CNC vs 3D Printing vs Injection Molding

A lower piece price is not the same decision. The crossover later is hypothetical teaching maths: not a BrahmWorks price, not a shop rate, not a yield and not a lead time. A printed part, a machined part and a moulded part are three different objects. The arithmetic only says when the invoices cross if you pretend they are the same part.

A print, whether deposited, cured or sintered, builds the shape without a draw direction. It will happily make zero draft, a trapped undercut and a wall that varies for no reason. Strength depends on the process and, for many printers, on direction. A screw boss that held in the print may strip in the resin. A clip that survived may have survived because the stand-in polymer yields differently. Layer lines are not a texture. Use the print to learn the hand, the board envelope and the assembly sequence. Write down every cheat so a later tool does not inherit it.

CNC removes metal or plastic from a block with a cutter. The material can be the production grade, or close, which is its main advantage over a casual print. It does not need draft, so a machined enclosure teaches nothing about ejection. It does not produce weld lines, sink from a local thick section, or the shrink a cavity is cut to. Sharp internal corners are limited by the cutter radius. Thin ribs can chatter. A beautiful machined housing can be an unmouldable one. Use it when you need a stiff, stable part for a radio trial, a seal trial or a fit check in the real grade, and when the quantity is still small. Do not use a machined pass to sign off a moulding tolerance.

Injection moulding repeats a cavity. It will show you fill, sink, warp, weld lines and the dimensions of that resin in that tool. It will not show you a different resin, and it will punish a model that was never drafted. The tool is the slow, expensive place to discover that the parting line crosses the logo. Mould when the shape is the question and the quantity makes the tool rational, or when only moulded parts can answer a moulding question such as a living hinge or a textured show face. Do not mould to answer a firmware question.

The table sorts the decision. It is not a claim about a particular printer, shop or press. Ask the supplier what their process will not show.

Founders switch process because a spreadsheet crossed a quantity. That switch is legitimate when the part is the same design, in the same material duty, and the fixed cost is the tool you are willing to risk. It is not legitimate when the printed enclosure never had draft, the machined one was a different polymer, and the moulded quote assumes a shut-off you have not drawn. Crossing the number does not convert evidence. Tests run on the print do not transfer to the moulding. Say so in the record, or a later failure will be blamed on a factory that built a different part.

A fourth option is easy to omit: stay on CNC, or on a softer tool, until the latch, the seal and the standoffs stop moving. The piece price looks worse. The chance of scrapping a cavity looks smaller. Your quantity decides, not a rule that real products are moulded.

Invented enclosure for a small indoor instrument. The duty, assumed for the exercise, allows either a machined polymer or a moulded one of the same family. The printed route is a different polymer, used only to check the outline. No route here is a BrahmWorks build, and no route has a reported test result.

The team needs 30 units to finish firmware on a shape that may still move, then expects a few hundred if the trial holds, and talks about a few thousand later. The process choice for the 30 is printing or CNC, because a tool cannot teach them anything those 30 will not invalidate. CNC wins over printing if they need the real grade for a boss or a snap. Printing wins if they only need the envelope and will throw the shells away. The mould is a later decision, gated on the geometry freezing, not on the enthusiasm in the volume slide.

Assume, for teaching only:

• Printed shell: no setup, INR 860 each. Different polymer from the other two.
• CNC: setup INR 16,000 once, then INR 460 each. Same family as the moulded part.
• Mould: tool INR 2,80,000, then INR 60 each.

Print against CNC, ignoring the material difference for a moment: setup difference INR 16,000, piece difference 860 − 460 = INR 400. Crossover = 16,000 / 400 = 40 parts. From the 41st part, CNC is cheaper on these lines. That comparison is still dishonest if the print cannot do the job, because a cheaper part that answers the wrong question is not cheaper.

CNC against mould: fixed difference = 2,80,000 − 16,000 = INR 2,64,000. Piece difference = 460 − 60 = INR 400. Crossover = 2,64,000 / 400 = 660 parts. On piece plus one setup plus tool, CNC costs less below 660 and the mould costs less from the 661st. Tax, freight, rejects, assembly, colour, engineering and a second setup are excluded.

Print against mould, as pure arithmetic: 2,80,000 / (860 − 60) = 2,80,000 / 800 = 350 parts. Do not use 350 as a decision. At that count you would be comparing a stand-in polymer with a production tool. The money can cross while the evidence does not.

If the geometry changes once and the tool change is extra money, the 660 moves up, in the same way any added fixed cost moves it. Add INR 40,000 of tool change: (2,64,000 + 40,000) / 400 = 760 parts. One change is not free.

How those parts then get built, inspected and repeated by someone other than the designer is How to Take a Hardware Prototype to Production.

• The question this batch must answer is one sentence.
• The material is the production grade, or the record says what the stand-in cannot prove.
• Draft, wall and undercuts are either designed for moulding or labelled as cheats.
• Quantity is the buy you will place, split by what you will throw away.
• Setup, tool and piece are separate. Tax and freight are not buried.
• A crossover is not used to compare different polymers.
• Tests are tied to the process that produced the unit.
• A tool is gated on the features that have stopped moving.
• Secondary machining is counted in the CNC and the moulded routes.
• The route you reject is written down, with the reason.

Yes, when the quantity stays low, the material and the process control meet the duty, and you are willing to specify orientation, inspection and a supplier who can repeat it. That is a production decision, with its own drawing controls. It is not what most "prototype prints" are. Do not slide from one to the other because the machine looks the same.

You can, at the quantity where the tool never pays, and when the geometry does not need moulded features such as a living hinge, a designed texture or a closed thin shell a cutter cannot reach. At higher quantity the piece price and the spindle time dominate. Also check whether the machined cosmetics and the sharp internal corners are actually what you intend to sell.

Only for what it can run. If the resin, the texture, the cooling and the cycle are not the production ones, say which conclusions are withheld. A few dozen shots that fill and hold a dimension are real evidence of those shots. They are not a life test of the steel and not a piece-price proof.

Keep them as the record of what was faked, not as the reference for inspection. Inspect the moulded part against the drawing. Inspecting it against a print reintroduces the cheats. Scrap or quarantine the prints so they cannot be assembled into a unit that is then called a production sample.

Bring the question the next batch must answer, the quantity, and whether the material has to be the real grade. BrahmWorks can say which process is allowed to answer it, and which evidence will not survive the switch.

Review the next stage with BrahmWorks