The Six Questions That Separate a Real Quote From a Guess
Most die casting service pages read like menus. Here are our alloys. Here are our finishes. Here are our certifications. All true, all useless when you're trying to work out whether the shop in front of you can actually make your part.
The gap is that a capability list tells you what exists, not what applies. Eight finishes are offered; for your part, probably two are appropriate and one is quietly a bad idea. That's the part nobody writes down.
So here's the version I'd want if I were sourcing: six questions, and what a good answer sounds like.
1 What tolerance can you hold as-cast, and where does that stop?
Published tolerance figures for die casting services usually appear as a single range, something like ±0.003" per inch. That number is real but incomplete, because as-cast capability and machined capability are different animals.
Two things make this practical rather than academic.
First, tolerance across the parting line behaves differently from tolerance within one die half. Features formed entirely in the cover half hold well. Features spanning both halves inherit die-close repeatability and any flash variation. If your critical dimension crosses the parting line, either move it or accept a looser number.
Second, blanket-toleranced drawings are the single most common cause of inflated quotes. Tolerance every hole at ±0.05 mm and the shop has no choice but to machine every hole. Identify the four or five dimensions that actually matter and open the rest to a general cast tolerance note. I have seen this alone drop a quote by a fifth.
A good answer distinguishes as-cast from machined, mentions the parting line, and asks which dimensions are critical to function. A weak answer quotes one range for everything.
2 Is your alloy recommendation about the part, or about the shop?
A380 is the default for aluminum die casting, and usually correctly — good fluidity, good strength, forgiving to run. But defaults deserve a sanity check.
A383 (ADC12 in JIS naming) flows better and suits thin, intricate aluminium Casting Parts where A380 struggles to fill. A360 offers better corrosion resistance and ductility at the cost of castability. A413 is the pressure-tightness choice, which matters for pipe fittings and hydraulic bodies where a leak path is a warranty event.
Then there's the material question one level up. Zinc casting — Zamak 3 in most cases — will hold thinner walls than aluminum, runs hot chamber with far shorter cycles, gives five times the tool life, and takes decorative plating properly. If your part is small, detailed, and doesn't need to be light, zinc is often the better answer even though the material costs more per kilo.
What you're testing is whether the recommendation is reasoned. A shop that says "we'll run A380" without asking about wall thickness, pressure requirement or finish intent is describing its own furnace, not your part.
3 Which finish, and do you understand what silicon does?
This is the question that catches people, and it belongs in every finish conversation about cast metal.
A380 contains roughly 8–9% silicon, which is exactly what makes it castable. Anodize it and the silicon doesn't convert to oxide the way aluminum does. It stays behind as dark particles suspended in the film, and the result is grey, faintly mottled, and nothing like the crisp anodized finish on a machined 6061 sample.
This surprises buyers constantly. They approve a colour against a wrought aluminum chip, then reject the first article because the die cast metal came out looking like a different product. It did.
Type II anodizing on a die casting is a corrosion treatment with an industrial appearance, not a decorative finish. If you need colour and consistency, powder coat. If you need corrosion protection cheaply and paint adhesion afterwards, chromate conversion under MIL-DTL-5541 does the job and stays electrically conductive. If you want genuine chrome brilliance, that's a zinc part, which is precisely why decorative hardware is cast in zinc.
A supplier who volunteers this before you hit it is worth keeping.
4 Who owns the tool, and what exactly does the tooling price include?
Tooling ownership should be explicit in writing. The healthy arrangement is that you pay for the die and you own it, meaning you can move it if the relationship sours. Some arrangements amortise tooling into piece price, which lowers the upfront number and quietly ties you to that supplier.
Then check what the tooling figure covers:
- The die itself — cavity, core, inserts
- The trim die, which is separate tooling and can add 15–25% on a complex part
- Any machining fixtures, if there are secondary operations
- Sampling and first article documentation
- A stated number of maintenance or refurbishment cycles
That third item catches people. Custom die casting with several machined faces needs dedicated fixtures, and those get quoted separately more often than not. Ask.
Also ask what happens at end of tool life. Dies wear — heat checking on the cavity face, dimensional creep as high-wear zones erode. A shop that tracks shot count and tells you at 80,000 shots that refurbishment is coming is managing your program. A shop that discovers it when parts start failing inspection is not.
