Automated Gummy Molds as Line Components

If you're speccing automated gummy production molds, you're speccing a line component: a tray a machine locates, a depositor fills, a tunnel sets, a station dumps, and a washer returns.

It isn't a soft sheet a person peels. Automation cares about pitch, flatness, a designed dump motion, and cavities that stay equal after a hot stack. Platinum RTV, rigid plastic, or a jacketed hybrid still has to copy the grid. Food-contact on RTV remains 21 CFR 177.2600 extractives and LFGB on the article. No fake cycle times.

Register, pitch, and the servo dump

Automation starts at location. Pins, edges, pockets, or vision fiducials have to be in the CAD. Rubber stretch isn't a locating strategy. A rigid jacket or dual-pour exists so the pitch the robot learned on Monday is the pitch on Thursday.

Dump automation: invert-and-knock on a stiff tray (PC or firm RTV in a frame); flex over a driven roller (needs a hinge in the tray, not a tear web); mechanical eject in a hybrid (hygiene of the eject path). A robot that peels ears like a person is usually a science fair. Design the cavity so the station can be boring. Servo depositors own shot. The mold owns aim point. A 1 mm cupping of the tray is a miss at every nozzle. Flatness after post-cure and after wash is an automation spec.

Belts accelerate. Soft trays wrinkle, mis-aim, and pool slurry. Spec back stiffness or a carrier, thickness that still allows the dump you chose, no floppy handles in the food area, stacking that doesn't dish the center. ASTM D395 screens compression set. The line test is a camera on the first row of nozzles. If shots sit on land, the tray moved. Cavitation equals nozzles. Extra cavities are wash load.

Sensors, wash, and boredom when it's right

Some lines sense plate present, some sense pockets, some checkweigh downstream. The mold can help: a non-food tab for a photoeye; opaque vs translucent (some sensors hate clear silicone); coordinate marks for vision that inspects fill; equal cavity volume so a checkweigher isn't chasing master scatter. Don't engrave QR codes into the show face. Vision that finds starved ears is a scrap tool — fix CAD if ears starve every cycle.

Fully automated CIP of an RTV gummy tray is uncommon compared with rack washers. Spec rack wash: fleet sized to washer takt, stacking ribs, chemistry SDS. Any CIP: designed flow, no crevices, insert seals that don't harbor gel, no tin-cure beads. 21 CFR 177.2600 expects a cleansed article; automation doesn't remove that. CIP chemicals that stainless loves may attack silicone — trial. Don't copy a chocolate PC CIP recipe onto RTV without a test.

Write the mold onto the line equipment list: envelope, mass of the tray (robots care), pin pattern, pitch, dump motion, washer recipe, Shore or dual-pour, article tests, spare count tied to takt, revision and SKU tab. One-up geometry trial before the production master is still required — automation doesn't forgive starved ears; it repeats them faster. Starch moguls are automated too, with different equipment. Don't spec an RTV tray onto a mogul P&ID.

Integration sequence that saves pain: lock envelope with the depositor OEM, freeze dump motion with the station OEM, then sculpt the mascot. Bring a one-up to the line FAT, not only to the mold shop. Safety and guarding change dump technique — an operator can't reach in to free an ear. "No peel assist" belongs in the mold spec. Software recipes should call a plate revision, not only a cartoon name.

Common mix-up: a kitchen-soft RTV sheet can be automated

A lot of RFQs treat "it's silicone, the robot can peel it" as an automation plan. In the shop that usually means the tray will miss pitch and tear. Use a framed / jacketed tray, then automate a designed motion. LSR vs HCR press talk is the wrong fork if what you need is a poured RTV depositing tray; an LSR injection tool with a cold-runner is a molded *part*.

What to ask when you're buying the line component

• Pin pattern, nozzle pitch, tray mass, tray size
• Dump motion the guarded cell can run with no peel assist?
• Shore + jacket so the belt doesn't wrinkle the array?
• Photoeye tab / opacity vs your sensors?
• Rack wash takt and spare count?
• 177.2600 extractives / LFGB on the article before takt?

