• English
  • 中文
  • Français
  • Deutsch
  • Español
  • Português
  • 日本語
  • 한국어
  • Nederlands
  • Türkçe
  • Bahasa Indonesia
  • ไทย

CNC Machining vs Die Casting: Cost, Tolerances and When to SwitchCNC 加工 vs 压铸:成本、公差与切换时机的完整对比

Choosing between CNC machining vs die casting is rarely a question of which process is "better". It is a question of which process is cheaper at *your* volume, with *your* geometry, at *your* tolerance, on *your* calendar. This guide gives you the numbers that actually decide it: tooling amortisation, per-piece cost curves, a break-even formula that accounts for design changes, and the tolerance and cosmetic limits that override the arithmetic entirely. It also covers the option most programmes end up using — cast the near-net shape, then machine the features that matter.

The short answer

If you need parts next week, or your design is still moving, or your tolerances are tighter than about ±0.05 mm, or the part has to look like a consumer product — CNC machining wins, and usually by a wide margin. If you need tens of thousands of identical parts with a frozen design and moderate tolerance, die casting wins on unit cost, sometimes by a factor of five. Everything in between is arithmetic, and the arithmetic is what this article is for.

SituationChooseWhy
1–500 parts, any yearCNC machiningNo tooling to amortise
Design still changingCNC machiningAn ECO costs a program change, not a new tool
First article needed in under 3 weeksCNC machiningDie tooling alone takes 4–10 weeks
±0.01 mm featuresCNC machiningAs-cast capability is ±0.1–0.3 mm
Cosmetic anodised faceCNC machiningCast alloys anodise grey and blotchy
Pressure-tight without impregnationCNC machiningCasting porosity is inherent
10,000+ parts/year, design frozenDie castingCycle time collapses unit cost
Thin-walled, ribbed 3D shell at volumeDie castingNear-net shape beats hogging from billet
Big aluminium housing, some tight facesBothCast near-net, machine the critical features

One caution before the numbers: the cost figures below are typical 2025–2026 ranges for aluminium parts quoted in Asia and North America. They are for *relative* decisions, not for budgeting a purchase order. Get real quotes at your volume before you commit.

What each process actually is

Before the arithmetic, it helps to be precise about what the CNC machining vs die casting decision is really comparing: one process adds metal to a cavity, the other removes it from a solid block. That single difference explains almost every downstream consequence — cost shape, tolerance, surface, and lead time.

High-pressure die casting

Molten aluminium alloy — most commonly ADC12 (equivalent to A383) or A380 — is injected into a hardened steel tool at 40–100 MPa and held under pressure while it solidifies. Cycle times for a palm-sized part run 30–60 seconds. The tool, usually H13 steel, is EDM'd, hardened to 46–50 HRC and fitted with slides, lifters and cooling channels. That tool is the entire economics of the process: it is expensive, it takes weeks to build, and every part it makes is cheap.

Because the metal is injected, die casting produces near-net shapes that would be wasteful to cut from solid: thin walls, ribs, bosses, internal cavities, and complex three-dimensional contours in one shot. Because the metal solidifies from a turbulent fill, it also traps gas. That is where porosity comes from, and porosity is the root of most of the problems people have with cast parts.

CNC machining

CNC machining removes material from a solid billet, plate or bar using rotating cutters on 3-, 4- or 5-axis centres. There is no dedicated tooling beyond workholding and a CAM program, so the first part costs roughly what the hundredth part costs. Setup is the dominant fixed cost, typically a few hundred dollars per operation.

What machining buys you is precision and material integrity. Wrought 6061-T6 has no porosity, a uniform grain structure and a tensile strength around 310 MPa, versus roughly 240 MPa and 1–3% porosity for ADC12 as-cast. Tolerances of ±0.01 mm are routine on a CNC machining tolerance basis; ±0.005 mm is achievable on specific features. And because the surface is wrought metal, it anodises cleanly — which is why every cosmetic audio panel we make is machined, never cast. Machining also handles geometry casting cannot: deep pockets, cross-drilled holes, sharp internal corners and true undercuts without slides. The cost side is covered in detail in our CNC machining cost guide.

The economics: tooling, per-piece cost and break-even

Where the money goes in each process

The two processes have almost opposite cost structures, which is exactly why the crossover in any CNC machining vs die casting comparison is so sharp: one front-loads everything into tooling, the other spreads a small fixed cost across every single part.

Cost elementDie castingCNC machining
ToolingUSD 12,000–120,000USD 0–2,000 (programming + fixtures)
Setup per runLow once runningUSD 150–500 per operation
Cycle time30–60 s per shot5–45 min per part
Material utilisationNear net shape, little waste50–80% becomes chips
Engineering changeUSD 1,500–18,000 per changeProgram edit, near zero
First-part lead time4–10 weeks (tool build)3–10 days
Economic MOQ500–1,000 pcs1 pc

Break-even, worked three ways

Static break-even is simple: tooling cost divided by the per-piece saving.

