Short answer: on polyester the antennas stay inside; on stainless they go outside, always. A plastic or GRP wall costs a cellular antenna a couple of decibels, which stings but survives on any usable signal. A stainless wall costs 100 to 10,000 times less power at the modem, and no margin survives that. On a metal box you mount each antenna outside on its own bulkhead connector, and the steel goes back to its real job of shielding the electronics. Below are the numbers by wall material, the arithmetic to check your own link, and what to put on the order.
What a decibel of wall costs you
Decibels compare powers, and the conversions you need are short. Every 10 dB takes another factor of ten off the power that reaches the modem, and the table below does the rest. Throughput follows the power down, because a cellular modem steps to simpler modulation as its signal falls. This is the whole story at mid cell.
| Wall loss | Power at the antenna | What you notice |
|---|---|---|
| 3 dB | one half | one modulation step down, about half peak speed |
| 10 dB | one tenth | slow data, dropouts at the cell edge |
| 20 dB | one hundredth | data dead except on a very strong signal |
| 30 dB | one thousandth | no service |
| 40 dB | one ten-thousandth | no service |
Near a tower you carry margin you never spend. At the edge of coverage, every dB in this table is money.
The loss by wall material
The numbers below are design values at the common cellular bands from 700 MHz to 2.6 GHz, for plastic walls of 2-4 mm and sheet metal around 1.5-2 mm.
| Wall | Typical loss | Antennas inside? |
|---|---|---|
| Polycarbonate or ABS | under 1 dB | yes |
| GRP or polyester | 1-3 dB | yes, keep margin |
| Painted steel, 304 or 316L stainless, aluminium | 20-40 dB, often more | bulkhead only |
Loss in plastics climbs with frequency, so expect close to double the 2.4 GHz figure at 5.8 GHz through the same wall. GRP runs a little worse than clean polycarbonate, because the glass fibres absorb at these frequencies. So a polyester box on a wall, with a DIN-rail router inside, sits firmly in that polyester row: measurable, budgetable, fine.
Why steel is total and plastic is not
Conductivity decides this, not thickness. At 1 GHz, radio current penetrates about 13 micrometres [0.0005 in] into stainless steel, so a 2 mm wall is more than a hundred skin depths thick and the wave has nowhere to go. Thickness stops mattering once the wall is metal; conductive shielding fabric sold to block Wi-Fi is about 0.1 mm thick and works on the same physics. What leaks through a real steel box comes from the door seam and the gasket, which is why a gasketed enclosure lands in the bottom rows of the first table instead of the total blockage a welded body would give. For the measured version of this on a stainless box, read Wi-Fi signal through a stainless steel enclosure.
Living with antennas inside a plastic box
The wall is cheap; two other things still cost you. Proximity comes first: the router hangs on a DIN rail, often beside a steel bracket or mounting plate, and an antenna folded against metal detunes and hands back a few dB you never budgeted. Stand the whip up into free space above the rail. Noise comes second: a switching power supply or a small drive on the same rail lifts the local noise floor, and the modem reports a worse SINR even with full bars of RSRP. If the polyester box also carries a battery and lives outdoors, a cellular antenna and battery inside a sealed outdoor enclosure covers the layout that keeps both dry and connected.
The link budget: a DIN-rail router at -95 dBm RSRP
Inputs, each with its source:
- Outside signal at the mounting point: -95 dBm RSRP, a fair mid-cell reading you take with a phone or the modem's status page (-80 dBm is strong; -110 dBm is where data turns ugly).
- Wall loss, GRP polyester 3 mm: 2 dB, the middle of the polyester range in the table above.
- Wall loss, 316L stainless: 30 dB, the middle of the metal range in the table above.
- Bulkhead path, antennas outside: about the polycarbonate figure, for a 0.3 m flylead and its two connectors.
- Usable floor: -110 dBm RSRP for data; most modems drop registration below about -120 dBm.
The rule in words: signal at the antenna equals outside signal minus path loss; margin equals that result minus the data floor; you want the margin at least double the loss, so an equal fade still leaves you working.
- Polyester, antennas inside: -95 - 2 = -97 dBm. Margin: -97 - (-110) = 13 dB, against a 2 dB wall. Inside works with room to spare.
- Stainless, antennas inside: -95 - 30 = -125 dBm. That is 5 dB past the -120 dBm dropout. Same router, same wall, no service.
- Stainless, antennas on bulkheads: -95 - 1 = -96 dBm. Margin 14 dB, and the wall contributes nothing because the antennas sit on the far side of it.
Decision: with 6 dB or less of wall loss and a mid-cell signal, keep the antennas inside the plastic box. In any metal box, take the bulkhead whatever the phone says today, because that wall loss erases a 13 dB margin more than twice over.
