Prototype vs Production Tooling for RTV Silicone Molds

If you're quoting prototype versus production on a custom RTV silicone mold, you're buying two different objects. A prototype (printed master, tin glove, crude flask) exists to learn geometry, stretch-off, and dump shrink on a few casts. A production RTV mold (usually platinum, keyed split, rigid jacket, recast rights) exists to hold a process window for the SKU's real dump life. If the RFQ doesn't say which one you want, one shop will bid a learning glove and another will bid a depositing tray, and you'll compare the numbers like they were the same cavity.

Silicone doesn't change that split. A platinum RTV mold can be a prototype of a *casting* process. It isn't a prototype of an LSR steel tool. Don't use a poured mold to "save" production LSR engineering, then act surprised when shrink and flash disagree. That LSR path is a different quotation for a molded *part*.

What a prototype RTV mold can skip (if you write the skips)

• Hardened production masters and full key packages
• Multi-cavity balance and depositing nests
• Chemical brand texture (a B-finish or printed layer may stand in)
• Spare skins and a recast program
• A rigid jacket, if you accept slump on a short run

Don't skip a measurement state, a mix sheet, or honesty about life. A prototype that's silent on life will be asked to run SOP. That's how tin gloves get stored as library tools.

Prototype goals should be explicit: confirm fit, confirm stretch-off of an undercut, generate a dump coupon. If the goal is a sales sample in the right pectin color and a food-contact letter, say that; it may still be a prototype mold with production-intent *dump compound*.

Common prototype paths: single-cavity tin or platinum off an SLA master; printed hybrid in a standard flask; CNC single cavity when you need a show-face polish the print won't hold but you still don't want 12 cavities. Finish the master; the rubber copies layer lines.

What production RTV is actually paying for

Production is paying for repeatability: keys, intact splits, a jacket that locates, platinum that holds size in storage, finish and texture to chips, and a traveler. SPI class language is how steel people say that; on RTV, say platinum + jacket + recast instead. It's still an intent, not a warranty.

Production should include the conversation on cavitation, depositing fixtures, and spare skins. Skipping them to "match the prototype quote" buys a prototype with better photos.

Food programs add compound freeze and extraction method. A prototype can use the same certified grade for a first-dump-like pack; it can't skip the later freeze if you launch.

Common mix-up: launching on the bridge because the date slipped

A lot of RFQs treat a successful prototype pour as SOP. In the shop that usually means flash grew, tin drifted, and the jacket wasn't built for the depositor. Valid bridge: learn on a print, freeze geometry, recast platinum against a CNC master with CAD-safe lessons applied. Invalid bridge: launch on tin because the retailer window moved.

Squeeze the bridge only when the SKU may die, quantity is truly low, and the flash spec is honest. Write a kill date so it isn't still running two seasons later.

China cluster pricing makes the mix-up more tempting: a production-looking photo of a prototype stack. Ask for chemistry, jacket type, and recast terms in writing.

What to ask when you're buying

• Prototype / bridge / production as a closed choice on the RFQ?
• Platinum or tin, jacket yes/no, recast yes/no?
• Expected dump band (not a fake cycle warranty)?
• Is a second, production master already budgeted?
• Will first dumps use production dump compound, or a written waiver?

If a supplier answers a production RFQ with a tin glove, that isn't a bargain. It's a misread.

When this split is the wrong argument

If you needed LSR/HCR *parts*, neither RTV prototype nor RTV production is your production tool — new engineering pass, new quote. A one-cavity platinum + jacket can be production for a low-volume chocolate even without 24 pockets. Geometry can override labels: a hollow mandrel mold is production-intent hardware even at 200 dumps.

Related: low-volume casting, how we make the mold, custom RTV molds.

Worked example

A 4-cavity chocolate, 90 × 45 mm, started as tin RTV off an SLA master for 40 show samples, then recut a CNC master, poured platinum with a keyed split and plaster jacket for ~600 winter dumps. The prototype taught stretch-off of the logo; it didn't teach season-two size. They budgeted the second pour at RFQ. Coating the tin glove "to make it production" was offered and declined.

RFQ language that stops apples-to-oranges

• "Prototype / bridge / production" closed choice
• Chemistry + jacket + recast (not SPI class as a substitute)
• Cavitation and depositing fixture yes/no
• Dump band
• Whether a second production master is already in the program

Send which of the three you want, dump band, and dump compound. We'll quote that intent — not a cathedral when you asked for a glove, or a glove when you asked for a season.

We would not recommend launching on the tin bridge because the retailer window moved

We would not recommend coating a prototype glove “to make it production,” or treating a successful print-master pour as SOP. We would not use a poured mold to skip LSR engineering if the article is actually a molded *part*. We would not skip a measurement state just because the tool is a prototype. Write a kill date so the bridge is not still running two seasons later.

RUUIPON Mold Structure Decision: prototype may skip multi-cavity balance, brand texture, spare skins, and even the jacket if you accept slump. Production pays for keys, platinum, jacket, chips, recast rights. A 1-cavity keyed, jacketed platinum stack can still be production for a low-volume chocolate. Hollow mandrel at 200 dumps is production-intent hardware.

flowchart TD
  A[RFQ closed choice] --> B{Prototype / bridge / production}
  B -->|prototype| C[SLA or tin glove; mix sheet; kill date]
  B -->|bridge| D[Learn on print, recut CNC, recast platinum]
  B -->|production| E[Pt + keys + jacket + recast]
  C --> F{Launch because the date slipped?}
  F -->|yes| G["Invalid — recast to production intent"]
  E --> H{1-cavity vs 24?}
  H -->|construction decides, not pocket count| I[Jacket + platinum can be production at 1 cavity]

Typical shop values

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

| Item | Typical shop value | | --- | --- | | Prototype cavity count | 1–4 | | Production food / gummy band | 8–24 with jacket nests if a line is real | | Cavity Shore | 10A–20A; prototype may skip the jacket if you accept sag | | Mix | Platinum 1:1 or 10:1 by grade (tin only with a written kill date) | | Platinum shrink after full cure | About 0.1–0.3%; tin will not teach season-two size | | Loaded-tray edge sag if production skips the jacket | About 0.5–1.5 mm | | Outer-row gummy vs center if the tray sags | Edge often 3–6% heavy |

FAQ

Can a prototype tin mold become production with a coating? Rarely. Coatings help release. They don't add keys, a jacket, platinum stability, or recast rights. If volume appeared, recast to the production intent. Using coating as a class upgrade is how splits still tear.

Should prototype first dumps use production compound? If the lesson is shrink, certificates, or color, yes. If the lesson is only "does this undercut demold," a substitute dump can be cheaper, with the waiver written. Mixing those lessons on one unlabeled sample is how you freeze the wrong size.

Is a 1-cavity platinum mold a prototype or production? It can be either. Construction and life hardware decide, not cavity count. A 1-cavity keyed, jacketed platinum stack can be production for a low-volume SKU. A 1-cavity unkeyed tin glove is a prototype even if the CAD is frozen.

Who pays for the production recast after a successful bridge? The program, as a planned second mold. It isn't a warranty claim on the prototype shop unless they poured the bridge wrong against a frozen file. Budget it at RFQ. Surprise recasts feel like failure; planned recasts are the method.

Can I prototype in RTV and produce in LSR without a new design pass? You can learn styling and some stretch-off. You can't transfer flash, shrink, and vestige. Plan an LSR engineering pass on a parts OEM. Treating RTV castings as LIM first-article is the usual category error.