readease Chapter 35, First Slash
Chapter 35, First Slash
2019 12 Month 9 Day.
Wuxi. Microsensor Semiconductor.
Su Chen has been on the production line of Micro-Sensitive for eleven days.
For the first five days, he didn't say anything. Every morning at eight o'clock, he would go to the observation window outside the clean room, wearing the disposable dustproof suit provided by Microsensor, and observe the process from the deep hardening process to the stress release of the wafers and then to the encapsulation. He watched each step at least three times.
Hongyuan's on-site engineer, Xiao He, accompanied him. Xiao He was responsible for recording every temperature, humidity, and time parameter during the production line's operation. Every night at nine o'clock, Xiao He sent the day's data report back to the Shenzhen headquarters, while also copying Su Chen's notebook.
Su Chen spent his days inspecting the production line. In the evenings, after returning to his hotel, he would enter the virtual disassembly lab.
In the virtual environment, he completely replicated the deep hard-working process of the micro-sensor—not from imagination, but reconstructed based on the hundreds of actual parameters recorded by Xiao He every day. Then he began to do something impossible in reality in the virtual environment: gradually reducing the temperature control precision from plus or minus two points and one degree.
±1.8 degrees – the stress distribution on the disc surface improved by 4%.
±1.5 degrees Celsius – the temperature drift data begins to show a visible downward trend.
±1.2 degrees Celsius – a sudden mismatch in humidity parameters during a certain packaging process leads to the scrapping of the entire batch.
In the virtual environment, scrapping is free. Su Chen rolled back the parameters to the 1.5-degree node and readjusted the encapsulation humidity. After three rounds of simulation, he found a feasible window of encapsulation parameters.
Then he continued to press down.
±0 degrees Celsius – the temperature drift data has been reduced to less than three times that of Bosch.
±0.8 degrees Celsius—twice that of Bosch.
However, the required precision of 0.8 degrees exceeds the capabilities of existing microsensor equipment. This means that hardware modifications are necessary.
Su Chen spent two sleepless nights testing four hardware modification schemes in the virtual environment. Three of them required replacing the core heating module—too costly and time-consuming. The fourth involved a minor modification to the existing heating module's control circuit—adding a PID feedback loop to compensate for insufficient hardware precision with software. The cost was approximately 150,000 yuan. The modification took three days.
On the morning of the sixth day, Su Chen handed the first version of the improved plan to Huang Zhihua in the small conference room on the third floor of Weigan.
The plan was only seven pages long, but each page was filled with specific parameters, modification locations, and expected results.
Huang Zhihua read it page by page. It took him forty minutes.
As he put down the proposal, he tapped his fingers lightly twice on the table.
"Mr. Su, where did you verify the PID feedback loop design on page three of this plan?"
"I have my own verification method."
Huang Zhihua stared at him for three seconds. He didn't ask any further questions. He turned to Zhao Jiancheng.
"Mr. Zhao, the first step in this plan involves temperature control in the deep hardening process. It requires modifying the heating module control circuit of Furnace No. 2. Furnace No. 2 will be shut down for three days during the modification period. Furnace No. 1 will not be affected."
"Modification costs?"
"Approximately 150,000."
"risk?"
Huang Zhihua remained silent for two seconds.
"If the PID loop parameters are set correctly, the temperature control accuracy can be reduced from ±2.1 degrees to approximately ±1.2 degrees. If the parameters are incorrect—the heating module of Furnace No. 2 may need to be recalibrated, requiring an additional one to two days of production shutdown."
What's your opinion?
Huang Zhihua glanced at Su Chen. Su Chen's expression remained calm.
"The technical logic of the solution is correct. I've seen a similar principle to the PID loop design in Bosch's literature. However, the micro-sensor's device is different from Bosch's. We'll only know the actual effect after running it."
He paused for a moment.
"I recommend doing it. If anything goes wrong—I'll take responsibility."
Zhao Jiancheng nodded.
"implement."
