You need a precision injection molding supplier when a plastic part has tight dimensional limits, small mating features, difficult engineering resins, demanding assembly requirements, or production volumes where minor variation becomes expensive. A 0.05 mm dimensional shift may be acceptable on a large housing but unacceptable on an 8 mm connector, fluid channel, gear, or sealing feature. At 1 million parts per year, even a 2% reject rate produces 20,000 unusable pieces. Precision molding is mainly about keeping dimensions and material behavior repeatable across thousands or millions of cycles, not simply machining a more accurate mold.
A useful starting point is the tolerance on the drawing. Plastic parts do not leave a mold at the exact cavity dimensions because polymer contracts during cooling. ISO 294-4:2018 specifically addresses molding and post-molding shrinkage in directions parallel and perpendicular to melt flow, showing why shrinkage cannot always be treated as one uniform number. A nominal 25.00 mm feature with ±0.05 mm tolerance has only a 0.10 mm total acceptance window, so tool dimensions, packing pressure, mold temperature, resin condition, cooling time, and measurement method all need enough consistency to stay inside it.
That requirement becomes harder when several dimensions work together in an assembly. A housing may pass individual dimensional checks but still cause excess insertion force if a locating boss, pin hole, wall, and mating surface all shift toward the same side of their permitted ranges. If four dimensions can each vary by ±0.05 mm, a simple worst-case stack can reach 0.20 mm. Product engineers therefore need to identify which dimensions affect fit, sealing, alignment, electrical contact, or movement instead of placing unnecessarily tight tolerances across an entire drawing.
A precision supplier should ask what a dimension controls in the finished product, not only whether the mold shop can machine that dimension.
Material choice adds another layer because shrinkage and dimensional stability differ widely among polymers. Semi-crystalline materials such as POM, PA, PBT, PPS, and PEEK generally require more attention to cooling and crystallization than many amorphous resins. Fiber reinforcement adds directional behavior because fibers tend to align with flow. A 30% glass-filled grade, for example, can respond differently along and across the melt-flow direction, so a single published shrinkage figure is rarely enough to predict every feature on a complex molded part.
Moisture also matters with hygroscopic polymers. Nylon and several engineering resins must be dried according to the material supplier’s processing recommendations because excess moisture can affect melt behavior and, for susceptible polymers, contribute to molecular degradation during processing. A factory making a few decorative covers may tolerate wider variation, while an Industrial plastic molding manufacturer producing precision assemblies should document drying temperature, drying time, material exposure, regrind policy, lot identification, and machine settings when those conditions affect the finished dimensions.
Machine size needs similar attention. A small part does not automatically belong in any available injection molding machine. If the shot uses only a very small portion of barrel capacity, material may remain in the barrel longer than intended, while repeatability can become more difficult to maintain. A 2 g part molded with an 8-cavity tool requires only about 16 g of part material before runners are counted, so machine selection should consider total shot size, residence time, clamp force, injection pressure, screw geometry, and the resin’s recommended processing range.
Small components make the relationship between absolute error and feature size easier to see. A 0.05 mm deviation equals only 0.05% of a 100 mm feature, but it equals 2.5% of a 2 mm feature. Miniature connectors, diagnostic components, sensor bodies, small gears, fluid passages, and optical mounts therefore often need better control of mold alignment, vent depth, gate dimensions, cavity balance, ejector position, and temperature than a larger nonfunctional enclosure.
Cavity count can increase the same problem. An 8-cavity mold is effectively producing parts from eight related but not perfectly identical cavities. Runner length, gate size, cooling-channel position, venting, steel dimensions, and local pressure losses can produce measurable cavity-to-cavity differences. If cavity 1 averages 10.02 mm while cavity 8 averages 9.96 mm on a 10.00 ±0.05 mm requirement, both averages appear acceptable, but one cavity is already much closer to the lower specification boundary.
For that reason, dimensional records should retain cavity identity when cavity variation can affect product performance. Mixing 500 measurements from several cavities into one dataset can produce an acceptable overall average while hiding one cavity that regularly approaches a limit. A better review separates cavity results, compares their mean and spread, and then checks the combined production distribution. Multi-cavity molds used for annual volumes above 1 million pieces benefit considerably from this level of inspection because a small systematic difference repeats many times.
Process capability provides a more useful view than a handful of approved samples. A supplier can produce 30 good parts during a tool trial and still have difficulty holding the same dimensions across multiple shifts, resin lots, mold cleanings, and production weeks. Capability studies commonly examine whether the process distribution stays comfortably within specification rather than merely whether individual measured parts pass. Where customers require Cp or Cpk targets, the required value should be agreed before validation because different companies and industries may apply different acceptance rules.
Tool temperature deserves equal attention because cooling occupies a large portion of many injection molding cycles. A cavity does not cool uniformly merely because the mold controller displays one set temperature. Thick ribs, bosses, inserts, deep cores, and areas far from cooling channels can remain hotter than surrounding surfaces. If one side of a 120 mm flat component cools faster than the other, differential contraction can create bow or twist even when length and width remain within tolerance.
Warpage is especially common when geometry, flow orientation, and cooling work against one another. Increasing packing pressure may improve one dimension but worsen another; extending cooling time may reduce post-ejection deformation but raise manufacturing cost. If cycle time rises from 20 to 22 seconds, output per machine hour falls from about 180 cycles to roughly 164 cycles, a reduction of nearly 9%. A precision supplier therefore needs to balance dimensional stability against production rate rather than adjusting one molding parameter without checking the rest of the part.
Inspection has to match the tolerance being requested. A ±0.50 mm feature can often be checked with straightforward equipment, while a ±0.02 mm geometric requirement may need a CMM, optical system, controlled fixture, or dedicated gauge. The measurement method also needs repeatability small enough to distinguish real part variation from inspection variation. Recording five decimal places does not improve a result when fixturing, operator method, part temperature, or equipment uncertainty contributes more variation than the tolerance permits.