Break-even quantity = tooling cost ÷ (CNC unit price − die cast unit price)

Worked example, a 150 × 120 × 45 mm aluminium bracket:

Annual/order quantityCNC unit priceDie cast unit price (incl. amortised tooling)Cheaper
1–10 pcsUSD 180not economicalCNC
50–200 pcsUSD 85USD 420CNC
500 pcsUSD 55USD 48Marginal
2,000 pcsUSD 42USD 16Die casting
10,000 pcsUSD 36USD 7.50Die casting
50,000 pcsUSD 34USD 3Die casting

With USD 20,000 of tooling and a USD 43 per-piece saving, static break-even lands at about 465 pieces. That is the number most comparison articles quote, and it is the number that gets programmes into trouble — because it assumes the drawing never changes.

Real-world break-even depends heavily on part complexity. Published figures range from about 500 pieces for a simple housing with cheap tooling, through 1,200–2,500 for a structural bracket with slides, to 3,000–8,000 where tooling is complex or the per-piece gap is narrow. Treat 500 and 8,000 as the edges of a band, not as two competing facts.

The ECO tax: why your real break-even is higher

Every hardware programme issues engineering change orders between pilot build and scale-up. In die casting, the cost of a change depends on whether it is *steel-safe*:

Change typeWhat it meansCostLead time
Steel-safeAdd metal to the part by removing steel (bigger boss, thicker wall)USD 1,500–4,0005–10 working days
Non-steel-safeRemove metal from the part by adding steel (thinner wall, moved groove)USD 6,000–18,0003–5 weeks
New insert / re-cut cavityGeometry will not weld and re-machineFull tool or sub-insertTool lead time again

Fold that into the decision with a simple expected-value correction:

Effective break-even = static break-even × (1 + p(ECO) × penalty index)

Likelihood of a geometry change before SOPPenalty indexEffect on a 1,800-pcs static break-even
Design frozen, released to production01,800 pcs
Minor revisions expected (25% chance)0.6~2,070 pcs
Active NPI, likely one revision (60%)1.2~3,100 pcs
Early concept, two or more revisions likely (90%)2.5~5,850 pcs

The practical rule: do not release die casting tooling until the design freeze is real, not aspirational. For programmes still in NPI, staying on low volume CNC machining to 2,000–5,000 pieces is often cheaper in total cash than tooling early and paying for changes.

Tolerance, geometry and what the drawing can honestly ask for

Cost decides most of the CNC machining vs die casting decision, but tolerance overrules cost. If the drawing asks for something the process cannot hold, the cheap option stops being cheap the moment the parts do not assemble.

Capability comparison

AttributeCNC machiningHigh-pressure die casting
Linear tolerance, as produced±0.01–0.05 mm±0.1–0.3 mm
Achievable on critical features±0.005 mm±0.05 mm after machining
Flatness0.02 mm / 100 mm typical0.1–0.3 mm, poor on large spans
Surface finish, as producedRa 0.8–1.6 µmRa 1.6–3.2 µm
Minimum wall thickness0.5 mm (cost rises steeply)1.2–1.5 mm practical
Draft angleNone required1–3°
Minimum internal radiusSet by cutter, can be sharp-ish0.5 mm+ to avoid hot spots
UndercutsFree with 5-axisNeeds slides/lifters, adds tooling cost
ThreadsCut directly, full strengthUsually machined or rolled after

That table is the reason most "die cast" parts still visit a machining centre. A cast aluminium part rarely ships as-cast: sprue and runner trimming, flash removal, machining of mating faces and threaded holes, and finishing typically add 20–50% to the as-cast number, and sometimes more. Our CNC parts finishing article covers what those secondary operations actually cost.

Walls, draft, parting line and radii

Die cast parts need uniform wall thickness to avoid hot spots and sink marks, a draft angle of 1–3° so the part releases, and generous radii at every internal corner. They also carry a parting line — a visible witness mark where the two tool halves meet — plus ejector pin marks on a non-cosmetic face. If your drawing does not say where the parting line may and may not be, the toolmaker will decide for you, and you may not like the answer.

CNC machined parts need none of that. Draft is irrelevant, walls can vary, and internal corners are bounded only by cutter diameter. This is why a 5-axis CNC machining part can be a single monolithic component replacing a cast-and-assembled subassembly — no fasteners, no sealing joint, no leak path.

Datum strategy on a cast-then-machined part

This is where hybrid parts succeed or fail. A casting is dimensionally loose; the machined features on it must be tight relative to each other. So the drawing needs two datum systems:

  1. Casting datums — three loosely toleranced targets used to locate the raw casting in the machining fixture.
  2. Machining datums — the finished features that define the functional relationship (bores, sealing faces, mounting pads).

