Balancing Pours in Multi-Cavity RTV Silicone Molds

If you're pouring a multi-cavity RTV silicone mold, pocket 1 is full and pocket 8 is light. The “runner” is not almost balanced. There isn't a runner.

Uniform fill on a custom RTV tray is even pour head, last-fill vents, vacuum, and a jacket that does not bulge. You can CNC eight identical masters and still ship eight different dumps if the sheet sags, the tilt starves a corner, or air sits in pocket 8. LSR cold-runner systems and HCR transfer pots fail in different ways and belong on a parts RFQ. Do not copy a thermoplastic fishbone into a chocolate slab and call it done.

If you ignore balance, the defects are shorts, fat outer rows, and a dimensional spread you will try to blame on ISO 3302-1. That standard is a drawing class for a rubber *article*, not a promise sixteen pockets weigh the same.

What “balanced” means in a poured tray

Equal pocket drawings are not equal fill. Balance is:

• Same fill time (or a known, tiny offset you designed) as the dump or the mold rubber reaches each pocket.
• Same head when the pocket is full, so density and flash at the split match.
• Same jacket height so cavity 8 is not a balloon when cavity 1 is backed.
• Same peel at open, so pocket 8 is not tearing when pocket 1 is already out.

Platinum RTV mix 1:1 or 10:1, vacuum degas until the foam collapses. A short, fat pour path to pocket 1 and a long, thin path to pocket 8 will not “even out because silicone flows so well.” It will fill 1, pack 1, flash 1, and starve 8 — of *dump*, not of injection pressure.

Common mix-up: pasting LIM runner math onto a chocolate slab

A lot of RFQs treat “16-cavity = 16× output” as a tray fact. In the shop that usually means the jacket wasn't flat and the outer row shipped 3–6% heavy. Weigh all pockets. Don't paste cold-runner math onto a flexible sheet.

RTV multi-cavity is gravity, air, and the jacket

Eight jewelry masters in one pour: tilt, last-fill vents, and a jacket that does not bulge. There is no injection pressure to balance. Vacuum helps. A “balanced sprue” carved in clay is a pour path. Useful. Not an LSR cold-runner. If someone later wants 8-cavity LSR *parts*, the RTV eight-up only proves the shape, not the manifold.

Family molds (two different pockets in one tray) are unbalanced by intent unless the volumes and dump windows match. A 2 cm³ gummy next to a 15 cm³ bar is not “close enough.” Either dedicate trays or accept a process window that is a tightrope.

Overfilling the easy pockets to fill the hard one is how you drown the split in flash and still have a short — and how a loaded tray sags 0.5–1.5 mm at the edge.

How to know you are not balanced

Shop checks that do not require a branded simulation suite:

1. Short-pour series. Stop early. Which pocket fills last? That is your restriction.
2. Flash map. Which pocket flashes first when you add a little dump? That one is overfed.
3. Weight scatter of as-cast dumps (same trim). More than your fill-weight study can swallow means head scatter.
4. Cavity-to-cavity dimensions on a critical. ISO 3302-1 M2 on a 16-cavity sagging tray is a wish.
5. Peel / stick differences. Sticky in one hole only is inhibition, finish, or trapped air, not “that silicone.”

Fix jacket and pour path before you “tune it in the process” forever. Process can hide a mild imbalance. It cannot hide pocket 8.

What to ask when you're buying

• Cavity count versus dumps per month, after freeze?
• Will first article weigh every pocket, not the hero?
• Jacket material and how you keep the sheet flat?
• Family mold: same dump window, or two skins?
• Two 8-cavity trays vs one 16: which risk split is priced?

Worked example: 8-cavity then 16-cavity gummy sheet

A gummy brand wanted a 22 mm bear, 4.0 g target, food-contact, about 10,000 dumps a month. First production tool was an 8-cavity platinum sheet, 10A–20A skin, mix 1:1, vacuum degas, rigid jacket, so outer-row sag could be measured before copying to 16. Edge pockets that ran 3–6% heavy were a jacket problem, not a reason to add more cavities. A parallel LIM 16-cavity quote for a silicone toy of the same outline doesn't deposit pectin.

