Mechanical × RF Co-Design · Access Points

The Aluminum Case Is an Antenna’s Worst Landlord

Published by Zukaka · Last updated August 22, 2026 · 18 min read

Wrap a radio in metal and you quietly steal decibels, shift its resonance, and bend its pattern into knots. This guide is the mechanical-and-electrical handshake that keeps a great AP looking premium without sounding bad.

Why the Enclosure Is a First-Class RF Variable

Metal enclosures look and feel right—they shield, they cool, they sell. But a metal shell does several unsettling things to the antennas inside it. Understanding those effects, and the clearances that defuse them, is the difference between a flagship product and a slow, self-interfering one.

The deeper point is that the chassis is not a passive container; it is a bound. The material choice sets a hard ceiling on how much of the radio’s power can ever leave the product. Once the industrial design locks in a full metal shell, every later choice—antenna type, placement, aperture, standoff—is an exercise in begging back some of the efficiency that the metal took away. This is why mechanical design must be treated as a primary RF variable rather than a cosmetic afterthought: it is typically the single largest variable an AP team can control, and it is the one they control last.

The consequences are rarely subtle. A ceiling-mount AP that shows a healthy EIRP on the bench can lose several decibels the moment it is assembled into its chassis, and the failure mode is so reproducible that RF engineers can predict it before the first mechanical sample arrives. Metal reflects, detunes, bleeds, and sometimes radiates in ways that have no analogy in free-space antenna data sheets. The rest of this guide turns those effects into numbers and rules you can actually design against.

What the Metal Actually Does to the Antenna

Five distinct physical mechanisms, in rough order of how often they bite:

• It reflects. The conductive wall turns an omnidirectional pattern directional; gain away from the metal drops by several decibels. A metal enclosure behind a PCB antenna behaves like a crude reflector dish that is only helpful if the pattern was going to be directional anyway.
• It detunes. Metal in the near field drags the antenna’s impedance off 50 Ω, costing mismatch loss. The matching network you tuned in free space simply stops being right once the wall is a few millimetres away.
• It bleeds efficiency. Power coupled into the box becomes ohmic heat and induced currents, not radiated signal. Efficiency is the silent killer: a radio’s link budget is set by gain and efficiency together, and efficiency is the part the enclosure attacks first.
• It swirls eddy currents. RF induces circulating currents on the metal that add ohmic loss and local heating, and those currents can re-radiate in unpredictable directions.
• It worsens VSWR. Close metal pushes the standing-wave ratio up, the net mismatch grows, and everything downstream—linearity, noise figure, effective transmit power—suffers in turn.

The numbers are telling. A 3 dBi antenna in free space can fall to near-zero gain with the box a few millimetres away, while efficiency tumbles from roughly 75% to under half, and community-verified guidance is consistent with these magnitudes: RF test measurements in production design notes repeatedly report efficiency falling by more than 50% when metal sits within the near field, and recommend a minimum spacing on the order of 5 mm. In practice, auditors often insist on far more, because the 5 mm floor recovers only the catastrophic loss, not the insidious deciBels:

Treat these as indicative rather than exact; every antenna and chassis is different, which is precisely why the design must be simulated with the real enclosure model rather than trusted to a rule of thumb. But the shape of the problem is universal: the closer the metal, the faster the fall.

Clearance: The Cheap Fix That’s Always Cheaper Than Respinning

The single most rewarding habit is keeping metal out of the antenna’s near field. Clearance is nearly free on the drawing board and prohibitively expensive after mould tooling is cut—every millimetre of distance you budget up front is worth decibels you will otherwise beg back with a $5,000–15,000 mold modification later. The rules-of-thumb that hold in practice:

Several chip vendors and module makers publish their own floors that trend the same way. Espressif, for example, recommends at least 15 mm of clearance in all directions between a module’s antenna and any metallic structural element of the housing, and explicitly includes the enclosure among the things you must keep out of the keep-out zone. Treat the datasheet floor as a beginning, not a ceiling—the datasheet is usually tuned to the vendor’s own evaluation board, not to your metal chassis.

When the industrial design simply will not give you the space, restore performance deliberately:

• Plastic antenna window — swap the metal right above the antenna for a plastic insert (about 15×15 mm for 2.4 GHz). The plastic transmits RF with only modest loss while the rest of the shell keeps its premium metal feel.
• Standoffs — lift the antenna off the board or chassis; every few millimetres returns about a decibel.
• Slot antennas — cut the enclosure itself as the radiator, the trick behind many high-end APs, and something we return to below.

