How Buyers Should Evaluate Custom RTV Silicone Molds

If you're evaluating a custom RTV silicone mold before you freeze dumps, you're asking whether this stack will hold a process window for the duty you bought — not whether one chocolate looks glossy. Score the master copy, keys, jacket, mix identity, and a real first-dump pack. If you only look at cosmetics, you'll freeze a glove whose split is already tired.

This is a buyer's score of *intent versus delivery*: did you pay for a production platinum skin with a rigid jacket and receive a tin prototype, or the reverse? SPI 101–105 language is steel construction intent. It doesn't grade a rubber cavity.

What to score on the bench

Master copy. Layer lines, sealer blush, incomplete logos, bubbles at the parting. The rubber copies the pattern. If the master wasn't finished, the mold isn't "quality."

Keys and split. Pips, bushes, or CNC bosses that locate before the dump. Nicks at the split are future flash. Look at the worst pocket, not the hero.

Jacket / mother mold. Does it actually constrain Shore 10A–20A rubber, or does it rock? A floppy RTV tool with pretty first castings is a lucky pour.

Chemistry vs duty. Platinum when you need recasts and storage hold. Tin when the master is dirty and the run is short. Scoring a tin glove as a library tool is a fail even if first dumps are pretty.

Mix traveler. Grade, lot, ratio, vacuum degas, pour date. A naked stack with no traveler is a future orphan.

If the quote listed a fiberglass jacket and you received tape on a plywood flask, that's a specification miss now, not a quality problem later.

Sample dumps as a mold report, not a brochure

A quality evaluation uses the first-dump pack: all cavities, mix sheet, measurement state (23 °C after a named wait), flash at named splits, texture versus chip, hardness of the skin. If the shop sends six hand-trimmed bars from cavity 2, you've evaluated a trimmer, not a mold.

Look for process heroics: extra clamp on a rocking jacket, extra bake to chase a sticky face, a technician who "knows how to fill cavity 7." Those are deductions. A good mold runs inside a written window.

Dimensional evaluation has to respect the dump. ISO 3302-1 class on a rubber *article* drawing is a tolerance class, not a license to brutal-CMM a gummy lip, and it isn't another word for "precision." If size is critical, the pack should show a method incoming can repeat.

Common mix-up: pretty dump = production mold

A lot of RFQs treat one glossy first article as a freeze. In the shop that usually means the land of the split is already nicked from "sampling," or the jacket was never rigid. Coatings and food-release films are extras you score as extras: present, food-safe, recoat plan. They don't substitute for a jacket.

Certificates (FDA 21 CFR 177.2600, LFGB, ISO 9001) evaluate the *program*, not the mill finish of the master. Don't give a mold a food score because the lobby has a flag. Score the rubber, then score the QMS separately.

If the evaluation is remote, require key photos, jacket marks, and a video of the dry fit with the mold number in frame. A factory tour video is marketing. Your mold number is evaluation.

What to ask when you're buying

• Guidance/keys, jacket rigidity, and land of the split — can I see them, not only the show face?
• Platinum or tin, and is the traveler complete enough for night shift?
• First-dump pack: all cavities, wait/temperature, chip photos?
• Recast path and who holds the unsanded master?
• Pass / accept-with-ECO / reject in writing — silence after a pretty sample isn't approval.

Spare skins are part of the score on a production-intent tray. A 24-cavity gummy stack with no recast plan is a risk even if first dumps are clean. You may still accept the mold, but don't freeze SOP until the spare plan exists or is explicitly waived.

When this scorecard is the wrong tool

Steel LSR stacks (heaters, cold-runners, shutoff lands) need a different incoming form — that's a molded-*part* quote. A 20-dump tin art glove can skip spare-skin scoring if you write the kill date. Geometry can override quantity: a three-piece hollow chocolate with a mandrel fails evaluation if the core won't extract, even when the show face is A-range.

Related: quality inspection, prototype vs production RTV, custom RTV molds.

Worked example

An 8-cavity muffin, 70 mm, food-contact platinum, fiberglass jacket, about 1,200 bakes in season. Evaluation walked keys, jacket rock, mix sheet, and all eight first bakes at 23 °C after a 24-hour wait. Cavity 5 had a nicked pip from "already sampled"; they recast that skin before freeze. Cosmetics on 1–4 would have hidden it. A parallel LSR bakeware-*part* quote wasn't scored — wrong object.

Evaluation checklist to copy

• Intent: production platinum + jacket vs tin prototype (written)
• Keys, split, jacket dry-fit, master copy, sticky-face check
• Mix traveler + first-dump pack (all cavities, 23 °C, wait X)
• Recast / spare-skin plan or explicit waiver
• Written accept / accept-with-ECO / reject

Send key photos, the mix sheet, and the first-dump pack. We'll score intent versus delivery and tell you freeze, ECO, or reject — not whether the photo looked glossy.

We would not recommend freezing on one glossy first article

We would not recommend scoring a tin glove as a library tool, or grading a rubber cavity with SPI 101–105 language. We would not give a mold a food score because the lobby has a flag. We would not raise Shore A first to “look more production.” Keys, jacket rock, mix traveler, and an all-cavity pack decide.

RUUIPON Mold Structure Decision: production intent is platinum + jacket + recast path. Tape on a flask, missing keys, and a land already nicked after “sampling” are prototype behavior. A three-piece hollow chocolate fails evaluation if the core will not extract, even when the show face is A-range.

flowchart TD
  A[Score] --> B{Intent written: Pt + jacket vs tin prototype?}
  B -->|mismatch| C[Reject or reclassify]
  B -->|matches| D[Keys, split, jacket dry-fit]
  D --> E[Mix traveler + all-cavity pack at 23 C]
  E --> F{Heroics: extra clamp / extra bake / cavity-7 wizard?}
  F -->|yes| G[Deduct; do not freeze SOP]
  F -->|no| H[Accept / accept-with-ECO / reject in writing]

Typical shop values

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

| Item | Typical shop value | | --- | --- | | Production-intent Shore | 10A–20A platinum skin, rigid jacket | | 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% | | Jacket rock / edge sag | About 0.5–1.5 mm is a fail against a production jacket | | Outer-row gummy vs center if the tray sags | Edge often 3–6% heavy | | First-dump pack | All cavities; 23 °C after a named wait | | Spare skins on a 24-cavity tray | Score the plan, or write an explicit waiver |

FAQ

What's the fastest way to tell a prototype RTV mold from a production one? Look at keys, jacket rigidity, platinum vs tin, and the traveler, then at the quote's intent. Tape on a flask, missing keys, and a land already nicked after "sampling" are prototype behavior. A polished cavity face alone doesn't make a production skin.

Should I reject a mold if first-dump flash exists? Not automatically. Flash at a named split within the spec, from a written window, can be acceptable. Flash that needs hero clamp, or flash that grew during the sample lot, is a key or jacket fail. Use the limit sample, not disgust.

Can I evaluate mold quality from dump cosmetics only? No. Cosmetics miss a rocking jacket, a missing mix lot, and a cavity that wasn't the one in the photo. Use rubber inspection plus a full-cavity pack. Cosmetics are one chapter.

How do food-release coatings affect a quality score? Score whether the coating you paid for is present, whether it's compatible with the dump, and whether a recoat path exists. An unexpected coat that fills texture is a fail against the chip. A missing coat that was on the PO is a fail.

Is a higher SPI polish on the master always higher mold quality? No. Quality is fitness. A-1 on a sticky low-durometer skin can hurt release from the master. A sealing lip wants an intact specified face, not jewelry. Score polish against the drawing's face-by-face callout.