ISO 20457 provides a formal framework for dimensional and geometrical tolerancing of molded plastic parts. The 2018 edition covered manufacturing tolerances for molded thermoplastics, thermoplastic elastomers, and related materials; ISO records show that edition was withdrawn in August 2026 and replaced by ISO 20457:2026. Suppliers working from older drawings should therefore check which revision, customer standard, or drawing-specific tolerance requirement governs the project rather than assuming a historical general tolerance remains appropriate.
Medical-device work adds process documentation to dimensional control. U.S. FDA guidance states that manufacturing processes must be validated when their results cannot be fully verified by subsequent inspection and testing. Injection molding programs may therefore require documented process qualification, material traceability, approved parameter ranges, equipment records, inspection plans, and change control. A part can measure correctly in a 100-piece trial and still be unsuitable for regulated production if the manufacturing method cannot show consistent control over later lots.
Automotive, electronics, fluid handling, and mechanical assemblies have different documentation needs, but the cost logic is similar. At 500,000 parts annually, a 3% rejection rate equals 15,000 rejected components. If each rejected molded part costs only $0.80, direct molded-part waste is $12,000 before sorting labor, machine time, assembly disruption, freight, testing, or replacement material is counted. Reducing rejects from 3% to 0.8% removes 11,000 rejected pieces from the same annual volume.
| Production condition | What should be checked |
|---|---|
| ±0.02–0.05 mm functional dimensions | Tool capability, measurement method, process capability |
| 8, 16, or 32 cavities | Cavity balance, cavity identification, local cooling |
| 20–40% fiber-filled resin | Flow orientation, shrinkage direction, warpage |
| More than 1 million parts/year | Preventive maintenance, automated checks, scrap trend |
| Regulated medical use | Validation, traceability, controlled changes |
| Insert or overmolding | Insert position, substrate tolerance, adhesion, thermal effects |
Insert molding can make tolerance control more demanding because the finished dimension depends on both the molded polymer and a previously manufactured component. If a metal insert location can vary ±0.05 mm and the molded feature around it can vary another ±0.05 mm, the assembly can already see 0.10 mm of combined location range before other tolerances are included. Fixture repeatability, insert seating, mold closure, thermal expansion, and plastic shrinkage therefore need to be considered together.
Overmolding adds material compatibility and substrate movement. A soft elastomer molded over a rigid thermoplastic may require controlled substrate temperature and accurate placement to maintain wall thickness and bond consistency. A nominal 1.00 mm overmold layer with ±0.10 mm tolerance gives much less room for insert movement than a 3.00 mm cosmetic layer. Testing should therefore measure the finished interface rather than relying only on dimensional records from the first molding operation.
Tool maintenance begins to matter more as production grows. A gate, vent, sliding component, shutoff surface, ejector pin, or textured cavity can change after hundreds of thousands of cycles. A mold that produced acceptable parts in 2025 may not reproduce identical dimensions indefinitely without cleaning, inspection, repair, and replacement of wear components. Preventive-maintenance intervals should be based on tool design, resin abrasiveness, cycle count, observed wear, and production history rather than one universal number.
Glass-filled materials make wear especially relevant. A 30% glass-filled resin repeatedly moving through gates and across cavity surfaces can be more abrasive than an unfilled polymer. Gate wear can change fill behavior; worn shutoffs may create flash; worn slides can affect feature position. When a tolerance is only ±0.03 mm, small mechanical changes in the mold can eventually become measurable part changes, so tool condition belongs in the quality record alongside molding parameters.
Supplier selection should therefore be based on demonstrated work with similar geometry, resin, tolerance, cavity count, and annual volume. Ask for examples involving comparable dimensional ranges, the inspection equipment used, how cavity data are separated, how resin preparation is recorded, and how production settings are protected after approval. A supplier that successfully runs 5 million simple PP caps per year does not automatically have experience with a 12-cavity glass-filled PPS component containing ±0.03 mm alignment requirements.
A useful supplier discussion can stay practical:
-
Which dimensions are hardest to hold and why?
-
What sample size will be measured during qualification?
-
Will every cavity be measured separately?
-
How will shrinkage be established before final steel adjustment?
-
Which molding parameters will have approved operating ranges?
-
What happens after a material lot or machine changes?
-
How often will the mold be inspected?
-
Which measurements will be recorded during normal production?
Answers should connect the drawing to an actual production method. If a supplier promises ±0.02 mm everywhere without asking about material, feature size, measurement temperature, gate position, cavity layout, or annual quantity, that claim provides little engineering information. Plastic dimensions depend on the complete manufacturing setup, and a tolerance proven on 50 parts from one cavity is not the same as a tolerance demonstrated across 16 cavities and several hundred thousand production cycles.
Early supplier involvement is most useful before tool steel is finalized. Gate position, wall thickness, draft, rib layout, cooling access, datum selection, insert location, and steel-safe dimensions are much easier to revise in CAD than after machining. Moving a gate by 10 mm or changing a rib from 1.2 mm to 1.0 mm may affect fill pattern, packing, sink, or warpage enough to change later dimensional behavior.
The need for precision molding can therefore be judged by the cost of dimensional variation in the finished product. When a 0.10 mm change has no measurable effect on fit or performance, spending more to hold ±0.02 mm may add manufacturing effort without improving the product. When 0.03 mm changes seal compression, connector alignment, gear engagement, optical position, or assembly force, working with a Industrial plastic molding manufacturer that can document tooling accuracy, process stability, material handling, cavity performance, and measurement capability is commercially easier to justify.