If My Samples Fail, Should I Change the Tooling or Change My Drawing When I Import Custom Die Cast Parts from China?

We see it happen on almost every new die cast program we manage for clients — T1 samples arrive, something is out of tolerance, and the pressure to "fix it fast" leads to the wrong decision. The wrong fix wastes money, delays production, and sometimes makes the problem worse.

When die cast samples from China fail, the answer depends on root cause. If measurements are consistently offset in one direction across all samples, the tooling is likely wrong. If measurements scatter widely, the process needs adjustment. If the tolerance itself exceeds NADCA die casting capability, the drawing needs revision. Identifying which of these three causes applies before spending money is the most important step.

There is no universal answer. But there is a clear diagnostic process. The sections below walk through each question you need to answer before making your decision.

How Can I Decide Whether My Issue Is a Design Problem or a Tooling Problem?

Every failed sample we review tells a story in the measurement data — if you know how to read it. The pattern of deviation across your sample set is the primary diagnostic signal, and it points clearly to one of three distinct root causes.

A design problem shows up as a tolerance requirement that no tooling modification can consistently achieve for that feature and process. A tooling problem shows up as consistent, directional deviation across all measured parts. The key diagnostic step is to plot your sample measurements, calculate the mean deviation and scatter, and compare the tolerance callout against NADCA standard capability for that feature type.

The Three Root Causes of Sample Failure

Every T1 sample failure in die casting ¹ falls into one of three categories. They require completely different responses.

Confusing these three causes is the most expensive mistake you can make on a Chinese die cast program. Modifying the tool when the drawing is the problem wastes $3,000–$20,000 and several weeks. Relaxing the drawing when the tooling is wrong lets a defective tool enter production and generates quality problems for the life of the program.

Reading the Measurement Pattern

Take your T1 measurement data for the failing dimension and ask two questions.

First: are all samples on the same side of the nominal? If ten samples measure 48.2, 48.3, 48.4, 48.3, 48.2mm on a nominal of 49.0mm with ±0.2mm tolerance, every single part is low by roughly 0.7mm. That is consistent, directional deviation. The cavity is cut undersize at that feature. This is a tooling problem.

Second: do the measurements scatter widely around the nominal? If ten samples measure 48.6, 49.3, 48.9, 49.5, 48.4mm, the mean is close to nominal but the spread exceeds the tolerance band. No consistent direction, high scatter. This is a process problem — die temperature variation, inconsistent injection speed, or cooling time instability is producing dimensional scatter. No tooling modification will fix scatter.

If neither pattern fits — if the measurements are clustered in tolerance but your drawing calls for ±0.05mm and the best you are seeing is ±0.18mm — you may have a drawing problem. The tolerance is tighter than what the process can physically hold for that feature.

NADCA Standard Capability as Your Reference

Before concluding anything, check your drawing tolerance against NADCA Product Specification Standards for die casting. The table below summarises standard and precision tolerance for aluminium high-pressure die casting.

If your drawing calls for ±0.05mm on a parting-line dimension of a 200mm aluminium housing, that is outside NADCA standard capability. No tooling modification will achieve it consistently as-cast. The correct response is to revise the drawing — either relax the tolerance to a level the process can hold, or designate that surface for post-cast CNC machining ² and add 0.3–0.5mm machining allowance to the tooling design.

Why Chinese Suppliers Default to Tooling Modification

From a Chinese supplier's commercial perspective, proposing a tooling modification is the path of least resistance when T1 samples fail. It avoids the uncomfortable conversation about whether the drawing is realistic, it generates billable work, and it looks like responsiveness. A supplier who says "we need to modify the tool" has not necessarily performed a real root cause analysis ³.

Before approving any tooling modification, require the supplier to show you the raw measurement data from all T1 samples. The data should support the proposed diagnosis. If they cannot show you consistent directional deviation, be cautious about approving tooling work.

Will a Drawing Change Be Faster and Cheaper Than a Tooling Modification for Me?

Speed and cost depend entirely on which direction the correction needs to go — and on what the physical cause of the failure actually is. A drawing revision can be completed in hours. A tooling modification that requires welding, re-machining, and re-hardening can take two to four weeks and cost thousands of dollars.

A drawing revision is faster and cheaper when the tolerance exceeds what the die casting process can physically achieve for that feature. A tooling modification is necessary when the cavity geometry is confirmed to be machined incorrectly. Choosing the wrong option costs both time and money — sometimes more than starting a new tool from scratch.

The Cost Asymmetry of Tooling Corrections

Not all tooling modifications cost the same. The direction of correction matters enormously, and understanding this asymmetry helps you evaluate supplier quotes more critically.

Removing metal from a cavity — to make a part dimension larger — requires only careful re-machining. It is relatively fast and low-risk. Adding metal to a cavity — to make a part dimension smaller — requires welding the cavity steel, re-machining the weld to geometry, re-hardening the affected area, and re-polishing the surface. This is expensive, time-consuming, and carries risk: weld repairs can introduce cosmetic defects, residual stress, and reduced tool life at that location.

