Custom Silicone Grommets, Gaskets, and O-Rings Mold Design Guidelines

An O-ring mold is a gland problem. A grommet mold is a stretch-and-seal problem. A gasket mold is a land-and-load problem. If you tool all three as “a doughnut in rubber,” you will flash the O-ring, tear the grommet on eject, and ship a gasket that takes a set in the flange.

This is DFM for HCR compression/transfer and LSR injection of industrial seals. It is not a chocolate-mold article and not a Shore-A catalog.

Name the part by how it fails

O-rings. The functional surface is the torus that sits in a gland (AS568 / ISO 3601 type thinking). Flash on the ID or OD becomes a leak path or a cut. Parting-line placement and land quality dominate. Cross-section must be round after cure, not oval from a poorly filled or over-clamped cavity.

Flat gaskets and cover seals. The functional surface is a face that is squeezed a designed amount. Thickness tolerance, parallelism, and compression set of the grade dominate. A pretty edge with a thick flash skirt is a scrap gasket even if it “seals on the bench” under extra bolt load.

Grommets. The functional features are the groove that captures a panel and the ID that captures a cable or tube. Demold stretch through a thin membrane, concentricity of ID to groove, and tear at the membrane dominate. Hardness that is right for an O-ring squeeze is often too stiff to assemble a grommet.

Write the failure mode on the tool kickoff. The steel will follow that sentence.

Process choice before steel

Tin-cure RTV is a prototype path for a gasket layout. It is not a production O-ring process.

O-ring cavity rules

Parting line. Put it at the equator if you must split a two-plate tool, and budget a defined, tiny flash that you can tumble or punch. Better: a core/cavity split that keeps the sealing lines (the ID/OD contact bands in the gland) free of witness. A longitudinal flash across the sealing line is a leak.

Land and clamp. Compression O-ring tools live or die on land width and plate stiffness. Too little clamp, flash. Too much clamp on a weak plate, the cavity loses roundness and you still flash at the corners. See the flash article in this cluster for the clamp mistake. Here the rule is: design the land for the compound’s overflow, do not “add clamp” in tryout as the design.

Fill. LSR: gate into a non-sealing region or use a film gate you can destage. Avoid a gate scar on the torus. HCR: preform weight is a dimension. A heavy preform flashes; a light one knits and leaks.

Shrink and post-cure. Heat-cured silicone shrinks more than platinum RTV. Apply the grade’s shrink factor to the cavity, then post-cure the rings if the spec requires low compression set, *on a mandrel or in a tray that does not oval them*. Loose rings in a hot pile come out egg-shaped.

Finish. A smooth torus demolds and inspects cleanly. Texture is rarely useful on an O-ring. Do not polish only one half; mismatch shows as a step.

Gasket cavity rules

Thickness is the seal. Call a thickness tolerance that matches DIN ISO 3302-1 (or your drawing class) and then tool to it. A ±0.3 mm rubber title block on a 1.5 mm gasket is a shrug, not a spec.

Crush beads and lips. If the gasket uses a molded bead, the bead is the sealing line. Flash next to the bead is a shunt. Vent into overflow wells away from the bead, not across it.

Bolt holes. Molded holes need draft and enough land that the hole does not tear on eject. Punched holes in a flat sheet gasket are a different process; do not mix punched-hole tolerances into a molded-hole conversation.

Inserts and carriers. Overmolded metal or plastic carriers need the silicone to knit around the insert without a knit line on the sealing face. Place gates so flow meets behind the bead.

Compression set is a design input. The flange gap, the bolt load, and the temperature decide the grade. A 50A general HCR that looks fine at room temperature can take a set in a 150 °C cover and leak on cool-down. Put squeeze (often a 15-30 percent working band for many static silicone faces, not a universal law) on the drawing with the temperature. Then pick a low-set grade and a post-cure. Do not “make it thicker” as a substitute; extra thickness at the same bolt length is extra squeeze and more set.

Grommet cavity rules

Membrane thickness. The panel groove creates a thin web. If that web is too thin for the hardness, LSR will still fill it and then the part will tear on eject or on customer assembly. If it is too thick, the grommet will not seat. Model the stretch over the panel thickness and the cable OD. Prototype in the production hardness, not in a soft RTV that flatters assembly.

Undercuts. Grommets are undercuts. LSR tools need a collapse core, a split, or enough elongation and draft to strip. HCR compression often uses a split or a collapsible core. Forcing a 70A grommet off a solid core is a tear generator.

Concentricity. Cable ID to panel groove must be concentric or the cable pulls the groove off the sheet. Tool the core from one datum. Do not accept a two-piece core that can shift with flash.

Eject. Pins on a thin membrane print and tear. Prefer a stripper ring on the hub, air assist, or a plate that pushes on a thick shoulder. Pin marks on a sealing lip are leak paths.

Drawing notes that save a tool spin

• Process: LSR or HCR, and whether flash is as-molded, tumbled, or precision-trimmed.
• Parting line location, shown.
• Shore A, grade family (platinum LSR vs millable HCR), and compression set method (ASTM D395 Method B, with time and temperature), not “low set.”
• Finish on sealing faces versus non-sealing faces.
• Post-cure yes/no, and on what fixture.
• For O-rings: standard dash size or actual ID/CS with the gland.
• For grommets: panel thickness range and cable/tube OD range, not just the rubber dimensions.

If those notes are missing, the molder will optimize for easy fill. Easy fill is not a seal.

FAQ

Why can't I tool a grommet like an O-ring doughnut?

An O-ring fails on flash and roundness in a gland. A gasket fails on thickness, parallelism, and set in a flange. A grommet fails on membrane tear, concentricity, and stretch through a panel. Hardness that is right for O-ring squeeze is often too stiff to assemble a grommet.

Where must an O-ring parting line not sit?

Not as a longitudinal flash across the sealing contact bands in the gland. Equator split is a compromise with a defined tiny flash you can tumble or punch; better is a split that keeps those bands clean. A gate scar on the torus is the same class of leak.

Why is gasket thickness the seal instead of a pretty edge?

The face is squeezed a designed amount. A thick flash skirt is scrap even if extra bolt load "seals on the bench." Call a real thickness class, vent overflow away from a crush bead, and do not substitute extra thickness at the same bolt length for a low-set grade; that is extra squeeze and more set.

Why prototype a grommet in production hardness, not soft RTV?

Soft RTV flatters assembly. The panel groove is a thin web that LSR will fill and then tear on eject or at the customer if it is too thin for that Shore A. Too thick and it will not seat. Pins on that membrane print and tear; push on a thick shoulder with a stripper or air.

What drawing notes stop the molder from optimizing only for fill?

Process and destage method, shown parting line, Shore A and grade family, ASTM D395 Method B (time and temperature) not "low set," finish on sealing versus other faces, post-cure fixture, and for grommets the panel thickness range and cable OD range. Easy fill is not a seal.