Human beings sometimes get strange ideas out of nowhere.
In my case, that happens quite often.
At the moment the idea appears, I’m absolutely convinced it’s the best idea I’ve ever had. But then… time passes. The excitement slowly deflates. And before I know it, the “greatest idea ever” has quietly disappeared.
I realised that if I kept going like this, I would never actually build any of my “best ideas.”
So I decided to do something drastic.
I would make it impossible to go back.
That’s how this marble run began.
This marble run is 1200 mm (47.2 in) tall and 600 mm (23.6 in) wide. It is mounted directly on the wall next to our dining table.
Yes. Once it was attached to the wall, there was no turning back.
Lifting the Ball: The First Challenge
The first problem was simple in theory:
How do you lift a ball up 1200 mm?
There are many possible methods, but for this height, I thought the most efficient way would be to use metal balls and lift them with magnets attached to a moving chain.
So, without thinking too much about the overall design, I started by building the chain.
Before attaching the main board to the wall, I milled shallow holes into it using a CNC router, creating a grid pattern almost like graph paper.
This was an idea from my woodworking mentor, Juno.
It makes positioning parts and placing screws much easier.
Three-Gear Vertical Lift (a.k.a. The Hardest Part)
Once the base was ready, I wanted another lifting mechanism besides the chain. Using only a chain felt a bit boring.
So I decided to use three large gears.
The material? Leftover wood from making puzzles.
If you look at marble runs on YouTube, you’ll often see gear-based lifters where the gears are tilted, or where boards cover the holes so the ball doesn’t fall out.
But doing the same thing as everyone else is not very interesting.
So I stood three gears vertically? with no guards at all? and tried to design a system where the ball would travel all the way up.
This was the beginning of a long period of trial and error.
Making the ball fall only at a specific point was surprisingly difficult. And not only that ? it had to land precisely into the hole of the next gear.
This was the most challenging part of the entire marble run.
The Wall Marble Run, January 2023
Adding a Shishi-odoshi
One element I really wanted to include was a shishi-odoshi.
If you have visited a Japanese garden, you may have seen one.
It’s a bamboo tube that fills with water, tips over to empty itself, then returns to its original position. When it tips, it makes a distinctive sound. Originally, it was used to scare wild animals away from crops, but today it is often decorative.
In my marble version, the ball replaces the water.
The Momentum Loop and My First Time Using a Lathe
If the ball is going to roll with energy, I thought it should also feel a bit like an amusement park.
That’s how the loop was born.
However, I had no idea how to machine this shape using a CNC router. Once again, my mentor Juno came to the rescue.
Where the ball jumps, it lands in a trumpet-shaped piece that I turned on a lathe. It was my first time using a lathe, so I was slightly nervous.
More Parts, More Experiments
At the bottom, I created a storage area for the balls. Since I was making it anyway, I decided to use marquetry.
There are many ball-distribution parts in this marble run, but the most numerous elements are the rails.
Straight rails are functional but not very exciting, so I experimented with curved rails. Luckily, I happened to have just the right router bit and cut them horizontally with the CNC.
Next came the spiral drop section? slightly reminiscent of the Guggenheim Museum. I named it Helix Honey Dipper.
This was machined using a rotary attachment on the CNC router. Once again, Juno helped.
In the upper right corner is a vertical mechanism made of arrow-shaped pieces (Cascading Arrow Tilts) that pass the ball downward one by one. Surprisingly, transferring the ball smoothly between parts was quite difficult.
Actually… almost everything was difficult. 😅
The Wall Marble Run, May 2024
Vertical Splitter distributes the ball left and right. It’s similar to a typical Y-shaped splitter, but oriented vertically. At first, it didn’t work well until I placed two balls inside. Then it behaved perfectly.
The Wall Marble Run, September 2025
Open Frame Rail is a rail located high up. If it were a normal rail, it would be hard to see from below, so I designed it so the ball rolls visibly inside a frame.
Mosquito Coil Spiral is similar to the round dish element, but since it is also positioned high, I left openings so the movement of the ball can be seen from below.
Rotary Catcher receives balls from two directions. To prevent jamming, I used bearings for the centre shaft to ensure smooth motion.
Three Years and Eight Months
If I described every detail, this marble run would include more than 30 parts.
I worked on it whenever the workshop was available, so progress was slow. In total, it took three years and eight months to complete.
If I hadn’t mounted it on the wall at the beginning, I might never have finished it. 😆
From here on, I would like to continue exploring not only marble runs, but also automata and kinetic art.
This project started as a strange idea.
Now it lives on the wall.
And it moves.
