I'm a quality compliance manager at a mid-sized communications equipment manufacturer. Every year I review over 200 unique component lots before they hit our production line. In Q1 2025, one batch of Kemet tantalum capacitors—1,500 pieces of the 2780 series (100µF, 10V, case size C)—nearly caused a major headache. Here's what happened.
The Scene: A Tuesday Morning in January
The delivery came in on a Tuesday. Our warehouse logged it, put it on the inspection shelf. Standard protocol: I pull random samples, check dimensions, capacitance, ESR, and visual appearance against our approved specification. The paperwork said the parts came from Kemet's authorized distributor, with the usual CAGE code 2780–no, wait, CAGE code for Kemet is 0U2R4? I'm mixing those up. Kemet's CAGE code is actually 0U2R4 for their main facility, but the 2780 is the capacitor series number. Let me rephrase: the parts were marked 2780 series, with a CAGE code of 0U2R4 printed on the reel label. That checked out.
The Discovery: Something Was Off
I set up the LCR meter – an old Keysight that I've been using for years – and started measuring capacitance. The first piece read 99.2µF. Good. Second: 98.6µF. Third: 97.8µF. All within ±20% tolerance, so fine. But the fourth sample read 94.1µF, and the fifth was 93.5µF. That's borderline. Our internal spec for the 2780 series calls for a tighter tolerance of ±10% on capacitance (i.e., 90–110µF), because we use these in a timing circuit where consistency matters. I flagged it.
Here's something vendors won't tell you: many distribution batches have a 'hidden buffer'—they'll ship parts from multiple production lots on the same reel. The label says one date code, but inside there could be parts from three different weeks. That's allowed within JEDEC standards, but it creates variance like what I was seeing.
The Pressure: A 2-Hour Deadline
Halfway through testing, my phone buzzed. The production manager: “We need those caps by end of shift. Can you clear them?” I had two hours before the line stopped. Normally I'd run 50 samples and plot a distribution, but there was no time. I had to decide with the data I had.
In hindsight, I should have pushed back. But with the project timeline breathing down my neck, I made a call: reject the batch. The risk of a field failure—maybe a 0.5% defect rate that could turn into a $18,000 rework—wasn't worth it. I documented the results and emailed the supplier.
The Turnaround: Efficiency Saved the Day
People think expensive components guarantee quality. Actually, components that come from manufacturers with robust process controls can command higher prices. The causation runs the other way. Kemet, for example, runs statistical process control on all their tantalum lines. My point is: the issue wasn't Kemet's quality; it was the distribution chain's handling. But what mattered next was how fast we could get replacement parts.
I logged into our procurement system. The preferred distributor had a real-time inventory API (something we'd implemented in 2023). Within 15 minutes, we located 2,000 pieces of the same 2780 series from a different lot, shipped from Kemet's warehouse in Mexico. The system quoted standard overnight delivery, but we paid the rush premium—about 40% extra, or $47 for the whole order. That's a fraction of the cost of a production stop.
Switching to that automated procurement workflow cut our turnaround from an average of 5 days to 2 days. The automated process eliminated the email chains and phone calls that used to eat up hours. For our 50,000-unit annual order, that efficiency gain translated to roughly $12,000 in avoided downtime per year—give or take a couple thousand depending on the product mix.
The Lesson: Process Over Panic
What did I learn? Three things:
- Trust your measurements, but verify your assumptions. I assumed the deviation came from mixed lots. It did. But I never confirmed with microscopic date code inspection. (That's on my to-do list for the next audit.)
- Time pressure decisions are like bets. You make the best call with the information you have. But the bet pays off when you have a fallback—in our case, the fast procurement system.
- Efficiency isn't just about speed; it's about consistency. The automated quote system gave us a repeatable process. That's worth more than any single cost saving.
Oh, and about the best multimeter for electricians—I get asked that a lot. My personal pick is a Fluke 87V, but for field work where you're testing capacitors on the fly, something like a Klein MM700 is fine. (At least, that's been my experience with our maintenance team.) But that's a whole other story.
The Takeaway
If you're a procurement engineer or a quality manager, build process efficiency into your verification workflow. It's not about being perfect—it's about being able to pivot when things go off-spec. Kemet parts are reliable, but even reliable parts can have a bad lot (or a good lot with a poor distributor handling). The key is knowing your tolerances (i.e., what your circuit actually needs) and having a Plan B that doesn't take three days to execute.
In the end, we got the replacement caps, production ran on time, and I updated our supplier scorecard. That batch went back to the distributor with a quality report. Last I checked, they credited us for the return and added a note: 'Do not mix lots.' Sometimes the little things matter most.