5 What does your sampling process actually look like?
Sampling separates metal manufacturing operations that run a process from ones that run a program.
The sequence should be: DFM report before any steel is cut, T1 samples off the production tool, dimensional report against your drawing, and functional testing if the part is pressure-bearing or safety-relevant. Then, only then, production release.
Two details worth insisting on. T1 samples must come off the production tool, not a prototype or a machined stand-in — otherwise you've validated something you aren't buying. And samples should be produced with production process parameters, not a slow careful one-off. A part made lovingly at half speed tells you nothing about what shot 5,000 looks like.
For anything requiring pressure integrity, specify the leak rate numerically in mbar·L/s. "Pressure tight" is not a specification. It's an aspiration, and it will be interpreted generously.
6 What's actually in-house?
"One-stop metal casting services" appears on a great many websites and means different things behind each one.
The reason it matters isn't convenience — it's accountability. When a machined casting fails a leak test, the caster says machining opened a subsurface void and the machinist says the casting was porous. Both can be right. Neither pays. A single-source supplier internalises that argument instead of forwarding it to you.
There's a scheduling dimension too. A part routing through four vendors accumulates queue time at each, and queue time typically exceeds process time by a wide margin. That's where lead times quietly double.
The counter-argument is real: a shop that casts well and machines adequately may not beat a dedicated machinist on a demanding feature. So ask specifically. Casting, machining, finishing, assembly — which of those happen under your roof, and which are subcontracted? Either answer can be fine. An evasive answer is the problem.
Where this lands by industry
Electronics buys die cast housings and heat sinks, usually cast, faced, tapped and chromated, with thermal performance driving the alloy choice. Fluid power takes valve bodies and manifolds where a single hidden void fails a leak test after all the machining value is already invested. Automotive pulls structural fabrication nodes and motor housings under IATF discipline with full lot traceability. Plumbing runs cast fittings and flanges, and a serious pipe fittings manufacturer keeps separate melt records per alloy and its own thread-gauging cell.
Industrial equipment is the interesting one, because it mixes castings with custom metal fabrication routinely. A supplier offering both casting services and custom fabrication can move the cast-versus-fabricate line based on total processed cost rather than on which department happens to be free — which is worth more than it sounds when volumes are uncertain.
The short version
Ask about as-cast versus machined tolerance and the parting line. Make them justify the alloy against your wall thickness and pressure requirement. Raise silicon before you approve a finish. Get tooling ownership and scope in writing, including trim dies. Insist T1 comes off the production tool at production parameters. Find out what's genuinely in-house.
Six questions. A shop that answers them in numbers has read your model. A shop that answers them in adjectives has read your industry.
One last thing worth saying plainly. The best supplier conversation I've had started with the shop telling me the part as drawn couldn't be made economically, and why. That's not a shop being difficult. That's a shop that opened the file.
FAQs
Standard as-cast work typically holds around ±0.10 mm on smaller features, with tighter process control reaching roughly ±0.05 mm. Below that you're into machined features. Tolerances crossing the parting line are looser than those formed within a single die half.
Because A380 contains 8–9% silicon, which doesn't anodize. It remains as dark particles in the film, producing a grey mottled appearance quite unlike anodized wrought aluminum. For colour-critical parts, powder coating is the reliable route.
Aluminum for light weight, thermal performance and corrosion resistance. Zinc for small detailed parts, thinner walls, much longer tool life, faster cycles and genuine decorative plating. Zinc costs more per kilo but the tooling lasts roughly five times as long.
Frequently not, and it's a common source of surprise. A trim die is separate tooling and can add 15–25% to the tooling bill on a complex part. Ask explicitly, and ask about machining fixtures too.
You should, but confirm it in writing. Some suppliers amortise tooling into the piece price instead, which reduces the upfront cost but makes moving production difficult. Ownership terms should be explicit before you commit.
T1 is the first article produced from the production tool. It must come off that tool at production process parameters — not a prototype and not a slow careful one-off — otherwise you've approved something different from what you'll receive in volume.
Generally above roughly 1,000 pieces a year, though the real crossover depends on part size and how much machining the alternative needs. Below that, sand casting or machining from billet usually wins on total program cost.
Numerically. State the maximum acceptable leak rate in mbar·L/s and the test method. The phrase "pressure tight" carries no defined meaning and will be interpreted in the supplier's favour.