Specifying an injection molded part in 2026 is governed by four dials — material grade, wall-thickness ratio, draft angle, and gate layout — and every one of them must be locked before the mold is cut, otherwise defects such as flow lines, sink marks, and weld lines routinely push scrap above 5% on lines running 20,000 pcs/day [S2][S3].
The decision is not which CNC shop is cheapest. It is which combination of resin family, nominal wall, draft per side, and gate type survives Design for Manufacturability (DFM) review on the first tooling iteration, and which shop has the tonnage and shot size to actually run that recipe. The reference framework below maps part geometry to material, draft, gate, and machine class so a buyer can shortlist two or three vendors in one pass instead of running ten RFQs blind.
Match Resin Family to Mechanical, Thermal, and Chemical Duty
Material selection drives mold steel, processing window, and part cost in roughly equal weight, so the first gate is a duty-based cut: commodity polyolefins (PP, PE) for non-load housings below 80 °C; ABS and PC/ABS for impact-rated enclosures in the 80–105 °C band; PA66 (nylon 6,6) with 30% glass fiber for structural parts that must hold a dimensional tolerance under heat; POM (acetal) for low-friction gears; and PEEK or PPS for under-hood automotive or oil-and-gas service above 150 °C [S2].
Glass-filled resins shrink less (typically 0.2–0.5% vs 1.0–1.5% for unfilled grades) but they are abrasive and demand hardened tool steel or HRC 60+ insert surfaces. A buyer who picks the cheapest unfilled resin for a structural boss will pay twice: once in resin and once in warped parts that fail Cpk on flatness. For chemical exposure, cross-reference the resin's resistance chart against the actual reagent at service temperature — a generic "chemical resistant" claim on a datasheet rarely survives contact with hot hydrocarbons or hot detergents.
Wall Thickness and Rib Ratios Control Sink Marks and Cycle Time
Wall-thickness uniformity is the single largest predictor of sink marks, voids, and warpage in an injection molded part; the rule of thumb that holds across the major defect catalogues is a 1.5:1 ratio between adjacent wall thicknesses, with a hard ceiling near 4 mm nominal wall for unreinforced thermoplastics and up to 6 mm for glass-filled grades [S2]. Thick sections cool from the outside in, leaving a vacuum at the core that pulls the surface inward — the visible sink mark.
When a structural rib is needed, the standard rib-geometry envelope is rib base = 0.5–0.6 × nominal wall, rib height ≤ 3 × nominal wall, and a 0.5–1° draft per side on the rib itself; ignoring the draft is the most common reason an Autodesk Inventor "Rib" feature fails to merge cleanly between two drafted features such as a tapered screw boss and an exterior sidewall, as documented on the Autodesk Inventor forum [S1]. Coring out thick sections into ribbed lattices is almost always cheaper than reducing nominal wall, because stiffness scales with the cube of thickness and only linearly with rib count, so a 0.5 mm wall reduction across 50 mm of length saves more material and more cooling time than a single rib.
Draft, Ejection, and Surface Finish as a Single Stack
Draft angle is not a polishing detail; it is the ejection release, and the 1–2° per side default for ABS/PP rises to 2–3° for glass-filled or textured surfaces, with deep cavities (depth > 50 mm or aspect ratio > 4:1) pushing toward 3–5° per side to prevent drag marks and part sticking [S2]. Under-spec the draft and the part drags, the ejection pin marks grow, and the cosmetic class drops a tier; over-spec the draft and the part looks wedge-shaped, the assembly stack-up shifts, and the snap-fit cantilever loses its expected deflection.
Surface finish drives the same stack from the other side: SPI-SPI-A2 or SPI-A3 finishes need 1–1.5° minimum draft; SPI-B2 and finer (polished mirror, optical, or clear-coat surfaces) need 2°+ because micro-scratches become visible and the polishing direction must align with the part draw. For textured surfaces (MT11000–MT13000 molded-in texture), hold 1.5° per side minimum, or the texture will shear on ejection. The cost increment between SPI-A1, SPI-A2, and a molded texture is usually smaller than the rework of a stuck or scarred part, so specifying texture is a real engineering trade, not a marketing add.
Gate, Runner, and Tooling Class Versus Annual Volume
Gate type sets both the visible witness mark and the achievable cosmetics on the part, and it has to be picked together with annual volume. Edge gates (fan or tab) are cheap to machine and trim but leave a witness at the parting line; submarine gates auto-degate but limit material flow; hot-runner systems eliminate runner scrap (typically 5–30% of shot weight in a cold runner) and improve cavity-to-cavity balance, which is why a 20,000 pcs/day production line in a multi-cavity family mold almost always specifies a hot runner [S2][S3].
Match tool steel and cavity count to the program lifetime: prototype or bridge tooling in P20 or aluminum (1 000–10 000 shots) suits pre-launch builds of 5 000–50 000 parts; production tooling in H13 or S7 (≥1 million shots) suits sustained annual volumes above 100 000 parts. The decision tree is short: under 10 000 parts/year, single-cavity soft tool; 10 000–100 000 parts/year, 2–4 cavity family mold; above 100 000 parts/year, multi-cavity hot-runner production mold, with shot weight held below 80% of machine rated shot size to leave headroom for melt homogeneity.
