB. The wall has not changed; the field has

The same wall, three kinds of field

The wave impedance is the whole story. Reflection loss comes from the mismatch between the field’s impedance and the metal’s, and close to a current loop the field’s impedance is already low, so there is little mismatch left to exploit. This is why a screened room made of steel sheet does nothing about the 50 Hz field from the transformer inside it.
Case	Wave impedance	What dominates
Plane wave, or any source beyond λ/2π	377 Ω	Reflection, by a wide margin, until the wall gets thick in skin depths
Electric source inside λ/2π	η0/βr, so kilohms to megohms	Reflection, overwhelmingly. Almost any metal foil is an excellent electric shield
Magnetic source inside λ/2π	η0βr, so milliohms at audio frequencies	Absorption, which is tiny in a thin non-magnetic wall. This is the hard case

1 mm of wall against a 50 Hz magnetic field 30 cm away

A millimetre of copper — which gives well over 100 dB against a plane wave in the HF band — is worth 12.9 dB against a 50 Hz magnetic field at 30 cm, and only 0.93 dB of that is absorption. Mu-metal turns the same wall into 36.6 dB, almost all of it absorption, because μr of 20,000 shrinks the skin depth by a factor of 25. Note that raising μr also LOWERS the reflection term, from 21.4 dB to 13.7 dB, so the net gain is less than the absorption gain. And a real mu-metal can does better still, by shunting flux around the volume — an effect of the enclosure’s shape that no plane-wall model contains.
Material	A	R	B	SE
Copper	0.93 dB	21.37 dB	-9.36 dB	12.9 dB
Mu-metal (μr 20,000)	22.88 dB	13.70 dB	-0.01 dB	36.6 dB

Skin depth, in millimetres

Absorption loss is 8.686 decibels per skin depth of thickness, whatever the material and whatever the field. Read down a column: a 1 mm wall is a fraction of a skin depth at mains frequency and hundreds of skin depths at 1 GHz, which is why the same box behaves like two different objects. Steel’s permeability buys it a small skin depth and good absorption, at the cost of reflection loss and of a permeability that depends on how hard the field is driving it.
Frequency	Copper	Aluminium	Cold rolled steel
50 Hz	9.3459	11.9662	1.6895
1 kHz	2.0898	2.6757	0.3778
10 kHz	0.6609	0.8461	0.1195
100 kHz	0.2090	0.2676	0.0378
1 MHz	0.0661	0.0846	0.0119
10 MHz	0.0209	0.0268	0.0038
100 MHz	0.0066	0.0085	0.0012
1 GHz	0.0021	0.0027	0.0004

Why the field type decides the answer

A wave arriving at a conducting wall loses energy three ways. Some of it never gets in, because the wall’s surface impedance is nothing like the impedance of the wave — that is reflection loss. What does get in is attenuated as it crosses the metal, exponentially, one neper per skin depth — that is absorption loss. And what reaches the far surface partly reflects back, crosses again, reflects again, and leaks out in a decaying series — that is the multiple-reflection correction, and it is negative. Schelkunoff wrote this down as a transmission line with a lossy section in the middle, and the three terms add in decibels.

B is the term that gets dropped, and dropping it is not conservative. It only matters when the wall is thin compared with a skin depth, which is precisely when absorption is small and the reader most needs an honest number. At a fifth of a skin depth it is worth about −6.7 dB; at a twentieth, about −25 dB. A page that leaves it out tells you a thin foil is a better shield than it is, in the direction that makes a leaky enclosure look sealed. Once absorption passes 15 dB or so the term is under half a decibel and you can forget it.

The magnetic case is the one that catches people. Reflection loss is a mismatch effect, and a mismatch needs two dissimilar impedances. Close to a current loop — a transformer, a choke, a switching loop, a motor — the wave impedance is η₀βr, which at audio frequencies and a fraction of a metre is milliohms. The metal’s own surface impedance at those frequencies is micro-ohms. The two are not far apart on a logarithmic scale, so there is little reflection to be had, and with a thin wall there is little absorption either. That is the whole of it: a millimetre of aluminium that gives well over a hundred decibels against an HF plane wave gives a handful against a nearby 50 Hz magnetic field. Mu-metal exists because a high permeability shrinks the skin depth and turns absorption back on — and because a high-permeability box also shunts flux around the volume, which is a shape effect this model does not contain at all.

