Design for Manufacturing (DFM) Reviews: How Chinese Factories Do Them

A DFM review is an engineering check on your design files before tooling or production begins. The factory's engineers examine your CAD model and drawings for features that are hard, expensive, or impossible to manufacture as drawn — thin walls, missing draft angles, undercuts, unreachable tolerances — and return a report of suggested changes. In China, this review usually happens right after you submit 3D files and before the factory finalizes its quote.

Design problems are cheapest to fix on paper, and most Chinese factories include the review at no charge. For buyers who cannot read engineering feedback in Chinese or negotiate tradeoffs across time zones, CN Ally manages this stage: collecting DFM reports, clarifying what the factory is really asking for, and helping you decide which suggestions to accept.

What does a DFM review actually check?

A DFM review answers one question: can this part be made reliably, at the target cost, on the equipment this factory owns? Everything the engineer flags traces back to that. Five groups of checks cover it.

Geometry. Can the process physically produce every feature — sharp internal corners, zero-draft walls, undercuts it cannot release? These are hard blockers: they force a design change or a costlier process.

Dimensions and tolerances. The drawing says what the part needs; the factory checks what it can hold. Over-tight tolerances are among the most common and expensive findings, because they change the process plan — more setups, slower cycles, secondary operations.

Material and finish. The resin or metal grade has to flow, fill, or bend the way the geometry demands. Finish callouts must be achievable with the factory's polishing or blasting grades.

Assembly fit. When parts mate, the engineer checks that critical interfaces still work after suggested geometry changes. A thinner wall is easy; one that no longer grips its mating component is a failure.

Cost drivers. The report flags features that are manufacturable but needlessly expensive — deep ribs that slow the cycle, an undercut requiring a slider, engraving that could be a label.

A serious review touches all five. A cursory one only confirms the part fits on the machine.

How do Chinese factories conduct a DFM review?

1. You submit the design package. At minimum, a 3D CAD file in STEP or IGES format plus a 2D drawing — which matters more than most buyers expect, since it carries the tolerances, material specification, finish requirements, and notes a 3D file cannot. Include the expected quantity: a feature acceptable at 500 units may need redesign at 50,000.

2. An engineer, not a salesperson, opens the files. At competent factories the package goes to a mold designer or process engineer who works with the relevant equipment daily. Larger shops also run automated DFM tools that scan the CAD model — radii, hole sizes, internal corners, wall thickness — the way instant-quoting platforms do.

3. You receive a DFM report. Usually a PDF with annotated screenshots: arrows at flagged features, each with the problem and a proposed fix. For complex molded parts it may include mold-flow analysis — predicted filling, weld lines, air traps — plus a proposed gate and parting-line layout. Expect it within a few days of sending usable files.

4. You confirm or push back. The factory revises its quote after the design discussion, since accepted changes alter tooling cost and cycle time. Nothing proceeds until you sign off on the agreed design. Never shortcut this step: confirming without reading the report means paying for a design you may not have understood.

One warning sign: a factory that quotes within hours of receiving only a photo or a vague description did no real review. A meaningful DFM review requires the files and the engineering time to study them.

What do engineers check first: wall thickness and ribs?

Wall thickness is the first thing a molding engineer looks at, because nearly every cosmetic and structural defect traces back to it. The rule is simple: keep walls as uniform as possible.

Thick sections next to thin ones cool at different rates, producing sink marks, internal voids, warpage, and dimensional variation. Where a thick section is unavoidable — a mounting boss, for instance — the engineer will suggest reducing it or designing around it rather than fighting the physics.

Ribs get the same attention. Ribs add stiffness without thickening the whole part, but a rib too thick relative to the wall it sits on creates a sink mark on the opposite surface. The standard rule of thumb: rib thickness around half the adjacent wall thickness, with a generous radius at the root. Reports regularly flag ribs drawn at full wall thickness.

