EMI Shield Stamping: Shielding Cans and Spring Fingers

EMI failures never announce themselves politely. One radiated-emissions run - 30 MHz to 1 GHz per FCC Part 15 or CISPR 32 - and a board that passed every functional test walks out of the chamber with a 46 dBµV/m spike at 312 MHz, 8 dB over the Class B limit. The culprit is rarely the silicon. It is almost always the shield: a stamped can whose seam leaked like a slot antenna, a grounding finger whose contact force sagged from 120 gf to 25 gf after 5,000 actuations, or a tin-nickel surface that wore through where it rubs a chassis edge. By then you own the re-spin, the re-certification, and the field returns. Thirty years of stamping these parts tells me one thing: those failures are design decisions made before the die was cut - and every one of them is preventable on paper first.

This guide covers the full decision path for a stamped EMI shield program: how stamped shields fail, which alloy and plating to pick for the can versus the fingers, the aperture and seam rules that actually decide shielding effectiveness, the spring-force math that keeps a grounding clip alive, what the press and the tooling have to deliver, how to verify the finished part, and how to evaluate a supplier that is not going to hand you a 46 dB spike at your certification run. The short version up front: shielding effectiveness is made or lost in three places - the apertures, the ground path, and the contact surfaces - and all three are decided by geometry and surface choices you control before the die is cut.

• Shielding effectiveness of a properly grounded stamped can runs 40-60 dB - 40 dB cuts radiated power by 99%, 60 dB by 99.9%.
• Any aperture or seam longer than λ/20 at your highest frequency leaks; at 2.4 GHz that limit is 6.3 mm.
• C17200 beryllium copper (22% IACS, 1,100-1,380 MPa spring temper) is the reference for spring fingers; C5191 phosphor bronze is the budget alternative at ~13% IACS.
• Springback on high-temper BeCu 90° forms runs 3-8° of over-bend; ignore it and your grounding finger misses the pad by half a millimeter.

Why Stamped Shields Fail

Treat a shielding can as an antenna problem, not a box problem. A stamped enclosure with a 10 mm slot at 2.4 GHz - where λ/20 is 6.3 mm - radiates through that gap like a leaky waveguide, no matter how thick the wall is. The slot does not have to be a hole; a 90° bend that leaves a 0.3 mm gap between the lid edge and the sidewall is the same antenna. The seams between can and lid, the vent holes, the ejector-pin marks, even the burr side of a sheared edge that lifts 0.05 mm off the PCB ground plane - all of them are radiating edges.

The second classic failure is the grounding finger. A spring finger that started at 120 gf of normal force can relax to a third of that when the material is over-bent past yield, when the temper is softer than the drawing called for, or when the finger is loaded beyond its elastic limit during assembly. Below roughly 30 gf, contact resistance climbs, the ground path goes inductive, and the shield becomes a floating piece of metal that couples noise instead of shunting it. The third failure is plating. Pure tin on a finger that slides against a chassis edge gall-wears after a few thousand cycles; the nickel underplate then oxidizes and the contact turns into a diode. These are the three failure families we see in returns: geometry, force, and surface. If you are chasing a shield failure in the field rather than in the chamber, our guide on stamped shield can design and assembly walks the enclosure side, and spring contact design and force budgeting are covered in depth further down in this guide.

Material Selection: Conductivity vs. Spring Tension

Shield parts pull in two directions. The can body wants high conductivity and easy forming; the fingers want high yield strength and fatigue resistance. No single alloy wins both, so choose by function and plate the rest.

Alloy	Spec / Composition	Conductivity (% IACS)	Tensile, spring temper (MPa)	Typical job
C17200 beryllium copper	UNS C17200, 1.8-2.0% Be	22	1,100-1,380	Spring fingers, grounding clips, cantilever contacts
C5191 phosphor bronze	JIS C5191 (CuSn6P), ~6% Sn	~13	590-780	Can walls, low-cost spring contacts
C2600 brass	70/30 Cu-Zn	~28	390-540	Can bodies, EMI frames, non-contact shields
SUS301 stainless	17Cr-7Ni, full hard	~2.4	1,200+	High-temperature or corrosive environments

BeCu earns its price on anything that must carry current while flexing. At 22% IACS it conducts well enough for grounding duty, and its fatigue endurance around 240-345 MPa at 10&sup8; cycles keeps a 0.15 mm finger alive past 100,000 actuations. Phosphor bronze trades some conductivity and fatigue life for cost, which is why you see it in can walls and simple clips. Brass is the cheapest body material but work-hardens unpredictably in tight bend radii and sags under repeated heat cycles. And if you stamp a shield that has to survive 150°C reflow plus vibration, SUS301 full hard beats all of them on strength - you just have to ground through fasteners because 2.4% IACS will not carry a ground plane. For the alloy-family trade-offs beyond these four, the brass versus phosphor bronze comparison gives the full conductivity, strength, and cost picture.

