DIY Laptop CoolerDIY 压风式笔记本散热器

Airflow and Temperature风道与温度

This cooler began with my laptop overheating and dropping frames in CSGO. I had already adjusted the machine and cleaned its cooling system. Looking at an ordinary cooling pad, I could see plenty of air moving without knowing how much entered the laptop's bottom intake. I wanted to seal the surrounding gaps and make the fan respond to the computer's temperature.

做这个散热器的起因很直接，玩 CSGO 时，笔记本高温掉帧。我已经调过机器、清过灰，再看普通散热底座，总觉得风扇吹得很热闹，进到笔记本底部风口的气流却未必有多少。我想试着把四周封起来，让风少绕一些路，同时让风扇跟着电脑温度调整。

I built an acrylic base with foam against the laptop underside to reduce air escaping around it. For temperature, I chose the readings already available inside the computer. An external probe would add cost and measure farther from the CPU and GPU. Using the internal readings meant finding a way to bring them onto my own controller board.

结构上，我做了亚克力底座，用泡棉贴合笔记本底部，减少气流从四周逃走。温度则直接从电脑内部读取，不再加外置探头。外壳附近的温度与 CPU、GPU 的状态隔着一段距离，还要增加器件成本；既然电脑已经能提供内部读数，我就想把它接到自己的控制板上。

My First Desktop Controller第一套上位机

I had no WinForms or desktop-control experience. With CSDN material and AI assistance, I assembled a C# application targeting .NET Framework 4.7.2. The application source uses LibreHardwareMonitor for temperature monitoring, then provides COM-port scanning, connection, display, and automatic transmission. Lines such as CPU62.7 and GPU71.3 leave through a 115200-baud virtual serial port and reach an ESP32 using BluetoothSerial.

此前我没学过 WinForms，也没有写上位机的经验。我边查 CSDN 边借助 AI，做出一个面向 .NET Framework 4.7.2 的 C# 程序。上位机源码调用 LibreHardwareMonitor 读取温度，提供 COM 口扫描、连接、显示和自动发送，再把 CPU62.7、GPU71.3 这样的报文经 115200 波特率的虚拟串口送出去。接收端是使用 BluetoothSerial 的 ESP32。

I did not want using a cooling stand to involve opening VS Code, attaching a USB-UART adapter, and typing commands. A cooler should not come with that much homework. My original plan put every control in the GUI, but I finished only connection and temperature transmission there. Five physical buttons handled mode, PWM frequency, and manual duty. The incomplete GUI plan had one useful consequence: the fan could be adjusted directly after power-on.

我不想让一个散热底座的使用过程变成打开 VS Code、接 USB-UART、手敲命令。散热器还附送一份操作作业，实在没必要。原本我想把所有操作都放进 GUI，最后只完成了连接与温度发送，模式、PWM 频率和手动占空比交给五个实体按键。这个没做完整的界面方案倒也有实用之处，通电以后，伸手就能调风扇。

The first successful Connect click was exciting. Windows found ESP32_Bluetooth, temperatures appeared in the window, and the fan on the desk changed speed. A program on my computer was controlling a board I had soldered. The 3 min 57 s demonstration preserves the operation across Windows, Bluetooth, OLED, buttons, and the physical fan.

第一次在 Windows 点下 Connect，找到 ESP32_Bluetooth，看着温度出现在窗口里，随后桌上的风扇真的变速了，我很兴奋。写在电脑里的程序终于控制到了自己焊的板子。3 分 57 秒的演示保留了 Windows、蓝牙、OLED、按键和风扇之间的完整操作。

Temperature and Fan Control温度与风扇控制

The ESP32 firmware generates PWM on GPIO 5, initially at 15 kHz with 8-bit resolution. Automatic modes use simple proportional temperature mapping, alongside manual duty control. I spent roughly half a week trying PID, without completing an implementation. In retrospect, I had reached for tuning before defining a target temperature, selecting the exact sensor, or recording the thermal response.

ESP32 固件用 GPIO 5 输出 PWM，初始频率是 15 kHz，分辨率为 8 位。自动模式采用温度的简单比例映射，另有手动占空比设置。我也花过大约半周尝试 PID，最终没有实现。那时连目标温度、传感器选择和热响应记录都没定义清楚，现在回看，急着调 PID 确实太早。

I also reconsidered the OLED label Fan Speed. Its variable was duty cycle; the program never measured RPM. A fan manufacturer's PWM and speed-monitoring guide makes the distinction between the command input and tach feedback explicit.

OLED 上的 Fan Speed 也是我后来才意识到的问题，显示变量实际是占空比，程序没有测 RPM。风扇厂商的 PWM 与转速监测说明把控制输入与 tach 反馈明确分开。

The application attempts one temperature transmission every five seconds, alternating CPU and GPU; the firmware updates its output from the higher peak in a 30-second window.

上位机每五秒尝试发送一条，CPU、GPU 交替，固件再用 30 秒窗口中的较高峰值更新输出。

Two Hardware Versions两版硬件

For the ESP32 version, I completed the schematic, EasyEDA project, and Gerber files, then had the board fabricated, soldered it, powered it, and installed it in the acrylic base. It carried two fan interfaces, power conversion, an OLED, and five buttons. There was no separate fan-driver IC; PWM went to the fan's own control interface. This version completed the computer-temperature-to-fan integration, and I used it for about two or three days.

我为 ESP32 版完成了原理图、EasyEDA 工程与 Gerber，随后制造、焊接、上电并装入亚克力底座。板上有两路风扇接口、电源变换、OLED 和五个按键，没有另加风扇驱动芯片，PWM 接到风扇自身的控制接口。这一版完成了从电脑温度到实体风扇的整机联调，我实际用了大约两三天。

I later redrew the controller around an STM32F103C8T6, reorganizing and widening some traces, changing placement, reducing board size, and adding a CH340N USB-UART interface. I soldered and powered this board too, and wrote its firmware. Serial commands from VS Code produced replies and changed the fan speed. I stopped this version at serial control, with the external Bluetooth link still awaiting reconnection to the Windows temperature application.

后来我用 STM32F103C8T6 重画控制板，重新整理并加宽部分走线，调整元件布局、缩小板子，再加入 CH340N USB-UART。我也把这块板焊好、上电，并写了固件。从 VS Code 经串口发命令，它能够回复命令，让风扇变速。这一版做到串口控制就停了，外接蓝牙还没连回 Windows 温度程序。

I explored another enclosure for the Bambu Lab A1 mini, modeling and splitting the body across nine build plates in the 3MF project. I compared automatic and manual tree supports, orientation, brim, and skirt settings. That route stopped at slicing previews.

外壳也试过另一条路线。我为 Bambu Lab A1 mini 建模、拆件，在 3MF 中排成九块切片板，比较过自动与手动树状支撑、零件朝向、Brim 和 Skirt。做到切片预览就停了。

Using the Prototype实际使用

During those two or three days, I felt that cooling improved. I remember roughly 20 to 30 additional FPS in CSGO and a maximum temperature falling from about 92 to 93°C to around 84°C.

使用那两三天，我觉得散热有改善。按现在记得的量级，CSGO 大约增加了 20 到 30 FPS，最高温度从约 92 到 93°C 降到约 84°C。

The cooler I built could go under my own laptop and adjust its fan to the computer's temperature, just as I had first wanted. I was thrilled the first time I clicked Connect and watched the fan on my desk change speed with those readings.

我把最初那个让风扇跟着电脑温度调整的想法，做成了能放到自己电脑下面使用的散热器。第一次点下 Connect，看着桌上的风扇跟着电脑里的温度读数变速，我特别兴奋。