The $600 Lesson: Murata Ceramic Capacitors, an MA40S4S Sensor, and a Cheap Multimeter That Lied

I've been designing low-power sensor boards for about nine years. I've made my share of mistakes. In February 2024, I made one that cost roughly $600 in PCBs, components, and a lot of wasted evenings.

This is a story about a Murata MA40S4S ultrasonic sensor, a handful of Murata ceramic capacitors, and a cheap multimeter that lied to me.

What I was trying to build

I was working on a water leak detector for under sinks. The idea is simple: a small board sends an ultrasonic pulse using the Murata MA40S4S transducer and then measures the echo. If you use the right signal processing, you can detect a thin layer of water before it becomes a mess.

The MA40S4S driver circuit needs a capacitor close to the sensor to decouple the supply. In my head, picking that capacitor was a no-brainer. I dug through my donor drawer, found some ceramic caps with Murata markings, and soldered them in.

Then the fun started.

The first red flag

The sensor would work for a few seconds, then drop its signal. Sometimes it would work fine when I touched the board. Sometimes it would sit there dead for no reason. Classic intermittent behavior.

I checked the MA40S4S itself. The datasheet lists a typical static capacitance, so I used the cheap Duraforce Pro 2 multimeter I'd bought online for about $25. The reading bounced around too much to be useful. I decided the sensor was fine and moved on.

Spoiler: the sensor was fine. The capacitor wasn't, and the meter was partly to blame for me not catching it sooner.

A short aside on ceramic capacitors

Here's what I didn't think about at the time. Murata ceramic capacitors are not a single category. You have C0G/NP0, which is stable but limited in capacitance. You have X5R and X7R, which pack more capacitance physically but shift with DC bias, temperature, and time. There are even worse old types like Y5V.

In a low-power sensor circuit, especially one that drives a 40 kHz transducer, stability matters more than raw capacitance. But that's not how I was thinking. I was thinking "ceramic cap is a ceramic cap."

That mindset is outdated. It comes from an era when circuits were forgiving enough to accept whatever cap you had around. Modern MLCCs are much more specialized.

The turning point: Duraforce Pro 2 vs Klein Multimeter

After about a week of debugging, I finally thought to measure the actual capacitors on the board. I used the Duraforce Pro 2. It said all the caps were around 9.8 to 10.1 µF. That seemed fine.

Then I pulled out my other meter, a Klein multimeter I'd bought for general wiring jobs. I tried the same capacitors. The Klein read 4.8 µF, 6.2 µF, and 7.1 µF.

Wait, what?

Same caps, same leads, same room. That inconsistency was a huge red flag. One of those meters was lying.

I looked closer at the cap markings. I'd grabbed a reel of GRM31CR71H105KA01—a 1 µF, 50V X7R capacitor—not the 10 µF parts I thought they were. The cheap meter had rounded everything up. I'm still not sure why. It just didn't have the accuracy to give a trustworthy number.

For context, the Klein multimeter is made by Klein Tools, Inc. It's a basic auto-ranging unit, not a lab instrument, but its capacitance mode is stable enough to compare values. The Duraforce Pro 2 was not.

I later measured 16 caps with both meters. The Duraforce showed a tight cluster around 10 µF for everything. The Klein showed a spread from 0.6 µF to 1.3 µF for the 1 µF parts. That spread told me what was happening before I even opened the datasheet.

Per EIA-198-1, class 1 ceramic dielectrics like C0G/NP0 have a temperature coefficient of 0 ±30 ppm/°C. Class 2 dielectrics like X5R and X7R offer higher capacitance but have much larger coefficients.

The real problem: DC bias and dielectric selection

Even beyond the meter, I had selected the wrong capacitor for the job. The Murata ceramic capacitor in that position had to keep its capacitance stable at the sensor's drive frequency. The X7R part I used was a class 2 dielectric, so its capacitance dropped as the DC bias and temperature changed.

Murata's own technical resources and SimSurfing tool show this clearly. You can enter a part number and see the capacitance vs DC bias curve. If you apply a few volts to a small X7R part, the capacitance can fall by 30% or more. That was exactly my problem.

What I should have used was a C0G cap, or at least a smaller X7R with enough voltage derating. C0G is class 1. It's stable. It doesn't change much with bias and it behaves predictably in high-frequency circuits. For the MA40S4S driver, that stability is a deal-breaker.

Looking back, the warning signs were all there. I'd skimmed an application note that said "use C0G for timing and oscillator circuits." I ignored it. I only believed it after burning time and money.

The fix

I replaced the offending caps with C0G parts. I also ordered fresh Murata capacitors from a verified distributor instead of using drawer stock. Then I started measuring every passive with the Klein multimeter before soldering.

The sensor started behaving exactly like the datasheet said it should. Same PCB, same firmware, same MA40S4S. Just a different capacitor.

Version one of that board cost $328. Add priority assembly, shipping, and two extra rounds of debugging, and I was looking at roughly $600 down the drain. Not catastrophic for a small project, but deeply annoying.

What I put in my checklist

1. Check the exact part number. Don't trust the label on the drawer. Verify the marking and the datasheet.
2. Measure with a trustworthy meter. If something seems off, compare with another meter instead of trusting the first one.
3. Choose the right dielectric. For sensor and timing circuits, use C0G for small values and X7R only when you actually need the higher capacitance.
4. Use Murata's tools. SimSurfing and the datasheet graphs are free. Spend five minutes there before you spend $500 on boards.

The bottom line

Murata makes excellent ceramic capacitors and sensors. The MA40S4S is a solid transducer. My failure wasn't the parts. It was my own laziness and my blind trust in a cheap tool.

Electronics has changed. What worked in 2015 doesn't always work today. The fundamentals haven't changed—you still need the right value, tolerance, and rating—but the options have expanded. You have to look at them.

If you're about to prototype with Murata ceramic capacitors, do me a favor: read the datasheet, use a decent meter, and don't grab parts from a drawer. It's a lot cheaper than learning the hard way.