Leave 0.5–1.5 mm of stock on every surface that will be machined, and note that machining too deep cuts through the dense casting skin into the porous core. If a sealing face ends up in porous material, no amount of impregnation will save it. Raise stock allowance on faces that must pass a leak test.

Material reality: strength, porosity and cosmetics

The third axis of the CNC machining vs die casting decision is metallurgy. Even at identical geometry and identical price, the two processes deliver different metal — and some applications simply cannot accept the cast version.

Porosity and pressure tightness

Gas porosity of 1–3% by volume is normal in HPDC. For a bracket it is irrelevant. For a housing that must hold pressure or keep water out, it is the whole problem. Mitigations, in order of effectiveness:

Specify the leak rate and the acceptance criterion on the drawing. "Impregnate as required" is not a specification, and impregnation on a surface that will later be anodised or plated creates its own defects.

Mechanical properties

PropertyADC12 / A383 as-cast6061-T6 wrought7075-T6 wrought
Tensile strength~240 MPa~310 MPa~570 MPa
Yield strength~150 MPa~276 MPa~503 MPa
Elongation1–3%8–12%5–11%
Thermal conductivity~96 W/m·K~167 W/m·K~130 W/m·K
Typical useHousings, covers, bracketsStructural, thermal, cosmeticAerospace, weight-critical

Note the elongation row. A cast part is brittle: it cracks rather than bends. If your part sees impact, vibration or press-fit assembly, that 1–3% figure matters more than the tensile number.

Anodising and cosmetics — the hidden deal-breaker

This is the one that surprises people. Die casting alloys are high in silicon (ADC12 is around 10–12% Si) precisely because silicon improves fluidity and fill. Silicon does not anodise. An anodised die cast part comes out grey, mottled and darker than a machined 6061 panel, and it will not colour-match across a batch. Black anodising hides some of it; clear or coloured anodising does not.

For a hidden internal bracket, nobody cares. For a brushed and anodised audio front panel, a console face or anything a customer will look at, it is disqualifying. If cosmetics matter, the choice is not really CNC machining vs die casting — it is machined wrought aluminium, and our aluminium CNC machining and anodising pages go through the finish options in detail.

Lead time, tooling risk and the design-freeze gate

MilestoneCNC machiningDie casting
Quote turnaround24–48 h typical3–10 days (tooling must be scoped)
First article3–10 days4–10 weeks (tool design + build)
T1 samples then correction1 week2–5 weeks per iteration
Production rampImmediateAfter tool tryout and approval
Design change after toolingProgram editUSD 1,500–18,000 + 1–5 weeks

Time is the fourth input, and it is the one most often left out of the CNC machining vs die casting comparison. A workable sourcing sequence that avoids the traps:

  1. Concept and prototype — CNC prototyping, 3–7 day samples, design still free to move
  2. Pilot / bridge production — CNC at 50–500 pieces, validate the market and the fit
  3. Design freeze gate — a written milestone. No geometry changes after this date.
  4. Tooling release — only after the gate, with volume commitment that justifies it
  5. Mass production — cast near-net, machine critical features, finish, inspect

Programmes that skip step 3 are the ones that end up with an 8,000-piece effective break-even on a part they thought would break even at 1,800.

The hybrid route: cast near-net, machine the critical features

Framed as CNC machining vs die casting, the question sounds binary. In practice most mature programmes are not either/or. They cast the body and machine the features that carry function — and this is also where a CNC shop like ours fits in, whether the blank comes from your foundry or from a billet.

What to machine after casting

FeatureAs-castPost-machinedComment
Mating / sealing face±0.2 mm, wavyFlatness 0.05 mmAlmost always machined
Bore for bearing or shaftDraft tapered, ±0.2 mmH7, ±0.01 mmBoring or reaming
Threaded holesNot cast to sizeCut or roll-formedCast-in holes are usually cored then tapped
Mounting padsCoplanarity poorMachined coplanarCritical for EMI gaskets
Thermal interface face0.2–0.4 mm wavy0.05 mm flatnessDirectly sets thermal resistance
Cosmetic facePorous skin, parting lineMachined + brushed + anodisedThe only route to a retail finish

Stock allowance rules

A hybrid route is also the natural answer when a design is thermally demanding: cast or extrude the fin array, then machine the base flatness and the mounting features. Our heat sink CNC machining guide covers that specific case.

Requesting comparable quotes: a 12-line RFQ

The most common mistake in this comparison is comparing an as-cast price with a machined price. They are not the same thing. Ask every supplier, for both processes, to quote these twelve lines:

  1. Tooling cost and who owns the tool
  2. Tooling amortisation method and the quantity it is spread over
  3. As-produced unit price at your actual order quantity
  4. Secondary operations itemised: trim, deburr, post-machining, finishing
  5. Scrap and reject allowance, by operation
  6. Impregnation, if required, and whether it is included
  7. Finishing specification and cost, including masking
  8. Packaging and freight to your dock
  9. First article lead time and FAI documentation level
  10. MOQ and economic order quantity
  11. Cost and lead time of one engineering change
  12. Payment terms and any tooling milestone payments

If a supplier will not fill in lines 4, 5 and 11, you do not have a comparable quote. This checklist is the practical version of what we describe in choosing a CNC supplier.

