做 DFMEA 时，失效分析是不是写一句"电机不转"就交差了？AIAG-VDA FMEA 手册把失效分为七大类别，只写"功能丧失"的 DFMEA，恰恰漏掉了售后投诉里最多的那几类。下面以电动车窗升降器为例，把七类失效一个个说透。
▌ 直观的两类：功能丧失与功能退化
功能丧失：完全零输出——按开关，车窗纹丝不动。这是直观、容易写的一类，严重度通常也最高。
功能退化：性能随时间衰退——新车 2 秒升窗，一年后变成 8 秒。根因是碳刷磨损、润滑脂老化、密封条摩擦增大。这类失效要分析退化速率和寿命边界，而不是简单打个勾。
▌ 最难抓的两类：间歇性与部分功能
间歇性功能：时好时坏——晴天正常，雨天卡住；拍一下门板又好了。电磁干扰、接触不良、传感器阈值漂移是主因，台架试验要重点覆盖探测度。
部分功能：有但不全——升到一半卡住，或只能升不能降。分析时要区分是机械卡滞还是软件限位逻辑出错，两条路径对应的纠正措施完全不同。
▌ 最危险的三类：过度、非预期、延迟
- 过度功能：做过头了——升窗到顶不停止，继续顶上去，导致玻璃卡扣碎裂。后果可能比"不动"更严重；
- 非预期功能：错误触发——熄火拔钥匙后，车窗自己降下来。触发条件错了，BCM 软件逻辑、CAN 信号干扰、防夹误触发是排查重点；
- 延迟功能：晚了一拍——按开关 3 秒后车才开始动。继电器触点氧化、启动电容老化、控制模块延时是主因，放在安全系统里可能是致命的。
▌ 实操：每个功能都过一遍“七连问”
实际做 DFMEA 时，对每个功能都要问七遍：它会不会以这种方式失效？超过八成的 DFMEA 只写了功能丧失，退化、间歇、过度、非预期这四类几乎没人写——而它们恰恰是售后投诉里最多的。
失效类别写不全，后面的严重度打分、探测措施、设计改进全是空中楼阁。
记一句：DFMEA 的深度，不在于表格填得多满，而在于七类失效问得多全。
想让团队把 DFMEA 从"填表"做成真正的"失效分析"？续创的 FMEA 失效分析与失效库系统，帮你把七类失效和历史案例沉淀下来。欢迎联系续创信息技术咨询。
When your team does a DFMEA, does the failure analysis stop at one line — "motor does not run"? The AIAG-VDA FMEA handbook defines seven failure categories, and a DFMEA that only covers "loss of function" misses exactly the categories that dominate warranty claims. Using a power window regulator as the example, here is each of the seven, explained.
▌ The Two Most Obvious: Loss of Function and Degradation
Loss of function: zero output — press the switch and the window does not move at all. This is the most intuitive and easiest to write, and usually carries the highest severity.
Degradation of function: performance decays over time — a new car raises the window in 2 seconds; a year later it takes 8. Root causes are brush wear, aged grease and rising seal friction. This category demands an analysis of degradation rate and life limits, not a simple checkbox.
▌ The Two Hardest to Catch: Intermittent and Partial Function
Intermittent function: on again, off again — fine on sunny days, stuck in the rain; a knock on the door panel brings it back. Electromagnetic interference, poor contacts and drifting sensor thresholds are the usual suspects, and the test bench must focus on detection.
Partial function: working, but not fully — the window jams halfway, or goes up but not down. Separate mechanical sticking from a software limit-logic fault; the two paths lead to completely different corrective actions.
▌ The Three Most Dangerous: Excessive, Unintended and Delayed
- Excessive function: too much of a good thing — the window reaches the top and keeps pushing until the glass clip shatters. The consequence can be worse than "not moving";
- Unintended function: triggered at the wrong time — the window lowers itself after the ignition is off and the key removed. The trigger condition is wrong; BCM software logic, CAN signal interference and false anti-pinch triggers are the focus;
- Delayed function: a beat too late — the window starts moving three seconds after the switch is pressed. Oxidised relay contacts, aged start capacitors and module latency are the causes — potentially fatal in a safety system.
▌ In Practice: Run Every Function Through the Seven Questions
For every function in a DFMEA, ask seven times: could it fail this way? More than 80% of DFMEAs only document loss of function, while degradation, intermittent, excessive and unintended are almost never written — yet they are the most frequent categories in warranty claims.
If the failure categories are incomplete, the severity scoring, detection controls and design improvements that follow are castles in the air.
One line to remember: the depth of a DFMEA is not how full the form is, but how thoroughly the seven categories have been asked.
Want your team to turn DFMEA from form-filling into real failure analysis? CINO's FMEA failure analysis and failure library system helps you capture all seven categories together with your historical cases. Contact CINO Information Technology Consulting.