The Adapter Nobody Blamed

The client came to us with a device and a legal problem, in that order.

A customer had returned a mobile product with the USB charging port blackened and partially melted. There were photographs. There was a demand letter. There were references to medical bills, fire damage, and the CPSC. The client's legal team wanted to know one thing before they decided how to respond: was this a design defect, or had the customer done something to cause it?

It was, on its face, a routine question. I had seen the same damage pattern dozens of times. The micro USB connector is a simple construction — four conductors, a shell, a receptacle that gets plugged and unplugged hundreds of times over a product's life. When it fails thermally, it usually fails for one of two reasons. Either a fault develops between the power pin and ground, turning the connector itself into a resistive load and dumping the adapter's output into a few cubic millimeters of metal and plastic. Or the contact interface degrades over time — friction, fretting, the slow accumulation of corrosion product at the pin surface — and what was once a negligible resistance becomes enough to generate meaningful heat under charging current. The damage looks similar either way. The forensics, however, are not.

We opened the device. Under the microscope, the receptacle told a familiar story: green corrosion product at the pin interfaces, dendritic growth consistent with electrochemical reaction, the kind of residue that forms when moisture finds its way into an energized connector and stays there. The x-ray showed bridging between the power pin and the ground shield. The evidence pointed toward liquid contamination — customer-induced, most likely. The case looked straightforward.

We decided to replicate it anyway.

The replication was simple enough. We prepared a five percent sodium chloride solution — table salt and distilled water — and introduced it drop by drop into the receptacle of an exemplar unit while it was connected to the client's power adapter under normal charging load. Then we waited.

Within a few hours the temperature at the connector had climbed past sixty degrees Celsius. By morning it had exceeded one hundred. There was a smell first — the particular smell of heated plastic that anyone who has done this work long enough learns to recognize before they see anything. Then discoloration. Then charring at the receptacle mouth, consistent with the customer's photographs down to the pattern of the damage.

We had our replication. The mechanism was confirmed. Liquid contamination, electrochemical reaction at the pin interface, resistive heating under load. Customer-induced. The client's legal team was relieved. We were nearly done.

Then one of my colleagues asked a question that had nothing to do with the connector.

What does the adapter actually do when the resistance drops that low?

It was the kind of question that sounds almost too basic to ask out loud in a room full of engineers. We had been so focused on the connector — on the thing that was visibly damaged, the thing the customer had photographed, the thing the legal team was asking about — that nobody had thought carefully about the other half of the circuit. The adapter was a standard USB charger. It had shipped with the device. It was, as far as anyone had considered it, unremarkable.

We pulled out a decade resistance box and started at the beginning.

The foldback test is straightforward in principle. You connect the adapter's output across a variable resistance and sweep from near-zero — effectively a short circuit — upward through increasing resistance values, measuring the current and voltage the adapter delivers at each point. Every adapter has a characteristic curve. At some resistance threshold, a well-designed adapter will detect a fault condition and fold back its output — reducing current, cutting power, protecting whatever is downstream. That threshold is a design choice. It reflects how the manufacturer has balanced protection against nuisance tripping, how conservatively they have defined a fault.

We ran the client's adapter. Then we went to the supply room and pulled every third-party adapter we could find — units from the major players, the ones sitting in desk drawers and conference rooms, the ones that had shipped with phones and tablets and speakers from across the industry. We ran them all.

The client's adapter continued supplying current at resistance values where every competitor had already folded back. Not by a small margin. The protection threshold was meaningfully wider — the adapter would drive power into fault conditions that other adapters had already decided were not worth powering through. In a contaminated connector, where the resistance between power and ground might sit anywhere along a wide and unpredictable range as the corrosion evolved, this mattered. It meant the client's adapter would keep the fault energized — keep the heating going — in scenarios where another adapter would have simply stopped.

The connector was identical to every other micro USB connector in the industry. The adapter was not.

We now had two distinct findings on the table, and they did not agree with each other.

The first said this was a customer abuse case. The contamination was real, the mechanism was confirmed, and the connector itself was industry-standard — there was no meaningful design differentiation at the receptacle. The second said this was an elevated-risk product — not because of the connector, but because of what was driving it. The adapter's protection curve was an outlier. In the specific failure scenario the customer had experienced, the client's own charger would have made things worse than a competitor's charger would have.