For CNC parts the concern flips: thin walls deflect under cutting forces and vibrate, ruining finish and accuracy. Factories flag walls they cannot hold flat and suggest thickening, temporary support, or a different strategy. Sheet-metal reviewers watch flange lengths and bend zones, where stretching and cracking concentrate.

Why do draft angles and undercuts matter so much?

Draft is the slight taper on vertical walls that lets a molded part release from the steel. Without it, the part drags against the mold on ejection, causing scuff marks, distortion, or a part stuck in the tool. The widely cited guideline is a minimum of about 0.5 degrees per side, with 1 to 3 degrees preferred — and more on textured surfaces, which grip harder during ejection (manufacturing.net).

CAD defaults to perfectly straight walls, so reviewers flag every zero-draft surface — or ask you to accept the ejection risk in writing.

Undercuts are the more expensive problem. An undercut is any feature that prevents the part from pulling straight out of the mold: a side hole, a snap hook, an internal thread form. Molding them requires extra tooling — sliders, lifters, collapsible cores — each adding cost, slowing the cycle, and creating maintenance points. DFM feedback often proposes eliminating the undercut by redesigning the feature, splitting the part, or moving the parting line so the feature forms naturally.

The savings compound: one undercut rarely breaks a budget, but each adds a mechanism to the mold, and mechanisms are where molds fail and maintenance bills accumulate.

What tolerances will a factory flag, and why?

Every process has a natural capability — a tolerance band it holds comfortably in routine production. Pushing tighter means slower feeds, more rigid setups, extra measurement, sometimes secondary grinding. Injection molding follows the same curve: general dimensions are easy, but tight tolerances on plastic parts fight shrinkage variation and demand harder tool steel and tighter process control.

Factories flag two situations. The first is a tolerance the process genuinely cannot hold. The second, far more common, is one it can hold but that serves no function: a ±0.05 mm callout on a non-mating surface, a cosmetic edge dimensioned like a sealing face. The reviewer's question is always the same — what does this tolerance do for the product? If the answer is nothing, relaxing it cuts cost without touching quality.

Mark which tolerances are truly critical — sealing surfaces, bearing fits, mating interfaces — and which are general. A drawing where every dimension carries a tight tolerance tells the factory nothing, and the quote will price the worst case.

How does DFM differ for injection molding, CNC, and sheet metal?

Each process has its own failure modes, so the review focuses on different things:

Check · Injection molding · CNC machining · Sheet metal

• Wall consistency: Uniform walls; ribs ~50–60% of wall; radiused roots · Thin walls that deflect or chatter under cutting forces · Uniform thickness; watch flange and bend zones
• Release / access: Draft on vertical walls; undercuts need sliders or lifters · Internal corners need radii matching the tool; deep pockets limited by tool length · Bend radii; features too close to bend lines distort
• Tolerances: Tight tolerances fight shrinkage; demand harder steel and process control · Tight tolerances add setups and secondary operations · Tolerances limited by bending repeatability; holes near bends drift
• Material fit: Resin flow length vs. wall thickness; shrinkage of the grade · Alloy machinability; grain direction and hardness · Bend radius vs. thickness and hardness; cracking risk
• Typical cost driver: Undercuts and sliders; texture; cosmetic requirements · Small internal radii forcing small tools and slow feeds · Many bends, tight radii, secondary hardware
• Reviewers propose: Gate and parting-line layout; mold-flow results; rib redesign · Larger radii, looser non-critical tolerances, split setups · Moved holes, larger bend radii, self-clinching fasteners

Two patterns cut across all three. Sharp internal corners are a problem everywhere: stress concentrators in molded parts, unreachable for a round milling tool, crack starters in bent metal. And the reviewer always asks what is functional — a feature that does nothing but look engineered is a candidate for simplification.

For products combining processes — a machined insert overmolded in plastic, a sheet-metal bracket on a molded housing — expect the review to cover the interfaces too. Assembly tolerances across processes are where multi-part products most often fail.

