A Three-Second Silence as the CPK Report Was Revealed
I still remember several years ago, when our factory first landed that order from a European car manufacturer, everyone was incredibly excited. Not long after, the client reported that a batch of chips had exceeded frequency drift specifications during testing at their assembly plant, causing them to be rejected outright. At that moment, the production line deputy manager's face turned green; he pulled a few of us engineers into a review meeting. I recall him looking at that CPK report, seeing a red '1.08' on it, and the entire room truly went silent for three seconds. The deputy manager sighed and said, "This is trouble. The client is now demanding that our Tier 1 supplier halt all operations for inspection, and we, as Tier 2, also need to provide a clear explanation."
Where Did the Problem Occur?
To put it plainly, the automotive supply chain is a clearly stratified system, with each layer having its specific role. At the top are the car manufacturers (OEMs), responsible for vehicle design, assembly, and branding. Below them are the so-called Tier 1 suppliers, major companies like Bosch, Continental, and Denso. They supply directly to the car manufacturers, providing modules or subsystems, such as entire braking systems or infotainment systems. Companies like ours, which make chips and sensors, are typically Tier 2, selling our products to Tier 1. Some even lower tiers, providing raw materials or standard components, are Tier 3 or even lower.
So, here's the crucial point: when a problem is discovered by the car manufacturer, they directly approach their 'parent'—the Tier 1 supplier. Tier 1, naturally displeased, then turns to its 'children'—us, the Tier 2 suppliers. Tracing it down layer by layer, who bears the greatest responsibility? Theoretically, whoever caused the problem is responsible. But in reality, Tier 1 suppliers hold the power of direct communication with the car manufacturers, and their quality demands on their suppliers are extremely stringent. In our frequency drift case, even though our Tier 2 chips were just one component in the chain, because they were the source of the issue, Tier 1 demanded that we provide a complete Failure Analysis (FA) report, propose an improvement plan, and even trace the issue back to our Tier 3 suppliers to check if raw materials were at fault.
How Is It Done in Practice?
Frankly, the automotive industry's quality requirements are several notches higher than those for consumer electronics. A DPMO (Defects Per Million Opportunities) level of 6210, which might be acceptable in consumer electronics, is an absolute disaster in the automotive sector. Car manufacturers routinely demand a 6 Sigma level, meaning DPMO must be within 3.4, and Cpk must be at least 1.33.
So, when a problem occurs, how do you determine responsibility?
- Consult the contract, especially the Quality Agreement: This document explicitly defines each supplier's quality responsibilities, inspection standards, and anomaly handling procedures. In our case, we were constrained by the contractual clause requiring a Cpk of 1.33, as our 1.08 was simply unacceptable.
- Review the Failure Analysis (FA) Report: This is the most compelling evidence. We had to retrieve the chips returned by the client and analyze them using various instruments and equipment to determine if it was a design issue, a process issue, or a raw material issue. If the FA report clearly indicated that our process parameters had drifted, then responsibility was undeniable.
- Examine Traceability: Every batch of material and every process step must have complete records. Information such as the material lot number, production date, machinery used, and operator must be traceable. If we can trace the returned chips to a specific batch, it allows us to narrow down the investigation scope and identify the truly problematic stage.
The Most Common Pitfalls
To be frank, the most common pitfalls are 'cutting corners' and 'incomplete data'. Many engineers, in a rush to meet mass production deadlines, might slightly relax process monitoring, thinking, 'It probably won't make a difference.' The result is that only when a quality issue arises do you discover that what should have been measured wasn't, and what should have been recorded wasn't.
I remember once, a new colleague manually recorded a process parameter, mistakenly writing 0.5V as 5V. Consequently, the chip voltages for subsequent batches all exceeded specifications. When the client investigated, tens of thousands of units were rejected. We spent several days clarifying the issue then, only to discover it was due to a data entry error. What was worse, there were no digitized records at the time, nor any second confirmation, making quick traceability impossible. This was a classic 'weak link' scenario: a small oversight led to a cascade of problems. Therefore, data integrity and accuracy are truly more important than anything else.
One Thing You Can Do Today
Check your quality documentation and ensure all parameters are accurately recorded!