Industry playbooks

Audio and professional audio equipment

Cosmetics and tactile quality dominate. Machined 6061 or 6063 with brushed grain plus black anodising is the standard for front panels, knobs and console chassis; die castings cannot match the colour consistency. Volumes are usually 100–5,000 per SKU, well inside the CNC band. See audio mixer console machining.

Automotive and EV

This is die casting's home territory: engine brackets, transmission housings, sensor bodies, structural nodes at 50,000+ per year. But prototypes, pre-production builds and low-volume derivatives stay machined, and IATF 16949 documentation is required either way. Our automotive CNC machining and EV motor housing articles cover the automotive case, and IATF 16949 explained covers the documentation.

Consumer electronics and 3C

Short product cycles — often 9–18 months — mean the design freeze gate comes late and lifetime volume is uncertain. Machining wins through the pilot phase; casting only pays if the SKU clearly runs past the break-even band. More in consumer electronics machining.

LED, thermal and power electronics

Thermal performance is a flatness and conductivity problem, and ADC12's 96 W/m·K is roughly half of 6061's 167 W/m·K. Where the housing is also the heat path — LED fixtures, inverters, power modules — machining usually wins on performance even when casting wins on price.

Medical and robotics

Volumes are low, traceability requirements are high, and surfaces must be cleanable and non-porous. Machining is the default. See ISO 13485 machining and robotics CNC parts.

FAQ

At what volume does die casting become cheaper than CNC machining?

There is no single number, which is why die casting vs CNC machining comparisons that quote one are misleading. For a simple aluminium part with modest tooling, the published crossover sits around 500–1,000 pieces. For a structural bracket requiring slides and lifters, expect 1,200–2,500; for complex tooling or a narrow per-piece gap, 3,000–8,000. If design changes are likely, multiply that figure by the ECO correction in this article — an 1,800-piece break-even easily becomes 5,000 in an active NPI programme.

Can a die cast part hold ±0.01 mm?

Not as-cast. Typical HPDC capability is ±0.1–0.3 mm. ±0.01 mm is reached by machining the feature after casting, which is the standard practice for bores, sealing faces and mounting pads. Budget the secondary operation and the stock allowance in the casting design.

Is die cast aluminium as strong as machined 6061-T6?

No. ADC12 is around 240 MPa tensile with 1–3% elongation; 6061-T6 is around 310 MPa with 8–12% elongation. The elongation gap matters more than the strength gap: cast parts are brittle and crack under impact or press fits where a machined part would deform.

Why do die cast parts fail leak tests?

Gas porosity of 1–3% by volume is inherent to high-pressure fill. Interconnected porosity near a sealing face creates a leak path, and machining that face too deep removes the dense skin and opens more of it. Control it with vacuum-assisted casting, adequate machining stock, a specified leak rate and — where acceptable — vacuum impregnation with a written acceptance criterion.

Can you anodise die cast aluminium to match a machined panel?

Not reliably. Casting alloys carry 10–12% silicon, which does not anodise, so the result is grey, mottled and darker than anodised 6061, with batch-to-batch colour variation. Hidden parts are fine; any visible or retail-facing surface should be machined from wrought aluminium.

How long does die casting tooling take compared with a CNC first article?

Die tooling is typically 4–10 weeks for design, steel cutting, EDM, heat treatment and tryout, plus 2–5 weeks per correction iteration. A CNC first article is 3–10 days, and low-volume production can start immediately after approval.

Is a hybrid cast-plus-machined part worth the extra step?

Usually yes, above the break-even band. Casting handles the bulk shape cheaply, machining delivers the ±0.01 mm bores, flat sealing faces and cosmetic surfaces. The extra cost is one additional operation; the alternative is either an unaffordable all-machined part or a casting that does not assemble.

Do I need different drawings for CNC and die casting?

Yes. The die casting drawing adds draft angles, uniform wall thickness, fillet radii, parting line location, ejector pin zones, machining stock allowance and a separate casting datum system. The CNC drawing drops all of that and simply calls out tolerances per feature. Never send the same file to both unless it is a fully dimensioned model with process-neutral tolerancing.

Getting a decision on your part

If you send us a 3D model, a dimensioned drawing and your annual quantity, we will run the CNC machining vs die casting arithmetic with you and tell you honestly whether machining is the right answer — and if it is not, we will say so. Where the volume clearly justifies tooling, we machine the cast blanks and hold the critical features, so you get one supplier across the transition instead of two.

