I'm a sourcing engineer at an electronics distributor, and I've handled 200+ rush orders in the past 18 months—including same-day turnarounds for manufacturers with a down production line.
The component is rarely the real problem. The process around it is.
That's not a warm marketing line. It's the only conclusion that explains the pattern I keep seeing: the same types of emergencies, at the same types of companies, for the same three reasons.
Why I'm Saying This
When I'm triaging a call about a failed TDK transformer, my first question isn't the part number. It's: "did you measure anything before you committed to this build?" The silence on the other end usually answers for me.
Look, I'm not saying every failure is preventable. Ferrite cores crack during wave solder. Bad batches slip through—even from good factories, including TDK's own. I've personally stopped a board assembly line because I skipped a final verification I "knew" would pass. So this isn't about blaming the person who's already having a bad day.
But here's the thing: the companies that call me every month are missing the same few checks. And the companies that call me once a year—they're doing those checks. It's not luck.
The First Problem: Voltage Drop Is Treated As a Footnote
About a third of the "defective" transformers I receive for failure analysis test perfectly fine on the bench. Not "fine for what it is." I mean they meet datasheet specs with zero issues.
The real problem is the circuit around the transformer. Voltage drop—the difference between what your supply outputs and what your load actually receives—is the usual culprit. Thin wires, loose terminals, corroded connectors, a trace that's slightly too narrow. These don't show up on a datasheet. They show up under load.
The easiest way to catch this is a $30 multimeter and a few minutes of measuring. Measure at the supply output, then measure again at the transformer's primary, while the circuit is under full load. The difference is your voltage drop. If it's above about 3–5% of nominal, you're looking at a wiring or connection fault—not a transformer fault.
If you're new to using a multimeter for this kind of check, it's simpler than it sounds: source, load, compare. What you're looking for is a sag under load that the transformer can't smooth over.
In March 2024, a client called at 2 p.m. needing an emergency transformer rebuild for a servo drive. They'd already blamed the TDK part. I asked them to check voltage at the transformer terminals under full load. They found a 9% drop caused by a corroded connector. The fix cost $14. The rush rebuild they'd requested a quote for prior to the call? $3,800.
The Second Problem: N93 Is a Material, Not a Magic Word
Procurement lists love the shorthand "N93 ferrite." I love N93—it's one of TDK's workhorse MnZn power ferrites, and it performs well across a broad range of power converter applications.
The trouble starts when the shorthand replaces the thinking. N93 has a specific core-loss curve, and it behaves optimally within certain frequency and flux-density boundaries. I've seen it specified in designs running at frequencies where another material would have clearly lower core losses. Result: higher losses than intended, rising core temperature, even more losses. And then I get a call about a "thermal mystery" that's really a material-selection error.
Let me reframe this as an efficiency argument, because that's how I think: choosing the right ferrite material is a 15-minute datasheet review. Re-doing a transformer after thermal failure is a 3-day qualification cycle plus a rush order. You don't need a magic material. You need to check the loss curves at your operating point—not just the material name.
I'll add a caveat: my experience is mainly in mid-power industrial equipment—servo drives, motor controllers, AC/DC supplies in the 30W–3kW range. If you're in aerospace or medical with extreme reliability requirements, some of this will sound oversimplified. In the market segment I serve, the pattern holds.
The Part That Surprises Engineers: Where Your Transformer Is Made
Here's the argument you don't expect to see coming: supplier geography is an engineering parameter. Not the only one. But it matters more than most cost spreadsheets admit.
When I'm sizing up a rush request for a TDK transformer, I'm thinking about the map as much as the datasheet. Some TDK ferrite materials come from the company's Shawnee, OK facility, and a part like that can move through a U.S. assembly chain in days rather than weeks. Meanwhile, a component sourced from overseas—with perfect specs—is on a boat in the best case. Missed the port window? Now we're counting idle-line days.
In 2023, we lost a $47,000 contract because our client's preferred part had a six-week lead time from an overseas source. The alternative—comparable specs—could have shipped from Shawnee in three days. Their reason? "We've always specified this part." That's not engineering. That's inertia.
To be fair, the data pointed toward the overseas option. The unit cost was lower, and the spreadsheet looked clean. But my gut kept saying the domestic option was the safer bet. We went with the data anyway. The data cost us the contract.
I'd rather lose an argument than a client. Efficiency isn't just unit price. It's the speed of the next iteration. And faster iteration beats marginally lower cost—every time.
"We Don't Have Time for All That Testing"
I hear some version of this every week. "The product is already late. You can't add process steps now."
Okay. But the product is late because those steps were skipped earlier. The voltage drop check you're avoiding right now is exactly the check that would have prevented today's emergency if it was done six weeks ago. Rush fees exist for a reason: unpredictable demand is genuinely expensive. But the "unpredictable" demand that follows a skipped test isn't unpredictable. It's scheduled—just scheduled for later.
Why do teams skip the checks anyway? Because the person who fixes a crisis is a hero. The person who quietly prevents eight crises is sometimes called slow. I've been the hero. It stops being flattering when you realize the crisis was avoidable all along.
And I've made the same mistake. Last year, I recommended a weekend shipping option to a client without verifying the carrier's Saturday delivery guarantee. I knew I should've checked. But what were the odds? The odds caught up with me when the package missed the window by 12 hours and their assembly line sat idle. $4,000 in downtime. On top of $600 in "rush" shipping costs.
That mistake changed my approach. Now, when a client asks for a rush, I ask: "Is this an actual emergency, or is it just urgent?" An actual emergency requires speed. An urgent-but-not-emergency requires a process fix. They're different actions, and confusing them is expensive.
So Here's Where I Land
The industry talks about component quality as a fixed property, stamped onto the box at the factory. It isn't. An N93 core is only "high quality" if it's running where you actually run it. And a TDK transformer—yes, even a TDK transformer—is only as reliable as the voltage that reaches its terminals.
This isn't an argument for buying cheaper components. I still specify TDK transformers when the application justifies the cost, and I still recommend N93 for the right frequency band. But the difference between a smooth launch and a 2 a.m. panic call is rarely the brand of the ferrite. It's whether someone measured the basics before the deadline became a pressure cooker.
Three things to check this week: your voltage drop under full load; the core-loss curve of your ferrite material at your operating frequency; and the distance between your critical components and your assembly line. In that order.
If you've already done all three, you probably haven't needed me. If you haven't, I'd rather talk now than during the emergency. I can help with both. But one of those conversations costs a lot less—and it's a lot better for both of us.
