Guide to Multi-Jet Fusion MJF 3D Printing

• 1
  How Multi-Jet Fusion MJF 3D Printing Actually Works
• 2
  What PA12 and PA11 Give You
• 3
  Design Limits You Should Check Before You Send Files
• 4
  Tolerances and Surface Finish in Practice
• 5
  Where MJF Fits Against CNC and Injection Molding
• 6
  MJF vs CNC vs Injection Molding

The process runs in a powder bed. A recoater blade spreads a thin layer of nylon powder across the build bed, typically 0.08 mm thick. An inkjet array then prints a fusing agent onto the cross-section and a detailing agent around its perimeter. An infrared lamp passes over the bed.

Where the fusing agent sits, the powder absorbs infrared energy and melts. Where the detailing agent sits, the powder stays cool and resists melting, which keeps edges sharp. The bed drops by one layer thickness, fresh powder rolls on, and the cycle repeats. No support structures are printed.

The unsintered powder does the supporting. A part with a floating boss or an internal channel is held in place by the surrounding powder cake. That is why MJF can nest parts vertically and pack a build volume far more densely than fused deposition modeling.

After the build, the cake cools in the machine. The operator breaks it out, bead-blasts the parts, and recovers the loose powder. Roughly 80–90% of the powder is sieved and reused, which is one reason the process holds up at volume.

• 1
  Layer thickness0.08 mm is standard; 0.10 mm is faster but shows more stair-stepping.
• 2
  AgentsFusing agent absorbs IR; detailing agent suppresses edge melting.
• 3
  Powder reuseUnsintered powder is sieved and blended back into the feed.

PA12 is the workhorse. It is a semi-crystalline nylon with low moisture uptake, so parts hold dimensions in a humid shop. Tensile strength sits around 48 MPa, elongation at break around 15%, and it survives repeated impact without cracking. Those numbers are isotropic in the XY plane and slightly weaker along Z.

PA11 is bio-based and more ductile. Elongation at break runs higher, often above 30%, so it suits living hinges, snap-fit latches, and clips that flex thousands of times. It costs more and is less widely stocked.

Glass-bead-filled PA12 raises stiffness and heat deflection temperature. It is the choice for brackets that must hold shape under a warm hood or near a motor. The trade-off is brittleness: a glass-filled part snaps where a neat PA12 part would bend.

TPU grades cover the soft end. They print with the same agent chemistry but give a rubbery feel, useful for grips, gaskets, and vibration pads. Shore hardness depends on the grade, so specify the number you need rather than saying soft.

• 1
  PA12Balanced strength and toughness; the default for functional prototypes.
• 2
  PA11Higher elongation; use for hinges and repeated flexing.
• 3
  Glass-filled PA12Stiffer and more heat-resistant, but more brittle.

Wall thickness is the first thing to verify. A 0.8 mm wall prints but flexes and may warp during cooling. Walls of 1.0–1.5 mm give a stiff, predictable part. Anything under 0.6 mm may not form at all.

Minimum feature size is around 0.4 mm for a raised rib or a slot. Small text engraved into a face needs at least 1.5 mm character height to stay legible after bead blasting, because the blast rounds the top edges.

Holes print undersized. A nominal Ø6 mm hole often comes out near Ø5.7 mm. Design the hole at Ø6.3 mm if you need a clearance fit for an M6 bolt, or drill and ream after printing if you need a precise bore.

Large flat panels warp. A 200 mm × 200 mm plate will bow as it cools. Adding ribs or a slight crown controls the distortion. If the part must stay flat, plan a secondary machining pass on the mating face.

• 1
  Wall thickness1.0–1.5 mm is safe; below 0.6 mm may not form.
• 2
  Text heightKeep at least 1.5 mm so bead blasting does not erase it.
• 3
  Hole sizingAdd roughly 0.3 mm to the diameter for a clearance fit.

MJF holds about ±0.3 mm on a well-oriented part, or ±0.3% of the dimension, whichever is larger. On a 300 mm part, that means the tolerance band widens to about ±0.9 mm. The build chamber itself adds variation: parts near the edge of the bed cool slightly differently from parts in the center.

The as-built surface is grainy, around Ra 10–15 μm. Bead blasting brings it down to Ra 6–10 μm and removes the loose powder skin. That is fine for most brackets and housings, but it will not pass for a sealing face or a bearing bore.

For tight features, print oversize and machine. Adding 0.5–1.0 mm of stock on a critical face lets a 3-axis mill face it to ±0.005 mm and Ra 0.8–1.6 μm. Hybrid workflow costs less than a full CNC part when the geometry is complex and the tight features are few.

Dyeing changes color but not dimensions. A dye bath adds a thin surface layer, so recheck a press-fit after dyeing if the fit is already at the low end of the tolerance.

• 1
  General tolerance±0.3 mm or ±0.3%, whichever is larger.
• 2
  As-built finishRa 10–15 μm; bead blasting reaches Ra 6–10 μm.
• 3
  Hybrid approachLeave 0.5–1.0 mm stock and machine critical faces.

MJF has no tooling cost. That makes it the cheaper route from one part to a few hundred. The build is packed with many parts, so the cost per part drops as the build fills. A single part pays for a lot of unused powder volume.

CNC wins when you need metal, a mirror finish, or tolerances below ±0.05 mm. It also wins when the part is a simple shape that a mill can cut in one setup, because there is no powder handling or post-processing.

Injection molding wins above a few thousand parts, once the tool is amortized. Between roughly 100 and 2,000 parts, MJF usually beats molding on total cost, especially if the design may still change.

The decision is not permanent. Many programs print MJF parts for functional testing, then cut a mold once the design freezes. The same CAD file feeds both routes.

• 1
  1–500 partsMJF is usually the lowest total cost.
• 2
  Tight tolerance or metalCNC, or MJF plus a machining pass.
• 3
  Above 2,000 partsInjection molding starts to win on unit price.