Aperture Rules: The λ/20 Line

Every opening in a shield is a compromise between assembly and attenuation. The rule that has not changed in forty years: keep every opening smaller than one-twentieth of the wavelength at the highest frequency you must suppress. Above that, the aperture couples energy out at nearly full strength.

Frequency	Wavelength λ (mm)	Max aperture λ/20 (mm)
1 GHz	300	15
2.4 GHz	125	6.3
5.8 GHz	52	2.6

Apply that number to every feature: vent holes (keep them under the limit or add a stamped louver), the gap under a snap-fit lid, the spacing between grounding tabs on a seam, even the distance between dimples that stitch two halves together. At 2.4 GHz, seam tabs must be closer than 6.3 mm apart - which means a 40 mm can edge needs at least seven grounding points, not two at the corners. Stamped dimples, lanced tabs, and coined bosses are cheap ways to stitch a seam; solder or weld is better, but you can rarely weld at board-assembly cost. This is where 40-60 dB of shielding effectiveness is actually made or lost - the metal does not attenuate, the apertures decide.

Grounding surfaces get the same discipline as apertures. The shield is only as good as the impedance of its ground path, so specify the plating that keeps that path stable: tin 2-8 µm over a 1-3 µm nickel strike for solder-down cans, and tin-nickel (65/35, ~700 HV) on fingers and edges that rub. Tin-nickel resists wear and keeps contact resistance low where pure tin smears and oxidizes. If your frequency plan goes above 5 GHz, review every seam and every via density against the table before you approve the drawing - the margin between 6.3 mm and 2.6 mm is exactly where a "minor" design change turns a passing board into a failing one.

Spring Fingers and Grounding Clips: Force Is Everything

A grounding finger is a cantilever beam with a job: hold 50-150 gf of normal force against the mating surface for the life of the product, without yielding. The three variables are length, thickness, and deflection - and the engineering error we see most is designing for the force at assembly instead of the force after 10,000 actuations. Keep the operating deflection under half the elastic limit of the material, and BeCu fingers will hold over 90% of initial force indefinitely. Push them past yield once, and the force curve is gone.

Fatigue life follows the same logic. A 0.15 mm C17200 finger flexed within its elastic range survives millions of cycles; the same finger over-deflected by 0.2 mm fails in the tens of thousands. If your clip must survive repeated insertion - battery contacts, card guides, shielded connector housings - either lengthen the beam, thin the stock, or step up to BeCu. On the surface side, tin-nickel plating is the difference between a clip that slides 5,000 times and stays at 80 mΩ and one that wears to bare nickel and drifts to 500 mΩ. Specify it on every sliding contact.

The contact-force budget has to include the assembly stack, not just the finger. A finger designed for 50-150 gf assumes a defined deflection at assembly; if the shield sits on standoffs that vary by ±0.1 mm, the finger force varies with them. This is why we push customers toward a coined or dimpled contact land on the mating surface - it fixes the wipe point and the deflection - and why the spring contact finger and gasket guide is the reading we hand out with every new finger drawing. The finger geometry, the land, and the stack tolerance are one system; quote them as one system.

Plating for RF Grounding and Wear

The base metal sets the spring; the plating sets the contact. A bare C17200 finger oxidizes over time and the oxide film raises contact resistance even when the mechanical fit looks perfect. The plating thickness and chemistry decide the life of the interface, and for RF the requirement is stricter than for DC power: an oxidized or worn contact that still conducts at DC can already be adding insertion loss and reflection at gigahertz frequencies.

