# Multi-Cavity Injection Tooling: Core Shift and Neck Thread Tolerances (18/410–28/410)
TL;DR — A multi-cavity mold moves tolerances from the drawings into the steel: cavity balance and core support decide how straight the molded neck threads come out, and the thread window the catalog records runs from 18/410 through 28/410 across the fine-mist, lotion-pump, trigger, and cap models. The same catalog records measured outputs from 0.13 ml to 4.00 ml per stroke in the lotion-pump family and a 0.22–1.35 ml per stroke window on trigger outputs. The mechanics below are engineering guidance — cavity counts, tolerance values, and gate layouts are configured per project, with mold/tooling development and dosage/output testing as the verification hooks.
Why: The Mold Decides What the Part Can Be
Most of a pack's quality is decided before the resin is ever melted — it is decided by the shape of the steel cavity the resin fills. A multi-cavity tool multiplies that decision across every shot:
- To get tolerances right in the steel, not in rework. Injection molding produces what the tool produces. Two cavities fed unevenly mean two different parts, no matter how good the material is.
- Threads are the first seal. The neck finish of a pump or bottle is a thread system; the molded thread is the datum every closure, collar, and capsule pair must agree with. A straight thread window is a tooling result, not a drawing wish.
- Cavity balance is output consistency. In a multi-cavity mold, output per stroke and thread geometry are symptoms of the same cause: how evenly each cavity fills, packs, and cools.
- Tooling lead time is project lead time. The capabilities chain of our manufacturing system — injection molding, mold/tooling development, and mass production — means the tool is developed and run inside the program, so geometry decisions are negotiated before the first trial shot, per project.
- Dose consistency is a delivered promise. The pump, actuator, or sprayer that meters the product takes its repeating behavior from parts that repeat part to part; a dose that drifts by cavity is a tooling story before it is a dosing story. Measured output windows in the catalog — 0.13–4.00 ml per stroke in the lotion-pump family, 0.22–1.35 ml per stroke on triggers — are the reference the tool is expected to hold.
The reading of a mold, then, is not a single number but a set of dependencies. Pick any two parts out of a batch and the question worth asking is not "are they within tolerance" but "are they the same from the same cavity, and the same across cavities." That is the discipline tooling brings: consistency is engineered before the first part is counted, and every part after that is tested against it.
None of this is exotic. It is the ordinary lifecycle of a molded part — design the tool, balance the cavities, run the trial, measure the first articles, then hold the window in mass production. What this article adds is where the neck thread and the dose output plug into that lifecycle, and which of those steps the program records as data rather than as guidance.
How: From Melt Flow to Molded Threads
3.1 Multi-cavity layout and cavity balance
A multi-cavity tool routes one melt stream through a runner system into several cavities at once. The economics are immediate — one cycle produces several parts — but the engineering burden moves to balance: every cavity must see the same melt temperature, fill rate, and packing pressure in the same window of time. In practice, balance is managed by the geometry of the runner and gate arrangement, and it is validated by trial shots, not assumed from the design.
A few cues read the balance of a running tool without tearing it down. The distribution of flash across cavities, the weight of parts pulled from each cavity on a trial shot, and the reproducibility of measured output down each cavity row all hint at who is filling hot and who is filling cold. The customer-side symptom is normally quieter: a sprayer that loses its dose consistency only on certain runs, or a pump whose thread catches clean on some assemblies but binds on others. Tracing the symptom to a cavity is the diagnosis.
Cavity count is an economic and risk decision, not a capability claim. A higher cavity count stretches cycle economics across more parts but narrows the tolerance band each cavity can hold relative to its neighbors. The number itself — how many cavities a given tool carries — is configured per project; our manufacturing system records injection molding, mold/tooling development, and mass production as its capabilities, without publishing equipment tonnage, machine counts, or cycle times.
Gate placement belongs to the same set of per-project decisions, and it carries a feature-level consequence: the gate feeds the neck, the wall, or both, and weld-line position lands where the flows meet. For threaded parts, a weld line that crosses the thread flank is a strength and flash risk that no later process fully repairs — it is managed upstream by gating and venting choices agreed in tool development, then confirmed on trial shots.
3.2 Core shift: what moves, and what it costs
Core shift is the displacement of the mold core under filling pressure. During injection, the melt pushes against unbraced core surfaces; if the core yields, the molded wall — and any geometry tied to the core — lands off-center. For neck finishes this matters twice: the thread pitch circle shifts relative to the part axis, and the wall at the neck can thin on one side. Either outcome breaks the pairing logic of the finish — the closure or collar that threaded on straight in the drawing cannot thread on straight in the hand.
