Resources
DFM & DFA Guidelines
Design guidelines for plastic injection molded and die cast parts, plus design for assembly. Written so you can check your own model before sending it — most of what we flag in DFM review is on this page, and every item caught at CAD stage is one that doesn't have to be cut out of steel later.
These are general engineering guidelines and typical starting figures, not part-specific advice. The right number for your part depends on resin or alloy, geometry, and tolerance, and gets confirmed during DFM review. Wrex does not select materials or alter customer designs in-house — you own the design and remain responsible for verifying any change against fit, form, function, and application. See Engineering Support for where that line sits.
Plastic injection molding DFM
Almost every molding defect traces back to one root cause: material cooling at different rates in different places. Uniform walls, drafted faces, and radiused corners are all the same idea applied in different ways.
Uniform wall thickness
Hold a consistent nominal wall. Typical ranges: 0.040–0.140 in (1.0–3.5 mm) for most engineering resins. Where thickness must change, transition gradually over at least 3× the wall.
Why: Plastic shrinks as it cools, and thick sections cool last. A thick section next to a thin one shrinks differently, which is what produces sink marks, voids, internal stress, and warp.
Cost impact: Thick walls also cool slowly — and cooling is roughly 80% of cycle time, so a heavy wall costs you on every single part, forever.
Draft on every vertical face
1–2° per side is a normal starting point for untextured surfaces. Deep draws want more. Textured surfaces need substantially more — commonly 1° per 0.001 in of texture depth, on top of the base draft.
Why: Without draft the part drags on the steel as it ejects, causing scuffing, ejector-pin push marks, stress whitening, or a part that sticks in the tool.
Cost impact: Adding draft after the tool is cut means re-cutting steel. It is nearly free at CAD stage.
Rib proportions
Rib thickness 50–60% of the nominal wall at the base. Height no more than about 3× the wall. Space ribs at least 2× the wall apart. Draft the rib and radius its base.
Why: A rib thicker than ~60% of the wall creates a thick spot at the junction and pulls a visible sink mark on the opposite show surface.
Cost impact: Ribs are the cheap way to add stiffness — far cheaper than thickening the wall, which costs cycle time on every shot.
Boss design
Outer diameter about 2× the hole diameter. Wall of the boss 50–60% of the nominal wall. Support tall bosses with gussets rather than thickening them. Never run a boss straight into a thick base.
Why: A solid or thick-walled boss is a mass of material that sinks, voids, and traps stress exactly where a screw will later load it.
Radius every corner
Internal radius at least 0.5× the nominal wall; external radius = internal radius + wall thickness. Avoid sharp internal corners entirely.
Why: A sharp internal corner is a stress concentrator and a flow restriction. It is the most common origin point for cracks in service.
Avoid undercuts where you can
Features that trap the part in the mold need a side action, lifter, or cam. Ask whether the undercut can be redesigned as a through-feature or moved to the parting line.
Why: Each action is extra steel, extra motion, and extra maintenance in the tool.
Cost impact: Every side action adds tool cost and a wear point. Designing one out is usually the single largest tooling saving available to you.
Think about gate location early
Tell us which surfaces are cosmetic. Gate position drives fill pattern, knit-line placement, and where the gate witness mark ends up.
Why: Knit lines form where two flow fronts meet and are both a cosmetic and a strength feature. Their position is decided by the gate, not by luck.
Tolerance only what matters
Call out the few genuinely critical dimensions and let the rest run at standard tolerance. Molded tolerances are wider than machined ones and vary by resin shrink rate.
Why: Every tight tolerance on a molded part is a cost — in steel, in inspection, and in scrap.
Cost impact: This is the single most common source of avoidable cost on an incoming print.
One tool, one resin
A tool is cut to a specific resin's shrink rate. Decide the material before the tool is built.
Why: Swapping resin later changes part dimensions, because the steel was sized for the original material's shrink. It usually means reworking the tool.
Die casting DFM
Die casting follows the same logic as molding — uniform walls, draft, radii — but the consequences differ. In plastic a heavy section gives you a cosmetic sink. In metal it gives you internal porosity you cannot see until the part is machined or sectioned.
Thin, uniform walls
Aluminum typically 0.060–0.180 in (1.5–4.5 mm); zinc can go thinner, from about 0.030 in (0.75 mm). Keep the wall uniform and avoid heavy sections.
Why: Molten metal freezes fast. Thick sections trap the last liquid metal, and that is where shrinkage porosity forms — internal voids you cannot see until the part is machined or sectioned.
Cost impact: Thin, even walls fill and freeze predictably. Heavy sections are the main cause of scrap in die casting.
More draft than plastic
Typically 1–3° on outside surfaces and more on inside/cored surfaces — interior walls commonly need 2–3× the draft of exterior ones.
Why: The casting shrinks onto the core as it solidifies, gripping it. Interior surfaces therefore need more draft to release than exterior surfaces do.
Generous fillets and radii
Radius all internal corners. A fillet roughly equal to the adjoining wall thickness is a reasonable starting point.
