The Setup: A Rush Job and a Tempting Shortcut
Last January, I was handling a power supply prototype order for a client who needed a 24V/5A industrial unit. The timeline was brutal—six weeks from design to first article. I'd been in component procurement for seven years at that point, and I'd made my share of mistakes. But this one? This one still makes me cringe.
We needed 2,200µF electrolytic capacitors for the output stage. Our usual supplier for TDK caps quoted a lead time of three weeks. Fine. But the client pushed for an accelerated schedule, and I had to decide: wait for TDK, or grab a cheaper alternative from a distributor that could ship overnight. The price difference? About $0.18 per cap—$360 total. I went with the cheap option. I figured, capacitors are capacitors, right?
The Turning Point: Everything Worked… Until It Didn't
Assembly went smooth. The board powered up. We shipped the prototype to the client. Then the emails started: Output ripple is 150mV peak-to-peak, our spec calls for 50mV max.
I grabbed my Fluke 117—the best multimeter for electronics troubleshooting I've owned—and started measuring. The cheap caps had ESR readings of 0.8Ω at 100kHz, when we needed under 0.1Ω. That ripple was cooking the rest of the circuit. I got a thermal camera on the board and saw one cap at 85°C. Another was bulging.
I still kick myself for not checking the datasheet. If I'd caught the ESR spec earlier, I'd have known that budget caps rated for 2,000 hours at 85°C were never going to survive a 5A ripple current.
The Fallout: Time, Money, and Credibility
The rework cost $1,200 in labor plus a one-week delay. I had to explain to the client why the prototype failed. Not my finest moment. I remember saying, “We saved $360 on components and lost $1,200 on rework plus a week of trust.” The client was understanding, but I saw the doubt in their eyes.
That's when I switched back to TDK capacitors. I placed a rush order for their B41858 series—same footprint, same capacitance, but rated at 10,000 hours and 105°C. The TDK filters (ferrite beads on the input line) also went in as part of the redesign. The ripple dropped to 28mV. Problem solved.
The Lesson: Quality Is Brand Perception
Per the FTC guidelines on advertising substantiation (ftc.gov), claims about product performance must be backed by evidence. In our case, we claimed the supply would meet 50mV ripple. We couldn't prove it with cheap caps. That's not just an engineering failure—it's a marketing liability.
What I learned is that the perceived quality of your output is your brand. When a client sees a bulging capacitor, they don't think “bad component choice.” They think “this supplier cuts corners.” The $0.18 per cap I saved cost us disproportionately more in reputation.
Now I keep a pre-order checklist we developed after that incident. We verify ESR, ripple current ratings, and operating temperature for every critical component. We flagged 47 potential errors in the past 18 months using that list. Not all were as dramatic, but every one avoided a client-facing problem.
If you're designing a power supply—especially for industrial or medical clients—don't compromise on passive components. The difference between a $0.45 TDK cap and a $0.27 generic might seem small, but the total cost of ownership includes your time, your rework budget, and your client's trust.
One last thing: when I tested the revised prototype, I used a Keysight U1282A multimeter to confirm the ripple. (I'd borrowed it from a colleague—my Fluke was in for calibration.) The reading was rock-solid. That's the feeling you want: confidence, not fingers crossed.
Expensive lesson. But I've never made the same mistake twice.
