Injection Molding Design for Product Startups

The sums below are hypothetical teaching maths. They are not BrahmWorks fees, not tool quotes, not yields and not timelines. A moulder’s number replaces them the day you have a revision and a resin grade.

A tool buys a cavity that can make many similar parts. It does not buy a working product, a radio, a seal rating or a certification. First shots can tell you whether the part fills, ejects and holds the dimensions you measured, in that resin, on that press. They do not tell you the life of a snap unless you cycle it. They do not tell you about a resin you did not run. They do not forgive a drawing that still describes the printed cheat.

Write what this tool is allowed to prove. A usable list is short: fill and eject in the intended family; critical dimensions on a named cavity; cosmetic faces against a signed sample. Drop, ingress and radio are separate work unless you fund them as such.

List the cheats before you praise the print. No draft. A wall thickness the printer could vary and a mould will not. A drilled side hole. A boss glued on after the print failed. A clip bent back into place. A resin that is tougher, or more brittle, than the grade you will buy. Each cheat is a conclusion you may not carry forward.

Keep what the print taught about hold, board fit, cable exit and button placement. Translate those into mouldable geometry. Do not turn layer lines into a texture spec, or a sanded face into a polish the tool must hold.

Send a model that already contains draft, or mark draft as open and expect the quote to move when it is added. Send a drawing with datums and with tolerances only where a function needs them. Name the resin family, and say so if the grade is still open. Colour is a plaque or a sample, not a hex code on a render. State the quantity as an assumption, the show faces, the zones where a gate or an ejector is forbidden, and whether an insert or a second shot is in scope.

Without that pack the quote is for a different part. Ask each moulder to name cavity count, the actions, the exclusions, and which dimensions the first article will measure.

A softer, shorter-lived tool and a hardened production tool answer different questions. The softer tool can be the right buy while the snap is still moving and you need tens or hundreds, not tens of thousands. It is a poor teacher of a high-glass resin, a fine texture, or the cycle you will later use in a piece price. This article states no tool life and no lead time. Those are questions for the toolmaker at a named revision.

Adding plastic is usually a further cut. Removing plastic is usually a weld or a new insert. Ask which changes on this tool are still a cut. Bias a risky fit towards the cheaper direction, and write that bias down so nobody "corrects" it back to the print.

A second cavity copies a mistake and adds balance as a new variable. A family tool that puts a lid and a base in one shot couples their scrap, their colour and their engineering changes. You cannot revise the latch without touching the other part’s cycle. Startups often family-tool to save a frame. The saving is real only if both parts are stable and the volumes match. If one part is still a guess, give it its own cavity.

Colour is a lot and a process window. Regrind changes the melt and can change strength and the show face. For a pilot, forbid regrind or cap it in writing. A first article is a measured part: cavity, resin lot, the critical dimensions, and the shots kept. "It looks like the prototype" is not a first article.

A soft grip over a shell is a bond and a second tool. A moulded-in insert is a loading step and a missing-insert failure. Either can be right. A separate elastomer or a screwed insert is often the process you can still change. Choose the bond because the duty needs it. If you keep it, put it in the quote and in the assembly trial.

Invented product, not a BrahmWorks build. A desktop instrument, two moulded covers, assumed quantity 2,000 in the first year and a hope of more later. The snap changed twice in the last month. A spreadsheet says a four-cavity tool wins on piece price.

The disposition for the exercise: do not cut four cavities of a moving snap. Cut one cavity. Freeze draw, parting line and show faces. Let the moulder say whether this resin and this texture belong in a softer tool or in steel; do not let the spreadsheet say it. Measure the latch and the board standoffs on the first article before anyone orders cavity two. The hoped-for later volume is not evidence that the geometry is done.

Assume one production-class tool at INR 5,00,000, a moulded cover at INR 35, a machined cover at INR 520, and one machining setup at INR 15,000. Tax, freight, rejects, assembly, colour changes and engineering are excluded on purpose.

Fixed difference = 5,00,000 − 15,000 = INR 4,85,000. Saving per part = 520 − 35 = INR 485. Crossover = 4,85,000 / 485 = 1,000 parts.

On these lines only, machining plus one setup costs less below 1,000 parts, and the tool costs less above 1,000. Now assume one tool change at INR 70,000 before you have shipped, still hypothetical. New fixed difference = 4,85,000 + 70,000 = INR 5,55,000. Crossover = 5,55,000 / 485 = 1,144.3, so the tool pulls ahead only from about the 1,145th part.

The second sum is the one startups skip. A tool that moves once is not the tool in the first spreadsheet. Neither sum is a quote. If the machined part is a different resin, the crossover is not even a comparison of the same product. Use it to see the volume at which the tool can pay, then throw it out when the geometry is still moving.

Taking that geometry into a repeatable build, with drawings someone else can follow, is How to Take a Hardware Prototype to Production.

Each line has an owner, or it is not decided.

• The tool’s job is written: what these shots are allowed to prove, and what they are not.
• Draw direction, parting line and draft are in the model, or explicitly still open.
• Resin family is named. Grade and colour sample are named or listed as open.
• Show faces and forbidden gate and ejector zones are marked.
• Every undercut is an action, a shut-off, or deleted.
• Prototype cheats are listed so they cannot be mistaken for requirements.
• Cavity count is one until the critical features stop changing, or a reason otherwise is written.
• Inserts and second shots are in the quote or out of the product.
• Regrind is forbidden or capped for the pilot.
• First-article dimensions, cavity and resin lot are defined before the tool is released.
• Likely corrections are identified as a further cut or a weld, with the toolmaker.
• The quantity in the crossover is labelled as an assumption.

Only if the question is moulding. A first working unit should answer function, with its cheats written down. Cutting a tool to answer a firmware or a radio question freezes an outline you are about to change. Mould when the shape is the thing you need to learn, and when you can say which dimensions matter.

From stability and from quantity, not from the lowest piece price on a part that is still moving. One cavity until the latch, the seal and the board fits repeat. Add cavities when a change to the part is unlikely and the press time is actually your constraint. A family tool needs both parts to be stable and to share a colour and a volume.

They may be assuming different texture, different polish or a different draw. Make them mark it on your model. Disagreement you cannot see is two different parts. You pick a draw and a finish, then ask both to quote that, rather than averaging two silent assumptions.

Sometimes, inside a family, with a new trial and new measurements. A change of shrink, fill or chemical behaviour is a new part. Glass fibre, a flame-retardant package or a switch from an amorphous to a semi-crystalline resin can move dimensions and warp enough to scrap the fits you already accepted. Treat a resin change as a new first article, not as a drop-in.

Bring the model, the cheat list, the quantity you are actually willing to buy, and what the tool is supposed to prove. BrahmWorks can say whether that pack is ready to quote, or which decision is still a CAD change.

Review the next stage with BrahmWorks