I'm sitting in my apartment in Shekou. It's 10:30pm. I've been on the phone for three hours. And I still have to write this down before I forget.

We almost lost a $4.2 million order this week. Not because of the factory. Because of me. Because I ignored the warnings.

Let me tell you the whole story. And then let me explain why I'm never making this mistake again.

Three Months Ago – The First Warning

I was in my office in Seoul. The quality team brought me a report on our new PCB supplier—let's call them Supplier G.

The report said Supplier G's sample boards failed the thermal cycling test. They had passed visual inspection. They had passed electrical testing. But when we put them through the thermal cycle—from -40°C to +125°C, 500 cycles—they failed. The boards showed micro-cracking around the plated through-holes.

The quality engineer said: "We recommend disqualifying Supplier G."

I said: "Get them to fix it."

The quality engineer said: "They don't know how to fix it. They don't have the process controls to control the thermal expansion."

I said: "Get them to fix it anyway. We need a backup supplier."

The quality engineer didn't argue. But he should have.

Two Months Ago – The Second Warning

We had a meeting with Supplier G's technical team. They presented a solution. They would change the laminate material from standard FR-4 to high-Tg FR-4. They would change the surface finish from HASL to ENIG. They would change the drill parameters to reduce stress on the plated through-holes.

It was a comprehensive solution. It looked like they had done their homework.

I approved the changes. I didn't check the details. I didn't verify the material certification. I didn't ask to see the laminate supplier's test data.

I trusted them.

The technical solution was correct. But the execution was not.

One Month Ago – The Third Warning

We received a second set of samples. The quality team tested them. They passed the thermal cycling test.

But there was a problem. The impedance was inconsistent. Some boards were at 48Ω. Some were at 53Ω. The spec was 50Ω ± 10%.

The quality engineer said: "This is marginal. They're technically within the tolerance, but the variation is high. We should be concerned."

I said: "They passed the thermal test. Let's move forward."

I was focused on the big issue. I ignored the small one.

This Week – The Fourth Warning

We received the first production batch. 5,000 boards. We inspected them. The visual inspection passed. The electrical testing passed. The impedance testing passed.

But the quality team noticed something. The boards were warped. Not significantly. About 0.5mm across the 300mm length. Within spec, technically.

The quality engineer called me. "David, there's something wrong with these boards. They're not meeting our internal quality standard for flatness."

I said: "Are they within spec?"

He said: "Just barely."

I said: "Accept them."

He hesitated. But he accepted them.

Yesterday – The Disaster

The boards were installed in our customer's product. The customer was a European automotive manufacturer. They started testing their product.

And one of the boards failed. Just one.

The customer sent it back. We analyzed it. The failure was a cracked plated through-hole. The same failure we saw in the first sample. Three months ago.

The high-Tg material was correct. The ENIG finish was correct. The drill parameters were correct. But the manufacturing process wasn't controlled. The laminate supplier had made a small change in the material formulation without telling anyone. The change was within spec. But it changed the thermal expansion coefficient. The boards failed under thermal stress. Not all of them. Just the ones that were on the edge of the process window.

The customer was furious. "If one board can fail, others can fail. How do we know this is not a systematic problem? We cannot accept these boards."

They were right.

I got on a plane. I flew to Shenzhen. I went to Supplier G. I sat with their production manager. I said: "What happened?"

Wednesday – The Answer

I asked the production manager one question: "When did you change the laminate supplier?"

He looked at me like I'd asked the impossible question.

He said: "We didn't change the supplier."

I said: "Did you change the material grade?"

He said: "No."

I said: "Did the material batch change?"

He paused. Then he said: "Yes. The batch number is different. But the material is the same grade."

I said: "Did you test the batch before you used it?"

He said: "We tested a sample. It passed."

I said: "Did you test the thermal properties? The CTE? The Tg? The actual values?"

He said: "We tested the thickness and the copper weight. Those are the standard tests."

That was the answer. They had tested the wrong things.

Thursday – The Call

I called the European customer.

I said: "We have a systematic problem. The supplier changed the material batch without telling us. They did not test the thermal properties. We will recall all 5,000 boards and replace them with boards from a certified supplier. We will cover all costs. We will expedite delivery. And we will implement a new material approval process to ensure this never happens again."

The customer was silent. Then he said: "How long?"

I said: "Four weeks."

He said: "Four weeks is too long."

I said: "Then let me do it in three."

He said: "Okay. But this is your one chance."

Friday – The Lesson

I've been thinking about this all day. I made three mistakes. Each one was avoidable.

Mistake One: I overruled the quality engineer's first recommendation. He said: "Disqualify Supplier G." I said: "Get them to fix it." I was focused on the need for a backup supplier. I ignored the engineer's technical judgment.

Mistake Two: I approved the changes without verification. The technical solution was correct. But I didn't verify that the implementation was correct. I didn't check the material certification. I didn't check the laminate supplier's test data. I trusted the supplier instead of verifying.

Mistake Three: I accepted the warped boards. The warpage was within spec. But it was a sign. The boards were telling me something. I didn't listen.

What I Learned

This week taught me three things.

First: The quality engineer is always right. I know that sounds like I'm saying something obvious. But I didn't act like it. I overruled the engineer because I had other priorities. That was a mistake. The engineer's job is quality. My job is to listen to them.

Second: Verify everything. A supplier's word is not verification. A test report from the supplier's lab is not verification. The only verification that counts is third-party testing on a sample from the actual batch.

Third: Listen to the little signs. The warped boards were a sign. The inconsistent impedance was a sign. The marginal thermal test result was a sign. I ignored all of them. I won't make that mistake again.

What I'm Doing Now

I've made three changes to our system. I'm writing them down so I don't forget.

Change One: All new suppliers must be qualified through our quality engineering team. No exceptions. I cannot override a quality rejection.

Change Two: All material changes must be verified by third-party testing before production. We pay for the testing. But it's cheap compared to a failure.

Change Three: No more ignoring the signs. If the boards are warped, even slightly, we stop production. We investigate. We don't accept.

The Cost

I ran the numbers. Here's what this lesson cost.

That's more than my annual salary. And it doesn't include the trust we lost with the customer.

One More Note

I called the quality engineer this morning. I said: "You were right. I should have listened."

He said: "You're the boss. I just do my job."

I said: "You're right. But I should have listened. I'm going to change that."

He didn't say anything. But I could tell he was satisfied.

Maybe that's the real lesson. Not the technical one. The human one.

Listen to the people who know.

We had three warnings. We ignored all of them. Here's how we almost lost a $4.2 million order – and what we learned fro

Three warnings. We ignored them all. The fourth one almost cost us a $4.2M order. Here's how I screwed up – and what I'll never do again.

Jangan direproduksi tanpa izin：Blog-Pulviso » A Procurement Director's Notebook: The PCB Disaster We Saw Coming – and the One Phone Call That Cha