At the end of February 2024, I had 50 TE Connectivity industrial connectors lined up on my inspection bench and a spreadsheet on my screen telling me to get rid of them.
The spreadsheet came from procurement. It compared the TE part against a cheaper alternative that a new supplier had submitted for one of our telemetry control units. TE’s connector cost about 19 percent more per unit. On a 50,000-unit annual buy, that gap was roughly $27,000 in apparent savings. My purchasing manager called it a no-brainer. I called it a reason to slow down.
I’m a quality inspector for a manufacturer of industrial control and monitoring equipment. I review 200+ unique components a year, so I’ve learned to be suspicious of spreadsheets that don’t come with physical parts attached. Before the bench tests, I did what I always do with a supplier in a qualification review: I searched. First te-connectivity specs. Then the te connectivity industrial catalog. Then, because search autocomplete is a dangerous thing, te connectivity layoffs. The engine also offered transparent smartphone and networks vs cisco, two queries that had nothing to do with my job. I ignored them. The layoffs query stayed with me.
Restructuring coverage follows every large manufacturer, and I’ve seen what that noise does inside a buying organization: a headline becomes a rumor, a rumor becomes an unspoken reason to pick the cheaper part. My first instinct was to downgrade TE to approved-but-not-preferred and move on. I’m glad I didn’t.
The multimeter doesn’t read headlines
I started the evaluation the same way I start every supplier evaluation: on the bench, with parts in front of me. I put the multimeter in four-wire resistance mode and checked contact resistance across all 50 TE connectors. The spread from best sample to worst was roughly 0.4 milliohms. Then I ran the same test on 50 samples from the alternative supplier. Three fell outside their own printed specification. A fourth had a plating defect I could see at 10x magnification.
Four out of 50 is not a statistical proof—it’s a warning. I’m not saying the alternative vendor can’t make good parts. I’m saying the samples they chose to send us were inconsistent. Consistency is the real metric. Let me rephrase that. I don’t need a connector to be perfect. I need it to be predictable. A part that fails at a known rate can be designed around. A part that fails when it feels like it creates a warranty crisis.
I also checked the TE housings for mold flash, inspected the plating, and compared sample dimensions against the mechanical drawing. Everything matched. That sounds boring. In my line of work, boring is the highest compliment.
Two weeks of going back and forth
Even with those results, I hesitated. I went back and forth between the cheaper alternative and TE for two weeks. The alternative had a lower price and an energetic sales engineer. TE had a higher price and restructuring coverage. On paper, the decision was obvious. On paper, the cheaper part always wins.
So I asked to visit the factory. TE invited me to one of its industrial connector plants in Germany—southern Germany, near Nuremberg, if I remember correctly, though I might be mixing up the exact location. The visit wasn’t flashy. It was better than flashy. Shift logs. SPC charts on plating thickness. Calibration stamps on torque tools. Operators who could explain their process limits without reading from a slide. No factory is perfect. This one was controlled, and control is the thing I can’t fully test from a sample tray.
I asked the plant manager about the restructuring coverage. He didn’t dodge it. Some business units were being consolidated, he said; the plant I was standing in had gained responsibility rather than lost it. That matches a pattern I’ve seen more than once. People assume layoffs mean quality collapses. The reality is usually messier and less dramatic. Restructuring is a portfolio decision. The product lines that survive a consolidation are often the ones with the strongest process discipline. The weak processes get closed. What remains is more focused, not less. I’m not defending the human cost of restructuring—I’ve watched good coworkers leave with packages, and it’s brutal. But as a predictor of component quality, a headline is almost worthless.
A transparent phone and twenty years of reality
On the flight home, I watched a video about a transparent smartphone concept. It was beautiful, impractical, and about as relevant to my job as a concept car is to a fleet manager. But it made a useful point.
Consumer hardware gets attention because it fails in public. A phone that overheats becomes a video with a million views. An industrial connector fails inside a sealed cabinet at 2 a.m., and the only evidence is a production line that won’t start. The transparent smartphone will be forgotten in a year. A good industrial connector is supposed to be forgotten for twenty years. One of those forgetting periods is a success; the other is a failure.
