A consumer product team submitted a casing design: 0.5 mm walls, a 0.3 mm gap between the lid and the light guide post, and a 1.0 mm wide groove on the inner surface. All three features were printed on the test parts – two failed before acceptance, and the third failed during the first cleaning after mass production. The wall chipped when supports were removed, the gap expanded from 0.3 mm to 0.43 mm after sandblasting, and the groove became blocked during manual sanding of the exterior surface. None of these parts originally had design errors. Each failure resulted from the design not accounting for the as-shipped state, rather than the as-printed state.
All "minimums" disclosed by manufacturers describe what is "barely achievable under ideal conditions." "Margin" is what converts this "barely achievable" into "reliably delivered." Discipline is specific – thin walls, gaps, and channels each have their own failure mechanisms, and their solutions are also different.
Why Small Features Fail
Small features fail through four mechanisms; a good solution usually first identifies which mechanism is at play, then determines the numerical adjustments. Material behavior – swelling, cure shrinkage, thermal shrinkage – can cause a nominal 0.3 mm gap to shift by tens of micrometers in both directions, cumulatively exceeding the allowable fit range. Each post-processing step changes dimensions; the effects of vapor smoothing, sanding, and painting are predictable but often overlooked. Mechanics – support removal, handling, and the customer's first contact – can damage thin-walled features in ways not visible in CAD. Cleanability is the most often overlooked issue for channels and internal geometries: the print limit and the cleaning limit are two different things, the former found on datasheets, the latter learned from manufacturing experience.
| Failure Mechanism | Location | Recommended Margin |
|---|---|---|
| Cure / Sintering Shrinkage | Gaps shrink, walls shrink | Resin 0.3–0.8% of nominal; polymer powder 2–4% |
| Post-processing Thickening | Vapor smoothing, painting | +0.05–0.15 mm per side |
| Post-processing Thinning | Sandblasting, manual sanding | -0.05–0.15 mm per side |
| Support Removal Damage | Down-facing walls, rib roots | Localized chipping 0.1–0.5 mm |
| Powder / Resin Trapping | Enclosed channels, deep pockets | Nominally printable, practically unusable |
| Handling Deflection | Walls with aspect ratio > 20:1 | Permanent deformation without support |
Thin Walls Need Handling Margin, Not Process Minimums
The minimum wall thickness on a datasheet is a resolution number—describing the size that can be formed once on a test specimen under ideal conditions. It hardly describes "what the wall will experience from the build plate to the customer's hand." Every step of the journey is a load: bending during support removal, flexing during powder cleaning, compression during QA measurements, and pinching during the customer's first operation. Margin is not pessimism; it is engineering discipline that aligns the design intent with the entire manufacturing chain.
| Process | Datasheet Minimum | Recommended Design Wall Thickness with Handling Margin |
|---|---|---|
| SLA / DLP | 0.5 mm | 0.8 mm; 1.0 mm for sandblasting or coating |
| PolyJet | 0.3 mm | 0.6 mm; 0.8 mm for aggressive support removal |
| SLS / MJF (PA12) | 0.8 mm | 1.2 mm; 1.5 mm for spans > 80 mm |
| MJF / SLS Filled Nylon | 1.0 mm | 1.5–2.0 mm |
| FDM (Industrial) | 1.0 mm | 1.5–2.0 mm perpendicular to layer direction |
| DMLS / SLM (Metal) | 0.4 mm | 1.0 mm + machining allowance for critical surfaces |
Gaps: Design for the Finished State, Not the Printed State
The gap written in CAD is the gap in the "fresh out of the machine" state. Every post-processing step afterward is an opportunity for the gap to change. Vapor smoothing adds about 0.08 mm per side; painting adds 0.05–0.15 mm per coat; sandblasting removes 0.05–0.10 mm per side—cumulatively, a nominal 0.3 mm gap can easily drift by ±0.3 mm. The correct approach in CAD is to define the design gap as the "target gap after post-processing," then reverse-calculate the nominal value for the printed state.
