QC for Custom RTV Food-Contact Mold Lots

If you're writing QC for a food-contact RTV mold, you're lot-controlling an article, not gloss-checking at packing. Incoming A/B kits aren't incoming food-contact. In-process you still catch sticky faces, inhibition, and flash. At finish you still catch key register, jacket fit, and smell. Then, on a defined frequency, you prove the cavity against 21 CFR 177.2600 and/or LFGB / BfR XV. Platinum RTV is the food family. Tin-cure RTV doesn't belong on the food line. The food-contact face is a QC surface: uncut, unflashed, unsticky.

Incoming, pour-side, and lot flasks

Receive against a frozen spec: compound code, platinum vs tin, Shore target, color, kit date, and the food-contact report this pour last supported. Don't stamp "FDA received." Stamp "compound X, color Y, linked to report Z."

Pour-side: no sticky faces (inhibition or mix error) — quarantine. No detachable flash; trim then wash. Knit, voids, short pours: soil traps and weak logos. Dimensions: pan envelope, web thickness for ice, cavity volume if the customer bought a yield. Don't treat a molded-rubber tolerance class as a synonym for RTV accuracy; check master finish, box rigidity, and jacket. Post-cure: oven chart, loading that allows air, thickness family batched together. Nested cups that post-cure closed fail German sensory later.

You can't flask every muffin. Freeze compound, color, pour, post-cure; run 177.2600 and/or BfR XV at qualification; re-run on change. A spatula from another tool isn't a lot sample for a 24-cavity tray.

Common mix-up: packing gloss as food-contact QC

A lot of plants treat a pretty carton and a cousin certificate as control. In the shop that usually means scored logos, sticky cups from tin contamination, and a flask that assumed an intact surface you didn't inspect.

What to ask when you're buying a QC plan

• 100% visual on the food-contact face for chocolate and ice, or an AQL that lets scored logos through?
• Hold tags with the lot, or a verbal "don't ship those"?
• Post-cure oven as a gaged process (thermocouples, load photos) or a closet?
• Master identity and recast: is the pattern still the one in the report?
• Tin-cure parts: off the food line, different gloves?

Worked example: 3-piece hollow chocolate, keys drift

A confectioner runs a 6-cavity hollow chocolate, 80 mm, three-piece platinum RTV with an inner core, about 800 pulls a month. QC: 100% key register and core pull path, no sticky inner face, flash off the show surface, smell off post-cure, lot COC pointing at 177.2600 fatty plus BfR XV sensory. If keys round off, hold the lot and recast against the master — don't "let the flash line show you."

When this article is the wrong quote

LSR press-side cavity-pressure QC and HCR mill incoming are parts-plant systems. Don't paste them onto this pour. A non-food resin mold needs flash and register checks, not a 177.2600 lot flask. Geometry can override sampling: a single-cavity showpiece still wants 100% face inspection even when dimensions can be sampled.

Related reading: production challenges, verify certifications, durability testing, custom RTV molds.

Lot record a retailer can audit

• Compound and pigment batches; kit dates
• Master ID, box, pour, jacket
• Post-cure chart and production wash
• In-process defects and holds
• Link to food-contact report
• Care text that shipped: no abrasive; heat as tested
• Recast log: same master, or new CAM

Send your current incoming checklist and a photo of a "borderline" face. We'll mark reject vs recast vs flask, and write hold rules that travel with the lot.

We would not recommend packing gloss as food-contact QC

We would not recommend an AQL that lets scored chocolate logos through, releasing "good looking" cups from a sticky tin-contaminated load, or treating a missing post-cure chart as optional. We would not hardness-test every ice web instead of looking at the face.

Hold vs ship:

flowchart TD
  V[Visual food face] -->|cut sticky flash| H[Hold lot]
  V -->|clean| P[Post-cure chart present?]
  P -->|no| H
  P -->|yes| C[COC links report identity]
  T[Tin on bench suspected] --> H

Typical shop values: 6-cavity hollow chocolate QC

| Item | Typical shop value | | --- | --- | | Geometry | 80 mm, 6-cavity, three-piece plus inner core | | Monthly pulls | ~800 on a confectionery tool | | Shore | 10A–20A platinum RTV plus jacket | | Mix / shrink | 1:1 or 10:1; ~0.1–0.3% platinum | | Face inspection | 100% on chocolate and ice food faces | | 177.2600 fatty | 175 then 4 mg/in² n-hexane |

Caption these as typical shop values, not a sampling plan we ran on your lot.

FAQ

How often must we re-run 177.2600 on a stable muffin SKU? There's no universal calendar in the section. Qualify the SKU, freeze compound, color, pour, and post-cure, then re-run on change and on a frequency you can defend. A spatula from another line isn't that sample.

Is a smell check at packing a substitute for BfR XV organoleptic? It's a useful in-process screen. It isn't the lab sensory method. Dual-market qualification still needs the named stack.

Do we hardness-test every ice-tool cavity? No. Spot-check lots for wrong compound. 100% visual on webs, flash, and cuts. A Shore A gun won't see a knit tear.

Can QC accept a scored face if extractives already passed? No. The flask assumed an intact surface. A cut face holds fat and is a new article.

Where does tin-cure RTV sit in a food-plant QC plan? Off the food line. Different benches, tools, and gloves. Tin contamination is a platinum RTV hold. Don't include tin-cure parts in food lot identity.