April 16th, 2025

Today is a day to remember. The sensor instability issue that had been troubling me for nearly two weeks finally had a major breakthrough today.

Recap

Ever since the new controller was built, it has been extremely unstable, specifically exhibiting:

• Random IMU sensor disconnections (sometimes one, sometimes both)
• The timing of disconnections was completely unpredictable – sometimes it lasted a long time, sometimes it disconnected after just a few minutes
• Other components continued to work normally during IMU disconnections
• Restarting it would fix it temporarily, though you never knew how long it would last this time

It didn’t consistently fail together, nor did it fail constantly. Forget solving the problem, I couldn’t even figure out where to start troubleshooting.

While searching for similar cases, I saw a post from some guy that included the line: “Is there a special incantation I should utter before powering it on?” I deeply resonated and sympathized.

After Continuous Crashes

Decided to make one last desperate attempt today. Saw some hints on the sensor’s FAQ page and decided to add pull-up resistors to the sensors. While doing so, I also re-soldered some potentially loose wires.

Testing again, the situation seemed to improve:

• Complete freezes were significantly reduced
• When data transmission stopped, letting the umbrella remain still for about 2 seconds allowed it to recover
• Through continued testing, I discovered that perhaps particularly high acceleration might be the trigger of the problem

Considering that previously even slight movements could cause freezes and cannot recover, and the timing of issues was random, I think the pull-up resistors probably, maybe, definitely played a stabilizing role… right?

And Then It Got Weirder!

As I debugged further, I discovered an even stranger phenomenon during testing:

• Swinging it vigorously with the canopy removed caused almost no problems
• But after attaching the canopy, it was easy to make both sensors stop working together

Actually, after discovering that high acceleration tended to cause problems, I had been performing vertical slashing motions to observe the data. After multiple stops, to make it easier to investigate what was wrong, I removed the canopy for easier handling. That’s how I discovered this issue…

This… is too bizarre. This phenomenon is completely beyond my understanding. How does the presence or absence of the canopy affect my sensor operation? I absolutely cannot figure it out.

Just Then!

Just as I was repeatedly attaching and detaching the canopy, frantically performing vertical slashs like a fool in my home…

It broke!

The umbrella broke! And it successfully cut my right thumb!

🩸🩸🩸🩸🩸

After I Took Care of My Thumb

Facing the mess, I could only:

• Discard the broken umbrella
• Remove the handle and canopy assembly, and install them onto a brand new umbrella

Unexpectedly, perhaps because the re-installation was more secure, the glitches with this umbrella seem less frequent? I don’t even know what to make of today. Does this count as finding and solving the problem? I don’t know, but who cares, as long as it works now. Hopefully, when I wake up tomorrow, this umbrella will still be working properly.

Lastly, because the fixtures on the original umbrella were glued too securely, I reprinted new ones. This time I chose a vibrant red—partly because the printer currently had red filament loaded, saving me a color change time, but it also actually makes the umbrella look better, and fittingly commemorates my blood sacrifice ritual.

-End-