Design review checkpoints for enclosure shielding, aperture control, gasket design, and local shielding of noisy components
Shielding effectiveness (SE) quantifies the attenuation of electromagnetic fields by an enclosure. For a solid metal enclosure, SE is determined by reflection loss, absorption loss, and re-reflection correction. In practice, apertures (ventilation holes, seams, displays) limit the actual SE far below the material capability.
| Material | sigma_r (rel to Cu) | mu_r | SE_R at 100 MHz | Skin Depth 100 MHz |
|---|---|---|---|---|
| Copper | 1.00 | 1 | 88 dB | 6.6 um |
| Aluminum | 0.61 | 1 | 86 dB | 8.5 um |
| Steel (mild) | 0.10 | 200 | 103 dB | 1.8 um |
| Mu-metal | 0.03 | 20000 | 120 dB | 0.65 um |
| Tin plate (0.3mm) | 0.15 | 1 | 80 dB | 17 um |
| Conductive paint (50 um) | 0.001 | 1 | 50 dB | 260 um |
Die-cast aluminum enclosure (2mm wall): Material SE > 100 dB at all frequencies above 1 MHz. Apertures: ventilation holes (3mm diameter, array of 50 holes) covered with honeycomb vent (60 dB SE). Cable entry: all cables through filtered connectors (feed-through or Pi-filter). Display: ITO-coated glass (30 dB SE). Seams: finger-stock gasket on lid (70 dB SE). Measured system SE: 35 dB at 1 GHz (limited by display aperture with ITO coating). Adequate for FCC Class B with 10 dB margin for internal emissions of 75 dBuV/m.
Plastic enclosure with conductive paint: ABS plastic with internal conductive paint (3-5 ohm/square). Paint SE: approximately 30-35 dB at 100 MHz (marginal). But: paint does not connect reliably at seams between lid and base (no gasket). Effective SE at seam: 5-10 dB. Paint adhesion fails at mounting boss areas (no shielding around PCB mounting screws). Measured system SE: 8 dB at 500 MHz -- provides almost no shielding benefit.
HFSS Enclosure Simulation: Model the enclosure geometry including all apertures, seams, and cable penetrations. Use a plane wave excitation at various angles of incidence. Calculate the field inside the enclosure relative to the incident field to determine SE. This captures resonance effects that simple formulas miss (enclosure cavity can resonate, reducing SE at specific frequencies).
SE Measurement: Per IEEE 299 (large enclosures) or IEEE 299.1 (small enclosures < 2m). Use transmit antenna outside, receive antenna inside. Measure insertion loss with and without enclosure. Difference is SE. Test at multiple orientations and polarizations.
Apertures in shielded enclosures are the primary mechanism for EMI leakage. A slot or hole in a conductive surface radiates (or admits) electromagnetic energy with an effectiveness that depends on its dimensions relative to wavelength. The longest dimension of an aperture determines its worst-case frequency of maximum leakage.
Requirement: 40 dB SE at 1 GHz, need 50 cm^2 ventilation area
Option 1: Single large opening (7 cm x 7 cm):
SE = 20*log10(300/(2*70)) = 6.6 dB -- FAILS by 33.4 dB
Option 2: Array of round holes (3mm diameter):
Number needed: 50 cm^2 / (pi*(0.15)^2) = 707 holes
SE_single = 20*log10(300/(2*3)) = 34 dB
Waveguide bonus (2mm wall): 32*(2/3) = 21.3 dB
Array penalty: 20*log10(sqrt(707)) = 27 dB
Net SE = 34 + 21.3 - 27 = 28.3 dB -- still below 40 dB target
Option 3: Honeycomb vent (5mm cells, 15mm depth):
Cell size D = 5mm
SE_single = 20*log10(300/(2*5)) = 29.5 dB
Waveguide bonus: 32*(15/5) = 96 dB (!!)
Array penalty: area 50cm^2 / 0.19cm^2 per cell = 263 cells
Array penalty = 20*log10(sqrt(263)) = 24.2 dB
Net SE = 29.5 + 96 - 24.2 = 101.3 dB -- EXCEEDS requirement
Option 4: Metal mesh screen (1mm openings, wire 0.3mm, 2mm deep):
Opening size = 1mm
SE_single = 20*log10(300/(2*1)) = 43.5 dB
Waveguide bonus: 32*(2/1) = 64 dB
Number of openings per 50 cm^2: ~3800
Array penalty = 20*log10(sqrt(3800)) = 35.8 dB
Net SE = 43.5 + 64 - 35.8 = 71.7 dB -- Excellent!
