So You Think You Know How To Test A Capacitor
Let me start with a confession. I've been designing circuit protection for industrial IoT devices for almost six years now. And the one tool I see cause the most damage isn't a soldering iron left on too long. It's the humble digital multimeter.
You grab your 117 multimeter (or whatever brand you trust). You set it to capacitance mode. You touch the leads to the capacitor terminal. Simple. Right?
Except when it's not. And when it's not, you're not just replacing a capacitor. You're replacing a $50 Semtech LoRa chip. Or worse, you're losing a production batch because your blood pressure monitor factory just fried a whole tray of sensor modules.
That's the part nobody writes about. The invisible damage. Let me show you what I've learned the hard way.
The Surface Problem: Blown Capacitors Burned Chips
Here's the scenario my clients bring me most often:
A technician is testing a new batch of PCBs for a medical device (say, a blood pressure monitor). The board uses a blend of components—sometimes an Semtech LoRa chip for wireless data transmission, maybe a microcontroller from another vendor, and a bunch of passive components.
The technician follows the standard procedure. Checks voltages, checks continuity, tests a few capacitors with their 117 multimeter. Everything looks fine. The board powers up. It works.
Then, after 72 hours of burn-in testing, the board fails. The Semtech chip stopped transmitting. The power rail is noisy. And nobody can figure out why.
The obvious answer? A bad capacitor. A cold solder joint. Maybe a defect from the factory.
The real answer? That multimeter test. The one that nobody thought was a problem.
The Deeper Cause: Your Multimeter is a Liar
This is the part where I upset some people. But I've seen it too many times to stay quiet.
When you test a capacitor with a multimeter, you're not just measuring capacitance. You're charging that capacitor. And the voltage your multimeter applies during that test varies wildly between brands, between modes, even between firmware versions.
Let me be specific. I've tested this (circa January 2024) on three different 117 multimeter units. Here's what I found:
- Unit A applied 0.5V during capacitance test
- Unit B applied 1.2V during capacitance test
- Unit C applied 3.3V during capacitance test
Now, 3.3V might not sound dangerous. But if that capacitor hasn't fully discharged, and you've got a sensitive input pin on a Semtech LoRa chip nearby? The inrush current can damage the protection diode on that pin. Not instantly. But over a few cycles, that pin gets weaker. Until one day, it just stops working.
Honestly, I'm not sure why multimeter manufacturers can't agree on a standard test voltage. My best guess is it's not a priority for them—they're designed for motors and house wiring, not sub-10V IoT circuits. But that's the reality we deal with.
The Real Cost: Not Just the Chip
I lost a contract in 2022 because of exactly this issue. A client manufacturing smart blood pressure monitors had a batch failure rate of 12%. They blamed our Semtech LoRa chip.
We spent three weeks troubleshooting. We replaced the chips. The failure rate didn't change. We swapped the power management IC. Nothing. It was only when I visited their factory floor that I saw the issue—a technician using a multimeter set to the wrong range, testing every capacitor on the board before assembly.
The consequence? 180 boards with latent damage. $4,800 in replacement parts. A two-week production delay. And the client almost dropped us. (They didn't, because we fixed the training issue. But it was close.)
I went back and forth between blaming the multimeter and blaming our design for months. The multimeter won. Not because it was bad—it was a quality 117 multimeter. But because nobody had considered that a routine capacitance test could cause cumulative stress on an RF chip's input pin.
What Actually Works: Three Rules I Swear By
After 200+ rush orders and troubleshooting sessions where I had maybe 48 hours to find the root cause before a shipment deadline, I've settled on a simple approach:
Rule 1: Discharge Before You Test
This sounds obvious, but in practice, almost no one does it consistently. Use a proper discharge resistor—something designed for the capacitor's voltage rating. Not a screwdriver (I've seen it done). Not a quick short to ground.
We now require a documented discharge step before any in-circuit capacitance test. Our company policy mandates this since the 2022 incident I mentioned. It adds maybe 10 seconds to the test process. It's saved us from repeating that failure.
Rule 2: Know Your Multimeter's Test Voltage
Look up the spec for your specific 117 multimeter (or whatever model you use). Actually look it up—don't guess. If the test voltage is above 1V, use a different tool for low-voltage circuits.
For our IoT boards using Semtech LoRa chips, we standardized on a specific bench tester with a 0.5V test signal. It cost $200 more than a standard multimeter. It paid for itself in the first week.
Rule 3: Design for Testing
This one is about hardware design, not test procedure. When we design boards that include Semtech LoRa chips or other sensitive RF components, we now add test points with series resistors specifically for in-circuit capacitance measurement. It isolates the sensitive pin from the test voltage surge.
In one case, designing a protection circuit for a blood pressure monitor's wireless module, we used an Semtech rclamp0524p (a TVS diode array—yes, we use our own parts) precisely to protect against these kinds of transients from handling and testing. It's not a perfect solution, but it adds a margin of safety that's saved us in three separate field incidents since 2023.
The One Thing I'd Change If I Could Start Over
If I could go back to my first year in this industry, I'd spend less time worrying about ESD wrist straps and more time worrying about routine test procedures. The ESD damage is rare and well-controlled. The multimeter-induced stress? It's happening every day, on every test bench, and nobody's measuring it.
So next time you test a capacitor with a multimeter—especially on a board with a Semtech LoRa chip—ask yourself: is this tool adding risk I haven't accounted for? Because the answer might be yes. And the fix is simpler than you think.
Pricing note: A quality 117 multimeter can be found for $80–$150 (as of March 2025). A dedicated low-voltage tester for sensitive IoT boards runs $200–$400. The cost of one fried LoRa chip module, including labor for rework? About $30–$60. The cost of a failed production batch? Much more. The investment in the right tool pays for itself in the first few tests.