| Interface Type | Recommended Gap (PA12 MJF Sandblasted) | If Vapor Smoothed Additionally | If Painted Additionally |
|---|---|---|---|
| Slip Fit (Lids, Housings) | 0.3–0.4 mm | + 0.15 mm | + 0.1–0.3 mm per coat |
| Snap Fit Cantilever | 0.3 mm | + 0.1 mm | + 0.1 mm |
| Pivot Hinge | 0.5 mm | + 0.15 mm | + 0.15 mm |
| Press Fit (Interference) | −0.05 to −0.1 mm (less than nominal) | Re-verify after smoothing | Avoid painting mating surfaces |
| Sliding / Linear Rails | 0.4 mm | + 0.15 mm | Avoid painting rail surfaces |
| Sealing Surface (Static O-ring) | Groove depth designed based on compression | Re-check after coating | Avoid painting inside groove |
Channels, Grooves, and Internal Accessibility
An unclenable channel is not a manufacturable feature. This is the first question for any narrow internal geometry, and also the most often skipped. The process defines the minimum cleanable size, not the minimum printable size. SLS / MJF powder needs a path to exit; SLA / DLS resin needs a path to drain; DMLS metal needs a path for powder to fall out due to gravity or vibration. If a channel is printed but cannot be cleaned, the result is added weight (trapped powder), bleed-out after post-processing (trapped resin), or outright blockage.
| Process | Min. Print Diameter | Min. Cleanable Diameter | Required Drainage / Venting |
|---|---|---|---|
| SLS / MJF (Polymers) | 0.8 mm | Through hole 1.5 mm, blind hole 3 mm | 2 ≥ 4 mm powder escape holes per enclosed volume |
| DMLS / SLM | 0.4 mm | 1.0 mm + vibration-assisted powder removal | Powder removal direction + escape port |
| SLA / DLP | 0.5 mm | 1.0 mm + drainage | ≥ 3 mm drainage hole at lowest point |
| FDM | 1.0 mm | 1.5 mm | Supports must be soluble or accessible |
| Carbon DLS | 0.5 mm | 1.5 mm + drainage | Drainage needed for both washing and curing stages |
| PolyJet | 0.3 mm | 1.0 mm + accessible for jetting | Support material flush path for every channel |
Oversize Dimensions: Geometry Larger Than the Build Plate
Dimensionally limited designs can also be limited in reverse – the part itself is larger than the build plate. The solution is usually not to shrink the design, but to split it at structurally reasonable locations, print in two (or more) segments, and then reassemble using deliberately designed connectors. Adhesives, mechanical fasteners, and molded inserts each have their applicable scenarios; the key is to treat the connectors as design features, not as realities discovered after printing.
| Joining Method | Strength Relative to Base Material | When to Use | Trade-offs |
|---|---|---|---|
| Two-part Epoxy | 60–80% | Non-structural joints, aesthetics | Seam visible; requires surface treatment |
| Dovetail / Keyed Interlock + Glue | 80–95% | Structural joints requiring alignment | Keyed surfaces require precise printing |
| Threaded Inserts + Screws | Depends on screws | Repairable / disassemblable | Adds hardware and cost |
| Ultrasonic Welding (Polymers) | 70–90% | Mass production consumer parts | Only suitable for weldable thermoplastics |
| Snap Fit + Secondary Pin | 50–80% | Prototype assembly, low load | Not suitable for cyclic loading |
| Press Fit Across Seam | Metal 40–70% | Metal parts requiring field-disassemblable seams | Requires machined tolerances; difficult alignment |
Application Cases
Sealing Flange with Margin Found After Two Failures
A medical device housing required a 40 x 25 mm sealing flange with a 0.3 mm static gasket compression groove. The first version of the groove printed precisely to 0.30 mm; after sandblasting, it measured 0.38 mm—too loose, failed pressure test. The second version used a 0.26 mm nominal groove; after sandblasting, it was 0.33 mm, but the surrounding 0.6 mm wall chipped during support removal, losing the sealing edge. The third version had a 0.28 mm nominal groove (0.34 mm after sandblasting), the sealing wall was increased from 0.6 mm to 1.0 mm, and the orientation was chosen so supports would not contact the groove or surrounding wall. It passed the pressure test, with zero failures out of the remaining 60 parts. None of the three iterations were solved by making features smaller—all were solved by retaining sufficient margin to allow the features to survive the entire manufacturing chain.
Key Design Action: The process datasheet said 0.5 mm walls were feasible. The actual lower limit for this application—after sandblasting and support removal—was 1.0 mm. Following the datasheet would mean specifying the best-case scenario that "cannot withstand our own post-processing chain."