Send the files for a 24-hour quote, or read our DFM analysis service first if you want the drawing reviewed before it goes out. Everything is produced under IATF 16949, ISO 9001:2015 and ISO 13485, with 200+ CNC machines, 23+ years of machining experience and ±0.01 mm capability — see the about page for the equipment list and certifications.

在 CNC 加工 vs 压铸之间做选择,很少是"哪个工艺更好"的问题,而是"在你的产量、你的结构、你的公差、你的交期下,哪个工艺更便宜"。本文给出真正在做决定的那些数字:模具摊销、单件成本曲线、把设计变更算进去的盈亏平衡点公式,以及会直接推翻成本结论的公差与外观限制。同时也会讲大多数项目最终采用的方案——先压铸近净形,再加工关键特征。

一句话结论

如果你下周就要零件、设计还在改、公差紧于 ±0.05 mm,或者零件必须有消费级外观——选 CNC 加工,而且通常优势明显。如果你每年要几万个一模一样的零件、设计已经冻结、公差中等,压铸的单件成本会赢,有时是五倍的差距。中间地带全靠算术,本文就是为这部分算术写的。

场景选择原因
1–500 件,任意年度量CNC 加工没有模具需要摊销
设计仍在改动CNC 加工工程变更只是改程序,不是重开模具
首件三周内必须到CNC 加工光开模就要 4–10 周
±0.01 mm 特征CNC 加工铸态能力只有 ±0.1–0.3 mm
阳极氧化外观面CNC 加工铸造合金氧化后发灰发花
不浸渗也要密封CNC 加工气孔是压铸的固有属性
年量 10,000+,设计冻结压铸节拍把单件成本压到极低
薄壁带筋的三维壳体,量大压铸近净形远优于从棒料掏出来
大铝壳,局部有精密面两者结合铸近净形 + 加工关键特征

先说明一句:下面的成本是 2025–2026 年铝件在亚洲与北美报价的典型区间,用途是做相对决策,不是给你下采购单。真正要投入之前,请按你的产量拿实际报价。

两种工艺到底是什么

在讲算术之前,先把 CNC 加工 vs 压铸的本质区别说清楚:一个工艺是把金属填进型腔,另一个是把金属从实心块上削掉。这一个差异,几乎解释了后面所有的后果——成本结构、公差、表面、交期。

高压压铸

熔融铝合金——最常见的是 ADC12(相当于 A383)或 A380——在 40–100 MPa 下射入淬硬钢模,保压至凝固。掌心大小零件的节拍是 30–60 秒。模具通常用 H13 钢,经电火花加工、淬硬到 46–50 HRC,并配滑块、斜顶与冷却水道。这套模具就是整个工艺的经济学:它很贵,要几周才能做出来,但它做出的每一个零件都很便宜。

因为是射入成型,压铸能做出从实心料上切会很浪费的近净形:薄壁、加强筋、凸台、内腔、复杂的三维轮廓,一次成型。但也因为金属是在湍流充型中凝固的,它会裹进气体——这就是气孔的来源,也是绝大多数压铸件问题的根源。

CNC 加工

CNC 加工是在三轴、四轴或五轴中心上用旋转刀具从实心棒料、板材上切除材料。除夹具和 CAM 程序外没有专用工装,所以第一件和第一百件的成本大致相同。占主导的固定成本是装夹,通常每个工序几百美元。

加工买到的是精度与材料本体性能。锻轧态 6061-T6 没有气孔,晶粒均匀,抗拉强度约 310 MPa;而 ADC12 铸态约 240 MPa,还带 1–3% 气孔。按 CNC 加工公差的口径,±0.01 mm 是常规能力,特定特征可做到 ±0.005 mm。而且表面是锻轧金属,阳极氧化出来的颜色干净——这正是我们做的每一块外观音频面板都是加工件而不是铸件的原因。加工还能做压铸做不到的结构:深腔、交叉孔、较清根的内角,以及不需要滑块的真正倒扣。成本侧详见 CNC 加工成本。

经济性:模具、单件成本与盈亏平衡点

钱分别花在哪里

两种工艺的成本结构几乎相反,这也正是交叉点如此陡峭的原因。在任何 CNC 加工 vs 压铸的对比里,一个把所有成本前置到模具,另一个把小额固定成本摊到每一件上。

成本项压铸CNC 加工
模具12,000–120,000 美元0–2,000 美元(编程 + 夹具)
每次上机装夹跑起来后很低每工序 150–500 美元
节拍每模 30–60 秒每件 5–45 分钟
材料利用率近净形,几乎无废料50–80% 变成切屑
工程变更每次 1,500–18,000 美元改程序,几乎为零
首件周期4–10 周(开模)3–10 天
经济起订量500–1,000 件1 件

盈亏平衡点,三种算法

静态盈亏平衡很简单:模具费除以单件节省额。

盈亏平衡数量 = 模具费 ÷(CNC 单价 − 压铸单价)

