Die Casting Case Study / Dimension Control
ZAMAK 3 Pearl Chrome Die Casting Case Study: Assembly Step Dimension Analysis and Tolerance Review
A ZAMAK 3 die casting and hexavalent pearl chrome plating case combining production fitting inspection, dimensional sampling and a 148-piece three-level DOE to evaluate how 37.50, 37.55 and 37.60 dimension settings relate to loose-fit risk.
Case Background
| Item | Engineering Record |
|---|---|
| Material and process | ZAMAK 3 zinc alloy die casting with hexavalent pearl chrome plating. |
| Production route | Die casting, degating, deburring, drilling and tapping, grinding and polishing, pearl chrome plating, threaded insert assembly, inspection and packing. |
| Quality concern | Dimensional concern suspected to cause assembly appearance issues, especially step protrusion after fitting with the mating plastic part. |
| In-process lot | 3500 pcs were checked by fitting inspection; after fitting, the assembly step protrusion was confirmed acceptable. |
| Dimensional sample | 200 pcs were randomly selected from this batch for measurement of two key dimensions. |
Inspection Logic
Measurement Summary
| Dimension | Current specification | Current lower limit | Current upper limit | Sample minimum | Sample maximum | Result |
|---|---|---|---|---|---|---|
| 14.3±0.1 | 14.20 to 14.40 | 14.20 | 14.40 | 14.26 | 14.75 | Exceeds upper specification limit. |
| 37.34+0.15/-0.1 | 37.24 to 37.49 | 37.24 | 37.49 | 37.35 | 37.65 | Exceeds upper specification limit. |
Assembly Result
Although both measured dimensions exceed the current drawing upper limits under the present manufacturing route, the fitting inspection showed that the assembly step protrusion was acceptable after adaptation. Based on this batch evidence, the step appearance result did not show a significant correlation with these two measured dimensions alone. This finding applies to step protrusion appearance in that production batch and should not be generalized to the separate loose-fit response evaluated in the follow-up DOE.
Engineering Analysis
- The two metal-part dimensions are out of the current drawing limits, so the drawing and process capability cannot be judged only by nominal conformance.
- The accepted assembly step result indicates that the actual appearance risk is likely controlled by the full assembly stack-up, not by either dimension alone.
- Based on the dimensional distribution and the fitting result, the mating plastic part should be checked for shrinkage, because plastic shrinkage may shift the actual assembly relationship.
- A metal-part-only rejection rule could create unnecessary sorting if the final assembly function and appearance are still acceptable.
Follow-up DOE: 37.50, 37.55 and 37.60
A follow-up full inspection compared three nominal dimension levels against loose-fit results. Material defects and plating defects were recorded as separate dispositions so that assembly-fit behavior could be evaluated on both the total-inspection basis and the effective-fit basis.
Full inspection record
| Dimension level | Fit OK | Loose fit | Material defect | Plating defect | Total | Loose / total |
|---|---|---|---|---|---|---|
| 37.50 | 21 | 30 | 4 | 1 | 56 | 53.57% |
| 37.55 | 29 | 2 | 5 | 6 | 42 | 4.76% |
| 37.60 | 24 | 17 | 1 | 8 | 50 | 34.00% |
Effective fit comparison
Because material and plating defects do not provide an independent fit judgment, the primary comparison uses only parts classified as fit OK or loose fit.
| Dimension level | Fit OK | Loose fit | Effective fit sample | Loose-fit rate |
|---|---|---|---|---|
| 37.50 | 21 | 30 | 51 | 58.82% |
| 37.55 | 29 | 2 | 31 | 6.45% |
| 37.60 | 24 | 17 | 41 | 41.46% |
Statistical result: Pearson chi-square = 22.13, p = 0.0000156 and Cramer’s V = 0.424. Under these sampled conditions, dimension level and loose-fit result show a statistically significant, medium-to-strong association.
DOE Interpretation and Decision Boundary
- 37.55 produced the lowest effective loose-fit rate at 6.45%, compared with 58.82% at 37.50 and 41.46% at 37.60.
- The odds of a loose fit were approximately 20.7 times higher at 37.50 than at 37.55, and approximately 10.3 times higher at 37.60 than at 37.55.
- The difference between 37.50 and 37.60 is not conclusive on the effective-fit basis because its 95% confidence interval includes 1.
- The response is non-linear: the middle level performed best, so the data do not support a simple rule that increasing or decreasing the dimension continuously increases looseness.
