Binder Jetting vs. CNC for Low-Volume Stainless Steel Housings

CNC machining can hit 0.002 mm on a good day, though most shops quote stainless work at ISO 2768-m, which lands at ±0.1 mm as the standard baseline. Binder jetting's typical quoted tolerance is ±0.2 mm, already twice as loose before anyone touches the real culprit: shrinkage. Print resolution isn't what drives the error budget on a binder jetted part. Shrinkage is, and most people evaluating BJ against CNC are looking at the wrong number entirely.

Bronze-infiltrated parts under 75 mm shrink somewhere between 0.8% and 2%. Larger infiltrated parts creep closer to 3%. Sintered parts, which is the route most 316L structural housings take, shrink around 20%. That's not a rounding error you correct with a scaling factor. That's a fifth of the part vanishing between the green state and the finished one.

And it doesn't shrink evenly. Sintered 316L pulls in 18.5% to 19.5% along Z (the build axis) and 16.5% to 17.5% in XY, and even X and Y drift apart from each other by half a percent to 0.8%, since one follows powder spreading and the other follows binder injection. A housing with two bores that need to line up, or a flange that has to seal flush, can't get scaled by a single number and called done. Every axis needs its own correction, worked out on its own.

Put it in shop-floor terms: sintered BJ parts land in broad tolerance grades well above what functional interfaces demand. Functional interfaces, the bores and sealing faces that mate with something else, typically demand significantly tighter grades than sintering alone can deliver. That's not a near miss, that's a different grade of part entirely, and no amount of vendor optimism changes the physics.

None of this matters for a wall that just has to exist and stay out of the way. There, the ±0.2 mm gap is noise nobody will ever measure. But on bore spacing, thread bosses, gasket faces, and mounting patterns, CNC's ±0.1 mm baseline is a real structural edge, not a marketing line, and pretending otherwise is how first-article inspection turns into a redo. Simulation-based pre-compensation can close some of the sintering gap, but that's engineering labor, not something a service bureau just hands over for free. The workaround most shops actually use, print the body in BJ and machine the critical interfaces after, only works if it's decided at the design stage. Discover the problem after the print comes out of the sintering furnace, and the fix costs a lot more than the plan would have.

How surface finish requirements land differently on each process

As-sintered 316L comes off a binder jet printer rough, Ra 15 to 20 microns, well past what most housing exteriors are supposed to look or feel like. Bead blasting knocks that down to Ra 3 μm, with general post-processing typically landing around Ra 6 μm. CNC-machined stainless, meanwhile, comes off the machine at Ra 3 μm or better with no secondary step at all. That gap alone should settle most exterior finish arguments before they start.

BJ does hold one real advantage: on internal geometry no tool or laser can reach afterward, it starts at Ra 6 μm, against Ra 12 to 16 μm for laser powder bed processes like DMLS or SLM. So against other metal AM methods, BJ wins on interior surfaces. Against CNC, it's still rougher, just less rough than the laser alternatives.

Match the finish call to the actual feature, not the process brochure:

• External cosmetic surfaces typically require a fine finish, Ra 3 μm or better. BJ gets there only with bead blasting, at minimum.
• Sealing surfaces and O-ring grooves generally need a finish considerably finer than bead-blasted BJ can deliver. Neither as-sintered nor bead-blasted BJ touches that. Secondary machining isn't optional here, it's the whole plan.
• Internal fluid or gas passages might tolerate BJ's Ra 6 μm depending on what's flowing through. CNC bores get smoother, but only where a tool can physically reach in.
• Threads come off a CNC machine ready to use. BJ can print stainless threads too, but the pitch diameter has to get checked after sintering, since shrinkage doesn't spare threaded features any more than it spares anything else.

Neither process finishes every surface on a housing without some secondary work. Anyone who tells you otherwise hasn't run the part through inspection. The real question is which secondary steps a given design can absorb, and which ones it can't skip.

The geometry types where binder jetting's freedom is genuinely worth the tolerance trade-off

Binder jetting doesn't need support structures. The powder bed holds the part up during the print, layer by layer, which opens the door to internal cavities, undercuts, and re-entrant shapes that would demand a support strategy, and the removal work that comes with it, in nearly any other process. Build volumes run up to 800 x 500 x 400 mm, and since parts aren't fused to a build plate, the whole volume can get packed with multiple housings at once. That packing is the real cost lever on low-volume runs, more on that below.

Where does the freedom actually earn its keep on a housing? Four cases, and only four:

• Internal channel networks, cooling paths or fluid routing, that no tool could reach from any angle on a CNC setup.
• Hollow walls or lattice-filled sections for weight savings, as long as the wall thickness survives sintering without warping.
• Curved, organic exterior shapes that would otherwise demand 5-axis CNC and several setups.
• Consolidated designs, where several machined-and-welded pieces become one printed body.

Where it isn't worth it: flat, broad parts with uneven wall sections warp during sintering, plain and simple. Large solid cross-sections build up shrinkage gradients between thick and thin regions, which compounds error rather than averaging it out. Any feature longer than roughly 50 mm carries more post-processing dimensional risk the longer it runs.

Orientation still matters even with no supports to plan around, since the build direction carries the heaviest shrinkage, and that axis needs to line up with whatever dimension can afford to move. Here's the rule worth actually following: if the housing can be made in three or fewer CNC setups without special tooling, binder jetting's geometric freedom isn't buying enough to offset what it gives up on tolerance and finish. Most housings that "could go either way" belong on the mill, full stop. The ones where BJ genuinely wins are the ones CNC literally cannot cut, not the ones where BJ is merely a different way to cut them.