Enclosure with honeycomb vents: 20mm deep aluminum honeycomb with 6mm cell size bonded to enclosure wall with conductive epoxy around entire perimeter. Provides 80+ dB SE at 1 GHz while allowing forced-air cooling. All ventilation openings use this treatment. Combined with proper seam gaskets, system achieves 60 dB measured SE from 200 MHz to 6 GHz.
Enclosure with long ventilation slots: Designer specified 2mm x 80mm ventilation slots for aesthetic reasons. SE of each slot at 1.5 GHz: approximately 0 dB (resonant!). At 1 GHz: SE = 20*log10(300/(2*80)) = 5.5 dB. With 20 slots: penalty = 13 dB. Net SE = -7.5 dB at 1 GHz -- the slots actually ENHANCE radiation at this frequency! Measured emissions 12 dB above limit at 1 GHz band.
Slot antenna effect at seams: A 1mm gap running 20 cm along a panel seam is NOT a 1mm hole -- it's a 200mm slot! The longest dimension (200mm) determines the SE. At 750 MHz (lambda/2 = 200mm), this seam provides ZERO shielding. Treat all seams as potential slot antennas.
Display windows: A display opening is often the largest aperture. ITO (Indium Tin Oxide) coated glass provides 20-30 dB SE. Wire mesh over the display provides 30-50 dB but reduces optical clarity. For high-SE requirements, use specialized EMI glass with embedded mesh.
Enclosure seams (where panels, lids, doors, and covers meet) are typically the weakest shielding point. Without proper treatment, seams create slot antennas that can completely negate the shielding provided by the panel material. EMI gaskets maintain electrical continuity across seams while allowing mechanical assembly/disassembly.
| Gasket Type | SE (dB) | Frequency Range | Compression | Cost |
|---|---|---|---|---|
| BeCu finger stock | 80-100 | DC - 18 GHz | Low (1-3mm) | High |
| Knitted wire mesh | 60-80 | DC - 10 GHz | Medium (2-5mm) | Medium |
| Conductive elastomer | 50-80 | DC - 10 GHz | High (3-8mm) | Medium |
| Conductive foam | 30-60 | DC - 3 GHz | High (3-10mm) | Low |
| Spiral wound | 70-90 | DC - 18 GHz | Low (1-2mm) | High |
| Conductive fabric-over-foam | 40-70 | DC - 6 GHz | Medium (2-6mm) | Low-Med |
Enclosure lid with BeCu finger stock: Lid perimeter has continuous BeCu finger stock (3mm finger pitch, 2mm deflection) soldered into a groove. Finger stock contacts machined bare aluminum surface on base (alodine finish, no paint in gasket zone). Fastener spacing: 50mm (panel stiffness maintains contact between fasteners). Measured SE at seam: 85 dB at 1 GHz, 70 dB at 10 GHz. No leakage hot spots detected in near-field scan.
Enclosure lid with painted seam surfaces: Conductive foam gasket installed on lid, but mating surface on base is powder-coated (insulating layer). Even though gasket is compressed, it contacts paint, not metal. Effective gasket resistance: > 1 ohm per contact point. At 500 MHz, the seam behaves as an open slot -- no shielding improvement over an ungasketed joint. Fix: specify "bare metal zone" with conductive finish in gasket contact area.
Every cable that penetrates a shielded enclosure creates a potential EMI leakage path. Unfiltered conductors passing through a shield boundary act as antennas inside the enclosure, coupling to internal circuits and re-radiating on the outside. All conductors must be filtered or their shields bonded at the penetration point.