Plating	Typical Thickness	Best For	Watch Out For
Tin over nickel strike	2-8 µm tin on 1-3 µm nickel	Solder-down cans, static ground planes	Smearing and galling on sliding contacts
Tin-nickel (65/35)	2-5 µm	Sliding fingers and wiping edges	Harder to solder; keep away from solder joints
Silver	2-8 µm	Highest-conductivity interfaces	Tarnishes in sulfur atmospheres; use with nickel underplate
Gold (selective)	2-8 µm	Low-signal, low-cycling contacts	Cost; unnecessary on high-cycle door gaskets
Zinc	5-12 µm	Corrosion protection on steel cans	Not a low-resistance RF contact surface

The plating has to be specified by zone, not by blanket statement, and that is exactly what selective reel-to-reel plating is for: gold or silver only on the wipe and solder zones, tin on the solder land, nickel as the strike everywhere it is needed - all at controlled thickness, with salt-spray verification to ASTM B117 where corrosion is a risk. Two mistakes kill more shield programs than any alloy choice. First, plating the whole part with the finger metal: it costs money and it does not help, because the can body does not need 22% IACS or a gold flash. Second, post-plate forming: if you form after plating, the plating cracks at the bend and the ground path goes through a cracked film. We plate after forming on parts that demand it, and we design the tool so the critical surfaces are formed before plating lands on them. The stamping plating and surface finish guide covers the zone-by-zone approach, including how to pick between gold, silver, and tin for contact duty specifically.

Stamping Reality: Springback, Burrs, and Tolerance

None of this design survives contact with the press unless the tooling compensates for how these alloys behave. High-temper BeCu springs back hard: a 90° form at R/t between 0.5 and 1.0 needs 3-8° of over-bend in the die, and the exact number depends on grain direction - bend lines parallel to rolling direction spring back more than perpendicular ones. We compensate in the tool by air-bending past the angle, coining the bend zone, or both, and we verify with CMM on the first article, not by feel. The same applies to C5191 and brass, just with smaller angles.

Burr control is the second discipline. On a shield, the burr side of a sheared edge can lift a soldered seam or scratch a chassis surface it slides against. Standard practice on our floor: hold burr height under 10% of stock thickness - under 0.02 mm on 0.2 mm stock - and control burr direction so the sharp edge faces the scrap side, away from the contact face. Tolerances on critical dimensions like finger height and can depth hold ±0.05 mm in a progressive die; anything tighter starts costing real money in tool maintenance. Stock for shielding parts typically runs 0.15-0.5 mm, comfortably inside the 0.05-3.0 mm band our strip-fed fleet handles at widths up to 650 mm, and at that gauge a good progressive die with in-die forming, coining, and plating-ready flatness is the difference between a shield that snaps together on the line and one that fights the assembler. Our high-speed Aida presses run to 300 SPM for the simple can blanks, while the finger-and-form work runs slower on the same progressive die high speed stamping lines; the tolerances come from the tool, not from pressing harder.

Can Geometry: Seams, Dimples, and Corner Design

The aperture rules tell you the spacing, but the geometry of the seam itself decides whether that spacing is even achievable. A seam made by two overlapping flanges is a slot unless something stitches it; a seam made by a tongue-and-groove or a coined step is a labyrinth that keeps the slot length short even where the pieces meet. On our shields we use three stitch families: dimples coined into one flange that nest into the other, lanced tabs that fold over and trap the mating edge, and continuous coined ribs along the seam line that both stiffen the wall and hold the two halves in metal-to-metal contact. Each one is a stamped feature that costs nothing extra in a progressive die if it is on the drawing.

Corners are where stamped cans leak first. A drawn or folded corner with a tight inside radius holds metal-to-metal contact around the bend; a corner with a visible gap at the fold line is a vertical slot that runs the full can height. Specify the minimum inside radius, keep the corner tabs inside the λ/20 spacing, and check the actual folded corner on the first article with a feeler gauge, not with your eyes. The same discipline applies to the lid: a snap-fit lid is a seam all the way around, so the snap points have to be inside the spacing limit, and the lid stiffening ribs should push the lid edge down onto the can rim rather than letting it sit proud. If you are designing a shield can from scratch, the shield can design guide has the full drawing checklist; if you are sourcing one, send us the frequency plan with the drawing and we will check the seam spacing before we quote.

