The short version
If you're sourcing KEMET capacitors for a blood pressure monitor, don't approve anything until the full part number is on the table. The brand name alone tells you almost nothing. KEMET makes thousands of SKUs. A "KEMET 10 µF capacitor" could mean a ceramic X5R rated at 6.3 V or a tantalum rated at 50 V. The difference changes the BOM cost, the test results, and in the worst case, the patient's reading.
Here's the thing: I've spent six years managing component spend for a 35-person medical electronics company. I've tracked every invoice in our procurement system, and I've made mistakes that cost more than they should have. The most expensive lesson? Don't trust a short distributor description. Verify the full KEMET part number.
Why I stopped ordering by brand name
In Q2 2024, we were building a pre-production run of blood pressure monitors. Our design engineer wrote "KEMET T491" in the BOM, but not the full suffix. I sourced a part from a distributor that listed the same series and a close capacitance value. The price was good. I didn't verify the tolerance or ESR because I thought, "KEMET makes good caps, it'll be fine."
It wasn't fine. The alternate part had higher ESR. On the test bench, the blood pressure monitor showed intermittent low-battery warnings even with a new battery. The team spent two days troubleshooting. Someone opened a Cypress vs STM32 MCU debate, blaming the firmware. The problem was a capacitor that didn't meet the circuit's ESR requirement.
We paid $140 for the wrong parts, then KEMET charged us—well, an authorized distributor charged us—for expedited samples. The total rework cost was maybe $4,300, I'd have to check the exact labor hours. Not massive, but stupid. A 5-minute check of the full part number would have prevented it.
Yageo KEMET: same brand, different supply chain
Another thing I didn't initially understand: KEMET and Yageo KEMET aren't two competing brands. Yageo acquired KEMET and now owns the product lines. But that doesn't mean everything stayed the same. Some KEMET parts have been consolidated into the Yageo KEMET catalog. Others are no longer preferred. A few have moved to a different packing standard or lead time.
What this means for procurement: if you have an old KEMET part number in your system, check whether Yageo KEMET still lists it as active. Also check the last PCN. I found one our team had missed because we assumed "KEMET" was the manufacturer and the acquisition didn't affect us. It did. Lead time went from 8 weeks to 18 weeks.
In practice, for a blood pressure monitor, the safe move is to run a lifecycle check on every KEMET part once a year. Not because Yageo is doing anything wrong. Because the transition is still changing the catalog. I'd rather catch an EOL notice early than explain a last-minute redesign to a client.
The charged capacitor problem nobody talks about
When I say "KEMET charged," I usually mean the supplier added a fee for something we forgot, like a certificate of conformance or a rush order. In engineering, "charged" means a capacitor has stored energy. Both meanings have cost me money.
I'll use the technical meaning first. A capacitor can leave a test fixture charged, especially if the circuit has an output that doesn't discharge through a resistor. If your assembly line installs a capacitor that's still charged, you can get a transient that resets the MCU. A big enough charge can also damage the capacitor itself. In a blood pressure monitor, this can show up as a one-off startup failure that's almost impossible to reproduce.
We didn't have a formal discharge step in our bench-test process. After the third failure, I finally added "discharge caps before handling" to the checklist. That checklist has saved us more than the cost of a few cheap instruments. Prevention is cheaper than rework—that's the whole point.
The 7.1 million hours trap
One KEMET tantalum capacitor datasheet I reviewed lists an MTBF of 7.1 million hours at 40 °C. That sounds impressive. It's also misunderstood. MTBF is an estimate based on assumptions: recommended derating, correct soldering profile, proper storage, no ESD damage. Miss one condition and the reliability number is irrelevant.
I once saw a batch of KEMET capacitors fail visual inspection because the assembly house baked them too long before reflow. The datasheet still said 7.1 million hours. The parts on the board didn't care. The failure didn't happen in a blood pressure monitor, but it could have. The lesson is the same: reliability is a system property, not a brand property.
So don't quote MTBF to your supplier as if it's a guarantee. Instead, ask for test data from your actual assembly house. A simple solder-joint cross-section can tell you more than an MTBF calculation.
Checklist before you approve KEMET parts
Here's what I actually check now. It's not a long list, and it's saved me from the expensive kind of surprises:
- Full KEMET part number from the latest datasheet. Not the series name. Every suffix matters.
- Lifecycle status in Yageo KEMET. Active, new design, or NRND?
- Authorized distributor. If the quote is too good, check the source.
- ESR, leakage, and voltage derating vs. the circuit's real use. The BP monitor runs at 3.3 V, but peak loads matter.
- Discharge and handling step on the assembly/test floor.
- Certificate of conformance when the supplier says it's included. I've seen that phrase mean different things to different people.
The last point is another real story. I said "we need a certificate of conformance." The vendor heard "we need a shipment report." We discovered the mismatch when our quality engineer asked for it after receiving the parts. The vendor's certificate turned out to be a packing list. Same words, different meaning.
Oh, and one more: verify the date code. Old stock with a good part number can still have moisture sensitivity.
When you can skip this
This checklist isn't for every situation. If you're buying a few KEMET capacitors for a breadboard prototype, go ahead and order from an electronics retailer. If your blood pressure monitor is a consumer wellness device with no regulatory claim, you can probably reduce the documentation burden. If it is a regulated medical device, standards like IEC 60601-1 will require evidence that the component meets the intended safety requirements. Brand loyalty won't be enough.
The point isn't to make procurement slower. It's to catch the problems that only appear after you've committed. A little verification at the beginning is much cheaper than a board spin later.
That's the part I keep having to re-learn: 5 minutes of checking beats 5 days of rework. I only fully believed it after ignoring it and paying the price. Now it's my first rule.
