When RTV Casting Should Stop: Recast vs a Part Tool

If you're running custom RTV molds, stay on the glove until labor-per-dump and fill-weight scatter force a change. Don't switch because a slide deck said “production.”

RTV casting (silicone mold, vacuum, hand pour or deposit of a dump) and steel LSR (hot cavity, cold-runner, clamp, platinum liquid silicone *parts*) are different factories. The switch people mean is often the wrong one. On this site the next steps are usually recast the skin, duplicate the tray, or add a jacket. Jumping to steel LSR is a process change to molded *parts*: chemistry, shrink, flash, capital, and MOQ. Treat it as a date on a map only if the shippable object changed.

What you are leaving, and what you are buying

Leaving a tired RTV skin means leaving a tool that stretches off undercuts, a near-zero shrink platinum jacket (~0.1–0.3%), and a labor model where a technician or depositor is the press. Tin-cure jackets are cheaper and less stable. Neither is an LSR cycle.

Buying another RTV asset means a recast against the unsanded master, a twin 8-cavity instead of one crowded 16, or a stiffer mother mold so the sheet stops sagging 0.5–1.5 mm.

Buying steel LSR means heat-cured platinum silicone *parts*, production shrink ~2–3%, a metal shutoff, and a tool class. You also buy design freeze. Steel does not enjoy a weekly CAD tweak. Only do that when the product is a vulcanized article.

There is a missing middle people skip when they actually need *parts*: aluminum 1-cavity LSR. If you jump RTV → steel for parts, you are asking the most expensive tool to teach you LSR.

Common mix-up: a Kickstarter spike as a Class 103 trigger

Demand that vanishes leaves leftover steel as furniture — and still will not dump pectin. Inhibition on a dirty SLA master is a master-prep problem. Switching to steel will not un-inhibit a print, and it will not teach you RTV.

ISO 3302-1 does not pick the process. It is a drawing class if the article is rubber.

Four signals that *this* RTV skin is done

Check these in order. Volume is last on purpose.

1. The dump compound is no longer a stand-in — or the skin cannot take it. If the product needs an LSR *part* grade (medical, self-lubricating, conductive), every RTV-cast PU is a lie. Switch process *now* via a parts OEM, even at low quantity. If the product is still food/resin but the skin is torn at a logo, recast. Chemistry of a *part* cannot wait. A torn glove can.

2. Scatter is eating the fit. RTV dumps wander with pour mass, vacuum technique, jacket wear, and (for tin) aging of the mold. When your fill-weight assumes a capability the jacket does not have, more craftsmen will not help. A stiffer mother mold or a recast is a capability upgrade. Steel is a capability upgrade only for *parts*.

3. Labor per good dump is the business. Time the cell: mix, degas, pour, cure, demold, trim. If that stack already dominates material, scaling RTV is hiring — unless you duplicate trays onto two depositors. A steel 2- or 4-cavity does not have to be “cheap per tool” to win *parts*. It has to beat that labor curve *and* ship the right object.

4. Volume is real, and the CAD is frozen. Only now does annual dumps matter. Tens of thousands of a frozen food tray is still often RTV with recast. Tens of thousands of a frozen LSR *part* is steel. Hundreds may still be one jacket.

If (1) is a *part* chemistry problem and volume is tiny, you still leave RTV. You just don't have to buy Class 103 to do it.

Four signals you should not switch to steel yet

• The lip, the undercut, or the brand logo is still in argument. Recast the skin. Recuts on hardened cavities are the expensive kind — and the wrong object.
• You have never mapped fill-weight on the jacketed tray. You don't know sag. Buy the mother mold first.
• Demand is a spike, then silence. An 8-cavity LSR plate is how leftover steel becomes furniture.
• The “RTV problem” is inhibition on a dirty SLA master. That is a master-prep problem.

What to ask when you're buying

• Is the shippable object still a cavity you dump into?
• Recast vs duplicate tray vs jacket upgrade — priced?
• If someone is quoting steel, did they name a cold-runner *part* tool on purpose?
• CAD frozen including split and keys?
• Twelve-month dumps vs twelve-month *parts*?

Worked example: 10,000 dumps/month, still RTV

A gummy brand at ~10,000 dumps a month on a 16-cavity 22 mm bear wanted “steel so we can scale.” Outer row 3–6% heavy, sag 0.5–1.5 mm. We recast the platinum 10A–20A skin (mix 1:1, vacuum degas) onto a stiffer jacket and split into two 8-cavity twins. Steel LSR would have made silicone toys. Geometry can override quantity: the chin return still needed peel, not a cold-runner.