Good tooling design practice intentionally cuts new dies slightly undersize on external dimensions, so that first-off corrections go in the safer removal direction. If your T1 shows a dimension consistently undersize on the part — meaning the cavity is oversize — the weld correction direction is the more expensive and risky path.

When the Drawing Change Is the Right Answer

Die casting is a near-net-shape process ⁴. Its economic advantage is high volume at low piece cost, not precision tolerancing of critical interfaces. If your drawing was copied from a CNC machined part design without adjusting tolerances for the die casting process, some dimensions will specify tolerances that the process cannot hold as-cast.

The correct engineering response in those cases is not to keep modifying tooling and running trials hoping to beat the physics. It is to revise the drawing. Specifically:

• Relax the tolerance on that feature to a level within NADCA standard or precision capability, or
• Designate the surface for post-cast machining ⁵ and add a machining allowance of 0.3–0.5mm minimum on the affected face in the tooling design.

Post-cast machining adds piece cost, but it reliably delivers ±0.01–0.05mm on prepared datum surfaces. For bearing bores, sealing faces, threaded hole patterns, and assembly datums, this is standard practice in volume die casting programs worldwide.

Formal Deviation Requests as an Alternative

If a specific dimension cannot be held to drawing tolerance by any combination of process optimisation or reasonable tooling modification — but the out-of-tolerance condition does not affect fit, function, or assembly — the correct path is a formal deviation or engineering change request. This is a documented, signed agreement replacing the specified tolerance with an achievable one, with the rationale recorded.

Issuing a verbal waiver and hoping no one measures that dimension in production creates an undocumented quality risk. It can cause a supplier audit failure or a field quality issue years later. A formal deviation protects both parties and keeps the program clean.

What Technical Review Should I Request Before I Make This Decision?

The decision to modify tooling or revise a drawing should never be based on a verbal recommendation alone. Before authorising any corrective action, there is a specific set of technical documents and data you should request from your supplier — and specific things you should check yourself.

Before approving a tooling modification or drawing revision, request the full T1 measurement report for all sampled parts, a dimensional deviation summary showing mean and range for each failing feature, the supplier's proposed root cause with supporting evidence, and a cross-check of each failing tolerance against NADCA standard capability for the feature type and alloy.

The T1 Measurement Report

The T1 measurement report should list every controlled dimension from your drawing, the measured value for each sample, and the pass/fail status against your tolerance. Ask for raw data — individual measurements for each part — not just averages. Averages can hide dangerous scatter.

From the raw data, calculate for each failing dimension:

• Mean deviation from nominal
• Range (maximum minus minimum measurement)
• Whether all values fall on the same side of nominal (consistent direction = tooling indicator)
• Whether values scatter on both sides of nominal (high scatter = process indicator)

If your supplier cannot provide a complete measurement report with raw individual-part data, that is itself a quality control red flag. Require it before approving any corrective action spend.

The Drawing Tolerance Audit

For every dimension that failed, compare the tolerance callout on your drawing against the NADCA standard and precision tolerance table for the applicable feature type, alloy, and projected part area. This audit typically takes a few hours of engineering time and costs very little compared to a mistaken tooling modification.

Ask specifically: is this dimension across a single die half, across the parting line, or across a moving die component such as a slide or core? Each adds a tolerance increment on top of the base linear tolerance. A dimension that crosses the parting line of a large projected-area part can have an achievable tolerance of ±0.5mm or more under NADCA standards, even for precision tooling.

Ensuring your drawing uses geometric dimensioning and tolerancing ⁶ (GD&T) principles correctly is also critical here. GD&T controls form, orientation, and location in ways that linear ±tolerances alone cannot capture, and misapplied callouts are a common source of unnecessary failures.

The Process Sensitivity Study

Before concluding that a marginal deviation requires tooling modification, require the supplier to perform a process sensitivity study. This means varying one process parameter at a time — die temperature by ±15°C, injection speed, hold pressure, cooling time — across a range of shots, and measuring the effect on the failing dimension.

Features that are out of tolerance by less than 30% of the tolerance band and show moderate scatter may be correctable through process optimisation alone, with no tooling or drawing changes. Die temperature has a direct effect on cavity fill and the thermal expansion of both the die and the casting. Tuning die temperature by 10–20°C can shift some dimensions measurably.

A process fix costs nothing and takes days. A tooling fix costs money and takes weeks. A disciplined T1 review performs the process sensitivity study before recommending tooling modification for any marginal deviation.

What to Ask Your Supplier — in Writing

When you follow up with your Chinese supplier after T1, ask these specific questions:

1. For each failing dimension — is the deviation consistent in direction across all samples, or does it scatter on both sides of nominal?
2. What is the proposed root cause — tooling, process, or drawing?
3. What evidence supports that root cause — show me the data.
4. Has a process sensitivity study been performed? What were the results?
5. Is the specified tolerance within NADCA standard or precision capability for this feature type and alloy?
6. If tooling modification is proposed — is the correction in the removal direction (fast, cheap) or the weld direction (expensive, risky)?

