I coordinate emergency supply for an authorized Kemet Electronics distributor. In my role, I've handled maybe 200 rush orders in nine years. Maybe 180, I'd have to check the log. Most are straightforward: customer needs 5,000 capacitors, we book a courier, the line keeps moving. The one I keep replaying started with a blood pressure monitor.
We are an authorized Kemet Electronics distributor. That may sound like a marketing detail, but it matters during a failure investigation. A traceable lot removes suspicion from the component early in the process. I could tell the engineer exactly which reel, which date code, and which test data the parts came from.
In March 2025, at 4:47 p.m., a medical device manufacturer outside Minneapolis called. Their blood pressure monitor line was failing final test at a rate of about one in twelve. The test operator described the symptom: the display locks at '--', the cuff symbol blinks, and the unit needs a hard reset.
One thing that delayed the troubleshooting started with the display itself. Medical device interfaces use blood pressure monitor symbols to show status: the cuff, the pulse icon, the battery warning. When the operator said the cuff symbol had a slash through it, my first thought was that the pressure sensor circuit had lost its reference. That pointed back to the nearby Kemet capacitor. The symbol wasn't wrong. My interpretation was too narrow.
The technician's first guess was the Kemet capacitor on the pressure-sensor input. The part was a Kemet C1206C104K5RACTU, a 100 nF MLCC, X7R dielectric, 50 V, in a 1206 package. They pulled one from a failing unit, asked for 2,000 replacements, and wanted them by 6 a.m. Normal lead time through our channel was two days. This was a rush order, so rush fees applied.
The order was maybe $450—no, $470, I'm probably mixing in the test boards. The expedited courier added $380. We delivered at 5:05 a.m. Six hours later, the customer called back with the words no one in this role wants to hear: 'Are these the right parts?'
The capacitor checked out
I drove to their facility that same morning with a benchtop multimeter and a known-good reference board. The Kemet capacitors we'd sent were fine. Capacitance measured 98.6 nF against a 100 nF rating. ESR was in range. Insulation resistance was above the datasheet minimum. I also measured the original cap from the failing unit. It was the same. The failure rate stayed exactly the same after the swap.
Everything I'd read about MLCC failures said to expect cracked ceramic, low insulation resistance, or a shorted layer. This was an intermittent open in the analog path, which is much harder to catch. The multimeter said the capacitor was good, and it was. But that created a false conclusion: because the part was good, we assumed the part was irrelevant. The real issue was hiding two inches away.
The data said capacitor. My gut said sensor. Neither was right. The connector was outside both of my mental maps, and that's exactly why it survived for four hours of bench testing.
The connector that fooled everyone
On the bench, we found a 6-pin connector between the pressure sensor cable and the main board. The crimps on the 3.3 V and ground terminals looked fine from the top. The terminals were seated, the wire colors matched the harness drawing. But when I pulled on the wire with light force, the conductor slid out of the barrel.
That connector failure was deceptive. A good crimp holds the conductor tightly enough that resistance stays flat. A bad crimp can pass continuity when the cable is straight and then open up when the cable bends or warms up. That open looked like a failed capacitor to the monitor's firmware, because the voltage on the sensor rail drooped for a few milliseconds.
I didn't fully appreciate how to crimp connectors until that moment. The process is not complicated:
- Strip the wire to the barrel length, not a millimeter more.
- Use the terminal family specified by the harness drawing.
- Crimp with a tool designed for that terminal, not a generic crimper.
- Pull test the wire afterward. A good crimp holds; a bad crimp slides out.
According to IPC/WHMA-A-620, a crimp connection has to meet crimp height and pull force requirements, not just look tidy. The crimps on that connector looked tidy. They were still wrong. We spent the next three hours re-terminating 14 connectors and rerunning final test. The failure rate dropped to zero for the rest of the batch.
Why the mechanical path matters
I've been doing this long enough to remember when the first reaction to an intermittent failure was to replace all the passive components. The fundamentals haven't changed—contacts still need to be compressed around the wire—but the execution has. Capacitors like Kemet's are more consistent than they were twenty years ago. Connector assemblies, especially high-volume cable harnesses, still depend on how well a crimp tool was set up that morning.
I only started pulling on crimped wires after ignoring a connector that looked perfect. I know that sounds obvious now, but in the moment, the datasheet curve pointed at the capacitor. That's why I now do a pull test on every suspect harness before I order replacement parts.
Looking back, I should have asked the customer to ship the failed device to us before we booked a courier. We did ask for it, actually, but it arrived after our 2,000 parts were already in the air. That's when I changed how we handle intermittent failures: the mechanical path gets checked first, and the passive components get checked second.
A note on search terms and part numbers
In our search logs, people find these parts with very different strings: 'kemet', 'kemet electronics', 'kemet dugue', 'blood pressure monitor symbols', 'multimeter', 'how to crimp connectors'. I get it. The search term often comes from whatever label was on a supplier's page or from a memory of a symbol on a screen. What matters is the part number and the mechanical path around it. If the part number matches an authorized distributor's trace, and the connector crimps are verified, you'll save yourself a much more expensive phone call.
Prices and stock levels are as of March 2025; the supply chain changes fast, so verify current availability before you bake a date into your schedule.
