How to Inspect Quality on a Custom RTV Silicone Mold

If you're inspecting a custom RTV silicone mold, you're doing two jobs that buyers collapse into one photo. Job one: is the *mold* the mold you paid for (copy of the master, keys, jacket, mix identity)? Job two: do the *castings* from that mold meet the drawing in a stated wait and temperature? A pretty first chocolate doesn't prove the jacket is square. A CMM report on a 30 Shore A lip doesn't prove the keys will last.

This is an inspection memo for poured RTV tools — master, box, A/B, split, keys, inner core, support jacket — plus the first dumps they produce. Don't paste a steel checklist (heaters, cold-runner valves, water circuits) onto a rubber glove mold. If that's the checklist in your inbox, you're looking at a molded-*part* tool, a different quotation.

Incoming mold versus first dumps

When a mold arrives (or before it leaves the pour shop), inspect rubber as rubber:

• Cavity identity versus the master file.
• Keys, split shut, and whether the jacket locates without rocking.
• Incomplete copy of micro-texture, trapped bubbles at the parting, torn flash edges.
• Sticky faces (inhibition) — that's a failed mold, not a polish to inspect in.
• Inner core pull path; cores that don't extract are a cut mold.
• Mix identity: grade, lot, ratio, pour date on the traveler.

When dumps exist, inspect the article as the article. Don't use dump inspection to skip the mold inspection. A clever dump recipe can hide a bad split for a week.

The master can take a scan. The RTV cavity and the dump often can't, not in the same way. Probe sink, fixture strain, and temperature swamp any ISO 3302-1 conversation on a rubber *article*. For the dump: define measurement state (23 °C after wait X), use optical or dedicated gauges on lips, weigh cavities as a fill-and-flash proxy, durometer on a specified boss of the *skin* not on a 0.5 mm membrane. ASTM D412 / D624 only when the PO paid for tensile/tear; they're compound screens, not master screens.

Flash is an inspection item with a location and a method. “No flash” without a location isn't a spec.

Incoming RTV drums and cured-dump screens

The mold can be perfect and the lot still wrong. Incoming RTV drums: identity, mix behaviour as the shop's method requires. Cured dumps: appearance, and any regulated test the program named (food-contact extraction under FDA 21 CFR 177.2600 or LFGB, CPSIA for toys).

Platinum chemistry adds a contamination inspection that steel people forget: sulfur, tin-cure residues, some 3D-print masters, and dirty gloves inhibit cure. A sticky first dump is a chemistry fail, not a polish fail. Don't “inspect it in” with more time in the oven until you know the cause.

For tin-cure, record the age of the mold when you inspect dimensions of the *casting*. A “good” inspection on week one isn't a library approval for month six. Platinum is more stable but still fails on tears and sticky faces.

Common mix-up: carton AQL as mold incoming

A lot of RFQs treat visual AQL on finished chocolates or gummies as proof the tool is good. In the shop that usually means a lot can pass cosmetics and still come from a tray with a dying key that will fail next month. Export AQL is a third job. It isn't a substitute for mold incoming.

ISO 9001 status of the factory tells you whether they *have* a record system. It doesn't replace the actual sheets for your mold. Ask for the traveler, not only the wall certificate. A usable inspection record names tool number, cavity, RTV lot, mix-sheet revision, measurement state, equipment, and who signed. Photos of the keys at incoming and at a PM gate are worth more than a paragraph. Keep retains.

Multi-cavity inspection must keep cavity identity through the bag. Mixed cavities with one bad pocket will average into a false pass. Bag by cavity during first dumps. Production may rotate cavities on a plan; it shouldn't anonymize them.

What to ask when you're buying

• What's the incoming mold checklist — keys, jacket, bubbles, sticky faces, mix lot?
• How are first dumps measured, at what temperature and wait, with what gauge?
• Flash limit sample at which split location?
• Are we bagging by cavity for first article?
• Food-contact tests on which object (dump article vs grade letter)?
• For tin-cure, is mold age on the report?

If the shop only sends a photo of a pretty dump, you haven't inspected the mold.

A practical sequence and a worked example

1. Buyoff on the mold (the list above) plus photos of keys and the master.
2. Jacket dry fit: no rocking, core extracts.
3. First dumps, full pack.
4. Compare master/jacket issues versus pour issues. Recut the master only for the master. Change mix only for mix.
5. Correction pour, then freeze.
6. Incoming inspection on production lots against the retain and the flash limit sample.

If you skip step 1 because the airplane was booked, you'll inspect the keys with scrap dumps. That's a more expensive gauge.

Worked example. An 8-cavity hollow chocolate, 55 mm sphere, platinum RTV with a mandrel and a fiberglass jacket, a few thousand dumps a season. Incoming caught a bubble cluster at the equator of cavity 4 and a sticky patch on 7 (SLA residue on the master). They recast 4 and 7 against a recleaned master, then measured sphere OD at 23 °C after a 24-hour wait, bagged by cavity. A steel-tool CMM form from another vendor wouldn't have had a line for sticky faces or mandrel pull.

