故障很少一上来就报出自己的真名。屏幕停住、读数漂移、串口超时，这些都是现象。有用的习惯是：观察，一次只改一处，并把系统真正做了什么写下来。软件里的超时也是同一种东西。对话框只说明时间到了，没有说明原因。
以下正文为英文。
A fault rarely introduces itself by its real name. A screen freezes. A reading drifts. A serial port times out. A machine stops on a step that was fine yesterday. The first report is almost always a symptom. The work is to walk from that symptom back to a cause without inventing a story too early.
Engineering taught me to be suspicious of the first explanation, including my own. If I change three things and the problem disappears, I have not understood it. I have only been lucky. The useful habit is slower: observe, write down what the system actually did, change one thing, and see whether the behavior changes with it.
That habit moves from hardware into software without changing shape. A timeout in a LabVIEW VISA read, for example, is not yet an explanation. It might be the baud rate, the port, the termination character, a cable, a session that was never opened, or a device that was never asked the right question. The dialog box only says that time ran out. The notebook is where the real sentence gets written.
I keep this archive partly so those sentences have a place to go. Not every note needs to become a tutorial. A good one is allowed to be small: what I expected, what the system did, what I tried, and what I still do not know.
The longer I do this kind of work, the less I trust the feeling of being quick. Careful notes age better than a clever memory.