PA12 3D Printing: 7 Mistakes That Cost You Thousands

PA12 is hygroscopic. Powder left in an open hopper pulls water out of the air, and the pickup is fast in humid weather. When the fusing lamp or laser heats that powder, the water flashes to steam inside the melt. The result is porosity you cannot see from the outside.

The mechanical cost is not small. At roughly 0.5% moisture pickup, tensile strength and elongation at break both drop, and the part becomes unreliable in any application that has to hold pressure or vacuum. You usually find out at the leak test, after a full build cycle has already been spent.

Dry the powder before every build. A dehumidifying oven at 70–80 °C for several hours is the normal routine, followed by sealed storage with desiccant. If you run one machine in-house, buy a moisture meter and use it. A reading you did not take is a rejection you will pay for later.

One more point on reuse. Refresh ratios matter as much as drying. If you keep topping up with new powder without tracking the blend, moisture and thermal history both drift, and the next build behaves differently from the last one.

• 1
  Dry before build70–80 °C for several hours in a dehumidifying oven
• 2
  Store sealedMoisture-barrier container with desiccant
• 3
  MeasureHandheld meter on every batch, not once a month

In powder-bed fusion the part is not solid material. It is a stack of fused layers, and the bond between layers is weaker than the material inside a layer. A part built flat on its largest face has a different strength and a different surface than the same part tilted a few degrees.

That matters for any feature that carries load or seals against another surface. A snap fit printed so the joint opens along the layer direction will delaminate. Rotate the part so the load runs in-plane and the same geometry survives.

Nesting is the other half. Parts packed tightly reduce cost per piece, but they also trap heat. A dense nest cools unevenly and the parts closest to the center of the build keep more residual heat, which shows up later as warpage. Orientation is a tolerance decision, not a packing decision.

Ask for the build layout before the job runs. If your supplier cannot tell you which axis is Z and why the part sits that way, the orientation was chosen for their throughput, not your function.

There is a printable minimum wall and a functional minimum wall. A 0.5 mm wall may come out of the machine. It will not survive handling, tapping, or a drop test. Powder-bed PA12 walls below about 1 mm start to behave like sheet, and thin ribs warp as they cool.

The trap is cost. Thin walls look efficient on the quote because they use less material, then fail inspection or break in the field. You pay for the build twice.

For structural parts we usually want 1.5–2 mm as a working minimum, more where the wall carries a thread or a bearing. If the design needs a 0.8 mm living hinge, that is fine, but it should be a named feature with a stated function, not an accident of the CAD model.

Bosses and threads deserve the same treatment. A printed thread in PA12 is weak. Design a pilot hole and cut the thread, or use a metal insert. The printed boss should have enough wall around the insert to resist hoop stress.

• 1
  Under 1 mmPrintable but fragile; expect handling damage
• 2
  1.5–2 mmWorking minimum for load-bearing walls
• 3
  ThreadsCut them or use inserts, do not print them

PA12 shrinks as it crystallizes and cools. The amount depends on wall thickness, orientation, and how much heat the part saw during the build. Long flat panels and large rings are the classic casualties: they come out bowed, and no amount of sanding brings them back to a flat datum.

Compensation is done in the model, before the build. The supplier scales the part and applies local compensation where the geometry is thick. This is why a step file alone is not enough information for a first article.

If a part has a critical flatness or a bore that must fit a shaft, say so on the drawing. A general tolerance note does not tell the CAM engineer which surface matters. Mark the datum, mark the fit, and let the compensation follow the function.

For parts that must stay flat, consider whether powder-bed fusion is the right process at all. A machined PA12 or POM plate holds flatness far better, and we often run both routes and compare before committing a production batch.

A part pulled hot from the build chamber looks finished. It is not. Fast cooling locks residual stress into the semi-crystalline structure, and that stress releases later, often after the customer has installed the part and torqued the bolts.

Controlled cooling means letting the build cool at a rate the material tolerates, then annealing where the application needs dimensional stability at temperature. Annealing is not a universal step, but for parts used near the upper service range of PA12 it is the difference between a part that holds its shape and one that creeps.

Be careful about the trade. Annealing can relieve stress and also move dimensions. If a bore is already at the low end of tolerance, a post-build anneal can push it out. Decide the sequence before the first article, not after.

This is also where a supplier's process discipline shows. Ask what the cooling profile is and whether the parts are annealed in a fixture. A vague answer usually means the parts come out of the chamber and go straight into a box.

As-sintered PA12 has a matte, slightly grainy surface and open surface porosity. For a bracket that is fine. For a part that seals, slides, or gets painted, it is not.

Bead blasting evens the surface and removes loose powder. Tumbling smooths it further. Sealing or impregnation closes the surface for fluid contact. Dyeing gives a consistent color, though the depth of penetration varies with wall thickness.

The mistake is budgeting for raw SLS and then discovering the finish is required for the part to function. If the drawing calls for a sealing surface, that is a machining or a finishing operation, not something the printer will deliver.

Where a tight tolerance and a smooth surface both matter, we often print near-net in PA12 and finish the critical faces on a CNC. That keeps the organic geometry of the print and gives a real Ra 0.8–1.6 μm seal face where it counts.

PA12 powder is not an anonymous commodity. Particle size distribution, thermal history, and refresh ratio all shift the mechanical result. Two suppliers using the same brand name can deliver parts that test differently.

If a part is going into a regulated product, you need traceability. That means the powder lot, the refresh ratio, and the build parameters should be recorded against the parts. Without that, a field failure has no path back to a cause.

Ask for material certificates and a retained sample. Ask how many builds the powder has been through. A supplier that reuses powder indefinitely will quote lower and deliver parts with lower elongation, and you will only see it in testing.

This is where an in-house process with documented parameters beats a broker who subcontracted the build. GreatLight runs machining and finishing under one roof, so the print, the critical face machining, and the inspection report stay in one traceable chain. We hold ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022, which matters when the part is audited.