That’s why I kept returning to the te connectivity industrial product line. The same company makes micro-connectors and sensors for consumer devices, but its industrial line is where the engineering margin lives—vibration, temperature swings, decades of service. When your equipment sits in a substation or on a factory floor, you don’t want a connector designed for a phone. You want one designed for the twenty-year penalty.
The math I should have run first
Back at my desk, I ran the numbers the way I wish procurement had run them the first time. Not unit price. Total cost of ownership.
I kept the model deliberately simple:
- TE’s price on 50,000 units: about $27,000 more than the alternative.
- TE’s field history with a similar connector in our equipment, measured over five years: about one failure per 2,000 installed parts.
- The alternative’s sample lot: 4 of 50 parts didn’t meet their own specification. Even if the real production rate was ten times better than the sample implied, that still meant hundreds of questionable parts over a 50,000-unit run.
As soon as I wrote that down, the $27,000 argument collapsed. A price gap sounds large on a purchase order. It’s small next to the cost of one recall, one site visit, or one 3 a.m. phone call asking why a production line is down. I didn’t need exact probabilities. I needed to know which supplier had the narrower distribution of outcomes. TE had years of field history. The alternative had one carefully selected sample lot.
Unit price is what you pay to put a part in your product. Total cost is what you pay when it fails after the product ships. They are different numbers, and too many buying decisions only look at the first one.
This is the part I’m least proud of. I only learned total-cost thinking by ignoring it. A few years earlier, I pushed a lower-cost component through approval against a senior inspector’s advice. The plating quality was marginal—within the vendor’s tolerance, but not within what our storage conditions demanded. Eight thousand units sat in a warehouse long enough for the contacts to corrode. The rework cost us $22,000 and delayed a launch by a month. The supplier covered the replacement parts. They didn’t cover the schedule damage.
I approved TE as the qualified source for that connector. Not for brand reasons. Because the bench data, the factory audit, and the total-cost model all pointed the same direction.
Not another networks vs cisco argument
The purchasing manager pushed back once more. He said I was treating TE like the default safe answer—the Cisco of connectors. It was the same mistake as the old networks vs cisco argument, he claimed: pick a platform and defend it forever.
I told him the comparison was wrong in a useful way. A network platform is chosen for its software ecosystem, management interface, and the team that supports it. You can patch it, reconfigure it, and eventually migrate away from it. A connector is installed once and sealed. You can’t patch a contact beam. You can’t reboot a crimp. The decision lives or dies in the physical part, which means it should be made with physical evidence.
Did TE’s Tyco Electronics heritage matter? Honestly, it did—as an entry ticket. Heritage gets you invited to the audit. It doesn’t earn you the order. TE earned the order on the bench, in the factory, and in the spreadsheet that counted the cost of being wrong.
What I’d tell another inspector
If you’re qualifying a critical component supplier, this episode reinforced four things for me:
- A headline is not a failure analysis. Investigate the actual process before you write off a supplier.
- Samples are salespeople. They show you the best version of a vendor. Inspect the process, not just the part.
- Turn quality differences into dollars. If you can’t estimate the cost of failure, you’re comparing prices, not suppliers.
- The lowest unit price is a marketing result. The lowest total cost is an engineering decision.
I’m not telling you to put TE Connectivity on every design. If a part failure is cosmetic, the cheaper option might be perfectly fine. But if a failure means a truck roll, a production stoppage, or a safety hazard, choose with your multimeter instead of your price column.
(Should mention: nobody paid me to write this, and I have no commercial relationship with TE. I’m writing it because I nearly made a 50,000-unit mistake that a bench test prevented.)
As of this writing, we’re about ten months into the program. Just under 48,000 units have shipped, and I haven’t seen one field failure traced back to that connector. The spreadsheet still says we paid more. I still think we got the cheaper deal.