Process Window, Defect Diagnosis, and the First-Pass Yield Goal
The defect catalogue in any current injection-molding reference is consistent across vendors: flow lines trace back to low melt temperature, low injection speed, or an undersized gate; sink marks trace back to low holding pressure, short holding time, or thick sections; weld lines trace back to two melt fronts meeting at a cold insert or a long flow path; short shots and burns trace back to insufficient venting or excessive injection speed [S2]. The first pass on a new mold should target 90%+ first-pass yield on dimensionally critical features, and any defect rate above 5% on cosmetic class A surfaces is the trigger to re-cut the gate, re-balance the runner, or add conformal cooling.
Documenting the processing window in a single sheet — melt temperature, mold surface temperature, injection speed profile, holding pressure and time, cooling time — and stamping it on the part drawing is the cheapest QA tool available, because it lets the receiving inspector run Cpk against a known window instead of guessing. Most shops running IATF 16949 already do this, and the spec sheet is what separates a serious production line from a job shop.
Selection Criteria: 2–4 Realistic Options Compared
For a buyer comparing a 100 g housing part in ABS against PC/ABS, PA66-GF30, and PP-GF30, the criteria-based shortlist looks like this. ABS is the lowest-cost option (resin ~$1.8–2.2/kg, mold temp 40–60 °C) but softens above 100 °C and lacks chemical resistance. PC/ABS (~$2.8–3.5/kg, mold temp 80–100 °C) buys impact and heat to 110 °C at moderate cost. PA66-GF30 (~$4.5–5.5/kg, mold temp 80–100 °C) buys the highest stiffness-to-weight and 150 °C heat resistance, but it is hygroscopic and needs drying 4–6 h at 80 °C before molding, and it attacks standard screws unless they are replaced with captive inserts. PP-GF30 (~$2.4–3.0/kg, mold temp 40–80 °C) is the chemical- and moisture-resistant low-cost option, with lower surface quality than ABS. [S2]
For a buyer comparing tooling classes on the same geometry, the criteria-based shortlist runs: aluminum prototype tooling (lead time 2–4 weeks, $3 000–$10 000, lifetime 5 000–10 000 shots) for DVT and EMC builds; P20 soft steel production tooling (lead time 6–10 weeks, $15 000–$50 000, lifetime 100 000–500 000 shots) for low-volume programs; H13 hardened steel (lead time 10–16 weeks, $40 000–$200 000, lifetime >1 million shots) for sustained production. The wrong pick on this list wastes either 8 weeks of lead time or $50 000 of tool steel, and the [S3] shop profile — 100 g–4 000 g machines, 10+ machines, 20 000 pcs/day capacity, IATF 16949 quality system — is the kind of capacity envelope that should appear in the buyer's vendor prequalification.
Who Should NOT Pick the Mainstream Commodity Option
A buyer who specifies commodity PP or ABS for a structural load-bearing boss, a chemical-exposure under-hood component, or any service environment above 100 °C continuous is the profile that drives warranty scrap. The mainstream commodity option is correct for interior cosmetic housings, low-load covers, and prototyping, and it is wrong for anything that carries a fastener in shear, sees hydrocarbon exposure, or sits in a heat-soaked engine bay [S2].
Equally, a buyer who specifies soft aluminum tooling for a sustained 200 000 parts/year program is buying a tool failure: aluminum will erode at the gate and cavity surface long before the volume is reached, and cosmetic class A surfaces will oxidize and pick up. The aluminum route is the right pick for under 10 000 parts and the wrong pick beyond. The same logic holds for hot-runner systems: a hot runner is not cost-justified on a single-cavity prototype tool, and a cold runner is not cost-justified on a 16-cavity production tool running 20 000 pcs/day.
Shortlist Logic and Trackable 2026 Signals
To shortlist in one pass, run this gate: (1) material grade frozen from duty and temperature; (2) wall ratio ≤ 1.5:1 and rib geometry within the 0.5–0.6 × wall base / 3 × wall height envelope; (3) draft 1–2° per side minimum, 2–3° on glass-filled or textured parts; (4) gate type and tool class matched to annual volume; (5) processing window documented on the print. Pass all five, and two or three qualified shops will quote within a tight band; fail any one, and the RFQ becomes a roll of the dice. [S2]
Trackable signals for the next sourcing cycle: IATF 16949-certified multi-cavity capacity in the 100 g–4 000 g shot range, hot-runner-equipped production lines capable of 20 000 pcs/day, and DFM review turnarounds under 48 hours; a vendor who can show those three together is a serious candidate for 2026 production programs [S3]. For related selection logic on adjacent forming processes, see the spec-first map on die casting machine tonnage and alloys and the rubber extrusion profile selection guide, both of which share the same resin-and-tooling gate logic applied to a different forming process.
For component-level specifications, see embedded part, pressure transmitter, and flow meter.