Which Internal Antenna Suffers Most: FPC vs PCB vs Ceramic

Not every antenna reacts to a metal wall the same way, and the difference is a strong selection signal. The useful rule is simple: the freer an antenna is, the more clearance-sensitive it becomes, while the more embedded it is, the more tuning-sensitive.

FPC antennas: most clearance-sensitive

A flexible (FPC) antenna on a thin polyimide film couples easily to nearby ground, battery cans, and metal walls. Community and vendor guidance reports that a shift of only about 2 mm can move resonance by more than 80 MHz—enough to walk out of the 2.4 GHz band entirely. For a 20×30 mm antenna window, aim for a 6–8 mm air gap behind the foil and keep any EMI coating well away from it. FPC gives the most placement freedom in a cramped premium chassis, but it demands the most disciplined clearance budgeting.

PCB antennas: the predictable middle

A PCB antenna (IFA, PIFA, meander) performs more predictably when its edge ground cut follows the design note. IFA/PIFA needs a keep-out region on the order of a quarter-wavelength square—at 2.4 GHz that is roughly 15×20 mm or better—and the ground plane beneath a monopole-style element should extend to about 1.5× the antenna length so it can act as a proper reflector. The antenna itself must sit at the board edge with nothing—no traces, no copper fills, no silk-screen—inside the keep-out. Metal in that zone detunes it just as surely as it does an FPC, but the ground-referenced structure makes PCB antennas easier to model and re-tune than a free-floating foil.

Ceramic chip antennas: most tuning-sensitive

A ceramic chip antenna relies on a fixed ground plane and a tight tolerance on the local layout. Its failure mode is sharply different: spacing and metal errors mainly reshift the tuning of the matching network rather than collapse efficiency. The datasheet keep-out (often about 5 mm to the ground edge) is smaller than an FPC’s because the ceramic element is engineered to work on a specified ground configuration. The practical consequence is that a ceramic antenna is a good fit when board area is tight and the chassis is largely plastic, but it is the worst choice for a fully metal shell where it will detune unpredictably unless the window and standoffs are exact.

Plastic vs Metal: The Honest Material Trade-Off

The standing trade is stark. Plastic transmits RF with only modest loss but offers minimal shielding and can look cheaper; metal shields and builds beautifully but reflects nearly all the radio energy. Neither is “right–the decision belongs to product tier, market, and regulatory needs:

One often-missed detail: not all plastics transmit RF equally. ABS absorbs more RF energy than polycarbonate, and metallic or conductive paints can detune an antenna beyond easy recovery even when the base material is plastic. If you are forced into a metal look for marketing reasons, a conductive paint is the single worst choice for an RF engineer—it gives you a “plastic” shell that behaves like solid metal. When industrial design insists on metal, the RF side should involve the antenna design from the outset rather than treating clearance windows as a retrofit.

Apertures, Slots, and the Enclosure as a Radiator

Metal does not just hurt; under the right geometry it becomes the radiator. Aperture and slot positioning controls where the signal leaves the chassis, and skilful designers use it deliberately rather than suffering it accidentally.

Every opening in a metal enclosure—vent holes, seams, screw holes, a card slot—is a potential aperture through which RF can leak or couple. Long, thin openings in particular can resonate like slot antennas: a seam that is roughly a half-wavelength long at an operating band becomes a radiator whether you want it to or not, radiating unpredictably and coupling energy to nearby ports. The discipline is to make the intentional aperture the dominant one. A dedicated slot cut into the chassis, tuned to the band and fed by the antenna, turns the shell from an adversary into a high-performing, low-profile radiator—the technique behind many flagship enterprise APs that have no visible antenna at all.

Practical aperture rules:

• Keep vent and display openings away from the antenna volume unless you want them to radiate, and keep their longest dimension well under a half-wavelength at the highest operating band.
• Make intentional slots follow the pattern you need. A vertical slot radiates a pattern suited to mounting on a wall or ceiling; rotate the slot to rotate the pattern.
• Avoid random long seams. Where two metal panels join, a long narrow gap can become a parasitic slot antenna. Break it up with fasteners or overlapping flanges.
• Budget aperture location early. The opening location is a first-order control on the radiation pattern, so it belongs in the co-design review with the RF team, not in a late mechanical change order.

This is why “just cut more holes” is a poor instinct: holes are not a universal antenna fix. A correctly sized and placed slot can recover performance a window cannot, but a random array of vents can make matters worse. Simulate the enclosure with its real apertures; the pattern you see in free space is only the beginning.