Testing and Verification: What the dB Number Really Means

Shielding effectiveness is quoted in decibels, and it is a ratio of field strength with and without the shield, measured across frequency. Well-designed beryllium-copper finger stock achieves 60-120 dB of attenuation from roughly 100 MHz up through 10 GHz, with the strongest numbers - often 100 dB or better - around 1 GHz when the contact interface is clean and continuously loaded. Stainless-based seals with silver plating close the gap at lower frequencies but can lose 10-20 dB at the high end if the contact pressure is uneven. The dB figure assumes the gasket is compressed and the full perimeter is engaged; a 5 mm gap where a finger missed its land wipes out 30 dB in that sector regardless of the headline number.

Two verification points matter for a stamped shield program. First, test the assembled product, not the component: a can that measures beautifully on the bench but is soldered with two ground pads instead of seven has already failed by the aperture rule. Second, test for the end of life, not the first day: shielding effectiveness degrades with every open-close cycle as plating wears, so a seal rated at 100 dB when new might measure 80 dB after 50,000 cycles in a dirty environment. Qualify the part for the end-of-life number, or you will be re-testing in the field. On the manufacturing side, dimensional and surface verification is where the stamping shop earns its keep: CMM for the finger heights and can depths that set the force, optical measurement for the seam and aperture geometry, and in-line vision for burr direction and plating continuity on every strip. That is the QC stack we run on shield programs, and it is the stack you should ask any supplier for before you commit tooling.

Stamped Shields vs Machined and Die-Cast Enclosures

Stamping is not the only way to build a shield, and the honest comparison is against CNC-machined aluminum and die-cast zinc or aluminum boxes. Machined enclosures win on wall thickness and dimensional freedom - you can mill a 3 mm wall with cooling slots and threaded bosses - but they lose on cost per piece, material yield, and weight. A machined shield for a handheld device can cost five to ten times a stamped equivalent at volume, and the swarf from hogging out a solid block is pure waste. Die-cast enclosures give you complex 3D geometry and thick walls at moderate piece cost, but the tooling is expensive, the lead time is long, and castings need plating or conversion coating for RF anyway because bare die-cast surfaces oxidize and have high surface resistance.

Parameter	Stamped Shield Can	CNC Machined	Die Cast
Tooling cost	Low to moderate	None (per-part setup)	High
Piece cost at volume	Lowest	High	Moderate
Wall thickness	0.15-0.5 mm	1 mm and up	1-2 mm typical
3D complexity	Limited to forms and folds	Full	Full
Weight	Lowest	High	High
RF surface finish	Plated, controlled	Needs coating	Needs coating
Typical lead time	Weeks to tool	Days per part	Months to tool

The crossover is simple: if the shield is a box around a board with a few fingers, stamping wins at any real volume. If it needs thick walls, threaded inserts, or heatsink-integrated geometry, machining or casting earns its cost. A surprising number of products run both - a stamped can over the noisy section, a machined frame for the mechanical structure - and the electronics stamping guide covers how stamped shields fit into the wider electronics enclosure picture.

Evaluating a Shield Stamping Supplier

Five questions separate a shield supplier from a metal bender. First, do they measure springback and compensate in the tool, with CMM first-article data to prove it? Second, can they control burr direction - the difference between a seam that solders flat and one that lifts? Third, do they have in-house selective plating or a plating partner they can hold to zone thickness specs like 2-8 µm with ASTM B117 verification? Fourth, do they verify with CMM, optical measurement, and in-line vision rather than a caliper and a prayer? Fifth, will they review your frequency plan against the seam and aperture geometry before they quote - or do they just cut steel? Our own answer to all five: a 10,000 m² facility in Dongguan Chang'an running 21 presses up to 110 T with ±0.005 mm positioning accuracy, IATF 16949:2016 and ISO 14001:2015 certification, selective reel-to-reel plating lines, and a QC lab with CMM, optical, and in-line vision coverage on shield programs. The quality and inspection page shows the verification stack, and we will put the same questions to ourselves that you should put to us.

Your Action Roadmap

The pattern is the same in every failed-certification story: a shield designed by feel instead of by λ/20, a spring finger drawn without a force target, a plating spec picked from a catalog. Those are cheap to fix on paper and expensive to fix in a chamber. Send us your shield drawing - can, fingers, clips, or all three - with the frequency plan and the cycle count, and get a DFM review with real numbers: material recommendation with conductivity and temper, aperture and grounding-point check against your frequency plan, springback-compensated form angles, a zone-by-zone plating plan, and a tooling quote with tolerances that match your assembly. Send us your shield drawing and we will return it with a stamping feasibility review, so the only surprise at your certification run is that you passed on the first attempt.