Metal Enclosure & Shield Can Stamping: RFI Cases

Most RFI complaints we see at the factory do not start with a bad chip or a noisy clock. They start with a stamped shield can that leaks. Three failure modes account for the bulk of field returns: EMI leakage through seams you can see with the naked eye, poor grounding that leaves the can floating at RF frequencies, and assembly gaps that open the moment a PCB flexes or a lid snaps on a tenth of a millimeter out of tolerance. A designer will spend weeks screening the RF section, then drop a thin mild-steel cover over it and call it shielding. That cover is doing maybe 10 dB at 1 GHz and nothing useful at 2.4 GHz. After thirty years of stamping these parts for telecom, medical and industrial customers, I can tell you the can is half the battle, and the stamping details are where the other half is won or lost.

This guide is written for the engineer who has to put a shield can on a drawing and wants the numbers to survive a pre-compliance test: what SE target to set, which material and finish to spec, how long a seam can be, how many ground tabs are enough, and what tolerances the die actually has to hold. The manufacturing side matters more than most designers expect, because a shield can that works on a prototype with hand-soldered seams falls apart when the same geometry is stamped at 200 strokes per minute on a progressive die. We build these parts every week on 21 presses, from 25 to 80 ton Aida high-speed machines down to the heavy 45 to 110 ton line, holding ±0.005 mm positioning on strip from 0.05 to 3.0 mm thick, so the guidance below is grounded in what the die can actually hold rather than what a drawing hopes for.

What a Shield Can Actually Has to Do

A shield can is a Faraday cage in miniature. Its job is twofold: keep the circuit's own noise off the antenna and the rest of the board, and keep outside interference out of a sensitive front end. The number that matters is shielding effectiveness (SE), expressed in decibels. A good stamped metal enclosure on a populated board typically delivers 40-60 dB across 30 MHz to 3 GHz when the seams, vents and ground path are done right. Drop below 40 dB and you are one regulatory test away from a respin. Push past 60 dB and you are usually paying for welded seams and beryllium-copper gaskets you may not need.

SE is not a single number. It is the sum of reflection loss at the surface, absorption loss through the wall thickness, and the penalty you eat from every aperture, seam and cable crossing. The wall material and thickness decide absorption and reflection. The seams and vents decide whether the cage is actually closed. The stamped shield family is catalogued on the shield parts page, and the wider subject of shielding stamped parts is covered in the EMI shield stamping guide.

Material Choice Drives Cost, Solderability and SE

We stamp shield cans and enclosure shells from three materials most often, and each carries a different trade-off.

Material	Typical thickness	Relative conductivity	Stamping note	Best use
Tin-plated steel (SPCC + Sn)	0.15-0.30 mm	Good, ~15% IACS after plating	Cheap, solderable, rust-resistant	High-volume consumer, cost-sensitive
Aluminum (5052 / 1060)	0.20-0.50 mm	~60% IACS	Light, but oxide layer blocks contact	Weight-critical, vent windows
Beryllium copper (C17200)	0.08-0.20 mm	~22% IACS, high spring	Spring finger stock, form tools	Gaskets, contacts, finger stock
Stainless steel (304)	0.10-0.25 mm	Low, ~2.5% IACS	Hard to form, needs welded seams	Corrosive or outdoor environments

Tin-plated steel is the workhorse. The tin layer gives you a solderable, oxidation-resistant surface so the can can be reflowed straight to the board, and the steel gives you enough wall thickness for real absorption loss. Aluminum is lighter and more conductive, but that native oxide layer is a killer for electrical contact - you need conductive coating or a separate spring contact, and you cannot rely on a simple press-fit seam to carry RF current. Beryllium copper is the material we reach for when the part itself must act as a spring: finger stock, grounding tabs, board-to-chassis contacts. It holds spring force through thousands of cycles and survives reflow. Stainless steel appears when the enclosure must survive salt, spray and temperature swings that would eat tin-plated steel, but it forms hard, springs aggressively, and almost always demands welded seams, so it only wins on corrosion grounds. The alloy-level detail for the copper side of the family is on the phosphor bronze materials page.

The λ/20 Rule: Why Your Seam Is a Slot Antenna

Here is the single rule that catches most first-time designs. Any seam, gap or slot longer than roughly λ/20 starts to behave like a waveguide below cutoff that leaks energy at that wavelength. At 1 GHz, λ is 300 mm, so λ/20 is 15 mm. At 2.4 GHz, λ/20 drops to 6.25 mm. At 5 GHz it is 3 mm. That means a 10 mm unbroken gap that is harmless at 900 MHz is pumping RF straight through at 5 GHz.

