7 Questions I Get Asked (and the Answers I Wish Someone Had Given Me)
I’ve been managing procurement for a mid-sized telecom equipment manufacturer for about six years now. Over that time, I’ve processed roughly $180,000 in passive component orders—dozens of invoices, three vendor switches, and more than a few “I wish I’d known that” moments. This article is my attempt to put those lessons into a single FAQ. If you’re buying Kemet parts for cordless phone RF stages or network gear, I hope you find something useful here before you spend a dime.
1. Are Kemet MLCCs worth the premium over generic alternatives?
Short answer: In my experience, yes—for critical paths. For decoupling in a power rail that has 2x headroom? Probably not.
In Q2 2024, I had to choose between a Kemet C0805C104K5RACTU (0.1 µF, 50V, X7R) and a no-name alternative for a cordless phone base station’s RF filter network. The Kemet was $0.14/unit; the generic was $0.03. I almost went with the cheap option—until I ran a TCO. The generic had a ±20% tolerance on capacitance (vs. ±10% for Kemet), and a review of our past 18 months of returns showed that batches from that generic supplier drifted up to 30% after 1,000 thermal cycles. After factoring in rework labor, the “savings” evaporated. Total cost per board: Kemet $0.14, generic $0.18 after expected rework.
So yes, for anything near a timing or RF circuit, I’ll pay the Kemet premium. For a non-critical bulk decap? I’ll look at alternatives.
2. What’s the deal with Kemet and Yageo? Are they the same company?
Kemet was acquired by Yageo in 2020. It’s now a wholly owned subsidiary, but they operate with distinct product lines and, crucially, distinct supply chains. In practice, I treat Kemet and Yageo as separate vendors for quoting purposes—because their pricing and lead times can differ by 20-30% even for similar specs.
That integration has been smoother than I expected, but I’ve noticed that some older Kemet part numbers (especially legacy tantalum types) are being quietly phased out in favor of Yageo equivalents. Check availability before designing in a five-year-old BOM.
3. Why do cordless phones still use through-hole capacitors? Isn’t SMD better?
You’d think so. But in the RF output stages of some DECT cordless phone base stations, the vibration and thermal cycling from the handset cradle can crack small SMD MLCCs. I’ve seen it—literally cracked caps in a field-return unit that caused intermittent signal dropout. Through-hole parts (like the Kemet T350 series) have more physical robustness for that specific mechanical stress.
So “better” depends on your definition. For a stationary base station board with no moving parts? SMD all day. For something that gets slammed into a cradle 50 times a day? Through-hole may be the more reliable (and cheaper in the long run) choice.
4. What does “7.1” mean in a capacitor spec, and should I care?
If you’re looking at a datasheet and seeing something like “7.1 mm x 5.0 mm” or “Package 7.1,” it’s almost certainly a case size designation—specifically, a metric dimension that translates to an EIA standard (like 2815). I’ve seen engineers get tripped up on this when moving between datasheets from Kemet and a Japanese manufacturer because they use different naming conventions for the same physical part.
Here’s the practical point: Always measure the footprint yourself, and never assume two parts with the same “7.1” label are identical. I once spent an entire afternoon reworking five prototypes because I assumed a 7.1mm wide cap from Vendor A was the same footprint as Vendor B’s 7.1mm. Turns out Vendor A’s “7.1” was the overall length, and Vendor B’s was the width. (Mental note: never assume.)
5. Kemet vs. Cypress for capacitors—isn’t that comparing apples to oranges?
You’re right to raise an eyebrow. Cypress is primarily known for microcontrollers and memory, not passives. But I’ve seen the comparison come up in legacy BOMs from larger OEMs where a Cypress PSoC was used on the same board as Kemet capacitors. I think the confusion comes from older product catalogs where the component categories were adjacent.
If you’re literally choosing between a Kemet capacitor and a Cypress capacitor? You’re not. There’s no Cypress capacitor line (to the best of my knowledge). The real choice is between Kemet and another passive manufacturer like Murata or TDK. But if you’re comparing a Kemet cap to a Cypress microcontroller? That’s a different procurement bucket entirely.
6. Should I use Kemet MLCCs for the RF matching network in a DECT cordless phone?
Probably, but with caveats. The high-Q characteristics of Kemet’s C0G/NP0 MLCCs make them suitable for the MHz-range filter matching in DECT (around 1.9 GHz). I’ve used them in two designs now, and the return loss was within spec straight off the board on the first pass—no tuning required.
That said, I’ve also used a different brand (Murata’s GJM series) for the same application with identical results. The difference? Kemet’s were about 12% cheaper in volume. But they also had a 10-day longer lead time in Q3 2024. If your schedule is tight, that 10 days can kill the entire project. My rule: spec Kemet on the BOM, but list a pre-approved alternative with equivalent specs—and check both lead times.
7. What’s the biggest mistake you see people make when buying Kemet parts?
After tracking about 60 orders over four years in my procurement database, I found that about 15% of our “budget overruns” came from not verifying the voltage derating curve. Kemet’s datasheets specify maximum voltage at room temperature, but most MLCCs lose capacitance significantly when you apply voltage near their rated limit—especially with X7R dielectrics. I’ve seen teams spec a 50V cap for a 48V rail and then wonder why the capacitance dropped 40% under load. The fix is usually to oversize by 2x (e.g., a 100V cap for a 48V rail). That adds maybe $0.02 per unit, but it saves you from having to recharacterize the entire board.
So glad I caught that on a prototype once rather than in production. Dodged a bullet—was one purchase order away from ordering 10,000 of the wrong part.