以 150 × 120 × 45 mm 的铝支架为例:

数量CNC 单价压铸单价(含摊销模具)更便宜
1–10 件180 美元不经济CNC
50–200 件85 美元420 美元CNC
500 件55 美元48 美元临界
2,000 件42 美元16 美元压铸
10,000 件36 美元7.5 美元压铸
50,000 件34 美元3 美元压铸

模具 20,000 美元、单件节省 43 美元时,静态平衡点约 465 件。这是大多数对比文章给出的数字,也是让项目掉坑里的数字——因为它假设图纸永远不改。

实际平衡点高度依赖零件复杂度。公开数据从简单壳体配便宜模具的约 500 件,到带滑块结构支架的 1,200–2,500 件,再到模具复杂或单件价差很小时的 3,000–8,000 件。请把 500 和 8,000 看成一个区间的两端,而不是两个互相矛盾的事实。

工程变更税:为什么真实平衡点更高

每个硬件项目在试产到量产之间都会下工程变更单(ECO)。在压铸里,变更成本取决于它是否"加钢"(steel-safe):

变更类型含义成本周期
加钢(steel-safe)靠去掉钢料给零件加金属(凸台加大、壁厚加厚)1,500–4,000 美元5–10 个工作日
非加钢靠补钢来给零件减金属(减薄壁、移槽位)6,000–18,000 美元3–5 周
新镶件 / 重开型腔形状无法焊补后再加工整套或子镶件重新走一遍开模周期

用一个期望值修正把它算进决策:

有效平衡点 = 静态平衡点 ×(1 + 变更概率 × 惩罚指数)

量产前发生结构变更的可能性惩罚指数对 1,800 件静态平衡点的影响
设计冻结,已放行量产01,800 件
预计有小修订(25% 概率)0.6约 2,070 件
处于 NPI 活跃期,很可能改一次(60%)1.2约 3,100 件
早期概念,可能改两次以上(90%)2.5约 5,850 件

实用规则:设计冻结必须是真实的,而不是口头承诺的,否则不要开压铸模。 对仍处于 NPI 阶段的项目,继续用 小批量 CNC 加工做到 2,000–5,000 件,总现金支出往往比提前开模再付变更费更便宜。

公差、结构与图纸上可以合理要求的东西

成本决定大部分 CNC 加工 vs 压铸的选择,但公差能否决成本。如果图纸要求的是工艺做不到的东西,那么便宜的选项在零件装不上的那一刻就不便宜了。

能力对比

项目CNC 加工高压压铸
线性公差(原始状态)±0.01–0.05 mm±0.1–0.3 mm
关键特征可达±0.005 mm加工后 ±0.05 mm
平面度典型 0.02 mm / 100 mm0.1–0.3 mm,大跨距更差
表面粗糙度(原始)Ra 0.8–1.6 µmRa 1.6–3.2 µm
最小壁厚0.5 mm(成本急升)实际 1.2–1.5 mm
拔模斜度不需要1–3°
最小内圆角由刀具决定,可较小0.5 mm 以上以免热点
倒扣五轴下免费需要滑块/斜顶,增加模具费
螺纹直接切削,强度完整通常加工或挤压成型

这张表就是大多数"压铸件"仍然要上一次加工中心的原因。压铸铝件很少以铸态出货:浇口与流道切除、飞边去除、配合面与螺纹孔加工、表面处理,通常会在铸态价格上再加 20–50%,有时更多。这些二次工序的真实成本见 CNC 零件表面处理。

壁厚、拔模、分型面与圆角

压铸件需要均匀壁厚以避免热点和缩痕,需要 1–3° 拔模角才能脱模,每个内角都要有足够圆角。它还会留下分型线——两半模具合模处的可见痕迹——以及非外观面上的顶针印。如果图纸不写清楚分型线允许和不允许出现在哪里,模具厂会替你决定,而答案你未必喜欢。

CNC 加工件全都不需要这些。拔模无关紧要,壁厚可以变化,内角只受刀具直径限制。这也是为什么一个 五轴 CNC 加工零件可以是单块整体件,取代"铸造 + 装配"的分体结构——没有紧固件,没有密封接合面,没有泄漏路径。

铸后加工件的基准策略

这是混合工艺件成败的关键。铸件尺寸松散,其上的加工特征彼此之间必须精密。因此图纸需要两套基准体系:

  1. 铸造基准——三个公差宽松的目标点,用于在加工夹具上定位毛坯铸件。
  2. 加工基准——定义功能关系的成品特征(孔、密封面、安装凸台)。

每个待加工面留 0.5–1.5 mm 余量;注意加工过深会切穿致密的铸造表皮进入多孔的心部。如果密封面最终落在多孔材料上,再多的浸渗也救不回来。对必须过泄漏测试的面,余量要加大。