The practical consequence for stamping: the longest continuous unsealed seam on your can should be kept well under λ/20 at your highest frequency of concern, and every seam that must carry current needs either a welded joint or a compression interface with enough contact points to stay electrically continuous. A snap-on lid with a single line of contact around the perimeter is fine at low frequencies and a disaster at microwave bands.

RF Numbers to Put on the Drawing

Shielding effectiveness is set by physics before it is set by the material spec, and the two numbers that matter are skin depth and the longest aperture. Skin depth tells you how much wall you need for absorption; the λ/20 rule tells you how short every seam has to be.

Frequency	Skin depth, copper	Skin depth, aluminum	Skin depth, mild steel	λ/20 seam limit
100 MHz	≈ 6.6 µm	≈ 8.2 µm	≈ 1.3 µm	150 mm
1 GHz	≈ 2.1 µm	≈ 2.6 µm	≈ 0.4 µm	15 mm
2.4 GHz	≈ 1.4 µm	≈ 1.7 µm	≈ 0.3 µm	6.25 mm
5 GHz	≈ 0.9 µm	≈ 1.2 µm	≈ 0.2 µm	3 mm

Read the table twice. First, a 0.2 mm wall is hundreds of skin depths deep even at 100 MHz, so a solid stamped wall is never the weak point - absorption loss is effectively infinite for any wall thickness a die will form. Second, the magnetic permeability of steel gives it a smaller skin depth, which is why steel cans beat aluminum cans in magnetic-field-dominated environments despite lower conductivity. What the table cannot show is the aperture penalty, and that is the whole game: at 2.4 GHz a 6 mm slot costs you roughly 20 dB regardless of whether the wall is steel, aluminum or copper. A stamped can is a battle against slots, not against wall thickness.

Set the SE target from the test requirement, not from habit. A module that must pass FCC Part 15 class B radiated limits with 6 dB margin usually needs 30-40 dB of can SE in the offending band; a telecom front end living next to a transmitter needs 50-60 dB; an automotive ECU in an unscreened bay targets 60 dB or more. Everything above 60 dB starts costing real money per part, so the target band belongs on the drawing before tooling, alongside the frequency range of concern. The band-by-band treatment for radio modules is in the telecom and 5G stamping guide.

Seam Strategy: Compression-Type vs Welded-Type

We build two families of enclosures, and the choice is mostly a frequency and volume decision.

Compression-type (press-fit and snap-on)

The lid is held by spring fingers, embossed dimples or a rolled seam that presses the two surfaces together. The advantage is speed: no secondary process, fully automated assembly, and the can is reusable for rework. The catch is contact resistance. A compression seam only works if the mating surfaces stay in contact across every point, which means tight flatness tolerance on the stamped flange (we hold ±0.05 mm on the lip) and a clean, bare or tinned surface. For compression seams that must hit 40-60 dB, we often add an EMI gasket or a continuous beryllium-copper finger strip. Without it, a press-fit seam is lucky to reach 30 dB at 1 GHz. The finger-strip force budget is a spring design problem in its own right, and it is worked through in the spring contact and finger gasket guide.

Welded-type (laser and resistance seam weld)

For 60 dB targets or harsh environments, we laser-weld or resistance-seam-weld the can to the base or join two shells into a continuous wall. Welding kills the seam leak entirely because there is no seam - the metal is continuous. The cost is a secondary operation and a part that is not field-reworkable. We typically recommend welded seams only when the operating band goes above 2 GHz or the product sits in an unshielded, noisy environment like an industrial cabinet or a vehicle. Between the two extremes sits the hybrid: compression seams on the low-frequency edges of the can and one welded joint where the radiating structure sits. Hybrid builds are more common than either pure family, because real boards concentrate their noise sources.

Ground Tabs and the Floating-Can Problem

A shield can with a perfect wall and a poor ground is still a leaky can. At RF, ground is not a DC connection to a plane - it is a low-inductance path. A single thin solder fillet at one corner has enough inductance at 1 GHz to let the whole can float. We stamp multiple ground tabs, usually four or more, with generous pad area and sometimes raised dimples that pierce the solder mask for a direct metal-to-metal joint. The data from our test boards: a single corner ground gives about 25 dB SE at 900 MHz, moving to four evenly spaced tabs pushes it past 45 dB, and adding a continuous perimeter solder seam with finger contact reaches the 55-60 dB range.

The takeaway is simple. Ground early, ground often, and treat the ground path as part of the shield, not an afterthought on the schematic.