The modification began that afternoon. Huang Zhihua personally led the team. He opened the control cabinet of furnace No. 2 and, together with two process engineers, removed the control board of the heating module. Su Chen stood by and watched. He didn't lift a finger—this was Huang Zhihua's territory, and Huang Zhihua knew every screw on the production line better than he did.
But Su Chen will speak up at crucial moments.
"Mr. Huang, I suggest increasing the sampling frequency of the PID feedback loop from 100 Hz to 200 Hz. In my simulations, I found that 100 Hz sampling exhibits a 0.3-degree lag when the temperature changes rapidly."
Huang Zhihua looked up at him. Then he nodded.
"Two hundred hertz. Received."
The modification took two and a half days, half a day faster than expected—because Huang Zhihua's team soldered out an adapter board that originally needed to be customized using existing components at 3 a.m. on the second day.
On the morning of December 12, Boiler No. 2 was powered back on.
The calibration took six hours. At four o'clock in the afternoon, Huang Zhihua saw the new temperature control data at the control panel.
Plus or minus one point, one degree.
It was 0.1 degrees better than Su Chen's estimate of 1.2 degrees.
Huang Zhihua's hand paused on the control panel for a few seconds. He turned to look at Su Chen. He didn't speak. But the last vestige of reserve that had been present in his eyes since the exam—vanished.
"Next step," Su Chen said.
"Stress relief for the disc," Huang Zhihua continued, "on page five of your proposal."
"Yes. That modification is smaller. It doesn't require shutting down the furnace. Just adjusting the parameters."
"I'll bring a team to investigate tonight. The first batch of verification data will be available tomorrow morning."
"good."
12 month 13 day.
The parameter adjustments for stress relief on the disc were completed at 4:00 AM. Huang Zhihua and two engineers spent the entire night in the cleanroom.
The first batch of verification data came out at 8:00 AM. The consistency of stress distribution improved by 12 percent compared to before.
When Huang Zhihua showed the data to Su Chen, his hands trembled slightly. Not from nervousness—but from excitement.
"Mr. Su, if the humidity setting for the packaging is also adjusted according to the window specified in your plan—"
"We'll adjust it this afternoon. Tomorrow we'll conduct the first round of full-scale trial production."
"tomorrow?"
"tomorrow."
Huang Zhihua took a deep breath.
"good."
Meanwhile, the outside world was not quiet.
After Tianying's Falcon S2 Pro was officially launched at the end of November, its first month's sales figures were released in early December—38,000 units.
This figure caused a stir in the industry. Thirty-eight thousand units is a considerable number for a drone priced at 1999 yuan. 36Kr analysts estimated that Tianying lost at least fifteen million yuan on these thirty-eight thousand units.
But Haotai didn't care about the losses. Xia Kanghao once said something in an internal meeting (later relayed to Su Chen by Xu Lang): "Fifteen million to buy thirty-eight thousand users and market share, that's a good deal."
F4's November sales were visibly impacted – dropping from a stable range of 2,500 to 2,800 units to 2,100 units.
Meng Xiaoying noted this number in Friday's report. She didn't write any comments in the remarks section. But Su Chen knew what she wanted to say.
Zhou Ming asked Su Chen over the phone, "Should F4 lower its prices to counter this?"
Su Chen's answer was: "No price reduction. Small-scale testing of the F5 will begin in Q2 of next year. By then, we'll have broken through the performance ceiling of the S2 Pro. Lowering the price now would only sacrifice profits."
"clear."
Su Chen hung up the phone. He walked to the hotel window. The December night in Wuxi was already quite cold.
He mentally reviewed the current timeline.
On the microsensor side—tomorrow marks the first complete trial production of the process. If the improvements in the three stages combined can reduce the low-temperature drift to within three times that of Bosch, then the first step will be considered a success.
The Falcon S2 Pro will continue to attract entry-level customers of the F4 in the short term. However, the S2 Pro's open-source flight controller technology has reached its limits. The F5's release will be a game-changer.
Haotai's sensor offerings are putting pressure on Bosch and TDK—their joint procurement volume has already made them cautious. There's been no immediate action, but the undercurrents persist.
Three lines. Three time windows.
Tomorrow will be the first window.