Case Studies

The most useful thing a factory can show a buyer is not a glossy part photo — it is a problem it got wrong, found the cause of, and fixed. Below are three real cases from our floor, each with the measurement or photograph that proves the result.

Customer names are withheld under confidentiality, and the parts are described by type rather than by product. Everything else — the defect rates, the gauge readings, the corrective actions — is exactly as recorded on the floor.

Request a Quote Speak to an Engineer

An aluminium die casting fixtured in a CNC machining cell, cutting tool engaging a bore with coolant lines and chips visible

What do these die casting case studies show?

Three real production problems from our Dongguan floor and how each was solved: contamination inside plated threaded holes (8% batch rejects), an uneven hand-ground feed opening causing assembly interference (12%), and a cast boss coming out short because an ejector pin was deforming it (12%). Each case shows the diagnosis, the corrective action, and before/after photographic evidence — including gauge readings. Customers are not named, and no figures are claimed beyond what was measured.

At a glance

Three real production problems, each with the measurement that proves the fix
Root causes traced upstream — plating sequence, hand finishing, and an ejector pin
Batch inspection reject rates of 8%, 12% and 12% taken to zero on the features involved
Customers withheld under confidentiality; every number published was measured

What We Will and Will Not Publish

What is on this page: real defect rates from batch inspection, real gauge readings, real corrective actions, and photographs of the actual parts. Where a number appears, it was measured.

What is not on this page, and will not be: customer names without written permission, cost savings or volume figures we cannot show the working for, borrowed photographs, or success stories too smooth to have happened. A case that needs those to be impressive is not a case worth publishing.

When a customer does give permission, a named case will carry the full picture — the constraint, the volumes, the engineering decisions taken and rejected, the validation route from tooling trial to first-article approval, and what the customer would tell a peer. Until then, the parts above speak for themselves.

What These Three Cases Have in Common

Every one of them was solved upstream of inspection. Contamination was designed out of the plating sequence, variation was designed out by moving an operation onto a machine, and a dimensional defect was traced back to the ejection system that caused it. None of the three was fixed by inspecting harder or by asking operators to be more careful — because neither of those survives a production run.

The third case is the clearest example of why a factory that owns its own tool room behaves differently. The boss was short because an ejector pin was deforming it during ejection. Diagnosing that needs someone who understands the tool, and fixing it means taking the die apart and changing a pin. When the steel lives in another company's shop, that loop takes weeks and passes through a middleman; here it is a conversation with the tool room and a die that goes back on the machine.

Each fix was verified by measurement before production resumed, and the evidence is in the photographs above — the same hole, the same feature, the same gauge. That is the standard we would want to see from a supplier, so it is the standard we publish.

Three Problems, Three Fixes

Each of these went through the same loop: a batch inspection flagged a rate we would not ship at, engineering traced the cause, a change was made, and the result was verified by measurement before production resumed.

Case 1

Contamination inside plated threaded holes

Aluminium die-cast fixture · coffee machine component

Before and after macro photographs of a threaded hole in an aluminium die casting, showing plating residue inside the thread and the same hole clean after the fix — Same part, same hole. Left: residue trapped in the thread after plating. Right: clean threads after the process change.

The problemBatch inspection was rejecting 8% of parts. Threaded holes came out of the plating line with residue inside the thread, and the contamination interfered with assembly at the customer's end.

What we foundThe threads were open during plating, so solution and residue collected inside them and stayed there. Cleaning afterwards was unreliable because the contamination sat deep in the thread form.

The fixChange the sequence rather than add an inspection step: blow the threaded bores clean with an air gun before plating, then plug each hole with a stainless steel screw so nothing can enter during the plating process.

Verified resultDefect rate fell from 8% to zero in the batch inspections that followed. The fix costs a few seconds of handling per part and removed a rework loop entirely.

Case 2

Uneven feed opening causing assembly interference

Zinc die-cast housing · smart device accessory

Before and after photographs of a zinc die-cast housing feed opening, showing an uneven hand-ground edge and the same edge machined flat — Arrows mark the feed opening edge. Left: hand-ground, varying height. Right: CNC machined flat and repeatable.

The problemBatch inspection was rejecting 12%. The feed opening was being finished by hand, so its height varied from part to part. On assembly the component sat slightly proud, and the housing interfered with the mating part.

