Telecom Engineering

What Is Kemet? A $12,000 Sourcing Lesson About Ceramic Capacitors and Relays

2026-11-13 · Vishay Telecom Engineering
Telecom article technical bench

The Friday afternoon email that changed how I source components

It was 4:47 PM on a Friday in March 2023 when the purchasing director forwarded me the BOM for our next industrial control board run. Two thousand units. Tight timeline. And a line item that made my stomach drop: "Kemet components — 2660 flip capacitors, 2780 series relays — 4,800 pieces total."

I'd been a procurement specialist handling electronics orders for seven years at that point. I knew the drill. I'd been through two recessions, three ERP migrations, and one memorable incident involving a wrong reel of resistors that shut down a line for two days. But I'd also just sat through three consecutive budget meetings where "cost reduction" was the only phrase anyone remembered. The Kemet line alone was going to eat a noticeable chunk of the component budget.

So I did what a lot of procurement people do in that situation. I went looking for alternatives.

The $3,000 "savings" that looked too good to question

I found a distributor offering what looked like equivalent ceramic capacitors and relays from an offshore manufacturer. Same specs on paper — capacitance, voltage, tolerance, temperature range. The quote came in about $3,000 under the Kemet line for the full 4,800-piece order. Roughly 18% less, if memory serves.

I put together a comparison sheet, showed it to my manager, and got the green light. The logic was simple: same specifications, lower price, approved vendor list. What could go wrong?

I still kick myself for not asking a single engineer what they thought about the swap. I checked the datasheets myself. I compared the numbers. I even called the distributor to confirm lead times and availability. But I never asked the people who actually designed the board whether the parts were interchangeable in practice. That's a rookie mistake I should have grown out of by year three.

At the time, it felt like a clean win.

What "same specs" actually meant on the assembly floor

The parts arrived on schedule. The first production batch went out the door. Then the returns started coming in.

Not immediately. That's the part that threw me. If the boards had failed on the bench, we'd have caught it before shipping. Instead, the failures showed up in the field over about six weeks. Control boards were dropping offline in thermal cycling tests at a rate that our QA team flagged as statistically significant. A few customers reported intermittent issues after installs in high-humidity environments.

Our engineering lead pulled the failed units and dug in. His conclusion: the cheaper components likely drifted out of spec faster than expected under temperature stress. The 2780 relay replacements had contact resistance issues after cycling. The 2660 flip capacitor substitutes were showing capacitance shifts that the originals handled comfortably.

None of this showed up in the datasheet comparison. It showed up after 200 hours of operation. The quality of the replacement parts was acceptable in a controlled environment. Not great, not terrible. Serviceable — until it wasn't.

Going back to Kemet — and finally understanding the brand

That's when I got the education I should have sought six weeks earlier. I sat down with our senior engineer, and we went through the original Kemet datasheet line by line.

Kemet Corp, I learned, has been making passive components since 1919. The company built its reputation on tantalum capacitors, then expanded into ceramic capacitors, film capacitors, relays, connectors, and electromagnetic devices. Kemet components turn up in automotive, industrial, aerospace, medical, and telecom applications. The brand is now part of Yageo, but the product lines and spec sheets still carry the Kemet name.

The 2660 flip capacitors were selected for their thermal performance under the board's operating conditions — the "flip" in the part name has to do with the mounting configuration, though I'd be lying if I said I fully understood the technical details. That's engineering's job, not mine. The 2780 relays were chosen for contact reliability at the switching frequencies we were running. Our engineer explained it this way: "On paper, a lot of ceramic capacitors look the same. In practice, they're not."

I'm not a materials scientist, so I can't speak to the exact differences in dielectric formulations or electrode materials. That's outside my lane. What I can tell you from a procurement perspective is that the replacement parts met the written spec and still failed in the application. The Kemet parts had margin in areas the written spec didn't capture.

The real cost of chasing the lowest quote

Let me put actual numbers on this, because "hidden costs" sounds abstract until you're the one explaining them to your boss.

The failed production run involved 2,000 units. We had roughly 340 field returns and replacements. Engineering logged about 480 hours investigating, documenting, and re-specifying. A rework cycle on 1,200 finished boards that had to be re-tested and re-packed. Expedited shipping on replacement Kemet components to meet customer commitments.

All told, we spent somewhere in the range of $12,000 to $15,000 dealing with an issue that started as a $3,000 "savings." Take that with a grain of salt — the exact figure depends on how you allocate engineering time — but the order of magnitude is right. The boss's boss didn't love that math. Neither did I.

To be fair, I get why procurement teams chase lower prices. Budgets are real, and component costs add up fast. I've been on the other side of that conversation, explaining to a project manager why the BOM came in over target. It's not fun.

But the TCO calculation has to include more than the invoice price. It has to include the probability and cost of failure at your specific application conditions. For our general-purpose, non-critical boards, the cheaper alternative might have been fine. I'd probably still use it there. For this application — industrial control, temperature cycling, customer-facing reliability expectations — it wasn't.

What I do differently now

After that mess, I built a pre-sourcing checklist that I still use. It's not sophisticated. Three things, in this order:

First, check whether the part is application-critical or general-purpose. Second, get engineering sign-off on any substitution, even if — actually, especially if — the specs match on paper. Third, calculate the cost of a worst-case failure, not just the delta between quotes.

I've run this checklist on 47 substitutions over the past 18 months. We've caught six potential mismatches before they became production problems. That's a decent hit rate. Not perfect — I'm not 100% sure we'd catch everything — but it's better than the "compare datasheets and hope" method I was using before.

And when someone on my team asks "What is Kemet?" — or more likely, "Why do we pay more for Kemet components when this other one looks identical on the datasheet?" — I tell them the short version of this story. Usually while pointing at the $12,000 line in last year's variance report.

The lowest quote and the lowest total cost are rarely the same number. That lesson cost me more than I'd like to admit. But it's one I won't forget.

Protocol context: 3GPP TS 38.xxx, IEEE 802.3bt, ITU-T G.652.D, insertion loss dB, and PIM dBc assumptions should be validated against each carrier design pack.
V
Vishay Telecom Engineering

RF, optical, power, and reliability engineers reviewing component behavior for carrier infrastructure.