How quantity and cost structure interact to shift the crossover point between the two processes

CNC's cost is mostly cutting time plus a setup charge spread across however many parts get made. Setup cost doesn't change whether the order is for 1 part or 50, so per-part cost drops as volume climbs, but it flattens out fast once setup is paid off. Stainless also cuts slower than aluminum, which pushes per-part cost up across the board no matter the batch size.

Binder jetting's cost comes from machine time, powder ($50 to $200 per kilogram), machine rate ($100 to $300 per hour), and sintering. The number that actually moves the needle, though, is bin utilization: how many parts fit in one build. Pack the chamber and cost per part drops fast. Leave it half-empty and BJ loses its whole advantage, no matter how good the per-part economics look on paper.

BJ builds at up to 50 cm³ per hour, against 5 to 10 cm³ per hour for laser powder bed fusion, a comparison that matters more against other AM methods than against CNC. What matters more here is total cycle time: BJ runs 7 to 15 days once sintering gets folded in, while a capable CNC shop can typically turn a low-volume stainless housing with standard features considerably faster.

So where's the actual crossover? BJ's cost edge shows up most clearly on batches up to around 1,000 units, once the build is packed efficiently. For small runs of one to five complex parts, CNC's setup cost still holds its own, provided the geometry is machineable in the first place. And BJ carries zero tooling cost, which matters when a design is still likely to change: revising a CNC fixture costs real time and money, while reprinting a BJ part costs nothing beyond the reprint.

One more path worth naming: for box-like housings with even wall thickness at mid-volume, sheet metal fabrication competes with both processes on price. Not always on the table, but real when the form allows for it, and often overlooked because the conversation defaults to AM versus CNC.

The DFM decisions that determine which process a housing design actually belongs in

CNC rules that matter specifically for stainless housings:

• Internal corners need a radius. No end mill cuts a sharp inside corner, so specify a radius that matches a cutter that actually exists.
• Deep, narrow pockets limit cutter diameter and chip clearance, and stainless's stringy chip behavior makes this worse than the same cut in aluminum.
• Thin walls need lighter cuts and sometimes fixturing support, so set a wall thickness floor early instead of finding out during the first run.
• Tolerance only where function demands it. Over-tolerancing a stainless housing adds cost without adding anything useful. Stack several difficult features on one part and costs can swing 15% to 800%, with lead times stretching 25% to 1,480%. Handled well instead, DFM can cut manufacturing cost 15% to 40% and lead time 25% to 60%, and that leverage is highest at the concept stage, before drawings ever go out.
• Call out surface finish only on sealing faces, mating bores, and visible surfaces, not the whole part.
• Stick to standard drill sizes, thread forms, and common dimensions. Anything else triggers special tooling charges.

BJ rules that matter specifically for stainless housings:

• Orient the part so the build direction, the axis with the heaviest shrinkage, lines up with whichever dimension matters least.
• Avoid broad flat sections or sudden wall thickness jumps. Both invite warping during sintering.
• Decide up front which features get machined after sintering: bore diameters, thread bosses, sealing faces, and add stock for that machining from the start, not after the fact.
• Run simulation-based pre-compensation on any housing with more than one critical dimension. Don't gamble on the first article.
• Design revisions to nest together in the same build. That's where the bin-packing cost advantage actually comes from, and it disappears if every revision gets its own run.

The hybrid path, printing in BJ and finishing critical surfaces on a CNC machine, is a legitimate production method, not a stopgap for a process that fell short. But it has to get designed in from the start: surfaces slated for machining need extra stock, fixture points need to be defined on the model, and the order of operations changes total lead time. A DFM review at the quoting stage, not after the first part comes off the printer, is what catches tolerance stack-up before it turns into a remake.

A decision framework for choosing between binder jetting and CNC for a specific stainless housing

Four factors, worked through in order.

Geometric complexity. If the housing fits in three or fewer CNC setups, with a mostly prismatic or simply curved shape and no internal cavities, CNC is the default, full stop. Tolerance, finish, and lead time all point the same direction. If the design has internal channels, an organic form, re-entrant features, or consolidates several parts into one, binder jetting deserves a serious look, and probably the job.

Dimensional tolerance. A tolerance tighter than ±0.2 mm needs CNC, or BJ with post-machining built into the plan. BJ alone can't hold that reliably, not close. Sealing surfaces, precision bores, and threaded interfaces should get planned for machining no matter which process makes the rest of the part. If ±0.2 mm is fine on a non-mating structural feature, BJ's tolerance was never actually a problem in the first place.

Surface finish. Ra 3 μm or better on an external surface means BJ needs bead blasting at minimum, and any sealing surface needs secondary machining regardless of which process made it. Internal passages that no tool can reach might be the one spot where BJ's Ra 6 μm, or Ra 3 μm after blasting, is the only practical option short of EDM or specialized tooling.

Quantity and schedule. For one to five parts with machinable geometry and tight tolerances, CNC wins on speed (4 to 5 days against 7 to 15 for BJ with sintering) and on cost predictability. For ten to several hundred parts with genuinely complex geometry, that's where BJ's bin-packing economics and freedom from tooling costs start to beat CNC's per-setup charges, as long as the tolerance and finish requirements can actually live with what sintering delivers.

Run a housing through those four factors in order, and the right process is usually obvious before the quote even comes back. The mistake worth avoiding: picking the process first, out of habit or because one vendor picked up the phone first, and only checking tolerance and finish after the tooling's already been ordered. By then the fix is a remake, not a revision, and remakes are the most expensive lesson in this entire comparison.