Housing Split in Half to Fit the Build Plate
The drone fuselage was 520 mm long, exceeding the MJF 380 mm build envelope. The team did not switch to FDM (which could print it, but the surface layer lines were unacceptable). Instead, they split the fuselage in half at the bulkhead location, which was already stiffer than the surrounding shell. The joint used 5 mm keys and a 15 mm interlocking dovetail, then bonded with a 25 MPa tensile strength two-part structural epoxy. The joint showed no separation during testing at 8 G fuselage load. The seam was visible on the exterior, but industrial design integrated it into the design—a painted decorative line was added along the seam, treated as an intentional design element rather than a concealed defect.
Cooling Channel That Could Only Be Mass-Produced After Adding a Powder Removal Hole
A conformal cooling channel in a 3D printed mold core, spanning 90 mm with a diameter of 1.2 mm—within the SLS printing limit, but far below the cleanable limit. The first part arrived with the channel full of trapped powder, which could not be removed by sandblasting, ultrasonic cleaning, or manual probing. The solution was to drill a 4 mm powder removal hole in the mold backplate, extending into the channel from the middle; in the next build, powder was blown out from both ends through this hole in less than a minute. This hole was placed in a position that did not interfere with the mold's function, and the design change did not delay delivery. The only change was the recognition that the true lower limit is the cleanable limit, not the printable limit.
Do / Don't Comparison
| Do | Don't |
|---|---|
| Provide handling + post-processing margin for walls, gaps, and channels | Design to process minimums and hope for the best |
| Specify which post-processing state the gap corresponds to | Leave gap state ambiguous |
| Design drainage paths for every enclosed internal geometry | Assume the print shop's cleaning can reach anywhere |
| Split oversize parts at structurally sound locations | Redesign the entire geometry to fit a smaller bed |
| Verify critical features on actual mating hardware | Verify critical features only on isolated test coupons |
| Re-check gaps after vapor smoothing or painting | Assume post-processing is dimensionally neutral |
| Orient parts so supports do not contact critical features | Allow auto-orientation to place supports on sealing/mating/critical aesthetic surfaces |
Common Mistakes and How to Avoid Them
| Mistake | Why it Fails | How to Avoid |
|---|---|---|
| Wall thickness based on datasheet minimum | Prints fine, but chips during support removal or handling | Add handling margin; the table in this document provides practical defaults |
| Gaps not specified with post-processing state | 0.3 mm as-printed and 0.3 mm after smoothing are two different parts | Specify "gap after post-processing" on the drawing |
| Expecting to clean an 0.8 mm channel just because it printed | Printable ≠ cleanable | Design to the cleaning limit; add drainage paths |
| Stacking multiple near-limit gaps in one assembly | Each gap drifts; cumulative stack-up fails | Only keep the critical one tight, loosen others |
| Shrinking geometry to fit a smaller bed | Wall proportions and fits become illogical when scaled down | Split at structural joints |
| Painting over critical gaps | Each coat adds 0.05–0.15 mm; consumes the gap | Mask off gap surfaces before painting |
Pre-Print Checklist
Run through this once when CAD is finalized, and again before releasing for mass production. Each item takes 30 seconds and can prevent a class of failures that "blow up days later."
- Every wall has handling margin above the datasheet minimum; critical walls include post-processing allowance
- Every gap specifies the corresponding post-processing state (as-printed / sandblasted / smoothed / painted)
- Every enclosed internal volume has a drainage path designed for the cleaning limit (not the printing limit)
- Channels longer than 40 mm have two drainage paths, or a verified orientation favorable for drainage
- Oversize parts are split at structurally sound joints, not on aesthetic surfaces
- Support orientation has been checked against critical features; support contact does not fall on sealing / mating / critical aesthetic surfaces
- Stacked near-limit gaps have been resolved—only the critical one remains tight
- For painted or coated parts, mating / sealing surfaces are masked or specified as bare
Design Takeaways
Small feature failures are often caused by factors not visible in CAD: shrinkage, post-processing deviations, handling loads, and cleanability. Useful discipline involves designing for the as-shipped state rather than the as-printed state, adding handling margin above datasheet minimums, incorporating cleaning limits into channel design, and treating joints as design features rather than remedies. None of these require redesign—only incorporating what truly happens in the manufacturing chain during the design phase.