What we foundHand grinding cannot hold a consistent height across a production run — the variation is in the operator, not the part. Any feature that has to meet a mating component needs a repeatable process behind it.

The fixMove the operation off the bench and onto a machine: CNC mill the feed opening face flat instead of grinding it by hand.

Verified resultDefect rate fell from 12% to zero in the batch inspections that followed. The part now assembles without interference, and the result no longer depends on who finished it.

Case 3

Cast boss coming out short — an ejector pin was the cause

Aluminium die-cast main body · coffee machine component

Before and after digital height gauge readings on a cast boss, showing 8.41 mm before the tooling change and 9.72 mm after, against a 9.7 mm requirement — Digital height gauge on the same boss, measurement point circled. Left: 8.41 mm against a 9.7 mm requirement. Right: 9.72 mm after the tooling change.

The problemThe boss height on the main body was unstable and consistently short. The drawing called for 9.7 mm; the gauge read 8.41 mm. Parts would not sit level on assembly, and batch inspection was rejecting 12%.

What we foundThis was a tooling problem wearing a dimensional disguise. An ejector pin sat at the boss position, and every time the part was ejected the pin deformed the boss it was pushing against. No amount of process adjustment at the machine could fix a feature the ejection system was damaging on the way out.

The fixReplace the ejector pin at that position with a fixed insert pin, so the boss is formed against solid steel and nothing pushes on it during ejection.

Verified resultThe boss measured 9.72 mm against the 9.7 mm requirement, and stopped drifting between shots. The 12% reject rate went to zero on this feature in the batch inspections that followed.

These are our own production records, published with customer identities withheld. The defect rates are the batch inspection figures from those programs, and the gauge readings are from the parts photographed — not estimates or illustrations. Zero rejects describes the inspections that followed each fix; it is a record of what happened, not a guarantee of what every future run will do.

Where the Next Cases Come From

The programs under preparation span the territory this site describes: weldment-to-casting conversions for industrial equipment, thin-wall electronics enclosures balancing heat against cosmetics, plated zinc hardware at appliance volumes, and tooling transfers rescued from failing supplier relationships.

As customers give permission to name them and share program figures, cases will be published with those details attached. Until then they are published as above — real, measured, and anonymous.

Representative application illustration — not a record of a specific customer program.

Frequently Asked Questions

Why are the customers not named?

Because they have not given permission, and publishing a customer relationship without it is a breach of trust that no marketing benefit justifies. The engineering content of a case does not depend on the name attached to it — the thread contamination fix works the same way whoever the buyer was. When a customer does agree to be named, we will publish the fuller picture, including volumes and outcomes, with their team reviewing it first.

Are the defect rates and measurements real?

Yes. The 8%, 12% and 12% figures are the batch inspection reject rates recorded on those programs before the fixes, and the 8.41 mm and 9.72 mm readings are from the parts in the photographs, taken on a digital height gauge against a 9.7 mm drawing requirement. We publish what was measured. Where we do not have a verified number, we do not print one.

Does zero rejects mean you never produce a defect?

No, and any supplier who tells you otherwise is selling. It means the batch inspections after each fix found no rejects on the feature in question — the specific problem was eliminated rather than reduced. Casting is a process with variation in it; what a good factory controls is whether that variation reaches your assembly line, and whether the cause gets found when something does slip.

Could our project become a case study?

If the engineering is interesting and you are willing, yes — with your control over what is named. Customers who participate get a documented, peer-readable account of their program's manufacturing decisions, reviewed by their own team before anything goes public.

Can I see documentation for a program like mine?

Ask in your RFQ. Depending on confidentiality we can often discuss anonymised precedents close to your part's territory — same alloy family, similar geometry class, comparable volumes — and share a redacted control plan and first-article report so you can see how a program is actually run here. Specific questions get specific answers.

Where to Go Next

Precision Die CastingThe process decisions the parts above represent.CNC MachiningThe machine-what-matters half of the near-net story.Supplier Selection GuideHow to evaluate casting claims while named cases are pending.Industrial EquipmentThe conversion territory the first cases draw from.Request a QuoteStart the program — and maybe the next case.

Send the problem, not just the part

The Useful Conversation Starts With What Is Going Wrong

If a part is failing inspection, drifting between batches, or fighting you at assembly, send it with the symptom. Cases like the three above start exactly there.

Request a Quote Speak to an Engineer

Last reviewed August 2026 · Inox Die Cast, Dongguan