Skip to content
Electromagnetic Compatibility

Shielding: absorption, reflection and apertures

Shielding effectiveness, absorption 8.69 t/δ and reflection proved with copper examples, low-frequency magnetic fields, slots and cables passing through the shield.

A shield is a conductive enclosure separating the source from the victim: a device’s metal case, a cable’s braid, a foil over a sensitive circuit. It works in two ways, reflecting the wave and absorbing it, and the calculations show that even a very thin foil would suffice for most fields. The real problem is apertures: slots, seams and cables passing through the shield. In this lesson we look at shielding effectiveness, absorption and reflection with their proofs and examples, the difficult case of low-frequency magnetic fields, and apertures.

Shielding effectiveness

Shielding effectiveness is the ratio, in decibels, between the field without the shield and with it:

    \[ SE = 20\log_{10}\frac{E_{without}}{E_{with}} \]

According to Schelkunoff’s theory it splits into three terms: SE = R + A + B, reflection, absorption and a correction term for multiple reflections inside the shield, negligible when absorption exceeds about ten decibels.

A plane wave on a metal shield
The incident wave E i arrives from the left; most of it is reflected (E r). The part entering the metal of thickness t is attenuated exponentially, and only a small transmitted wave E t emerges on the other side.

Absorption

Inside a good conductor the field decays exponentially with depth, with a characteristic length called the skin depth (Electromagnetic Fields lesson on the skin effect):

    \[ \delta = \frac{1}{\sqrt{\pi f\mu\sigma}} \]

Theorem. A shield of thickness t introduces an absorption of

    \[ A = 8.69\,\frac{t}{\delta}\ \text{dB} \]

Proof. Crossing the thickness t, the field amplitude is reduced by the factor e^{-t/\delta}. In decibels: 20\log_{10}e^{t/\delta} = (t/\delta)\cdot 20\log_{10}e = 8.686\,t/\delta. ∎

Example. Copper, \sigma = 5.8\cdot 10^7 S/m: at 1 MHz \delta \approx 66 µm, and a 0.1 mm foil absorbs about 13 dB; at 100 MHz \delta \approx 6.6 µm and the same foil absorbs about 130 dB. Absorption grows with the square root of frequency: at high frequencies even a thin metal layer deposited on plastic is enough.

Reflection

A wave reaching a surface with an impedance different from its own is largely reflected (lesson on plane waves). The wave in free space has impedance Z_w = 377 Ω, the metal a very small impedance, |Z_s| = \sqrt{2\pi f\mu/\sigma}.

Theorem. For |Z_s| \ll |Z_w| the reflection is

    \[ R = 20\log_{10}\frac{|Z_w|}{4|Z_s|} \]

Proof. The transmission coefficient of the electric field from a medium of impedance Z_a to one of impedance Z_b is 2Z_b/(Z_a + Z_b). Entering the metal it is 2Z_s/(Z_w + Z_s) \approx 2Z_s/Z_w; leaving it, 2Z_w/(Z_s + Z_w) \approx 2. The product is 4Z_s/Z_w, and its inverse in decibels gives the formula. ∎

Example. Copper at 1 MHz: |Z_s| \approx 3.7\cdot 10^{-4} Ω, R = 20\log_{10}(377/(1.5\cdot 10^{-3})) \approx 108 dB. With absorption, even a thin copper shield gives more than 120 dB against a plane wave: far more than is needed.

The difficult case: near magnetic fields

Close to a magnetic source (a loop, a transformer) the wave impedance is much lower than 377 Ω, and reflection is reduced accordingly; at low frequency absorption is also small, because \delta is large (at 50 Hz in copper it is about 9 mm). A mains transformer cannot be shielded with a copper foil. What is needed is:

  • high-permeability materials (iron-nickel alloys such as mu-metal), which offer the field lines a preferential path and divert them around the area to be protected;
  • or, better, reducing the source and the victim loop (lesson on coupling).

Near electric fields, on the contrary, with high impedance, are easily shielded: a grounded conductive surface is enough.

Apertures

In practice a shield is never closed: there are ventilation slots, seams between panels, displays, connectors. A slot behaves like a slot antenna (lesson on aperture antennas), and what matters is its maximum dimension, not its area. For a slot of length L, less than half a wavelength:

    \[ SE \approx 20\log_{10}\frac{\lambda}{2L} \]

Example. A 10 cm slot at 1 GHz (\lambda = 30 cm) leaves only 3.5 dB of shielding; at 100 MHz, 23.5 dB. The thickness and material of the shield become irrelevant: the slot is what counts. Many small holes shield much better than a slot of the same area, because each has a small maximum dimension.

The resulting rules:

  • seams with frequent contacts (closely spaced screws, conductive gaskets), at a spacing much less than \lambda/2 at the highest frequency of interest;
  • ventilation with grids of small holes or honeycombs, which behave like waveguides below cut-off;
  • every cable passing through the shield must be filtered at the point of entry, or its braid must be bonded to the shield around its whole circumference (360°). An unfiltered cable carries the disturbance outside and radiates it, defeating the shield: it is the most common mistake. A braid connected with a short wire (pigtail) loses much of its effectiveness at high frequencies, because the pigtail has inductance.
Subscribe
Notify of
guest
0 Comments
Oldest
Newest Most Voted
0
Would love your thoughts, please comment.x
()
x