材料现实:强度、气孔与外观

CNC 加工 vs 压铸的第三个维度是冶金。即使几何形状和价格完全相同,两种工艺给出的金属也是不同的——有些应用根本无法接受铸造版本。

气孔与密封性

高压压铸中 1–3% 体积分数的气体气孔属正常。对支架来说无关紧要;对必须承压或防水的壳体,这就是全部问题所在。缓解手段按有效性排序:

请把泄漏率和验收判据写在图纸上。"按需浸渗"不是技术规范,而且对后续要阳极氧化或电镀的表面做浸渗,本身就会造成新的缺陷。

力学性能

性能ADC12 / A383 铸态6061-T6 锻轧7075-T6 锻轧
抗拉强度约 240 MPa约 310 MPa约 570 MPa
屈服强度约 150 MPa约 276 MPa约 503 MPa
延伸率1–3%8–12%5–11%
导热系数约 96 W/m·K约 167 W/m·K约 130 W/m·K
典型用途壳体、盖板、支架结构件、导热件、外观件航空航天、减重关键件

请注意延伸率那一行。铸件是脆的:它裂,不会弯。如果你的零件要承受冲击、振动或压装,这 1–3% 比抗拉强度那条数字更重要。

阳极氧化与外观——隐藏的一票否决项

这是最让人意外的一条。压铸合金含硅量高(ADC12 约 10–12%),正是因为硅能改善流动性和充型。但硅不参与阳极氧化。阳极氧化后的压铸件发灰、发花、比加工 6061 面板更深,而且批次之间无法对色。黑色阳极能掩盖一部分,本色或彩色阳极不行。

对藏在内部的支架,没人关心。对一块 拉丝阳极音频面板、调音台面板,或任何客户会盯着看的表面,这一条直接否决。如果外观重要,选择其实不是 CNC 加工 vs 压铸,而是锻轧铝加工——详见 铝合金 CNC 加工与 阳极氧化。

交期、模具风险与设计冻结门

时间是第四个输入,也是 CNC 加工 vs 压铸对比中最常被漏掉的一项。

里程碑CNC 加工压铸
报价周期典型 24–48 小时3–10 天(需评估模具)
首件3–10 天4–10 周(模具设计 + 制造)
T1 样品后修正1 周每次 2–5 周
量产爬坡立即模具试模并批准之后
开模后设计变更改程序1,500–18,000 美元 + 1–5 周

一套能避开陷阱的采购顺序:

  1. 概念与原型——CNC 打样,3–7 天出样,设计仍可自由调整
  2. 试产 / 过渡量产——CNC 做 50–500 件,验证市场与装配
  3. 设计冻结门——写成书面里程碑。此日期之后不再变更几何。
  4. 模具放行——仅在该门之后,并有足以支撑的产量承诺
  5. 量产——铸近净形,加工关键特征,表面处理,检验

跳过第 3 步的项目,就是那种原本以为 1,800 件能回本、结果有效平衡点变成 8,000 件的项目。

混合路线:铸近净形,加工关键特征

把问题框定为 CNC 加工 vs 压铸听上去是二选一,实际大多数成熟项目都不是二选一。它们铸出本体,再加工承载功能的特征。

哪些特征要铸后加工

特征铸态加工后说明
配合 / 密封面±0.2 mm,波浪形平面度 0.05 mm几乎总是要加工
轴承孔或轴孔带拔模锥度,±0.2 mmH7,±0.01 mm镗或铰
螺纹孔无法铸到尺寸切削或挤压成型通常先铸出底孔再攻牙
安装凸台共面性差加工至共面对 EMI 衬垫至关重要
导热界面面0.2–0.4 mm 起伏0.05 mm 平面度直接决定热阻
外观面多孔表皮、分型线加工 + 拉丝 + 阳极获得零售级外观的唯一路径

余量规则

当设计对散热有要求时,混合路线也是天然答案:铸造或挤出翅片阵列,再加工底面平面度与安装特征。散热器 CNC 加工一文专门讲这个场景。

拿到可比报价:12 行询价单

本对比中最常见的错误,是拿铸态价和加工价做比较。它们不是一回事。请让每家供应商就两种工艺分别报这 12 行:

  1. 模具费,以及模具归属
  2. 模具摊销方法与摊销数量
  3. 按你实际订单数量的原始单件价
  4. 二次工序逐项列出:切边、去毛刺、后加工、表面处理
  5. 各工序的废品与返工余量
  6. 浸渗(如需),以及是否包含在报价内
  7. 表面处理规范与费用,含遮蔽
  8. 包装与运到你码头的运费
  9. 首件周期与 FAI 文件等级
  10. 起订量与经济批量
  11. 一次工程变更的费用与周期
  12. 付款条件与模具里程碑付款