Mounting the antennas outside on an SMA bulkhead
Plan one bulkhead per antenna port. A typical DIN-rail cellular router carries two SMA ports for the radio, main and diversity, plus one or two for Wi-Fi, so count three or four holes of 6.5 mm [0.26 in], some cut with a small flat so the connector cannot spin. Have them cut when we build the box: you mark the positions on the approval drawing, and the seal at each hole is the connector's hardware rather than a field drill. Router enclosure antenna bulkhead layout and sealing works out where on the body each one goes.
Sealing lives in the O-ring under the locknut, and whether that seal is axial under the flange or radial on the shank changes which fitting suits your wall thickness; o-ring sealing for bulkhead fittings with locknuts covers the choice. Outdoors, a weatherproof SMA bulkhead connector with its O-ring seated and the nut torqued keeps rain out of the hole the same way the door gasket keeps it out of the door. Keep antennas at least 100 mm [4 in] apart; half a wavelength at 900 MHz is 165 mm [6.5 in] and buys real isolation between the two cellular ports. Disconnect the supply before you swing a wrench over the terminals; a wrench across 24 V DC still ruins the afternoon.
What the enclosure ratings say about radio
Nothing, and that is the clean answer. NEMA 250 and UL 50E describe hose, dust, rust and ice tests; IEC 60529 sprays jets and immerses. Not one of them measures radio transparency, and makers of plastic enclosures rarely publish an RF figure at all, so treat the values above as design numbers and measure your own box once. An enclosure RF shielding test that reads the attenuation of a closed box gives you the method. Shielded enclosures are built to the opposite goal on purpose, with conductive gaskets for EMC rooms; they are the wrong box for anything with an antenna inside.
What to specify
Work through the body, then the holes, then the antennas:
- Body: 304 stainless for a dry indoor panel, 316L for outdoor, coastal or washdown duty. The stainless enclosure size finder runs from 6 x 4 x 4 in to 60 x 36 x 16 in (H x W x D), and a router, its supply and a row of terminals want roughly 300 x 250 mm [12 x 10 in] of panel.
- Cutouts: three or four 6.5 mm SMA holes on the top face, at least 100 mm apart, positions marked on the approval drawing. The custom stainless cutouts page carries the drawing checklist that makes the quote fast.
- Antennas: one per bulkhead, whips clear of the wall and of each other. An SMA connector on a NEMA 4X enclosure wall covers torque and washer choice at the hole.
- Rating: every box we build is rated NEMA 4X / IP66 by the manufacturer; not UL listed.
Custom enclosures are delivered in 10 business days from your approval of the drawing. Still choosing a body? The enclosure selector filters by environment and inside size first.
The decision in one line each: plastic or GRP box with margin to spare, antennas inside, stood clear of the rail. Any steel, stainless or aluminium box, every antenna outside on its bulkhead. GPS outside as well whenever the box is metal, since its link runs on the thinnest margin of the three.
Enclosures for this job
Catalog sizes that suit the job this article answers, picked from its subject; open one for its full specifications.
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 10 × 8 × 6 in | 304 stainless | 10 x 8 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 12 × 10 × 6 in | 304 stainless | 12 x 10 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 16 × 12 × 8 in | 304 stainless | 16 x 12 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
How far should an inside antenna stay from the DIN rail?
Work to a quarter wavelength: 83 mm [3.3 in] at 900 MHz and 31 mm [1.2 in] at 2.4 GHz. Closer than about 20 mm [0.8 in] to a steel bracket or mounting plate, the antenna detunes and hands back several more dB on top of the wall loss. Where headroom above the rail is tight, an articulated adapter lets the whip lean clear of the plate instead of lying along it.
Will a viewing window in a steel door pass the signal?
A plain polycarbonate or glass window passes radio about as freely as the polycarbonate wall in the table above, so it barely registers. EMI-shielded glass is the trap, because its conductive mesh blocks as much radio as the steel around it. If the window came from a shielded or EMC product line, treat it as metal and put the antenna on a bulkhead instead.
Does the GPS antenna need to go outside too?
Yes on steel, every time. GPS arrives near -130 dBm, far weaker than cellular, so there is no margin to spend on a metal wall; under a steel lid the receiver never gets a fix. Under polyester expect slower first fixes and a few more dropped satellites. When position tracking matters, give the GPS puck its own bulkhead on the roof of the box.
How do I measure the loss through my own enclosure?
Set the modem in place and read RSRP with the door open, then close it and read again; the difference is your true wall loss, seals and hinges included. Take the weaker of two readings a few minutes apart, so a passing fade does not flatter you. A few dB means the antennas can stay; tens of dB means they belong outside.