- Combined material and plating defect rates also varied by group: 8.93% at 37.50, 26.19% at 37.55 and 18.00% at 37.60. This variation indicates possible batch, cavity, polishing or plating confounding.
Engineering conclusion: the current data support an association between the tested dimension levels and loose-fit outcome, but they do not prove that dimension alone caused the difference. The 37.55 level is the best candidate among the three tested settings and should be treated as a validation center point, not yet as a released production tolerance.
Recommended Confirmation DOE
Analyze the confirmation run with logistic regression including linear and quadratic dimension terms, with lot or cavity used as blocking factors. This will test whether the apparent middle optimum is repeatable and separate dimension influence from process variation.
Tolerance Review Direction
| Dimension | Current tolerance | Proposed review tolerance | Engineering note |
|---|---|---|---|
| 14.3 | 14.3±0.1 | 14.3+0.25/-0.1 | This raises the upper limit to 14.55, but the observed sample maximum is 14.75. Outliers and full distribution still require review. |
| 37.34 | 37.34+0.15/-0.1 | 37.34+0.25/-0.1 | This raises the upper limit to 37.59, but the observed sample maximum is 37.65. Additional validation is needed before formal drawing release. |
The proposed tolerance change should be treated as a drawing-review direction, not as a final approval by itself. Before release, XSD recommends confirming plastic-part dimensions, assembly stack-up, step protrusion gauge criteria, and Cpk/Ppk or batch-distribution evidence.
Improvement Actions
- Measure the mating plastic part and confirm whether shrinkage is present across different cavities, batches or suppliers.
- Create an assembly stack-up record connecting metal dimensions, plastic dimensions and accepted step protrusion results.
- Use a dedicated fitting gauge or visual limit sample for the step protrusion acceptance decision.
- Separate true dimensional outliers from tolerance limits that may be too narrow for the functional assembly requirement.
- Only revise the drawing tolerance after sample distribution, assembly validation and customer acceptance criteria are aligned.
Need die casting dimension and assembly stack-up review?
Send drawings, measurement samples, mating-part data and assembly acceptance criteria. XSD can help separate true dimensional risk from functional assembly tolerance.
压铸案例学习 / 尺寸控制
ZAMAK 3 珍珠铬压铸件案例：装配台阶尺寸分析与公差评审
