Overstressed over an erratically flickering check-engine lamp

Be prepared to respond to “electrical overstress” claims from semiconductor suppliers by citing their own datasheet entries.

During my automotive embedded career, I was heading a product team controlling auxiliary functions of a diesel engine powertrain for passenger vehicles.

As you are likely already aware, a “Check Engine” lamp illuminates on the driver dashboard in case of any malfunction with the engine or its control system. This activation alerts the driver, so that he or she can get the malfunction rectified before continuing the journey.

In one of our production batches, the vehicle assembly line technician started reporting “erratically and intermittently illuminating check engine lamp” as feedback. All of these vehicles were summarily rejected and blocked from further dispatch.

Do you have a memorable experience solving an engineering problem at work or in your spare time? Tell us your Tale

Tempers predictably flared all across the vehicle plant as the population of rejected vehicles grew rapidly. Curiously, our electronic control unit (ECU) associated with this function had been performing flawlessly for the last few years. It was a mystery to all of us why such a problem should crop up so suddenly.

When the failure was analyzed in our laboratory, it quickly became clear that the output driver IC pin used for driving the LED was pumping around 4 mA of current through the LED even in the “off” state! The datasheet had guaranteed a maximum leakage current value of around 0.5 mA. The output driver IC was apparently misbehaving, at least as far as the associated driving pin was concerned.

The IC semiconductor supplier’s application engineers were contacted and briefed about the problem. They later responded with a stock diagnosis of “electrical overstress” as the root cause.

“But what can ever ‘overstress’ your driver pin in such a simple circuit?,” we replied in attempting to counter the (lack of) logic behind their answer. “A simple short wire connects the ECU to the dashboard LED connector.”

“It must be electroshock discharge (ESD) events occurring during your electronic assembly, then,” they responded. “Sometimes ESD precautions are violated by the assembly technicians. We have faced such problems from a few customers in the past.”.

We clearly understood by this point that the semiconductor supplier team was trying to wriggle out of the situation without taking any responsibility or otherwise helping us. “How can so called ‘ESD events’ be partial only to a specific pin of your driver chip?,” we responded, in striving to alert them to the flaw in their reasoning. “It should have caused random failures across all units we’ve manufactured to date, and with all of the pins of your driver chip.” However, the supplier team clung to its reasoning, bringing the situation to an impasse.

Every passing day, our plant situation was getting more and more explosive, when I was struck by an idea. I asked our team to note down the alphanumeric production batch codes stamped on all of the culprit chips. To our surprise, all of them were identical! In contrast, the healthy chips all had other production batch codes.

We wrote an inquiry mail to the semiconductor company, asking them to decode the alphanumeric production batch code stamped on the culprit chips. Their answer was predictable: “It means: This chip was manufactured in country X and plant location Y, with month/year code as mm/yy.” Apparently, all of the culprit chips were manufactured at a specific plant location, where a production quality lapse was presumably leading to the disaster we were facing.

What was the way ahead? We pre-filtered out all of the culprit chips based on their production batch code, preventing them from being assembled on our boards. We also filtered them out of our supply chain by directing our distributors not to supply such chips to us. These steps solved the problem for once and all!

The lessons and insights behind this case are as follows:

Be thorough with your circuit analysis. This means, among other things, completely understanding all datasheet entries for all key components.
Prepare yourself to respond to “electrical overstress” responses from semiconductor suppliers by citing their own data sheet entries that support your design decisions.
“Electrical overstress” can occur due to numerous possible causes, such as ESD, over voltage, over-current, excessive power dissipation etc. Your prior circuit analysis during design verification should rule out these possibilities. Encourage semiconductor supplier engineers to sign off your circuit analysis. This will reduce their chances of later claiming “electrical overstress” as a diagnosis.
Be mindful of correlating production batch codes with culprit chips. During another of my experiences, this one more recent, I encountered randomly misbehaving CPU chips in one of the ECU failures I was analyzing. I was amazed to learn that all of these culprit CPUs also contained the same production batch codes. No semiconductor supplier will openly admit to quality issues in its own manufacturing plants, even if they’ll all hopefully take corrective action internally.
Always keep in mind that a random root cause, such as ESD mishandling, cannot cause a consistent failure such the exact same pin misbehaving every single time.

Fortunately, we hit upon the corrective solution to the seemingly intractable field issue quite early, which otherwise would have led to a substantive loss of business for our vehicles and company.

Vishwas Vaidya is a graduate of the Indian Institute of Technology in Delhi, India. Currently, he is self-employed as an engineering consultant and industry faculty member in the field of embedded systems for global automotive clients and high-repute academic institutions. Vishwas’ articles and research reports have appeared in many worldwide engineering publications.

Related Content

The post Overstressed over an erratically flickering check-engine lamp appeared first on EDN.

Source link