Cavitation versus class

Honest fill-weight wants fewer cavities until sag and pour path are known. Duplicate the tray after the first dumps. Jumping to 16 cavities to hit a piece price is how all sixteen are wrong the same expensive way.

A second 8-cavity twin often beats one crowded 16: a torn skin parks half the SKU, not all of it, and you can trial the dump window on the first eight.

When this article is the wrong quote

If you actually need LSR multi-cavity part tools with a balanced cold-runner, this RTV pour memo is the wrong asset. Quote a parts OEM. A tiny 2-cavity artisan chocolate can skip balancing entirely. Geometry can override quantity: one deep 40 mm hollow with a core is still a split job at 80 dumps; adding cavities doesn't fix the core pull.

Related reading: custom RTV molds, 1 vs 8 cavity piece price, gummy trays, keying two-part molds.

Cavity-balance lines to copy onto the RFQ

Copy this. A pocket count without these lines isn't output.

• Process: platinum or tin RTV tray (ban HCR/LSR substitution)
• Cavity count vs dumps/month; CAD frozen yes/no
• Jacket material; keys; how the sheet stays flat
• First article: weigh every pocket; short-pour map
• Family geometries: same dump window or separate skins
• Recast of the skin if one pocket tears

Send the pocket STEP, target fill-weight, and monthly dumps. We will mark 1–2 vs 8 vs 16 and whether two smaller trays beat one crowded sheet.

We would not recommend buying 16 cavities to decorate a quotation

We would not recommend qualifying only the hero pocket. We would not skip the jacket and then blame "bad rubber" for 0.5–1.5 mm edge sag. We would not mix two Shore A grades in one pour. We would not paste LIM runner math onto a chocolate slab.

RUUIPON Mold Structure Decision: extra pockets after freeze, on a jacket that stays flat. Flat back? open-back sheet. Deep undercuts that won't peel? split or core — extra cavities don't remove a lock. Soft 10A–20A or wet dump? mother mold before you add pockets.

flowchart TD
  A[Want more pockets?] --> B{CAD frozen?}
  B -->|no| C[1-2 cavities only]
  B -->|yes| D{Jacket keeps sheet flat?}
  D -->|no| E[Fix mother mold first]
  D -->|yes| F[Weigh every pocket on FA]
  A --> G{Need LSR or HCR parts?}
  G -->|yes| H[Stop: different part RFQ]

Typical shop values

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

| Item | Typical shop value | | --- | --- | | Mix | Platinum 1:1 or 10:1; vacuum degas | | Cavity Shore | 10A–20A skin with rigid jacket | | Platinum shrink | About 0.1–0.3% after full cure | | Loaded-tray edge sag without a jacket | About 0.5–1.5 mm | | Outer-row gummy vs center if tray sags | Edge often 3–6% heavy vs the center pocket | | 8-cavity then 16-cavity 22 mm bear | About 10,000 dumps/month after freeze is a common band |

FAQ

Why isn't equal pocket CAD a balanced RTV tray? Balance is equal fill time, equal head at full, equal jacket backing, and equal peel. A fat short pour packs and flashes while a long thin path starves. Matching branch drawings does not equalize dump at pocket 8.

Can a small gummy share a tray with a large bar? Family molds are unbalanced by intent unless the dump windows match. A few cubic centimetres next to a large bar is not close enough. Dedicate trays or accept a process window that is a tightrope.

How do I prove pocket 8 is the restriction without branded simulation? Run a short-pour series to see who fills last, map which pocket flashes first when you add dump, weigh as-cast pieces after the same trim, and check fill-weight scatter. Sticky in one hole only is air, finish, or inhibition, not "that silicone."

Does an RTV eight-up prove an 8-cavity LSR manifold? No. Eight masters in one pour are gravity, tilt, vents, and jacket bulge. There is no injection pressure to balance. The RTV eight-up can prove the shape, not the cold-runner. That LSR job is a different quote.

Should I jump to 16 cavities to hold ISO 3302-1 M2? Opposite. Honest fill-weight wants fewer cavities until sag and pour path are known, then duplicate the tray. Sixteen wrong impressions is an expensive way to miss class. ISO 3302-1 is a drawing class, not a balance spec.