Both findings were true. That was the problem.

The field return data did not cooperate with our finding.

Ten cases per month, roughly. Across a product line that had shipped in the hundreds of thousands. The rate was not nothing, but it was not the rate you would expect from an adapter that was meaningfully more likely to sustain a fault through to thermal damage. If the foldback curve was the differentiating variable, the field should have been showing us something. It wasn't.

We went back to the returns.

Not the numbers — the actual cases. The photographs customers had submitted, the notes from the service centers, the online reviews that contained images. What we were looking for was the adapter in the frame. Not the device. Not the cable. The adapter. The thing plugged into the wall.

It took time. Most customers don't photograph the adapter. Most service intake processes don't ask about it. The adapter is invisible in the failure narrative because nobody thinks of it as a variable — it ships in the box, it charges the device, it recedes into the background. But enough customers had photographed enough of their setup that a pattern emerged.

They were not using the client's adapter.

Not exclusively, and in many cases not at all. They were using whatever was nearby — the charger from their phone, the one that came with a tablet two years ago, the white brick from a different manufacturer sitting on the nightstand. The micro USB connector had become so universal by that point that interchangeability was assumed. Customers mixed and matched without thinking about it because there had never been a reason to think about it. A charger was a charger.

What this meant, statistically, was that the specific combination our foldback test had identified as elevated risk — the client's adapter driving a contaminated connector — was not occurring at the rate the product's install base would suggest. Customers were, without knowing it, self-selecting away from the dangerous configuration. They were borrowing protection from the more conservative foldback curves of their other devices' adapters. The field return rate was low not because the design was safe but because behavior had accidentally compensated for the design.

The client had been lucky. Not safe. Lucky.

The meeting where we presented this was a particular kind of quiet.

There is a specific silence that falls in a room when engineers and lawyers hear something they were not expecting and are not sure yet whether it is good news or bad news. This was that silence. On one hand, the field data was benign. On the other hand, we had just explained why it was benign, and the explanation was not reassuring. The protection the product had enjoyed was not engineered. It was accidental. It was contingent on customer behavior that could change — that would change, if the client ever ran a campaign encouraging customers to use only the included adapter, or if the industry shifted away from interchangeable standards, or simply if the mix of third-party adapters in customers' homes ever narrowed in the wrong direction.

The legal team asked whether we thought this needed to be reported.

The honest answer was that the adapter had a protection mechanism. It folded back eventually. The threshold was wider than the industry, but the threshold existed — this was not an unprotected output. The connector was standard. The failure mechanism required customer-introduced contamination. There were multiple conditions that had to coincide for the thermal damage to occur, and the field data suggested those conditions were not coinciding at a rate that implied imminent widespread harm.

The equally honest answer was that we had identified a measurable design characteristic that differentiated the client's product from the rest of the industry in a direction that increased risk. We had replicated the damage. We had a customer with photographs and a demand letter. And we were sitting in a room deciding how to characterize all of that.

What the client did was redesign the adapter. A new foldback curve, tighter protection, brought into line with what the rest of the industry was already doing. The change was introduced into production without announcement. Existing customers were not notified. The threatening customer's claim was settled quietly. No CPSC report was filed.

The case was closed.

I have thought about this case more than most.

The technical conclusion was defensible. The legal reasoning was defensible. The adapter did have protection. The field data genuinely did not suggest a pattern of harm at a reportable scale. None of the decisions made in that room were obviously wrong.

But the customers who had returned damaged devices — the ones with the blackened ports and the photographs, the ones who had experienced something frightening in their homes — they received settlements or replacements or silence, and none of them received an explanation. They did not know that the adapter they had been given was operating with a wider fault tolerance than the rest of the industry. They did not know that if they had happened to grab a different charger from a different drawer, the outcome might have been different. They went on using their devices, and presumably their adapters, without that information.

I do not know what the right answer was. I know what the defensible answer was, and I know they are not always the same thing.

What I learned from this case I have carried into every system-level investigation since: the thing that is damaged is rarely the thing that caused the damage. Everyone in that first meeting was looking at the connector. The connector was the victim. The question worth asking — the one that changed everything — was about the device nobody had thought to blame.

It usually is.