Method 1: Shielded connector with 360-degree bonding
- Cable shield connects to connector shell
- Connector shell bonds to enclosure wall
- No unshielded conductor inside enclosure
- SE = cable shield SE + connector transfer impedance
- Best for: coaxial cables, shielded data cables (HDMI, USB, Ethernet)
Method 2: Feed-through filter (Pi or C-type)
- Each conductor passes through a capacitive filter element
- Filter housing grounds to enclosure wall (360 degrees)
- Blocks conducted interference on each wire
- SE = filter insertion loss (typically 40-80 dB)
- Best for: DC power, low-frequency control signals
Method 3: Filtered connector (Pi-filter in each pin)
- D-sub, circular, or custom connectors with built-in filters
- Each pin has a capacitor to shell (typical: 100 pF to 10 nF)
- Shell bonds to enclosure via mounting hardware
- SE = 30-60 dB depending on capacitor value and frequency
- Best for: multi-conductor control/signal cables
Method 4: Bulkhead waveguide (for optical/pneumatic/fluid)
- Non-conductive penetration through a waveguide tube
- Tube diameter < lambda/3.4 at highest frequency
- Tube length provides waveguide-below-cutoff attenuation
- SE = 32 * (L/D) dB for D < lambda/3.4
- Best for: fiber optics, tubing, ventilation
Military enclosure with all penetrations treated: Power: feed-through Pi-filters (10 nF, 50V, 30A) for each conductor. Data: filtered D-sub connectors (1 nF per pin) for RS-422 interfaces. RF: bulkhead SMA connectors with direct panel-mount (inherent 360-degree shielding). Fiber: 15mm diameter tube, 50mm long (SE = 32*50/15 = 107 dB). Ventilation: honeycomb waveguide array. System measured SE: 65 dB at 1 GHz, 55 dB at 10 GHz. All penetrations treated consistently.
Shielded enclosure with rubber grommets: Aluminum enclosure with good 60 dB panel SE, but cables enter through rubber grommets (non-conductive). Each cable entry is effectively a slot equal to the grommet hole diameter (12mm). At 1 GHz: SE of each grommet hole = 20*log10(300/(2*12)) = 22 dB. With 6 cable entries: penalty = 7.8 dB. System SE limited to 14 dB -- enclosure provides almost no benefit despite expensive aluminum construction.
PCB-level local shields (board-level shields or BLS) are metal cans placed over specific IC sections to contain their radiation or protect sensitive circuits from interference. They are commonly used for wireless module isolation, clock generator containment, and sensitive analog front-end protection.
WiFi/Bluetooth module with two-compartment shield: WiFi transceiver and Bluetooth transceiver in separate shielded compartments (prevents desense). Shield fence soldered to continuous ground pad with vias every 3mm. Removable lids with spring fingers for production test access. Internal absorber material (Laird Eccosorb) on lid inner surface damps cavity resonances. Measured isolation between compartments: 45 dB at 2.4 GHz. WiFi sensitivity maintained to -90 dBm with Bluetooth transmitting simultaneously.
Clock oscillator shield with gaps in ground fence: 100 MHz crystal oscillator enclosed in a board-level shield can. But the shield fence has a 5mm gap where a power trace crosses under the fence (trace routed on Layer 1, breaking the ground pad). At 300 MHz (3rd harmonic): 5mm gap creates a slot with SE = 20*log10(300/(2*5)) = 29.5 dB -- the gap radiates 3rd harmonic energy that couples to a nearby USB connector. Effective shield SE at 300 MHz: 29.5 dB instead of expected 60+ dB.
Shields must be grounded to provide a reference for the shielding currents and to prevent the shield from acting as a floating antenna. The grounding method differs for DC/low-frequency (safety/ESD) versus RF (EMC shielding). Both requirements must be satisfied simultaneously.
Enclosure with RF bonding design: Machined aluminum enclosure with all panels bonded via BeCu finger stock (continuous). PCB mounted on 8 standoffs (M3 screws with serrated washers) at 30mm spacing around perimeter. Additional RF bonding clips (Wurth WE-SHC series) connect mid-board ground to enclosure floor at 4 interior points. Total ground connections between PCB and enclosure: 12 points, maximum spacing 30mm. Effective RF ground to 500 MHz+. Near-field scan shows no leakage at PCB mounting area.
Shield grounded by single wire: Metal enclosure lid connected to PCB ground via a 5cm wire (for "grounding"). Wire inductance: ~50 nH. At 100 MHz: Z = 31 ohm. At 500 MHz: Z = 157 ohm (effectively open). The shield is not grounded at RF -- it acts as a floating reflector that can redirect radiation in unpredictable directions. In some cases, adding this "grounded" shield INCREASES emissions because it creates a resonant cavity at specific frequencies without providing proper shielding current paths.
Paint/coating under fasteners: Even a few micrometers of non-conductive coating between the shield and ground surface can prevent RF grounding. At 1 GHz, the capacitance through a 10 um paint layer (1 cm^2 washer area) is only ~0.9 pF, which is 177 ohm -- essentially an open circuit. Always specify "bare metal" or "conductive finish" at all grounding contact surfaces.
Galvanic corrosion: Dissimilar metals in contact (e.g., aluminum enclosure + steel screw) can corrode, increasing contact resistance over time. Use compatible metals or plating (tin, nickel) and anti-corrosion treatments that maintain conductivity (NOT paint or lacquer).