I'm going to tell you about something that made no sense for four weeks.

July 2026. I'm a procurement manager for a battery assembly company. We buy lithium cells—cylindrical 18650s—from a Tier-1 Chinese supplier. The cells are good. Tested. Certified. Consistent.

We distribute these cells to three contract assembly factories. They weld them into battery packs. Simple process: weld nickel strips, connect to a BMS, pack into a housing.

Factory A has a 99% yield. Factory B has a 72% yield. Factory C has a 97% yield.

Same cells. Same BMS. Same housing. Same welding spec. But Factory B is scrapping almost 30% of the cells.

This made no sense.

The First Clue

I visited Factory B. The welding line was running. The operator was placing cells into the fixture. The machine was welding nickel strips. Everything looked normal.

I asked the quality manager: "What's the failure mode?"

"Internal resistance is too high after welding," he said. "The cells are good before welding. They fail after."

I checked the pre-weld data. The cells were within spec. I checked the post-weld data. The resistance had increased by 20-30%.

Welding shouldn't change the internal resistance. Something else was happening.

The Second Clue

I watched the operator place the cells. He was handling them quickly. He was stacking them in the fixture. He was pressing them down firmly.

I asked: "How much pressure do you use when placing the cells?"

The operator shrugged. "As much as needed."

"Show me."

He placed a cell. He pressed it down. I could see the pressure was high. He was pressing the cell into the fixture, compressing the cell.

The cell was being deformed. Not visibly. Micro-deformation. The pressure was changing the distance between the electrodes inside the cell. The internal resistance was increasing.

The Third Clue

I went to Factory A. I watched their operator. He was placing cells more carefully. He wasn't pressing hard. He was dropping them gently into the fixture and using a soft touch.

Factory A's process was the same as Factory B's. But the operator was different. The pressure was different.

Factory B was compressing the cells. Factory A wasn't.

The Fourth Clue

I asked Factory B's production manager: "Why do you press the cells so hard?"

"The fixture is designed that way," he said. "The cells need to be held in place."

I looked at the fixture. The cells sat in it. They didn't need to be pressed. They just needed to be placed. The fixture was designed to hold them. It didn't require pressure.

I said: "Did anyone train the operator to press?"

The manager said: "We taught him to place the cells. He presses because he thinks he's doing a better job."

The Fifth Clue

I checked the training records. The operator had been on the line for three weeks. He was new. He had been trained by the previous operator, who had also pressed the cells. The cycle of compression had been passed down through three operators.

None of them had been told not to press. None of them had been told about the micro-deformation issue.

The Full Picture

Here's what happened.

The cell is a mechanical structure. Inside, there is a jelly roll—layers of anode, separator, and cathode rolled together. The distance between the layers determines the resistance.

When you press the cell, you compress the jelly roll. The layers get closer. The resistance increases. The cell is damaged. Not visibly. Internally.

Factory A's operator had learned to be gentle. Factory B's operator had learned to be firm. Both were doing what they thought was right. One was destroying cells.

The Solution

We retrained the operator at Factory B. We showed him the data. We showed him the pressure sensor. We told him: "Do not press. Only place."

The failure rate dropped from 28% to 4% in two weeks.

We also installed a pressure sensor on the fixture. It didn't control anything. It just recorded the pressure. If the operator pressed too hard, the sensor logged it. The supervisor would check the logs.

What I Learned

Here's what I learned from this.

First: Training matters. The operator was doing what he thought was right. He was never told otherwise. The problem wasn't skill. It was knowledge.

Second: Pressure matters. A small change in handling can cause a big change in performance. The pressure wasn't visible. It wasn't measured. It was still affecting the product.

Third: The simplest explanation is often the right one. I spent three weeks looking at the welding machine, the BMS, the nickel strips. The answer was in the operator's hand.

Fourth: Operators pass on their habits. The previous operator pressed. The new operator learned to press. The bad habit had been passed down for three generations. Nobody had checked.

What I'd Tell Every Buyer

If you're buying lithium cells or battery assemblies from Chinese suppliers, ask these questions:

Question 1: "How do you handle cells during assembly? Do you have a pressure specification?"

If they can't answer, they probably haven't thought about it.

Question 2: "Do you monitor the pressure during assembly? Do you have sensors?"

If they don't have sensors, they're relying on the operator's feel. That's a risk.

Question 3: "How do you train new operators? Do you have a formal training program?"

A formal program catches bad habits. Informal training passes them on.

Question 4: "How do you track failure rates by operator? If a failure rate changes, can you trace it to a specific operator or shift?"

If they can't, they're not managing their process. They're hoping for the best.

The Aftermath

Factory B's yield is now 96%. Not quite as good as Factory A's 99%. But close enough.

The training program has been updated. The operator now knows why pressing is bad. The fixture now has a pressure sensor. The logs are reviewed weekly.

The cells are still the same. The process is different.

The Question I Still Have

I still wonder: how many other factories are compressing their cells without knowing it? How many operators are pressing harder because they think it's better? How many batteries are failing early because of micro-deformation that no one measured?

I don't know the answer. But now I always ask the question.

One batch of lithium cells. Three different assembly factories. Factory A had 99% yield. Factory B had 72% yield. The ce

One batch of cells. Three factories. 99% yield at one. 72% at another. I spent four weeks chasing the answer. It wasn't what I expected.

Jangan direproduksi tanpa izin：Blog-Pulviso » The Cell Mystery: Same Batch, Three Factories, Three Different Failure Rates