Why: Sharp corners restrict metal flow, cause turbulence and gas entrapment, concentrate stress, and erode the die faster — a die-life issue as well as a part issue.
Design for the parting line
Decide early where the die splits and keep critical features off it. Expect flash at the parting line, and put it where it can be trimmed without touching a functional surface.
Why: Flash is normal and gets removed in finishing. Flash across a sealing face or a datum is a finishing problem you pay for on every part.
Allow for ejector pins
Ejector pins leave visible witness marks. Tell us which faces are cosmetic so pins are placed elsewhere.
Why: Pins have to push on something solid. If every flat face is cosmetic, the tool designer has nowhere good to put them.
Cored holes over drilled holes
Cast holes in where you can, and add machining stock where the hole must be precise. Cored holes need draft; drilled holes do not.
Why: Casting a hole is free per part. Drilling it is a secondary operation on every part.
Leave machining stock where it counts
Where a face or bore needs a machined tolerance, add stock for it. As-cast tolerances are looser than machined tolerances.
Why: Trying to hold a machined-level tolerance as-cast produces scrap. Machining a designated surface is predictable and cheap by comparison.
Expect porosity — design around it
Conventional (non-vacuum) high-pressure die casting produces castings with some internal porosity. Do not place pressure-tight or heavily loaded features in the thickest sections.
Why: Wrex runs conventional non-vacuum casting. Parts requiring certified zero porosity — common in automotive, aerospace, and defense — are not a fit for this process.
Ribs instead of bulk
Add stiffness with ribs and gussets, not by thickening the wall.
Why: Same principle as plastic, for the same reason: bulk means slow freezing, which means porosity and shrink.
Design for assembly (DFA)
DFM asks whether a part can be made. DFA asks what it costs to put together — and since we assemble finished goods here, it is worth planning both at the same time rather than discovering assembly problems after the tooling is committed.
Reduce the part count
Before optimising an assembly, ask which parts can be combined or eliminated. Can two molded parts become one with a living hinge, or a snap feature replace a bracket and two screws?
Why: The cheapest part to assemble is the one that no longer exists. Every part removed also removes its inventory, inspection, and failure mode.
Cost impact: This is the highest-leverage DFA decision by a wide margin.
Design self-locating features
Add pins, bosses, lips, or steps so parts can only sit one way and drop into position without fixturing or measurement.
Why: An operator who has to align a part by eye is a source of variation. A part that locates itself is assembled the same way every time.
Assemble from one direction
Aim for straight-down, single-axis assembly — ideally with gravity helping. Avoid designs that require flipping the assembly mid-build.
Why: Each reorientation is handling time and a chance to get it wrong.
Make wrong assembly impossible
Use asymmetry deliberately. If a part must go in one way, give it a feature that physically prevents the other way. If orientation genuinely does not matter, make it fully symmetric.
Why: Poka-yoke beats work instructions. Half-symmetric parts — nearly but not quite reversible — are the worst case and a reliable source of defects.
Fewer, larger fasteners — or none
Standardise on as few fastener types as possible. Consider snap fits, heat staking, or ultrasonic welding where a permanent joint is acceptable.
Why: Every distinct fastener is another bin, another driver setting, and another chance to fit the wrong one.
Cost impact: Wrex runs ultrasonic welding and mechanical assembly in-house, so joining method can be planned alongside the molding.
Leave room for hands and tools
Check that a driver, gripper, or hand can actually reach every fastener and connector with the assembly in its build position.
Why: Access problems are invisible in CAD and obvious — expensively — on the bench.
Say if it is an ESD build
Tell us at RFQ if the assembly needs ESD control, and identify any ground or bonding contacts that must stay bare through finishing.
Why: ESD-controlled assembly is a different process with different handling. Powder coat is an insulator — a ground contact that gets coated is a failed assembly.
Design the test in
If the assembly gets checked, give the check something to grab — an access point, a visible indicator, a datum that survives to final assembly.
Why: Inspectability designed in at CAD stage costs nothing. Retrofitted onto a finished design it costs fixtures and time.
What to send for a DFM review
DFM review is included in quoting, not billed separately. It is only as good as what we receive, so:
- 3D model (STEP or native solid) — molds are built from the solid model
- 2D drawing for tolerances, material, surface finish, and critical dimensions
- Exact resin or alloy, including grade and any fill percentage
- SPI surface finish callout for cavity and core (required on every new plastic part)
- Which surfaces are cosmetic, and which faces may carry gate, ejector, or parting-line witness
- Annual volume and expected program life — this drives the tool class
- Any regulatory or compliance requirement (medical, RoHS, documentation)
Related
- DFM rules of thumb & pre-RFQ checklist — the short version
- Tooling Specifications — tool classes, steels, and surface finishes
- Material Guide — resins and alloys we run
- Finish & Coating Guide — masking, film build, and finishes that cannot be combined
Ready to talk about your project?
Send your drawings or describe your part — an engineer will follow up, not a sales queue.