Vent Windows: Let Air Through, Block RF

Power devices need to breathe, and a sealed can cooks the silicon. The fix is a stamped vent window - a honeycomb or expanded-metal aperture in the wall. A honeycomb vent made of thin-walled cells behaves like a bundle of waveguides below cutoff, so it passes air and blocks RF at the same time. The cell size sets the cutoff frequency: smaller cells block higher frequencies but cost more to stamp and restrict airflow. For a 2.4 GHz requirement we keep the cell pitch under 1.5 mm and the wall depth at least 3 mm. Done right, a honeycomb vent costs you only 2-4 dB against a solid wall while moving the air you need.

Watch the perimeter of the vent. If the honeycomb panel is simply glued into a hole, that glue line is a seam and a leak. We press-fit or weld the vent frame into the stamped shell so the RF path stays continuous.

Three RFI Cases From the Floor

Case 1: The 2.4 GHz Module That Failed Pre-Compliance

A customer brought us a Wi-Fi module that failed radiated emission at 2.45 GHz by 8 dB. The can was tin-plated steel, good wall, but the lid was snap-on with a single perimeter bead and one ground corner. We re-stamped the lid with four embossed spring fingers around the flange and added four ground tabs at the base. Same material, same thickness, no welding. SE went from ~28 dB to ~52 dB at 2.4 GHz and the module passed. The fix was geometry, not material.

Case 2: The Aluminum Lid That Would Not Ground

An industrial sensor used a 0.4 mm aluminum enclosure for weight. The lid was press-fit and the seams looked clean, but SE measured 18 dB at 900 MHz. The oxide on the aluminum meant the press-fit surfaces were not electrically touching. We added a continuous beryllium-copper finger strip around the lid flange and switched the ground pads to tinned steel inserts. SE climbed to 48 dB. Aluminum is fine, but you cannot trust its bare surface to carry RF current.

Case 3: The Welded Can for a Vehicle Telematics Unit

A telematics box sat in an unscreened vehicle bay with a 1.5-5.8 GHz radio. Snap-on was never going to hold 60 dB against that environment. We laser-welded the shell seams and resistance-seam-welded the lid, then stamped honeycomb vents for the power stage. Measured SE held 58-62 dB across the band. The part cost more and was not reworkable, but it passed automotive EMC first try.

Solderability and Surface Finish

A shield can is soldered to the board at the ground tabs, and the finish decides whether that joint is reliable in production and in the field. The finish requirements are different from a decorative coating: the surface must wet with solder, stay stable through one or two reflow peaks, and keep enough conductivity at the ground interface for the RF path.

Tin is the default finish for reflow-attached cans, applied after forming so the cut edges are covered. Matte tin is preferred over bright tin because brightener chemistry raises whisker risk, and a whisker bridging a 0.5 mm ground gap is an intermittent short that no test lab reproduces on demand. On steel, the tin is typically electroplated at 2-8 µm with a nickel underlayer; the nickel blocks iron diffusion into the tin so the solder joint does not grow a brittle intermetallic that cracks under thermal cycling. On aluminum, the oxide layer defeats solderability outright, which is why aluminum cans get conductive coatings or tinned copper inserts at the ground points instead of bare solder. The selective reel-to-reel plating line places gold, silver, tin, or nickel at 2-8 µm exactly where the contact and solder zones sit, and zinc at 5-12 µm protects non-solder steel hardware, with salt spray per ASTM B117 verifying the corrosion claim. The full finish-selection logic is in the plating and surface finish guide and the terminal plating selection guide.

Two field failures trace back to finish details. The first is solder dewetting on cans plated before forming: the bend at the ground tab cracks the tin, the joint wets poorly, and the tab lifts after the second reflow. The second is tin whiskers in humid automotive interiors, which is why automotive EMC enclosures specify matte tin, controlled composition, and sometimes a nickel barrier. Both are prevented at the plating spec, not at the test bench.

What the Die Must Do: Stamping Cans at Volume

A shield can is a box with a flange, dimples, tabs, and holes, and the die that makes it has to do everything at once: blank, draw, form, pierce, and often coin, in one stroke per part. That is a progressive die problem, and it is why shield cans are a stamping specialty rather than a sheet-metal job.

Thin material is the first challenge. Can stock runs 0.15 to 0.30 mm, and at that thickness the strip wants to buckle, wrinkle, and wander in the die. The solution set is well proven: pilot holes registered to the critical features, draw beads and pressure control at the forming stations, and in-die cam forming for the side features that would otherwise need a second hit. Our high-speed line runs these tools on Aida presses from 25 to 80 tons at up to 300 strokes per minute with ±0.005 mm positioning, and multi-up layouts put two, four, or six cans per stroke depending on the size, which is what makes a 5-cent consumer can possible. The process-level detail of how a progressive die builds a part station by station is in the progressive die stamping guide, and the line itself is described on the high-speed stamping service page.