Incoming fails: confirm then act

| Symptom | Likely cause | How to confirm | Fix | | Sticky cavity face | Inhibition on master or dirty pour | Face still tacky after extra bake | Reclean master, recast — don't inspect it in | | Bubble cluster at split | Incomplete degas or pour head | Photo vs mix sheet | Repour that skin; check vacuum | | Keys don't locate | Jacket warp or missing pips | Dry-fit rock test | New jacket / recast with keys | | Dump size off at 23 °C | Wrong wait, tin age, or master shrink | Compare retain vs new; check mold age | Coupon method; recast platinum if tin drifted | | One bad pocket in mixed bag | No cavity identity | Bag-by-cavity first article | Isolate, recast that skin |

When this inspection memo is the wrong form

Don't paste a steel incoming list (heaters, cold-runner valves, water circuits) onto a glove mold — that's a molded-*part* tool. Carton AQL on chocolates isn't mold incoming. A 30-piece art run in tin-cure doesn't need a full multi-cavity traveler; it does need mold age on the size report. Geometry can override quantity: one mandrel that won't extract fails the mold even if every other pocket photographs well.

Related: how to evaluate mold quality before freeze, sampling process, food-grade molds.

Incoming inspection list to put on the PO

• Mold: keys, jacket rock, bubbles, sticky faces, mix lot, pour date
• Dumps: 23 °C after wait ____, bag by cavity, flash limit sample at named split
• Texture/polish vs retained chip
• Tin-cure: mold age on the dimensional report
• Food-contact object: article vs grade letter

Send the incoming photos you want (keys in frame) and the measurement state. We'll inspect the rubber hardware and the first dumps as two jobs, then say recast, reclean, or freeze.

We would not recommend carton AQL as mold incoming

We would not recommend a contact CMM on every 30 Shore A lip, a WeChat pretty-dump as buyoff, or “inspecting in” a sticky platinum face with more oven time. We would not mix cavities into one bag at first article. We would not paste a steel incoming list (heaters, cold-runner valves, water circuits) onto a glove mold.

RUUIPON Mold Structure Decision: inspect the rubber hardware first (keys, jacket rock, bubbles, sticky faces, mix lot), then the casting at 23 °C after a named wait. Tin? mold age on the report. One mandrel that will not extract fails the mold even if every other pocket photographs well.

flowchart TD
  A[Incoming] --> B["Mold: keys, jacket rock, bubbles, sticky, mix lot"]
  B --> C[Jacket dry fit]
  C --> D[Dumps at 23 C after wait X, bag by cavity]
  D --> E{Fail type}
  E -->|sticky / inhibition| F[Reclean master, recast]
  E -->|bubble at split| G[Repour that skin]
  E -->|size off| H[Check wait, tin age, sag 0.5-1.5 mm]
  E -->|one bad pocket in mixed bag| I[Isolate, recast that skin]

Typical shop values

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

| Item | Typical shop value | | --- | --- | | Mix | Platinum 1:1 or 10:1 by grade; vacuum degas until the foam collapses | | Cavity Shore | 10A–20A skin; durometer on a specified boss, not a 0.5 mm membrane | | Platinum shrink after full cure | About 0.1–0.3% | | Measurement state | 23 °C after a named wait | | Jacket rock / edge sag | About 0.5–1.5 mm if the mother mold is missing or warped | | Outer-row gummy vs center if the tray sags | Edge often 3–6% heavy — weigh cavities as a fill proxy | | First-article bagging | By cavity; mixed bags average a false pass |

FAQ

Should I CMM every silicone lip on incoming lots? Usually no. Many lips are too soft for a contact CMM to mean what the print thinks. Use a dedicated gauge, optical method, or a functional fit check at 23 °C after the named wait. Save scanning for the master. If you insist on CMM on rubber, state probe, force, and fixturing, or you'll fight the fixture.

What's a reasonable flash inspection on an RTV split? A named location, a method, and a limit sample. “No flash” on a keyed chocolate mold is often a wish. “Flash not to exceed X at the split, no tear-trim into the logo” is a spec. Train the inspector on the sample, not on adjectives.

Does a factory ISO 9001 certificate replace mold incoming photos? No. The certificate says a QMS exists. Your mold still needs identity, key condition, and a first-article pack. Ask for the traveler that *includes this mold number*.

How do I inspect a texture or polish without a chip? You mostly can't, not in a way two people will repeat. Buy or retain the SPI plaque, VDI chip, or MoldTech chip used at first dump. Photograph under agreed lighting. Arguing from a phone screen isn't inspection.

Can visual AQL on cartons replace cavity-weight checks? Visual AQL catches gross cosmetics. It misses slow key wear and slow fill changes. At any serious dump volume, add weight or a flash gauge at the table. Carton AQL is downstream. Wear happens on the mold.