Plastic or aluminum PCB enclosure for a small batch run

Short answer: machine the carrier in ABS for a run of 10 to 50 pieces, and switch to anodized 6061 aluminum from about 50 pieces up. The deciding costs are fixed, so the batch size settles them. Plastic cuts in one setup with no finishing; aluminum adds an anodize lot plus a slightly slower cut. In the worked example below that premium is 32 minutes per piece at 10 parts, and ABS wins. At 100 parts it is 5 minutes, and the aluminum functions pay for themselves. You get a 0.02 in [0.5 mm] web between cutouts where ABS needs 0.06 in [1.5 mm], a shell that spreads a 1.5 W hotspot to a 15 K rise instead of 120 K, and a shield across the seams. Every number hangs on one example part: a 2.5 x 2.5 x 0.5 in [64 x 64 x 13 mm] board carrier built in batches of 10 to 200.

The example carrier, in numbers

Work the arithmetic on one concrete part and the comparison stops being a matter of taste. The example carrier measures 2.5 x 2.5 in [64 x 64 mm] inside, stands 0.5 in [13 mm] tall overall, and carries a 2.4 x 2.4 in board on four standoffs. One side wall takes two cutouts, one for power and one for USB; the other three sides each take a long slot; the top has a 0.08 in [2 mm] reset hole, four screws and a 0.02 in [0.5 mm] recess for a printed label.

Add the shell area, because every heat number below hangs on it. Two faces at 2.5 x 2.5 in give 12.5 in², four sides at 10 x 0.5 in give 5 in², so the shell sheds heat through about 17.5 in² [113 cm²] of surface. Assume the board dissipates 2 W and a single regulator accounts for 1.5 W of it. Assume a combined convection and radiation coefficient of 9 W/m²K, a typical figure for a small anodized surface running about 20 K above a 25 °C ambient: roughly 5 W/m²K of free convection in air plus radiation off a 0.8-emissivity finish. Those are the worked-example inputs; substitute your board's real dissipation and your product's real ambient before you commit the design.

The two routes side by side

Both routes cut from flat plate in a single fixture. The finishing lot is the extra step you buy out of by choosing plastic; every other row is what you trade away for it.

Where machined ABS wins

Below about 50 pieces, ABS wins on simplicity. You cut every part from flat plate in one setup and ship it as it comes off the machine: no anodize lot, no masking, no second vendor. A design change costs one re-programmed pass, so you can cut, test and re-cut inside a week. For a handheld product the weight gap is real too: the example shell with 0.06 in walls weighs about 18 g in ABS against about 46 g in aluminum, from 17.5 in² of wall at densities of 1.05 and 2.7 g/cm³.

Respect one floor. The rule of thumb quoted across CNC machining guides keeps machined plastic walls at 0.06 in [1.5 mm] or thicker, and the thin-wall behaviour of small plastic housings gets its own treatment in thin plastic enclosure materials for small prototypes. Design every web between cutouts to 0.06 in and the part cuts cleanly; push a plastic web to 0.02 in and it chips, gums or snaps on the first drop. Feeds, speeds and work-holding for ABS follow their own practice, set out in cutting methods, feeds, speeds and finishing for machined ABS.

For 20 to 50 plastic pieces, urethane casting is the middle path: one machined or printed master, a silicone mold, and 10 to 20 parts a day from a bench. The choice between casting, machining and molding is laid out in the small-batch plastic process comparison.

Where anodized aluminum wins

Aluminum buys three functions plastic cannot machine in. The first is thin metal: a careful shop holds a 0.02 in [0.5 mm] web in 6061, a third of the plastic floor, so a tight USB-plus-power layout fits the wall without growing the box. Wall thickness in milled aluminum boxes has its own numbers in wall thickness design for milled aluminum enclosures.

The second is heat. Aluminum conducts roughly 800 times better than ABS, about 167 against 0.2 W/mK in handbook values, so a hot regulator spreads across the whole shell instead of scorching one patch. The full arithmetic is the next section.

The third is shielding and survival. A metal shell with bonded faces contains radiated noise; a plastic shell radiates until you add a conductive coating, one more finishing step with its own scrap rate. Hard anodize also survives keys, shop solvents and alcohol wipes that craze bare ABS, and a dash or roof mount sees cabin temperatures in summer sun that commonly reach 60-70 °C. Typical ABS datasheets list heat-deflection near 90 °C at light load, and UV embrittles the plastic over months outdoors.