本案例针对ZAMAK 3锌合金压铸并电镀六价珍珠铬的产品，结合在制适配检测、尺寸抽样和148件三水平DOE，分析37.50、37.55与37.60尺寸档位对实配松风险的影响。
案例背景
| 项目 | 工程记录 |
|---|---|
| 材料与工艺 | ZAMAK 3 锌合金压铸，电镀六价珍珠铬。 |
| 生产路线 | 压铸、去水口、去毛刺、钻孔攻牙、打磨抛光、电镀珍珠铬、组装螺纹牙套、检验、包装。 |
| 质量关注点 | 因尺寸问题怀疑导致装配外观质量问题，重点关注与塑胶件适配后的台阶凸出状态。 |
| 在制批量 | 对在制 3500 件进行适配检测，经适配后确认台阶凸出合格。 |
| 尺寸样本 | 从本批中随机抽查 200 件组成样本，对两个关键尺寸进行测量和分析。 |
检测逻辑
测量数据摘要
| 尺寸 | 当前规格 | 当前下限 | 当前上限 | 样本最小值 | 样本最大值 | 判定 |
|---|---|---|---|---|---|---|
| 14.3±0.1 | 14.20 至 14.40 | 14.20 | 14.40 | 14.26 | 14.75 | 超出当前规格上限。 |
| 37.34+0.15/-0.1 | 37.24 至 37.49 | 37.24 | 37.49 | 37.35 | 37.65 | 超出当前规格上限。 |
装配结果
虽然按目前加工工艺，这两个尺寸均超出当前图纸规格上限，但本批产品经适配后，台阶凸出结果确认合格。基于本批适配证据，装配后台阶外观结果与这两个单独测量尺寸之间未表现出显著关联。该结论只适用于本批产品的台阶凸出外观，不能直接外推到追加DOE所评价的“实配松”响应。
工程分析
- 两个金属件尺寸确实超出当前图纸规格，因此不能只按名义规格直接判断工艺能力已经满足图纸。
- 台阶适配结果合格，说明实际外观风险更可能由整套装配尺寸链共同决定，而不是由某一个金属件尺寸单独决定。
- 根据尺寸分布和适配结果，建议重点复测与之配合的塑胶件，判断是否存在缩水导致的装配关系偏移。
- 如果只按金属件当前图纸上限进行拒收，可能会产生不必要的挑选和返工，但如果直接放宽公差，也需要数据证据支撑。
追加DOE：37.50、37.55与37.60尺寸档位
追加全检试验比较三个名义尺寸档位与“实配松”的关系。素材不良和电镀不良分别记录，避免把不能独立完成实配判定的样品直接混入装配松紧分析。
全检原始记录
| 尺寸档位 | 实配OK | 实配松 | 素材不良 | 电镀不良 | 总数 | 实配松/总数 |
|---|---|---|---|---|---|---|
| 37.50 | 21 | 30 | 4 | 1 | 56 | 53.57% |
| 37.55 | 29 | 2 | 5 | 6 | 42 | 4.76% |
| 37.60 | 24 | 17 | 1 | 8 | 50 | 34.00% |
有效实配口径
素材和电镀不良不能提供独立的实配松紧判定，因此主分析只使用“实配OK”和“实配松”两类样品。
| 尺寸档位 | 实配OK | 实配松 | 有效实配数 | 实配松率 |
|---|---|---|---|---|
| 37.50 | 21 | 30 | 51 | 58.82% |
| 37.55 | 29 | 2 | 31 | 6.45% |
| 37.60 | 24 | 17 | 41 | 41.46% |
统计结果：Pearson卡方值=22.13，p=0.0000156，Cramer’s V=0.424。在本次样品条件下，尺寸档位与实配松结果存在显著且中等偏强的关联。
DOE解释与决策边界
- 37.55的有效实配松率最低，为6.45%；37.50为58.82%，37.60为41.46%。
- 37.50相对于37.55的松动几率约高20.7倍；37.60相对于37.55约高10.3倍。
- 在有效实配口径下，37.50与37.60之间的差异尚不足以确认，其95%置信区间包含1。
- 结果呈非线性，中间档位表现最好，不能解释为尺寸单纯增大或减小就会持续增加松动。
- 三组素材与电镀综合不良率分别为8.93%、26.19%和18.00%，提示批次、模穴、抛光或电镀条件可能构成混杂因素。
工程结论：现有数据支持测试尺寸档位与实配松结果有关联，但不能证明尺寸是造成差异的唯一原因。37.55是三个测试档位中的最佳候选，应作为下一轮验证中心点，暂不直接作为已批准的量产公差。
下一轮确认DOE建议
确认试验建议采用包含尺寸线性项和二次项的逻辑回归，并将批次或模穴作为区组因素，从而验证中间最优现象能否重复，并区分尺寸影响与制程波动。
公差评审建议
| 尺寸 | 当前公差 | 建议评审公差 | 工程备注 |
|---|---|---|---|
| 14.3 | 14.3±0.1 | 14.3+0.25/-0.1 | 上限调整后为 14.55，但样本最大值为 14.75，仍需确认是否存在异常点或是否需要进一步评审尺寸链。 |
| 37.34 | 37.34+0.15/-0.1 | 37.34+0.25/-0.1 | 上限调整后为 37.59，但样本最大值为 37.65，正式改图前仍需增加装配验证和分布分析。 |
因此，14.3+0.25/-0.1 和 37.34+0.25/-0.1 更适合作为图纸公差优化的评审方向，而不是直接作为最终放行结论。正式变更前，建议补充塑胶件尺寸、装配尺寸链、台阶凸出限度样和 Cpk/Ppk 或批量分布证据。
质量改进动作
- 同步测量配合塑胶件，确认不同模穴、批次或供应商是否存在缩水差异。
- 建立金属件尺寸、塑胶件尺寸与台阶凸出结果之间的装配尺寸链记录。
- 针对台阶凸出建立专用适配检具或外观限度样，避免只靠单一尺寸判定。
- 区分真实尺寸异常点与当前图纸公差偏窄两类问题，分别采取修模、挑选或改图措施。
- 公差变更应在样本分布、装配验证和客户允收标准一致后再正式导入图纸。
需要压铸尺寸与装配尺寸链评审？
请提供图纸、测量样本、配合件数据和装配允收标准。XSD 可协助区分真实尺寸风险与功能装配公差边界。