How should you prepare design files for a useful DFM review?

The quality of feedback is mostly determined by what you send. A complete package gets reviewed once and thoroughly; an incomplete one produces clarifying questions instead of engineering.

Send STEP files. STEP is the universal standard. Avoid sending only STL meshes for machined or molded parts — they limit analysis and quoting.

Include a real 2D drawing. The 3D model shows the shape; the drawing carries the engineering intent: critical dimensions with tolerances, material grade, surface finish, threads, color and texture references, notes. A model without a drawing is a shape without requirements.

Specify the finish precisely. "Smooth" means nothing. Give a roughness value (Ra) or a texture grade, note which surfaces are cosmetic, and say where witness marks are acceptable. Referencing a standard like VDI or SPI removes ambiguity.

State material, color, and volume. The resin grade or alloy, color reference, required certifications, and expected order quantity. Material interacts with geometry, and quantity determines whether a suggested change pays for itself.

Mark what is critical and why. "This face seals against a gasket," "this bore is a bearing fit," "this surface is customer-visible." Engineers protect what matters and simplify what does not. Without those notes, everything is treated as critical — which produces a cautious, expensive quote.

Name an engineering contact. The fastest reviews happen when the factory's engineer can ask your designer a direct question instead of routing it through sales staff and time zones.

How do you read and respond to a DFM report?

Each item has three parts: the flagged feature, why it is a problem, and a proposed change. Items sort into two categories.

Required changes are geometry that blocks production as drawn — an unmachinable corner, a wall that cannot fill, an undercut with no release strategy. These are not negotiable. Accept the redesign, change the process, or accept that the part cannot be made this way at this factory.

Recommended changes reduce cost, cycle time, or defect risk without being strictly necessary — relaxing a tolerance, adding draft, simplifying a rib. Here you have real choices. Accepting most of them usually lowers the quote; declining some preserves your design at a known cost.

The feedback items below appear most often in DFM reports from Chinese factories:

Typical DFM feedback · What it means · Cost / quality impact

• "Add draft to vertical walls": Zero-draft surfaces drag or stick during ejection · Prevents scuffs and stuck parts; negligible design cost
• "Reduce rib thickness / add root radius": Thick ribs cause sink marks on the visible surface · Protects cosmetic quality; cost-neutral
• "Add radii to sharp internal corners": Sharp corners concentrate stress and are hard to produce · Stronger parts, easier manufacturing
• "Relax tolerance on non-critical features": Tight tolerances add setups and slow cycles · Often the largest single cost reduction in the review
• "Redesign to eliminate undercut": Undercuts need sliders or lifters in the mold · Removing one saves tooling cost and cycle time
• "Split part into two pieces": Geometry is unmanufacturable as one piece · Adds an assembly step but may unlock the design
• "Move or resize gate location": Current gate placement risks weld lines or air traps · Better strength and appearance; no added cost
• "Increase wall thickness in thin sections": Thin areas may not fill or may deflect · Prevents scrap and weak parts

Respond item by item: accepted, rejected with a reason, or needing discussion. Vague replies — "looks fine," "please proceed" — are where misunderstandings breed. If you reject a recommendation, say why: "this tolerance is a sealing surface, keep it" lets the engineer propose an alternative instead of guessing. Keep the exchange in writing, in the same thread as the report, so the agreed design is traceable when first articles arrive.

If a discussion stalls — limited English, suggestions that change the product's function, or feedback that looks aimed at simplifying the factory's life — get an independent read. A sourcing agent can pressure-test the feedback against other suppliers' reviews of the same files. If three factories flag the same feature, it is your design; if only one does, it may be their capability.

What are the red flags when a factory skips DFM?

The absence of a review is itself information. Watch for these signs.

A quote with no engineering questions. A firm price the same day you send 3D files, with zero questions about tolerances, material, finish, or quantity — no engineer looked at your design. The surprises arrive later in rework or parts that do not fit.