"Neck thread tolerance" sounds like one number, but it bundles several: the pitch and lead of the thread, its concentricity to the part axis, and the perpendicularity of the finish face. Core shift attacks the concentric and perpendicular members of that set first — the thread keeps its pitch but loses its place. That is why a plain thread gauge can pass a shifted neck while the assembly still fails: gauging thread form and gauging thread position are two different checks, and a multi-cavity tool puts both at risk.
The practical countermeasures are the standard tooling toolkit: supporting the core at both ends or with locking features, balancing fill so both sides of the core see similar pressure, and running validation shots that probe core position. How each is deployed is a per-project decision — the tolerance value for core shift is agreed with the project team, never borrowed from a publication. The check, when it matters, is largely concentricity on the molded neck and a stable thread catch across assemblies, and it stays a project-gated test rather than a global standard.
Two failure signatures are worth naming because they look unrelated but share a root cause. A lotion pump that fatigues at one side of its neck can be a packing-pressure story as easily as a materials story. A fine-mist bottle whose closure seats with a wobble on one assembly line but not another is the same upset elsewhere. When the same tool misbehaves intermittently in one feature on one run, core support is on the short list to interrogate before the process is re-tuned.
3.3 Neck threads are molded into the steel: the 18/410–28/410 window
A neck finish such as 18/410 or 28/410 encodes a diameter class and a thread profile; the molded thread becomes the datum for every cap, collar, and actuator that pairs with it. The pairing convention itself — how a finish class lines up with its closure, and where the compatibility edges sit — is covered in depth by the neck-finish compatibility insight; this article stays on the tooling side of that conversation: how the thread gets its geometry in steel, and how the catalog window documents what the molded record actually carries.
In the catalog, the measured thread window for the four core families runs from 18/410 through 28/410: fine-mist records carry 18/410 and 18/415; lotion pumps span 18/410, 20/410, 20/415, 22/415, 24/410, 24/415, and 28/410; triggers span 20/410, 24/410, and 28/410, with 28/400 and WAX entries; caps span 20/410 through 28/415. Around that window, dropper records run 16/410–24/400 and foam records 30/410–43/410.
The practical distinction is between the interface and its execution: the finish class is a chosen interface; the thread quality is a tooling result. Threads molded from steel come out with flash, lead, and concentricity that depend on tool finishing, venting, and how the tool is run — each confirmed per project, not inherited from the class number. The window above is what the catalog measures and records; the geometry inside it for a given tool is what the project must confirm.
Because the finish is both a diameter and a thread, the checks split naturally. A go/no-go mate with the paired closure confirms the diameter class seats; a thread-lead or concentricity check confirms the molded thread tracks true; a leak check confirms the pairing is actually a seal. Each belongs to a different stage of the testing sequence, and together they separate "the finish exists on the drawing" from "the finish works on the part."
The catalog window also carries edge entries worth reading carefully. Trigger records include 28/400 alongside 28/410, and WAX appears as an entry in the same family — the interface record is richer than a single thread class, and each variant carries its own pairing rules. When a finished model specifies one of these variants, the molded thread must be confirmed against that variant's pairing part, not against the neighboring class. The measured window is a map of what the catalog contains, not a license to interchange classes within it.
3.4 From many cavities to one consistent dose
Once the tool runs, the parts step into the test program. The dosage/output test is the output-side hook: whether a pump, actuator, or sprayer delivers its recorded output per stroke across many parts — including parts from different cavities of the same tool. A loud per-cavity signature in output testing is often a loud cavity-balance or core-support problem masquerading as a dose problem. Leak testing closes the loop on the thread and seal pairing, so a cavity imbalance that thins a neck wall cannot ship as a mystery.
The recorded outputs anchor the test targets: lotion-pump strokes record 0.13–4.00 ml per stroke, trigger strokes record 0.22–1.35 ml per stroke within a recorded ±15% window, and actuator records include 0.15 ml and 0.23 ml per stroke. These are measured data, published as the program's record, not as a guarantee for any specific formula. A multi-cavity tool that holds dose consistency across its own rows is the practical point — output spread attributable to cavity is exactly what the dosage/output hook is trained to surface.
Reading any single part rarely tells the full story; reading a population does. A first article proves the tool once; the same measurement repeated across a production lot proves the tool holds.
When the neighbor checks — weight, thread gauge, output — are measured on parts drawn from several cavities, the data starts to separate "the tool design" from "this shift this week." That separation is where multi-cavity debugging becomes routine rather than heroic.