A supplier who can answer these questions with data is a supplier who is doing proper engineering. A supplier who responds with "we need to modify the tool, cost is $X, lead time is Y weeks" without supporting evidence has not diagnosed the problem — they have guessed at a solution.

How Can I Avoid Repeated Sample Failures Caused by the Wrong Fix?

The most expensive situation in a Chinese die cast program is reaching T2, T3, or T4 still failing the same dimensions. At that point, each new trial costs tooling modification fees, sampling charges, and weeks of delay — and the downstream pressure from your customers is real. The way to avoid this is to apply the right fix the first time, which means getting the diagnosis right before T2 is authorised.

Repeated sample failures almost always mean the wrong corrective action was applied at T1. To avoid this, require a documented root cause analysis ⁷ with supporting measurement data before approving any corrective action, cross-check every failing tolerance against NADCA capability, and insist on a structured drawing review against process capability before authorising any further tooling modifications after T2.

Why Programs Get Stuck in a Trial Loop

If you have reached T2 or T3 with continuing failures on the same features, the problem is almost certainly one of two things: either the tooling modification at T1 corrected a symptom rather than the cause, or the drawing contains tolerances that exceed process capability and no amount of tooling work will achieve them as-cast.

The first scenario happens when a supplier performs a weld modification based on a visual assessment rather than data, or when a process-driven scatter problem is misdiagnosed as a tooling problem and a modification is made to the cavity that does not address the actual source of variation.

The second scenario is extremely common with buyers who source CNC machined parts for the first time in die casting. Tolerances that are routine for CNC — ±0.02mm on a bored hole, ±0.05mm on a milled face — are either unachievable or marginal for high-pressure die casting ⁸ on the same features without post-cast machining.

The Drawing Audit Before T2

If T1 failed and you are authorising T2, do the drawing audit first. Have a qualified die casting engineer review every dimension and tolerance callout on your drawing against NADCA standard and precision tolerance tables. This typically takes a few hours. The output is a list of every dimension that is within process capability (no issue), every dimension that is marginal (may be achievable with optimised process and precision tooling), and every dimension that exceeds process capability as-cast (requires post-cast machining or drawing revision).

Performing this audit before T2 ensures that if T2 samples fail the same dimensions as T1, you are not spending money proving the same physics twice.

Tracking Corrective Actions and Their Results

For every T1 failure, document the proposed root cause, the supporting evidence, the approved corrective action, and the expected outcome. When T2 samples arrive, compare the results to the prediction. If the corrective action was correct, the deviation should be resolved or significantly improved. If the same dimension fails again with a similar deviation, the T1 diagnosis was wrong.

This documentation also protects you in commercial discussions with your supplier. If a supplier performs two rounds of tooling modification without resolving a failure that was actually caused by an unrealistic drawing tolerance, the responsibility allocation becomes clear from the documented history.

When to Escalate to a Pre-Production Engineering Review

If your program has stalled in a trial loop with three or more failing trials, the correct escalation is a formal pre-production engineering review involving your design engineering team, the tooling engineer, and an independent die casting process engineer. A structured root cause analysis methodology ⁹ should guide this review. It should re-examine the part design for die casting producibility, revisit every critical tolerance callout, evaluate whether post-cast machining operations should be added to the process flow, and determine whether the tooling design itself has fundamental issues that repeated modifications cannot resolve.

This is a more significant intervention, but it is far less expensive than continuing to run trials on a program that cannot succeed without a design or process change. Ultimately, understanding the die casting process ¹⁰ capabilities and limitations at the outset of a program is the most reliable way to avoid the trial loop altogether.

Conclusion

Diagnosing correctly before spending money is the core principle. Check whether deviation is consistent or scattered, cross-check the tolerance against NADCA capability, and require data before approving any corrective action. The right fix the first time saves weeks and thousands of dollars.

Footnotes

1. Overview of high-pressure die casting: process mechanics, materials, and dimensional consistency. ↩︎
2. CNC post-machining of die castings for tight-tolerance bores, faces, and threaded features. ↩︎
3. Wikipedia entry on root cause analysis: methods for identifying the true source of failures. ↩︎
4. How high-pressure die casting produces near-net-shape components with minimal secondary ops. ↩︎
5. Key considerations when planning post-machining operations on aluminium die cast parts. ↩︎
6. Introduction to GD&T: how tolerances, datums, and feature controls are defined on drawings. ↩︎
7. Structured root cause analysis in manufacturing: steps, tools, and corrective action planning. ↩︎
8. Die casting process guide: capabilities, secondary machining needs, and design considerations. ↩︎
9. Root cause analysis in manufacturing: systematic process for eliminating recurring defects. ↩︎
10. Wikipedia: die casting process overview, alloys, tolerances, and industry applications. ↩︎