Two details belong on your drawing. Anodize is an insulating film, so call out the lid seat, tapped holes and any bonding pad as bare or conductive conversion-coated where a shield or ground path matters. And slots leak regardless of material: a cutout near a quarter wavelength radiates, about 1.2 in [31 mm] at 2.4 GHz, so long side slots need a ferrite, a filtered connector or a short cable stub either way.

The heat arithmetic

Here is the worked example, on the 17.5 in² [0.0113 m²] shell and the 9 W/m²K coefficient set above:

• 2 W spread over the board, aluminum: heat leaves through the whole shell, so the wall rises 2 ÷ (9 x 0.0113) = about 20 K over a 25 °C ambient. The carrier sits near 45 °C inside.
• 2 W spread, plastic: heat reaches only the top and bottom walls directly over the board, about 11.5 in² [74 cm²], so the wall rises 2 ÷ (9 x 0.0074) = about 30 K. Workable, but 50 percent hotter.
• The 1.5 W regulator alone, plastic: it couples through roughly 2.2 in² [14 cm²] of wall above it, so that patch rises 1.5 ÷ (9 x 0.0014) = about 120 K. The part cooks in its own corner while the rest of the shell stays cold.
• The same 1.5 W, aluminum: the shell spreads it across all 17.5 in², and the wall rises 1.5 ÷ (9 x 0.0113) = about 15 K.

Same watts, an 8-times lower rise, because 17.5 in² of aluminum sheds into 8 times the area the plastic patch can reach. That area ratio, not the 800-fold conductivity gap, is the honest core of aluminum's heat advantage: the outside air film, not the wall, sets most of the resistance.

Two design rules follow. In plastic, keep any single part under about 0.3 W, which raises its local patch about 24 K, or bond a metal spreader plate under it. In aluminum, take the full 2 W spread plus a 1.5 W spot without a spreader, and bolt a hot device straight to the floor or lid so the shell becomes its heatsink.

Where the batch size flips

Treat the crossover as arithmetic on fixed steps. Assume the anodize lot adds about 5 hours of handling per batch: pack, ship, receive, inspect and mask the faces that must stay conductive. Assume aluminum also cuts about 2 minutes slower per piece on this part, with gentler depths and a deburr clean enough for the anodize tank. The 5 hours divide by the batch size:

At 10 pieces the aluminum premium is 32 minutes on every part, and no thin web, heat spread or shield pays that back on a bench project; ABS is the call. From about 50 pieces the premium is 8 minutes a piece and falling, and that is the flip: the aluminum functions now cost less per piece than the failures they prevent, a snapped web on a shipped unit, a heat-stalled regulator, a scratched cover in a customer's hands. Both materials still cut per piece at this scale. An injection mold waits for a few thousand pieces and a frozen design, so a run of 10 to 200 stays in the machine shop either way, as described for the small-batch route for beta hardware.

The decision by environment

Below 5 pieces none of this economics matters: cut ABS or print a one-off and move on to the next revision.

What to put on the order

• The internal size with its corner miter dimensioned, plus a note on whether the 0.5 in [13 mm] height is overall or internal.
• A cutout schedule dimensioned from one datum: the two power-side cutouts, the three long slots and the 0.08 in [2 mm] reset hole, each with position and size.
• The webs: 0.06 in [1.5 mm] minimum in ABS, 0.02 to 0.06 in [0.5 to 1.5 mm] in aluminum, stated on the drawing so the shop quotes what you drew.
• Finish and maskings: anodize type and colour, with lid seats, tapped holes and bonding pads called out bare or conversion-coated.
• Fasteners: four tapped holes with thread, depth and countersink. A 0.06 in [1.5 mm] lid gives about three turns of an M3 thread, enough for a light cover.
• Board mounting: tapped standoffs in an aluminum floor beat plastic bosses at these temperatures; the method is laid out in mounting a PCB to an aluminum enclosure.
• The label recess at 0.02 in [0.5 mm] deep, so a printed label sits below the surface.

Send that package through custom stainless and aluminum enclosures or configure it in the 3D enclosure builder: you set material, size and every cutout from one piece, a drawing comes back for your approval, and the shell is delivered in 10 business days from that approval. Our route covers aluminum and polycarbonate shells, and custom enclosures for small PCB runs walks a board shell through the same process.

One boundary worth stating once: these machined shells are not UL listed. NEMA types and IP codes are manufacturer ratings for gasketed enclosures, and a milled carrier carries none of them.

Catalog dimensions to compare

Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.

Questions