MIMO Isolation Inside a Confined Metal Volume

A modern AP is not one antenna; it is a 2×2, 4×4, or denser MIMO array, and metal makes the crowding problem sharper. Every antenna couples to its neighbours through free space and through the shared chassis, and that coupling—isolation—directly caps how much spatial multiplexing the link can deliver. Poor isolation turns a supposed 4×4 array into an effective 2×2, silently halving aggregate throughput while keeping the marketing sticker intact.

The mechanisms inside a metal box: element-to-element spacing shrinks because the chassis consumes volume; the metal walls add a common ground coupling path; and reflections off the walls feed energy from one element to another. All three lower isolation and raise antenna correlation, which is the quantity that really limits MIMO gain, alongside the element efficiency itself.

Common isolation tactics inside a chassis: orient adjacent elements orthogonally (cross-polarised) to slash coupling; use the metal walls as deliberate dividers between elements; and keep symmetry so the array’s patterns stay coherent. What you cannot do is simply pack elements onto a shared ground edge and hope—inside metal, the walls do the coupling for you.

The Thermal-RF Handshake That Hides in Plain Sight

A metal enclosure is often chosen for heat, and heat is itself an RF variable. In a sealed metal box, the radio and its heatsinking are physically intimate, and the coupling is double-edged: the metal that cools also reflects, and the components you add to move heat—heat sinks, brackets, heat pipes—are often exactly the metal that detunes antennas.

Two effects compound in a real product:

• Thermal parts in the near field. A heatsink bolted next to the antenna behaves as a parasitic reflector, distorting the pattern and shifting resonance. Model the whole assembly—board, heatsink, bracket, shell—not the bare board, or your simulation will be optimistic on exactly the wrong axis.
• Temperature moving the tuning. As the chassis warms, the dielectric properties and physical dimensions shift. A matching network tuned at 25 °C at the design bench can drift when the sealed box reaches 70–85 °C in a ceiling plenum in summer. This is the same tempco discipline we explore in detail in the IIoT temperature-stability article; the enclosure magnifies it because it holds that heat in.

The co-design answer is to treat thermal and RF as one concurrent engineering problem: place the heatsink where it cools and does not crowd an antenna, verify tuning across the operating temperature range, and measure the mixed product hot—not a cold board on a lab bench. A design that passes cold and fails hot is a design that will fail in the field, because the field is warm.

A Practical Co-Design Rhythm for the Chassis and Board

Whichever material you choose, the winning move is co-designing structure and layout together, not letting one finish before the other starts. The mechanics of doing this well:

1. Define antenna locations and clearance zones first. Write them as explicit keep-out volumes in the mechanical model so the industrial designer literally cannot drop a boss or a screw there.
2. Share the model with mechanical early. The RF team needs the 3D chassis, not a photo of it. Freeze aperture and vent locations in the same review where antenna positions are approved.
3. Simulate with the real chassis model. Run the antenna and the full 3D metal shell together in the solver, including the heatsink and brackets, not the bare board in free space.
4. Place bosses, vents, and shields only where they neither feed energy out nor crowd antennas. Treat every mechanical feature as an RF feature until proven otherwise.
5. Verify with real S-parameter and pattern measurements on the first enclosure sample—before mould and tooling are committed. A functional prototype in a laser-cut box is the cheapest insurance you can buy.

The retrospective cost of skipping this is concrete. A mold modification to move a bracket or open a window runs $5,000–15,000 per change, before you re-test. Co-designing costs nothing extra on day one and saves orders of magnitude in tooling rework. The teams that ship premium metal APs on schedule are the ones that treat this rhythm as a requirement, not a nice-to-have.

A Worked Decision: The Ceiling-Mount AP

To make this concrete, walk through a realistic case. A ceiling-mount enterprise AP, a fully aluminum chassis for heat and aesthetics, four MIMO radios, and a hard requirement not to drop link margin under a $1,500 USD gate. The early decisions:

Every line is a co-design artifact: the slot answers the metal shell, the window answers the lower band, the isolation target answers the MIMO density, and the thermal line keeps heat from buying cooling at the price of radiation. The point of the worked example is that none of these decisions could have been made by the RF team alone or the mechanical team alone—each is a handshake. A team that tries to bolt a pre-designed antenna board into a finished metal chassis gets the hundred-decibel version of this story instead.

