What level of dimensional accuracy can MJF parts achieve?
MJF dimensional accuracy is often screened against supplier ranges near +/-0.3% with an absolute floor near +/-0.2 mm for selected PA12 routes, but no single range applies to every machine, material, orientation, size, or feature. That published range is a quotation aid, not a tolerance guarantee for holes, slots, thin walls, flatness, position, snap fits, or mating datums. The drawing should identify critical dimensions and geometric controls, the datum scheme, final finish, conditioning state, and inspection method. The supplier can then separate as-printed features from compensated, machined, or sample-validated features before release.
Typical Dimensional Tolerance
A percentage-plus-floor rule can estimate an early dimensional window, but it does not replace feature-level capability or drawing acceptance. The calculation and its limits should both be visible in the quotation.
For a 100 mm dimension, a +/-0.3% screening calculation gives 99.7 to 100.3 mm. That arithmetic does not prove the result: wall distribution, orientation, bed location, cooling, finishing, datum setup, and measurement uncertainty still determine whether the feature can be accepted.
For a 10 mm dimension, an absolute +/-0.2 mm screening floor gives 9.8 to 10.2 mm. A 10 mm external span and a 10 mm hole do not share the same risk; fusion around an internal edge, retained powder, and measurement access can change the hole result.
General size tolerances can support covers, ducts, fixtures, and selected parts for industrial equipment or consumer products. Position, flatness, profile, coaxiality, and sealing fit require a datum reference and a suitable measurement method. ISO/ASTM 52902 uses benchmark geometries to evaluate additive-manufacturing system capability; it does not guarantee the tolerance of an arbitrary production part.
Key Factors Influencing MJF Accuracy
MJF accuracy is governed by thermal history, material and powder condition, geometry, orientation, cooling, post-processing, and measurement state. Machine resolution or layer thickness alone cannot predict delivered-part tolerance.
Thermal gradients and cooling: Fusing and detailing agents control energy across the powder bed, but long flat spans, uneven mass, dense nesting, edge location, and cooling can still create shrinkage or warpage. A part can meet overall length while failing flatness or hole position relative to a datum.
Direction and compensation: MJF PA12 can show balanced behavior compared with strand-deposition printing, but X, Y, and Z results are not identical for every geometry. Compensation must be qualified for the named material, orientation, feature class, nesting map, and final conditioning state. Changing orientation can invalidate earlier dimensional evidence.
Feature geometry and powder interaction: Small holes can close or retain powder, short pins can grow or break, sharp edges can soften, and thin walls can move during cooling or cleaning. Fine powder supports detail, yet particle size does not by itself establish accuracy, minimum wall, or surface acceptance.
Practical Guidelines and Design Considerations
Accuracy improves when the drawing assigns each critical feature to an as-printed, compensated, post-processed, or machined route and defines inspection after the last dimension-changing operation.
Critical Features: Bearing seats, sealing faces, locating bores, insert pockets, and precise mating datums may need CNC Machining. The printed blank needs accessible datums and sufficient machining stock. Acceptance should reference the machined state, not a pre-machining print measurement.
Holes and Shafts: Internal holes, slots, and channels react differently from external pins or bosses. A critical diameter may need supplier compensation, a gauge-based first article, or post-process drilling and reaming. The RFQ should state fit, depth, through/blind condition, axis, datum, and final inspection tool.
Wall Thickness: Thin walls, tall ribs, broad panels, and abrupt thickness changes can distort or be damaged during depowdering. A supplier-specific minimum wall is only a printability screen. Functional release also needs stiffness, handling, flatness, and any assembly-load criteria in the final conditioned part.
Surface Finish: Blasting, dyeing, smoothing, coating, and tumbling can change edges, small openings, sealing surfaces, and measured size. The effect is not automatically negligible. Masking, process allowance, final-state measurement, and surface-texture acceptance should be agreed before finishing.
Comparison with Other Processes
MJF should be compared with other processes by the critical feature and final function, not by a single advertised accuracy number. Size, material, surface, geometry, quantity, and inspection burden can change the preferred route.
Vs. FDM: MJF can provide more consistent fine-feature and surface behavior for many nylon parts, while FDM may support larger or lower-cost coarse geometries. Either process still needs orientation-specific evidence for critical dimensions and mechanical loading.
Vs. SLA: SLA can provide fine detail and smooth surfaces, but resin grade, post-cure, aging, brittleness, and environment affect functional release. Choose SLA for its qualified feature and material behavior, not a generic claim that it is always more accurate.
Vs. SLS: MJF and SLS accuracy overlaps across many polymer powder-bed applications. Machine platform, material, thermal control, part orientation, cooling, surface route, and supplier compensation matter more than the process name. A first article under the planned production nesting is the useful comparison.
An MJF RFQ should include the STEP model and controlled drawing, exact material and color, quantity, critical datums, size and geometric tolerances, hole and slot details, fits, cosmetic surfaces, final finish, conditioning state, inspection stage, gauge or measurement method, sampling, and permitted machining allowance. The supplier should report which values are screening ranges and which features have route-specific capability or first-article evidence. Final acceptance must use the same orientation, nesting logic, cooling, post-processing, datum alignment, and measurement state planned for production.