When this article is the wrong quote

If the product has always been a vulcanized LSR gasket, you are late to the parts OEM. A tiny artisan chocolate can stay on RTV forever. Hostile steel geometry (deep undercuts with no slide budget) may stay on RTV even when labor looks ugly — price slides *before* you declare RTV dead.

Related reading: custom RTV molds, RTV vs LSR prototypes, TCO, gummy trays.

A crossover you can sketch without fake prices

Pick a horizon (12 months is enough). Add:

Stay RTV: jacket replacements + recasts + labor hours × lots + scrap from sag + freight of bulky hand-packed dumps.

Duplicate RTV: second tray + second sampling, lower risk of a torn 16 parking the SKU.

Go steel LSR: only if the object is a *part* — class-based tool + cavitation you can feed + setup fees at your real batch size.

Switch *objects* when the dump is no longer the product. Switch *skins* when RTV’s labor+scrap line crosses a recast plus jacket. If you cannot sketch those crossings with *drivers* (hours, cavities, expected dumps, scrap %), you don't have a switch date. You have a feeling.

Cavitation belongs on extra RTV trays, or on steel only after the object is a part. An 8-cavity LSR is not the switch. It is an option after you decided to mold parts.

The sentence that should trigger a steel parts PO: “Grade is locked. CAD is frozen including parting and gates. Aluminum LSR samples met flash, hardness after post-cure, and the fit. Twelve-month volume will feed at least a 1-cavity steel tool. Undercuts have a slide or a redesign. The shippable object is a vulcanized part.”

If any clause is false, you are not switching to steel. You are either recasting RTV, or you are buying a better jacket.

What not to buy: a steel multi-cavity quoted against RTV sample photos of chocolate.

Send monthly dumps, the failure (sag, tear, labor), and whether the object is still a cavity. We will mark recast vs duplicate vs “stop, that’s a parts RFQ.”

We would not recommend steel because a slide deck said production

We would not recommend an 8-cavity 103 against a gummy tray. We would not skip the jacket and then “graduate” to steel. Raising Shore A first is the wrong fix for scatter. We would not treat inhibition on SLA as a reason to cut steel.

RUUIPON Mold Structure Decision: object first. Cavity? recast / duplicate / jacket. Part? stop. Flat back? open-back. Deep undercuts that will not peel? two-part / core — steel slides are a different budget. Soft 10A–20A or wet dump? mother mold before any “switch.”

flowchart TD
  A[RTV hurting?] --> B{Shippable object still a cavity?}
  B -->|no, LSR/HCR part| C[Stop: parts OEM]
  B -->|yes| D{Skin torn or jacket sagging?}
  D -->|yes| E[Recast + mother mold]
  D -->|labor only| F[Duplicate trays]
  A --> G{CAD still moving?}
  G -->|yes| H[Don't cut steel]

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 jacket | | Platinum shrink | About 0.1–0.3%; LSR part contrast ~2–3% | | Loaded-tray edge sag | About 0.5–1.5 mm without a jacket | | Outer-row vs center if tray sags | Edge often 3–6% heavy | | 16-cavity food tray | Two 8-cavity twins often beat one crowded sheet |

FAQ

Is annual volume the first reason to leave RTV casting? No. Check whether the object is still a cavity, then scatter, then labor per good dump. Volume is last on purpose. If the product needs a real LSR *part* grade, every RTV stand-in is already a lie, even at tiny quantity.

When do I leave a tired skin without buying Class 103 steel? When the cavity is still the product: recast against the unsanded master, stiffen the jacket, or duplicate the tray. Steel can wait until the object is a part. Skipping that middle asks the most expensive tool to dump pectin it cannot dump.

Why shouldn't a Kickstarter spike trigger an 8-cavity 103? Demand that vanishes leaves leftover steel as furniture. Also stay off steel while the lip or logo is still in argument. Inhibition on a dirty SLA master is a master-prep problem, not a reason to cut steel.

What geometry fights a switch to steel even when labor looks ugly? Deep undercuts with no slide budget, glove-mold pulls, and one-piece skins. Price slides or a redesign before you declare RTV dead — and only if you needed *parts*. Thin flash-prone fins on a *part* want metal shutoffs; food trays want jackets.

What sentence should trigger the steel PO? Object is a vulcanized part, grade locked, CAD frozen including parting and gates, aluminum LSR samples met flash and fit, twelve-month volume will feed at least 1-cavity steel without silly batches, and undercuts have a slide or a redesign. If any clause is false, recast RTV or buy a jacket.