Here's the bottom line: when a home blood pressure monitor gives you a number you don't trust — or shows an Inf where a reading should be — the algorithm is usually the last thing to blame. The analog front end fails first. A decoupling cap that can't hold a rail under pump-motor transients, and a connector whose contact resistance has crept up since it left the factory, will wreck a measurement long before the firmware gets a chance to smooth it.
I've spent eight years on the sourcing side of portable medical devices — capacitors, interconnect, the unglamorous stuff. In that stretch I've made three mistakes big enough to get written up internally. Total damage: roughly $4,700 in scrapped boards and rework, plus one very uncomfortable call with a customer. The most expensive one wasn't a capacitor at all. It was a connector.
So here's what I'd tell anyone designing, sourcing, or just trying to get a reliable reading out of a blood pressure monitor: spend your debugging time on the power rail and the interconnect, not the measurement math. Below is why I believe that, what I got wrong along the way, and the places where my opinion shouldn't count for much.
Where this confidence comes from (and the two times I was wrong)
In 2019 I approved a substitution that looked like a no-brainer: swap a manganese dioxide tantalum cap for a polymer tantalum on the 3.3 V rail. Same case size, same nominal capacitance, lower ESR, lower price. Didn't touch the layout. Didn't re-run the loop stability sim, because in my head "same value is same value."
Two weeks later the regulator was ringing. Not enough to fail a quick bench check, but enough to show up as 2–3 mmHg of jitter on the pressure channel. The lower ESR had shifted the LDO's phase margin. That one cost about $600 in re-spins and roughly a month of schedule.
Lesson learned the hard way: lower ESR is not automatically better. It depends what's driving the rail. I'm not an analog designer — take my explanation with a grain of salt — but the behavior was repeatable enough that I stopped treating ESR as a one-way scorecard.
The second mistake hurt more. September 2021, I signed off on a battery-contact terminal with thinner gold plating to save something like 18 cents a unit. Ran the usual bench tests. Everything passed. Then we did a 500-hour damp-heat run at 40 °C / 90 % RH and watched contact resistance climb from about 12 mΩ to 89 mΩ on roughly one board in thirty. Pressure readings drifted past ±8 mmHg on those units. We scrapped 400 boards — call it $2,400, all in.
After that I built a pre-check list our team still uses. We've caught 31 issues with it since Q1 2024, mostly derating and plating questions that would otherwise have surfaced late or not at all. There's something genuinely satisfying about catching a $2,000 mistake on a checklist before anything ships.
What the KEMET T530 is actually doing in a blood pressure monitor
Most people searching for the KEMET T530 are looking at it for one of two reasons: the pump motor, or the valve. Both draw ugly transient currents, and both sit close enough to the pressure sensor's analog supply that a sagging rail turns into a wandering reading.
The T530 is a polymer tantalum — part of KEMET's KO-CAP line. The pitch is straightforward: high capacitance in a small case, low ESR, and no manganese dioxide to ignite if the dielectric breaks down. That last point matters more than the headline numbers when the device is strapped to somebody's arm thirty seconds at a time.
Where I'd push back on the marketing: the T530 is not a drop-in upgrade for every rail. Three things I'd check before specifying it:
- Derating, not the nameplate. The rated voltage on the label is not the voltage you should run it at. Read the derating curves for your actual ambient and ripple. Don't hold me to a specific percentage — it varies by series and case size — but the curve is not optional reading.
- It doesn't replace ceramic for high-frequency decoupling. Polymer tantalum handles bulk and mid-frequency well; the MLCC still does the fast work. In most blood pressure designs you'll see a mix, not a swap.
- Humidity. If the board isn't conformally coated and the device lives in a bathroom, that's a red flag no matter which chemistry you pick.
Worth saying out loud: KEMET has been part of Yageo since 2020, and the combined KEMET Yageo portfolio now overlaps in places it didn't used to. If you're comparing parts across the two brands, compare derating curves and qualification data — not the prefix on the part number.
The connector nobody checks
If you've ever chased down a measurement problem for two days only to find a loose terminal, you know that sinking feeling. If I had to bet on what actually kills a blood pressure monitor in the field, I'd bet on interconnect before I'd bet on any capacitor. There are at least three connector interfaces in a typical unit, and they fail in completely different ways.
1. Battery contacts
Spring terminals are the classic case. They look fine on a bench, then micro-motion plus humidity builds an oxide film, contact resistance climbs, and the pump browns out mid-inflation. You get a short measurement, an error code, or a low-battery warning on fresh cells. Nobody suspects the spring.
2. The cuff tube connector
This one is pneumatic, not electrical, and it gets ignored because it's "just a tube fitting." A slow leak at the luer-style connector means the pump runs longer and the deflation profile flattens. Firmware compensates for some of it — which is exactly what makes it hard to catch. The readings look plausible but trend high.
3. The sensor FPC
If the pressure sensor sits on a flex tail, the FPC-to-board connector is your drift source. Insertion cycles, solder wicking into contacts, a ZIF latch somebody opened too far during rework. I've watched contact resistance double after 200 mating cycles on a part rated for 30.
None of these present as a capacitor problem. All of them look like a "measurement accuracy" problem to the person holding the device.
That "Infinity" on the display, explained
If your monitor shows Inf, Hi, or pins the display at maximum, it usually means one of three things:
- Pressure exceeded the sensor range. Cuff over-inflation, a kinked tube, or a valve that didn't open. The firmware reports the overflow instead of inventing a number.
- The ADC saturated. The analog supply sagged or the reference drifted, and the converter hit full scale before the pressure did.
- An open in the sensor path. A connector that's fully open reads as effectively infinite resistance, and a naive conversion can render that as an overflow value.
The first is a user or mechanical issue. The second is power integrity. The third is interconnect. Only the first one gets fixed by re-reading the manual — and even then, only sometimes.
How to use a blood pressure monitor correctly (the 60-second version)
Half the people searching this are end users rather than engineers, so here's the short list. It's standard, and it's standard because each item measurably moves the reading:
- Sit still for 5 minutes first. Not "sit down and immediately press start."
- Back supported, feet flat, legs uncrossed. Crossing your legs is good for a couple of mmHg.
- Cuff on bare skin, one finger-width above the elbow crease, at heart level.
- Don't talk during the measurement. Talking adds a meaningful offset in most published comparisons.
- No caffeine, nicotine, or exercise for 30 minutes beforehand.
- Take two readings a minute apart and average them. One reading is a data point, not a diagnosis.
If you follow all of that and still get Inf every single time, stop adjusting your posture. Something in the hardware is out of range, and no amount of sitting still will fix a saturated ADC.
Where I stop being useful
KEMET doesn't build blood pressure monitors, doesn't write measurement algorithms, and can't tell you whether your device will clear ISO 81060-2 clinical validation. If you're heading toward a 510(k) or a CE mark under MDR, you want an experienced device manufacturer and a regulatory consultant — not a component supplier with an opinion.
I'm also the wrong person to ask about long-term drift in MEMS versus piezoresistive pressure sensors. I've heard decent arguments on both sides and I don't have the field data to pick a winner. If you've actually run that comparison, I'd like to hear it.
What I can offer is narrower and, I think, more useful: check the rail, check the derating curve, and check the connector before you blame the math. The capacitor gets the datasheet. The connector gets the failure.