如果供应商不肯填第 4、5、11 行,你手里的就不是可比报价。这份清单是 如何选择 CNC 供应商一文所讲原则的实操版。

各行业怎么选

音频与专业音响设备

外观与手感主导。加工 6061 或 6063 + 拉丝 + 黑色阳极是面板、旋钮与机箱的标准方案;压铸件无法达到这种颜色一致性。单个 SKU 通常 100–5,000 件,落在 CNC 区间内。参见 调音台金属件加工。

汽车与新能源车

这是压铸的主场:发动机支架、变速箱壳体、传感器本体、年量 5 万以上的结构节点。但原型、试制与小批量衍生件仍然走加工,而且无论哪种工艺都要 IATF 16949 文件。参见 汽车 CNC 加工、EV 电机壳体与 IATF 16949 解读。

消费电子与 3C

产品周期短,通常 9–18 个月,设计冻结门来得很晚,生命周期产量也不确定。试产阶段加工胜出;只有该 SKU 明确跑过平衡点区间时,压铸才划算。详见 消费电子 CNC 加工。

LED、散热与电力电子

散热性能本质是平面度与导热系数问题,而 ADC12 的 96 W/m·K 大约只有 6061 的 167 W/m·K 的一半。当壳体同时是散热路径时——LED 灯具、逆变器、功率模块——即使压铸便宜,性能上通常也是加工胜出。

医疗与机器人

产量低、追溯要求高、表面必须可清洁且无气孔,加工是默认选择。参见 ISO 13485 加工与 机器人 CNC 零件。

FAQ

压铸到多少数量才比 CNC 加工便宜?

不存在唯一数字,所以那些只给一个数的"压铸 vs CNC 加工"对比文章都具有误导性。对模具简单的铝件,公开交叉点约 500–1,000 件;需要滑块斜顶的结构支架,按 1,200–2,500 件估;模具复杂或单件价差小的情况,3,000–8,000 件。如果设计还可能变更,请按本文的 ECO 修正系数相乘——1,800 件的平衡点在活跃的 NPI 项目里很容易变成 5,000 件。

压铸件能做到 ±0.01 mm 吗?

铸态做不到。典型高压压铸能力是 ±0.1–0.3 mm。±0.01 mm 是靠铸后加工该特征实现的,这也是孔、密封面和安装凸台的标准做法。请把二次工序和毛坯余量一并算进铸件设计。

压铸铝和加工 6061-T6 一样强吗?

不一样。ADC12 抗拉约 240 MPa、延伸率 1–3%;6061-T6 约 310 MPa、延伸率 8–12%。延伸率的差距比强度差距更关键:铸件脆,在冲击或压装下会开裂,而加工件只会变形。

为什么压铸件过不了泄漏测试?

高压充型固有的 1–3% 体积分数气体气孔。密封面附近的连通气孔会形成泄漏通道,而加工过深会削掉致密表皮、暴露更多气孔。控制手段:真空辅助压铸、足够的加工余量、明确的泄漏率指标,以及在可接受前提下的真空浸渗并写明验收判据。

压铸铝能阳极氧化到和加工面板一个颜色吗?

不可靠。铸造合金含硅 10–12%,硅不参与阳极氧化,结果是发灰、发花、比阳极 6061 更深,且批次间存在色差。内部零件无所谓;任何可见或面向零售的表面都应该用锻轧铝加工。

压铸开模比 CNC 首件慢多少?

压铸模具通常 4–10 周,含设计、钢料切削、电火花、热处理与试模,每轮修正再加 2–5 周。CNC 首件是 3–10 天,小批量生产获批后即可开始。

铸后加工这种混合方案值得多一道工序吗?

超过平衡点区间后通常值得。压铸便宜地做出主体形状,加工交出 ±0.01 mm 的孔、平整密封面和外观面。多出来的成本只是一道工序;替代方案要么是全加工贵到不可接受,要么是铸件装不上。

CNC 和压铸要用不同的图纸吗?

要。压铸图要加拔模角、均匀壁厚、圆角、分型线位置、顶针区、加工余量,以及独立的铸造基准体系。CNC 图则把这些全部去掉,只按特征标注公差。除非是带中性公差的完整标注模型,否则不要把同一个文件发给两边。

让你的零件得到结论

把 3D 模型、标注图纸和年用量发给我们,我们会和你一起把 CNC 加工 vs 压铸的算术算一遍,并诚实地告诉你加工是否是对的答案——如果不是,我们也会直说。当量明确足以支撑开模时,我们加工你提供的铸坯并守住关键特征,让你在整个过渡期内只有一家供应商,而不是两家。

发文件,24 小时出报价;也可以先看 DFM 分析服务,在图纸发出前先审一遍。全部产品按 IATF 16949、ISO 9001:2015 与 ISO 13485 体系生产,200+ 台 CNC 设备、23+ 年加工经验、±0.01 mm 能力——设备与认证清单见 关于我们。

Get a QuoteFree DFM · 24h