Common Interference Crimes to Audit For

Measuring the Blended Product, Not the Bare Board

A design that performs in free space is only half the story; the disciplined team verifies the assembled product. The verification set that matters:

• Impedance and tuning across the band and temperature. S-parameter measurements on the first sample, repeated at nominal and hot soak, to confirm the matching survived the chassis.
• Passive and active pattern. An anechoic or chamber pattern of the full product, not the module, so the chassis-induced directionality and gain loss are visible directly.
• Efficiency. The figure metal attacks most; verify it against the free-space baseline so the link budget reflects the shipped unit.
• MIMO isolation and throughput. A multi-stream throughput test at realistic distances, because isolation that looks fine on the bench can degrade the multiplexing that sets real-world aggregate speed.
• Radiation and noise. Because metal concentrates energy and can couple into ports, a sweep for unintentional emissions and receiver desense closes the loop on the noise floor.

The recurring theme is to measure the temperature-stable, assembled, blended product. Every time that step is skipped in favour of a clean bench measurement, the enclosure is allowed to betray the radio undetected until customers report it.

The Balanced End State

A metal AP can absolutely deliver excellent Wi-Fi—the winners simply earn it. They budget clearance up front, open a plastic window or cut a tuned slot where the antenna lives, ground the shell on purpose, keep thermal parts out of the near field, isolate the MIMO array, and simulate and measure with the real enclosure. When industrial design and RF share the same drawing and the same deadline, you get a product that looks like a flagship and behaves like one too.

Frequently asked questions

How much can a metal enclosure hurt Wi-Fi performance?

A lot, and quietly. A 3 dBi antenna in free space can fall to near-zero gain with the box a few millimetres away, efficiency can tumble from roughly 75% to under half, resonant frequency can shift by 40–80 MHz, and range can drop 40–60%. The closer the metal, the faster the fall, so the chassis is a first-class RF variable, not a cosmetic one.

How much clearance should I keep between the antenna and a metal wall?

Keep about 10 mm minimum (ideally 10–15 mm) behind a 2.4 GHz PCB antenna, about 6 mm behind a 5 GHz one, ~8 mm around each MIMO element, and about 10 mm around nearby screws and bosses. Vendors such as Espressif recommend at least 15 mm in all directions from any metal structural element, and to treat the datasheet floor as a beginning, not a ceiling.

Which internal antenna suffers most in a metal chassis?

FPC antennas are the most clearance-sensitive: a shift of only about 2 mm can move resonance by more than 80 MHz, so they need a 6–8 mm air gap and no nearby EMI coating. Ceramic chip antennas are the most tuning-sensitive and are the worst fit for a fully metal shell. PCB IFA/PIFA antennas are the predictable middle, needing a ~15×20 mm keep-out and a ground plane about 1.5× the antenna length.

How do I keep a metal enclosure without wrecking the antennas?

Budget clearance up front, open a plastic antenna window (about 15×15 mm for 2.4 GHz) or cut a tuned slot above the antenna, use standoffs to lift the antenna, ground the shell on purpose, keep thermal parts out of the near field, isolate the MIMO array, and simulate and measure with the real enclosure rather than the bare board.

Why does thermal coupling matter as much as RF clearance in an enclosure?

A metal shell traps heat in the same small volume where the antennas need clearance, so a thermal solution routed across the antenna near field acts like a detuning metal plane that also shifts resonance with temperature. Design the thermal path to stay out of the near-field region and attach hot components to the chassis at the edge, not beside the radiating elements, so cooling does not trade away RF.

Glossary

Near field

— The region close to an antenna where metal objects shift impedance and detune; roughly the first wavelength or two. Keeping conductive parts out of it is the core clearance discipline.

Detuning

— A shift in an antenna’s resonant frequency caused by nearby material, most often metal, moving the impedance off 50 Ω.

Isolation

— The degree to which one MIMO element is decoupled from another; higher isolation supports higher-order spatial multiplexing.

Correlation coefficient

— A measure of how similar two MIMO element patterns are; low correlation is required for real array gain.

Slot antenna

— A radiator formed by a shaped opening in a conductive surface, which the metal shell itself can be made to be.

Keep-out zone

— The volume around an antenna that must remain free of copper, traces, and metal structure.

IFA / PIFA

— Inverted-F and planar inverted-F antennas, the common PCB antenna types for compact Wi-Fi devices.

Efficiency

— The fraction of delivered power actually radiated; metal coupled into ohmic loss reduces it directly.

Related Reading

• Where the money goes in an access point’s RF budget
• Choosing the right WiFi module for your product
• A 4×4 Wi-Fi 6E radio for dense enterprise rooms
• How temperature stability changes an industrial Wi-Fi module choice

The figures above are engineering estimates and community-consistent values intended to inform the design process; verify against your own antenna, chassis, and measurement. Zukaka’s technical content is produced as guidance and does not replace your own RF simulation and testing.