Straight to tooling payment. Factories pushing for the mold deposit before discussing the design are optimizing for cash flow. A serious molder settles the design first, because every change after steel is cut comes out of someone's margin.

"No problem" to everything. Any tolerance, any geometry — all fine, no changes needed. Real manufacturing has constraints. The most capable factories push back with specific, reasoned objections.

Verbal feedback only, no marked-up report. A proper review produces a document you can answer point by point. Feedback delivered only as chat messages cannot be tracked when first articles are inspected.

Suggestions that only simplify the factory's work. Recommendations should balance your product's function against manufacturability. If every suggestion removes a customer-facing feature, cross-check the same files with a second supplier.

No interest in the 2D drawing. A factory quoting from the 3D model alone, never asking for tolerances or finish, is estimating weight and machine time — not reviewing manufacturability.

Treat a missing or superficial DFM review as a supplier-selection signal. A factory audit before committing to tooling is the heavier version of the same due diligence — verifying that the engineering capability the sales team described actually exists on the shop floor.

Frequently asked questions

What does DFM stand for in manufacturing?

DFM stands for Design for Manufacturing (sometimes "Design for Manufacturability") — the practice of designing parts so they can be produced reliably and economically with the intended process. A DFM review is the formal check where the factory's engineers verify your design against that standard before production.

How long does a DFM review take?

A single CNC part with a clear drawing can come back in a day or two. An injection-molded assembly with multiple components and cosmetic requirements typically takes several days, longer with mold-flow simulation. Delays usually come from incomplete files — missing drawings, unspecified finishes, unknown quantities — not from the engineering work itself.

Do Chinese factories charge for DFM reviews?

Usually not. Most treat the review as part of quoting and product development — Yijin Solution, a Chinese CNC manufacturer, describes its DFM review and design modification suggestions as free services bundled with quoting. The incentive is straightforward: catching problems early avoids rework that costs far more than the engineering hours.

Should I accept all DFM suggestions from the factory?

Evaluate each one. Accept changes that do not affect function — added draft, relaxed non-critical tolerances, rib redesigns — since these cut cost without touching the product. Push back on anything altering a functional or cosmetic requirement, but expect the harder version to be priced accordingly.

What is the difference between DFM and DFA?

DFM concerns whether individual parts can be made economically. DFA (Design for Assembly) concerns whether the parts go together easily — fewer fasteners, self-locating features, sensible assembly sequence. They are often reviewed together, since a part that is easy to mold but impossible to assemble is still a bad design.

What file format should I send for a DFM review?

STEP (.stp) is the safest choice — nearly every CAD and CAM system reads it accurately. Include a 2D drawing in PDF with tolerances, material, finish, and notes. Avoid sending only STL meshes for machined or molded parts, since meshes lack the true surfaces engineers need.

When to push back, when to accept, and what to do this week

Accept changes that do not alter what the product does or how it looks: added draft, relaxed non-critical tolerances, rib redesigns, gate repositioning, radii on sharp corners. These are free or money-saving.

Push back when the suggestion touches function, fit, or appearance — but with engineering reasoning, not authority. "Keep this tolerance, it is a sealing surface" is productive. "Just make it as drawn" is not, and it usually buys a higher quote plus a quiet bet that the part fails inspection.

Walk away when there is no DFM review at all: same-day quotes on complex parts, no questions about your drawing, pressure to pay for tooling first. That is not efficiency — it is a factory telling you it has no engineering process, and the cost arrives later as rework or delays.

If your files are ready, assemble the package — STEP file, 2D drawing with tolerances and finish, material and quantity — and request a DFM review before a final quote. Send the same package to two suppliers and compare not just the prices but the reviews. The factory that finds real problems in your design before taking your money is usually the one you want making your product. If you would like someone to manage that comparison across time zones, write to hi@cnally.com — or see how quality control and the rest of the process protect the design once production starts.