Engineering a Custom RTV Silicone Mold Before the Pour

If you're engineering a custom RTV silicone mold, you're turning a part file into a pour that a table can repeat: parting, keys, pour gates in the box, vacuum degas, a shrink note for the *cast* article, CAD-safe logos, and a mother mold. It isn't a rendering. If the engineer can't say platinum vs tin, glove vs keyed split, or whether there's an inner core, they're decorating a STEP file.

Engineering happens before anyone mixes A/B. Changes after first dump are ECOs, not "we'll tweak it." Don't paste LSR runner math onto a poured tool. If the quote is full of cold-runners and heated cavities, that's a molded-*part* RFQ, not this mold.

From STEP to a box layout

Start with one revision: STEP plus 2D that states dump tolerances, measurement state (typically 23 °C after a named wait), finish, texture keep-outs, and core BOM. ISO 3302-1 class, if you use it, belongs on a rubber *article* drawing as a tolerance class — it isn't the shrink percent, and it isn't "precision."

Layout calls:

• Cavity count versus annual dumps and registration risk
• Family tray versus dedicated jacket
• Orientation for the dump, for show faces, and for later depositing
• Where the split hides, and where keys can actually shut
• Pour gates and vents in the mold box; vacuum degas plan; optional detail coat on the master

Hollow chocolates and ice spheres add mandrel locators and a pull path that doesn't tear the cavity. That's engineering, not a surprise at first pour.

Parting, keys, and the jacket as one system

Parting defines flash. Keys define whether the mold finds itself. The jacket defines whether Shore 10A–20A rubber holds size. Treat them as one sketch.

Silicone stretches, so some undercuts that would need slides in plastic can demold by flex. Some can't — thick, high-durometer, or sharp letters. Engineering is the documented call, not a hope that "rubber always comes out." Inner cores need a pull path; if they can't extract, you designed a cut mold.

Assign one shrink factor from a coupon of the *dump*, not from an LSR heat-cure percent. Platinum RTV after a full cure is near-zero; tin is time-dependent. Thick-thin dumps don't shrink isotropically; CAD-safe the critical logos so first dump can tell you which way to recut the master.

Simulation of LSR fill is the wrong purchase on this job. A charge trial or a first pour is cheaper and more honest for RTV.

Common mix-up: copying a LIM gate list onto a poured mold

A lot of RFQs treat "mold engineering" as gates, vents, and a cold-runner sketch. In the shop that usually means you've skipped the master finish, the box, and the mother mold. You engineer the master and the jacket, plus pour gates and a degas plan. There is no cold-runner on this PO.

Texture and polish engineering is face-by-face on the master: SPI or VDI 3400 or MT 11000, after barrier coat if coated, draft for grain depth. "Make it look nice" isn't an engineering note.

When the program might recavitate later, leave pour-head and nest room in the jacket. When it won't, don't pay for a cathedral.

What to ask when you're buying

• Glove, keyed split, or cored — and where's the pull path?
• Platinum or tin, and is shrink on the *dump* coupon, not an LSR percent?
• Jacket material vs Shore of the skin?
• Pour gates / vents in the box, vacuum degas, detail coat yes/no?
• CAD-safe logos until first dump?

If any of those are missing, CNC or SLA is guessing.

When this engineering pass is the wrong drawing

LSR/HCR part tools need parting, steel gates, and thermal zones — different shop, different quote. A tiny shallow tray can skip the jacket if you write it. Geometry can override quantity: one letter with negative draft forces a split even at 40 dumps.

Related: project workflow, texture on the master, what tooling means.

Worked example

A two-sided logo soap, 70 × 50 × 25 mm, 10 cavities, platinum RTV, keyed split, resin jacket, ~800 dumps/season. Engineering put the split through the untextured edge, kept leather grain off the keys, CAD-safe'd the logo depth, and called a vacuum degas plus a rigid jacket. First soaps were shallow at the split until pour head went up — process, not steel. They didn't add "slides" and they didn't apply an LSR shrink map.

Engineering build package to demand

• Cavity layout, split/key sketch, core pull path
• Shrink note for the dump + measurement state
• Finish/texture sheet on the master, keep-outs
• Box: pour gates, vents, degas, jacket type
• First-dump measurement plan
• ECO rule after the master exists

Send the STEP, dump material, and undercut photos. We'll return a split, jacket, and pour plan — not a cold-runner layout for the wrong product.

We would not recommend a LIM gate list as RTV engineering

We would not recommend paying for LSR fill simulation on a poured chocolate tool, or assigning an LSR 2–3% shrink map to a platinum master. We would not hope that “rubber always comes out” on a thick, high-durometer letter with negative draft. We would not skip the mother mold on Shore 10A–20A and then call sag a shrink error.

RUUIPON Mold Structure Decision: Flat back? → open-back. Else deep undercuts + peel? If no → two-part / plug / inner core. Large wet casting or soft Shore? → mother mold. CAD-safe the logos until the dump coupon exists.

flowchart TD
  F[Flat back?] -->|yes| O[Open-back RTV]
  F -->|no| U[Deep undercuts?]
  U -->|silicone can peel safely| O
  U -->|cannot peel| S[Two-part / removable plug / inner core]
  W[Large wet casting or soft Shore?] -->|yes| M[Mother mold / support jacket required]
  W -->|tiny shallow tray| O

Typical shop values

Caption these as typical shop values, not a named lot.

| Item | Typical shop value | | --- | --- | | Cavity Shore | 10A–20A platinum RTV, jacket behind it | | Mix | Platinum 1:1 or 10:1 by grade; vacuum degas until the foam collapses | | Platinum shrink after full cure | About 0.1–0.3% — coupon the dump, not an LSR percent | | RTV skin wall | Often 8–15 mm; plan pour head so end cavities fill | | Loaded-tray edge sag without a jacket | About 0.5–1.5 mm | | Outer-row gummy vs center if the tray sags | Edge often 3–6% heavy | | First food layout | 8–12 cavities until depositing nests are real |

FAQ

When is fill simulation worth the fee on RTV? Almost never as LSR software. Pay for a first pour or a coupon. On a chunky single-cavity resin master, a trial pour is cheaper and more honest. Never treat a generic 50 Shore A LSR card as a measurement of a gummy dump.

Who owns the shrink number in the engineering file? The process owner, from a coupon of the production dump and wait, with the measurement state on the drawing. Engineering applies it. Catalog midpoints are planning only. If no coupon exists yet, CAD-safe the logos and say so.

Can I add a split later if undercuts surprise us? You can, at the cost of a new pour. Stretch-off should have been an explicit call at layout. Adding a split to a packed multi-cavity tray is often a new mold. Prototype gloves exist for that surprise. Production jackets shouldn't.

Does engineering include the dump fixture? If warpage, depositing nests, or bonded inserts matter, yes. A free-state dump versus a nested jacket changes the article. Put the fixture in the build package or accept the free-state drawing.

How is RTV mold engineering different from LSR? You engineer the master and the mother, plus pour gates and a degas plan. Shrink of platinum RTV is near-zero; tin-cure is time-dependent. There is no cold-runner. Copying a LIM gate list onto a poured mold is the usual category error.