Deburring is the second challenge and the one that shows up on the test bench. Every pierced hole in a can wall or flange leaves a burr, and a burr on the flange lip lifts the lid seam by its own height - 0.05 mm of burr at one point is a 5 GHz slot opening at that point. We deburr or coin the flange edge in the die and keep burr under 10% of material thickness on every feature that touches the seam or a gasket. The same discipline applies to the vent window frame, where a burr traps solder and creates a fillet that does not wet. If the part is drawn rather than folded, the drawn shell rules apply, and they are covered in the deep draw stamping guide.

Tolerance and Flatness: The Quiet Killer

Stamping tolerance is where RF performance is quietly won. A flange that warps 0.1 mm out of plane opens a gap wider than λ/20 at 5 GHz across part of the perimeter. We control can-lip flatness to ±0.05 mm, hold draw depth to ±0.03 mm, and deburr every pierced hole so a gasket seats evenly. Progressive dies with in-die cam forming let us hold these numbers at production rate. If your supplier cannot hold flange flatness, no amount of material or gasket engineering will save the seam.

Flatness is verified on the optical measurement system and CMM in the quality lab, not on a surface plate with a feeler gauge. The inspection method belongs on the drawing alongside the tolerance, because a ±0.05 mm flatness claim measured differently is not the same claim. The tolerance system for stamped enclosures is laid out in the precision stamping tolerances guide.

Prototyping and Verifying the Shield

Before tooling, the geometry and the seam strategy can be proven with a small run of parts. Prototype quantities of shield cans are stamped on soft tooling or formed on the same progressive die philosophy at low speed, and the first articles go straight to SE measurement rather than to a fit check. The measurement tells you three things: the seam strategy works or does not, the ground tab count is adequate or not, and the flatness target is achievable at production tolerances. We recommend measuring SE on prototype parts with the actual PCB and the actual solder process, because hand-soldered prototypes routinely measure 10-15 dB better than reflowed production parts - the hand joint is a better conductor than a marginal paste joint. Rapid prototyping runs use the same material, finish, and forming logic as production so the SE data transfers.

Supplier Evaluation for Shield Cans

A shield can is a precision part wearing a commodity appearance, and the evaluation list reflects that split. Ask the supplier for proof, not promises:

• Flatness data on past production lots. Flange flatness per lot, measured on optical equipment, not a one-off first-article report.
• Plating thickness reports. XRF data on the tin band and the nickel barrier, per lot, with the coating spec on the drawing.
• Burr control records. Burr height on the flange and pierced holes, held under 10% of material thickness.
• Solderability evidence. Wetting test results on the actual finish, or a documented reflow history on a similar can.
• Salt spray per ASTM B117 for any corrosion claim on the finish.
• Die maintenance discipline. Sharpening schedules and punch replacement records, because burr and flatness drift with die wear and nothing else.
• Quality system fit. IATF 16949:2016 where the can lands in an automotive or EV assembly, ISO 14001:2015 for the environmental side of the program.

The quality system behind the claims is described on the quality page, and the industries that consume most of our shield work, radio modules and sensors, are profiled under telecom and electronics. For the connector-shell side of the same problem, the connector housing stamping guide covers shells, latches, and grounding springs in one place.

Design Checklist Before You Release the Tool

• Pick material by cost and frequency: tin-plated steel for volume, aluminum for weight, beryllium copper for springs and contacts.
• Keep every continuous unsealed seam under λ/20 at your top frequency; at 5 GHz that is 3 mm.
• Target 40-60 dB SE; use welded seams and finger stock only when the band or environment demands it.
• Ground at four or more points with bare or tinned metal-to-metal contact; never rely on one corner.
• Stamp vent windows as honeycomb with cell pitch under 1.5 mm for 2.4 GHz bands, and weld or press-fit the frame.
• Specify matte tin over a nickel barrier for reflow attachment, plated after forming.
• Hold flange flatness to ±0.05 mm so the seam actually closes.
• Put the SE target band and the frequency range of concern on the drawing before tooling.

Build the Can Right the First Time

RFI problems caught at pre-compliance cost a fraction of what they cost after certification or in the field. If you are specifying a shield can, enclosure, ground tab or vent window and want the stamping and the RF behavior designed together - not a cover dropped on after the fact - send us your board outline, frequency band and volume. We will quote the tooling, recommend tin-plated steel, aluminum or beryllium copper for your case, and prove the seam strategy with SE data before you commit to a production run. Request a quote with your board outline and frequency band, or send the drawing straight to our engineering desk for a DFM and SE review.