3.5 Trials, first articles, and holding the window in production
The first proof of a multi-cavity tool is the trial shot. Part weights pulled cavity by cavity, thread gauging on the molded necks, and a dimensional read of the critical features tell the team whether the tool is balanced before any finished assembly is made. The trial baseline — not the drawing, not the catalog — becomes the reference for the mass production that follows: same material, same settings, same output signature.
From there, production control is mostly a discipline of windows. The molded thread is checked against the paired closure; output per stroke is compared with the recorded family window; the leak check keeps the seal pairing honest. Every one of these criteria is configured per project — the program does not publish pass/fail thresholds for dimensional, dosing, or seal checks as fixed values. Mass production inherits the trial discipline instead of renegotiating it.
What changes across the program's life is not the physics but the confidence in it. The same mold that needed a weekend of trials at launch may need only an hourly window check in its steady state; the same part may demand far more scrutiny on its first production run than on its thousandth. The tooling development capability and the test capabilities exist precisely so that level of attention is a project decision, negotiated against the part's own record.
The flow above condenses that lifecycle into one path: design the tool with the cavity count the project chooses, balance it, mold the threads inside the window the catalog measures, check core position where shift can hide, and let dosage/output and leak testing carry the verdict on release.
<pre class="language-mermaid"><code>flowchart LR
A[Multi-cavity tool design<br/>cavity count per project] –> B[Cavity balance + core support]
B –> C[Molded neck threads<br/>18/410–28/410 catalog window]
C –> D[Core shift checks<br/>concentricity per project]
D –> E[Dosage/output + leak testing]
E –> F[Parts released<br/>per project criteria]
</code></pre>
What: Verified Program Data and the Verification Hooks
The published record covers what can be measured without a specific project: the capabilities of our manufacturing system, the thread classes recorded on catalog models, and the output window recorded per stroke. Everything about a specific tool — its cavity count, its tolerances, its gates — stays a per-project configuration.
- Manufacturing capabilities: injection molding, mold/tooling development, and mass production are the capabilities recorded for our manufacturing system — no equipment tonnage, machine counts, or cycle times are published.
- Thread window, measured on catalog records: fine-mist records carry 18/410 and 18/415; lotion-pump records span 18/410, 20/410, 20/415, 22/415, 24/410, 24/415, and 28/410; trigger records span 20/410, 24/410, and 28/410, with 28/400 and WAX entries; cap records span 20/410 through 28/415.
- The window is not the whole catalog: dropper records run 16/410–24/400 and foam records 30/410–43/410; the 18/410–28/410 range describes the four core families, not every thread class in the record.
- Measured outputs: lotion-pump strokes record 0.13–4.00 ml per stroke; trigger strokes record 0.22–1.35 ml per stroke within a recorded ±15% window; actuator records include 0.15 ml and 0.23 ml per stroke.
- Verification hooks: dosage/output testing anchors output consistency per stroke; the leak test family covers thread and seal pairing — both capability-level checks in the program's test set.
- Boundary: cavity counts, core-shift tolerances, gate layouts, and cycle parameters are configured per project and are not published as fixed values.
Gotchas
- Core shift is a mechanism, not a published number. The catalog publishes capabilities and measured thread classes and outputs; a core-shift tolerance is agreed with the project team for each tool.
- Thread class and thread quality are different things. 18/410 tells you the interface, not whether the molded threads are concentric, clean, and to lead — that is a tooling result gated by per-project checks.
- 28/400 and 28/410 are distinct entries. The trailing digits are part of the finish system; a "close but different" class that does not pair is exactly the failure a thread-gauge check is meant to catch.
- No equipment numbers are on the table. Our manufacturing system records injection molding, mold/tooling development, and mass production; tonnage, machine counts, and cycle times are not published — do not infer them from a cavity-count decision.
- Output windows are recorded data, not blanket claims. 0.13–4.00 ml per stroke and the 0.22–1.35 ml trigger window describe the catalog records; a specific formula-tool pairing is still validated before release.
- Balance shows up at the test bench, not the drawing. If dosage/output data splits down cavity rows, treat it as a balance or core-support symptom before re-tuning the process.
- The trial baseline is the reference, not the catalog. First-article measurements from the trial define the production window per project; a recorded family output window guides the target but never replaces the project's own validation.
- A molded thread is only half the pairing. Thread quality alone does not release a pump or bottle; the paired closure, the seal, and the output must be checked together downstream of the molding step.
Related Resources
- Categories: Pumps · Dispensers · Fine Mist
- Related insights: Neck Finish Compatibility · Metered Dosing · The 12-Test Lab Matrix
- Customization — tooling, thread, and finish configuration · Batch-5 related: Vacuum Leak Chamber Calibration