Example

Servo cell: 36-cavity platinum RTV, 5 mm pins, Shore 30A face in an aluminum jacket, ~20 mm flex over the dump roller, cavities equal to nozzles. One-up at FAT. Mixed v1/v2 spares would have been a pitch surprise — revision tabbed on a non-food lip.

When automation spec is the wrong page

If you don't have a depositor yet, freeze the grid before you buy "automated molds." If the mascot still needs a hand peel, don't put it in a guarded cell. If starch is the process, spec mogul equipment, not an RTV tray. Kitchen folklore doesn't belong on the P&ID.

Related reading: multi-cavity pitch, dump techniques, commercial fleets.

Line-component RFQ

• Envelope, mass, nozzle pitch, pins, dump motion, washer recipe
• Target grams, slurry family as fill/dump notes
• Process: platinum RTV + jacket / PC / hybrid (not an LSR part tool unless you want parts)
• Article tests; spare count tied to takt; revision tabs
• One-up at line FAT

Send grams, formula family, nozzle pitch, tray size, dump motion (including guards), and washer takt. We'll spec a jacketed RTV tray as equipment — not a kitchen sheet the robot is supposed to figure out.

We would not recommend automating a kitchen-soft RTV sheet

We would not recommend a robot that peels ears like a person. Design the cavity so the guarded cell is boring. We would not spec an RTV tray onto a mogul P&ID, or mix v1/v2 spares on a servo head. We would not treat LSR vs HCR press talk as the fork — that molds *parts*. Lock envelope with the depositor OEM, freeze dump with the station OEM, then sculpt the mascot.

flowchart TD
  E[Lock envelope + nozzle pitch with depositor OEM] --> D[Freeze dump motion with station OEM]
  D --> M[Sculpt mascot inside that map]
  M --> FAT[One-up at line FAT]
  FAT --> RUN[Jacketed tray: pins, flatness, cavities = nozzles]
  RUN --> W[Spare count tied to washer takt; revision tabs]

Typical shop values: servo-cell jacketed RTV tray

| Item | Typical shop value | | --- | --- | | Piece-weight target | 4.0 g (cavity volume × measured slurry density; fill-to-brim or fill-to-line stated) | | Outer-row vs center | 3–6% heavy if the loaded tray sags under the nozzles | | Loaded-tray edge sag | 0.5–1.5 mm without a flat jacket | | Cavity Shore | 10A–20A platinum RTV skin + rigid jacket | | Platinum linear shrink | 0.1–0.3% (cut the master so poured pitch lands) | | Mix | Platinum 1:1 or 10:1; rubber stretch is not a locating strategy | | Automation spec | Flatness after post-cure and after wash — 1 mm cupping misses every nozzle |

Caption these as typical shop values, not a measurement on your PO.

FAQ

Can we automate dump of a kitchen-soft RTV tray? Not in a way you'll like. Soft trays miss pitch and tear. Use a framed / jacketed construction, then automate a designed motion.

Do we need vision on every cavity? Only if scrap cost says so. First, equalize cavity volume and fill paths. Vision that watches a bad CAD is an expensive alarm clock.

Is LSR better than HCR for automated gummy trays? That's the wrong fork if you need a poured RTV depositing tray. LSR injection (heated cavity, often a cold-runner) molds silicone *parts*. For this line component, pick RTV Shore, jacket stiffness, flash trim, and register. All food-contact platinum, post-cured.

How does automation change food-contact testing? It doesn't change the method. It makes a wrong specimen more expensive because you'll run more pieces before you notice odor or stain. Test the article construction before takt.

Should spare trays be identical including revision? Yes. Mixed v1/v2 on an automated head is a pitch or volume surprise. Tab the revision. Automation won't read an email.