As far as marble runs go, few can hope to compete with [Wintergatan]’s amazing musical works. While undertaking the build of the Marble Machine X, timing problems became apparent. You can crank the machine faster or slower to alter the tempo, but the time it took marbles to drop due to gravity and hit an awaiting drum remains constant. You can’t control gravity, so you look to a mechanical solution in adjusting marble drops. In music, as in a jewel heist, timing is everything. Thankfully, [Machine Thinking] was on hand to collaborate, and a solution was found in the form of a beautifully machine indexable clutch.
The duo came together and designed a clutch, that would allow the timing relationship between various parts of the Marble Machine X to be varied. At 7:10 into the Wintergatan video below he gets down to brass tacks on how this brass music timing clutch actually works. At 90 BMP, the clutch allows the synchronization of the machine to be altered in approximately 10ms increments. Without this vital addition, drum beats would tend to slip out of time.
It’s a part that would challenge the novice, requiring the cutting of teeth and the inscription of accurate markings to allow it to do its job. However, it’s no trouble for [Machine Thinking], who shared a video of the machining process, including the outsourcing of the hand-engraved dial numbers.
Such a piece takes significant work to produce, and yet it is just one part of a much larger machine. We can’t wait to see the Marble Machine X finished, but if you’re unfamiliar with [Wintergatan]’s earlier works — you’re in for a treat.
The marble machine X is seriously one of the most underrated series on YouTube. Beautiful to see the machine slowly coming to be.
What I most enjoy is to what level of detail he shares every piece of the build process, including his failures. Sharing this experience makes it an outstanding channel to pick up knowledge as well.
Absolutely. Of you’re not watching ot, you should, if you’re watching ot, you should subscribe. If you’re subscribed, you should be a Patreon backer. Martin puts out a video each week that often looks like it took a week’s effort.
Magnificent!
90BMP = 90 Brake Music Power?
Yes, but it has a lot of low-end grunt with 180 M/M (Music per Meter).
Some clutch 101 for the terminology-challenged please. I thought a clutch is something you engage and disengage, and it passes or not the movement from one part of the machine to another. A kind of mecha-transistor gate. This beautiful shiny thing seems to provide variable reduction ratio, are such things also called clutches?
Nope, it does not affect the transmission ratio. It only engages and disengages, but allows for an indexing offset.
A clutch is something that mechanically connects an input shaft to an output shaft. An angular relation between the two shafts is not always necessary (see clutch in manual car, it doesn’t matter when the clutch engages).
The connecting action can be achieved by many means, one example would be pure friction (car clutch), another example would be tooth engagement (the device in the post).
The shiny thing up there does not provide variable reduction ratios. It only ensures a known and defined angular relation between the input shaft (handcrank) and the ouput shaft (rest of the machine) as in this case the timing of the crank and the dropped marbles is very crucial.
So it provides a convenient way of setting the offset angle, but in principle it could be a piece of pipe with a set screw?
Yes, pretty much.
Thanks everyone!
Epic. It is like the ad where the scientist has his life theorem on the chalkboard, looks away, and the janitor walks in and starts ‘cleaning’….
The difference being that one is a friction clutch and the other is a dog clutch to name them correctly. :)
I bet the next version will have a little flywheel in it that automatically changes the offset based on the rotational speed of the driving shaft. So as he speeds up or slows down, the downbeat will still stay in sync. I hope thinking machine is up for that challenge.
I don’t think it will. Having the drop drum match the set rpm gives the player a que. So that he can keep the rotation at the correct speed. So a 90bmp song actually get played at 90bpm. If the drop varied with the rotation speed it would still sound correct at 70bpm or 100bpm so the player would not know if he where cranking at the right pace without a metronome or similar.
But even if it was not functional, how cool would it be to have a fly-ball governor on the machine… :)
Wintergatan’s marble machines make life worth living. I’ve watched these thing operate dozens of times, and they still send a tingle down my spine.
Alas, I cannot say the same for HaD’s policy on editing comments—in that case, it’s more of a hot flash than a tingle.
Hmm. could he have not used a off the shelve differential gear were the drive shaft is de offset adjustment and one of the wheels the driving shaft and the other connected to the drum machine?
But all these problems came from the fact that he is not using marbles to trigger the drum sounds.
Watch the rest of his videos. He is using the marbles to trigger drum sounds.
The “paddles” that are attached to the crank are only used to check timing.
A differential would be ideal in this situation, particularly because it would allow changing the timing while it is running.
Beautifully machined but used a stack of washers instead of the correct length screws?
No, the washer stack is probably just what Machine Thinking just had on hand. Because, if you jump to 5:10 of [Wintergatan]’s video it shows the washer side of the clutch being mated to the pinion gears and proper lengthed screws.
The stacked washers are just placeholders because he did use the correct screw lengths.. You can see in the video that is where he attached the plywood gear.
Very pretty solution
But could you just raise or lower the drum to adjust the delay accordingly?
Definitely, but he’d have had to elevate the marble catch with it, as well as account for the shorter rebound hop the marble would make resulting from not having as much velocity upon impacting the drum. Perhaps just a bigger opening (or a funnel) on the catch tube would do it, but that seems less elegant IMO.
Or (come to think of it) he could also do some maths to find the exact angle for the drum such that the reduction in v(x) post drum strike is taken up by the added time of flight resulting from the drum being relatively higher than the catch tube. THAT would have been elegant!
My guess is switching songs with different tempos during a concert makes having a way to easily/quickly/reproducably recalibrate delay an absolute necessity. Turning a little wheel with clear markings to pre-tested calibration values between songs would be much simpler than unscrewing a set of drums, lowering/raising them, tightening them and still maintaining the marble return path so the machine doesn’t drop marbles all over the stage.