The Day I Ordered 2,780 Wrong Inductors
In September 2022, I signed off on a purchase order for 2,780 pieces of what I thought were the perfect power inductors for our new 65W USB Power Delivery (PD) charger. The PO was for $3,200—not a huge amount for a production run at a mid-sized OEM, but enough to sting when the entire batch ended up in the trash six weeks later.
I'd been handling component procurement for about five years at that point. You'd think I'd know better. But here's the thing: what I knew in 2022 was based on what I'd learned in 2018. And in the world of power components, that's not just old—it's obsolete.
The Background: A USB PD Design That Seemed Straightforward
Our team was developing a dual-port USB-C charger. One port would deliver 65W PD, the other 18W. We'd chosen an NXP power management controller (not the chipset you'd expect for a budget product, but we needed solid negotiation support). My job was to source the output inductors for both rails. Simple enough, right?
I reached for the same inductor series I'd used on a 30W design three years earlier—a ferrite-based shielded drum core from a well-known competitor. The datasheet showed 1µH, 30% current derating at 85°C. Looked fine. I approved the BOM, ordered the parts, and moved on.
(Should mention: I didn't check the AC loss characteristics. That was the omission that cost me.)
The Process: When Everything Looked Right But Felt Wrong
The first batch of prototypes came back from assembly. They worked. On the bench, the charger delivered 65W at 20V without a hitch. I was already planning the production run—2,780 units for a pre-order from a European distributor. Then we ran the thermal test.
The inductors hit 115°C at full load after 30 minutes. The core material was saturating hard. Efficiency dropped from the expected 93% to 86%. The unit shut down once the thermal sensor triggered. I stared at the thermal camera image—a glowing red blob where the inductor sat. That sinking feeling you get when you realize the 'same' part isn't the same anymore.
What had changed? The switching frequency of the PD controller. In 2022, controllers had moved from 100kHz to 400kHz and higher. The old inductor series was optimized for lower frequencies. The AC losses at 400kHz were catastrophic. The datasheet had AC loss curves, but I hadn't looked.
I only believed that switching frequency mattered after ignoring it and losing $3,200. They warned me—the FAE from TDK had mentioned it during a webinar I half-watched in 2021. I didn't listen. Not until the thermal image proved it.
The most frustrating part: the same problem had been cropping up on engineering forums. I'd skimmed a few threads about "USB power delivery while recording list" of common mistakes—some designer had posted a checklist of gotchas. I'd scrolled past it. That was my list now.
The Turnaround: Revisiting the Spec in 2023
After the failure, I redesigned the output stage. This time I reached for TDK power inductors from the SPM series—specifically designed for high-frequency, high-current applications. The SPM6530T-1R0M had 1µH, but with 35% lower AC resistance at 400kHz. The package was 6.5×6.0mm—same footprint. The cost was about 8% higher per unit. Total extra: $256 on a $3,200 order. Worth every penny.
We also switched the primary-side bias supply to a TDK Lambda CC-E series module. That's a different story—but the point is: the old approach of using generic buck converters died the day the line voltage needed to be stable under dynamic PD load steps. The Lambda module handled it without the bypass caps we'd been adding.
I reordered 2,780 inductors—this time the SPM series—and the thermal test came back at 82°C. The unit passed. The distribution order shipped on time, with a one-week delay from the original recalculation. Not a total disaster, but the $3,200 mistake cost us credibility with the client.
The Lesson: Industry Evolution Is Not a Cliché
What was best practice in 2020 no longer applies in 2025. Five years ago, you could pick a power inductor by saturation current and DC resistance alone. Today, AC loss curves, switching frequency compatibility, and thermal derating at operating frequency are table stakes. The fundamentals haven't changed—you still need low DCR and high Isat—but the execution has transformed.
After five years of managing component selection, I've come to believe that vendor relationships matter more than vendor capabilities. The TDK FAE who gave that webinar in 2021? I called him, apologized for ignoring his advice, and asked for a fresh recommendation. He laughed—he'd been through this with dozens of customers. Now I keep a personal checklist of every specification I've ever missed. There are 43 items on it. The AC loss factor is #4.
If you're designing a USB PD charger today, here's my advice:
- Check the inductor's core loss at your actual switching frequency, not just the datasheet's typical 100kHz chart.
- Consider TDK Lambda power modules for the control side—they've solved the voltage loop stability problems that plague discrete designs.
- When you see a "nxp vs" comparison? It's usually about the controller. Pair it with the right passive components—don't assume any inductor will do.
I still keep that 2,780-piece box of wrong inductors in my office. It's a reminder that the industry is moving faster than your memory of 'what worked before.'