Where the wave impedance comes from here. Most pages switch between three formulas at λ/2π and leave a discontinuity behind. This one uses the exact broadside wave impedance of an ideal dipole — η₀|P|/|Q| for an electric source, η₀|Q|/|P| for a magnetic one — which is smooth, needs no switch, and reduces to the three textbook cases where those are valid. The two impedances multiply to exactly η₀² at every distance, which is a useful thing to know and a good check on any implementation.

What this number is not. It is the attenuation of an infinite flat sheet. It has no edges, no lid, no seam, no ventilation, no display window and no connectors. Real enclosures are limited by every one of those, and the gap between the wall’s number and the box’s number is routinely 40 or 60 dB. If you want the number that decides whether hardware passes, it comes from the aperture and seam page, not from this one. What this page is good for is knowing whether the material and thickness you have chosen could ever be enough, and understanding why a magnetic problem does not respond to more metal.

And the cables. An enclosure with perfect walls and perfect seams still leaks through every cable that passes through it, because the shield current has to be given somewhere to go. That is the transfer impedance page, and on most hardware it is the dominant term in the RE102 margin.

Frequently asked questions

Why is my copper box useless against a 50 Hz magnetic field?

Because neither mechanism works there. At 50 Hz copper’s skin depth is about 9 mm, so a 1 mm wall absorbs under a decibel; and the wave impedance of a nearby current loop is a fraction of a milliohm, so there is almost no mismatch with the metal to reflect from. The sum comes to single figures, and the multiple-reflection correction takes some of that back. The answers are distance, a smaller source loop, or a high-permeability alloy — not more copper.

What is the multiple-reflection term B, and can I ignore it?

It accounts for the wave bouncing between the shield’s two surfaces, and it is negative — it reduces shielding effectiveness. Ignore it once absorption is above about 15 dB, where it is worth under half a decibel. Below that it is the difference between a useful estimate and a flattering one: at a fifth of a skin depth it is about −6.7 dB and at a twentieth about −25 dB.

Should I use the electric, magnetic or plane-wave case?

It is decided by the source, not by the frequency. A current loop — a transformer, a choke, a switching loop — makes a low-impedance magnetic field. A voltage node — an open wire, a heatsink at a switching potential, a high-impedance trace — makes a high-impedance electric field. Beyond λ/2π of the source both converge on 377 Ω and the distinction stops mattering, which the page tells you below. If you are not sure, assume magnetic: it is the pessimistic case by a wide margin.

Does this apply to my actual enclosure?

As an upper bound only. This is an infinite plane wall with no seams, no apertures, no fasteners and no cables through it. A real box is limited by all of those, usually by 40 to 60 dB relative to its own wall. Use this page to decide whether the material and thickness could be enough, then use the aperture and seam page to find out what the box actually does.

Why does raising permeability not help as much as I expected?

Because it cuts both ways. A higher µr shrinks the skin depth, which raises absorption — but it also raises the metal’s surface impedance, which reduces the mismatch with the incoming wave and therefore reduces reflection loss. Against a plane wave or an electric field, where reflection is most of the answer, a magnetic alloy can be worse than copper. Against a close magnetic field, where reflection was never going to help, it is a large win.

Is conductive paint or a plated layer any good?

Enter its real conducting thickness and its real conductivity and the page will tell you. A nickel-filled paint might be a few per cent of copper’s conductivity at 40 µm, which is a tiny fraction of a skin depth below the VHF band, so absorption is negligible and the answer is all reflection loss — fine against electric fields, nearly nothing against magnetic ones, and very sensitive to the multiple-reflection term. Coatings also have to make electrical contact across every seam, which is usually where they fail.

Related calculators

References

Schelkunoff SA. Electromagnetic Waves. D. Van Nostrand Company, New York, 1943. The transmission-line analogy for a conducting barrier, from which the absorption, reflection and multiple-reflection terms used here derive.
Ott HW. Electromagnetic Compatibility Engineering. Wiley, 2009. Chapter 6 Shielding, sections 6.4 Absorption Loss, 6.5 Reflection Loss (6.5.1 Reflection Loss to Plane Waves, 6.5.2 Reflection Loss in the Near Field) and 6.6 Composite Absorption and Reflection Loss. Section numbers and titles verified against the author’s own published detailed contents listing.
IEEE Std 299-2006, IEEE Standard Method for Measuring the Effectiveness of Electromagnetic Shielding Enclosures. The measurement this page’s number would have to be compared against; it measures an ENCLOSURE, apertures and seams included, which is not what a plane-wall model computes. Copyrighted; named here for the method, with no values reproduced.
MIL-STD-461G, paragraph 5.18 RE102, radiated emissions, electric field. The requirement a shielding calculation is usually serving, and the reason the aperture and cable pages matter more than this one.