Building Marble Machine 2
I hadn't actually planned on building a second marble machine, but in March 2000 I had an idea for building a better marble elevating device, plus I had enough time on my hand to actually build a machine using this principle. And this is the machine I built...
What inspired me to play with building a marble machine again was the idea for a marble elevating wheel, which would be very unlikely to jam.
Experimenting with wheels
|
I started experimenting by cutting a circular piece of plywood, and drilling 4 holes into it at
a slight angle. This I temporarily jigged up with some blocks of wood, and started turning it
to see how well it would pick up marbles. The whole thing turned out to work very well.
The wheel I had cut only had a 25 cm diameter, which would not provide enough head (elevation change) to allow the marbles to do much on the way back down. Rather than make a much bigger wheel, I figured I should use a second wheel to elevate the marbles even further. This would be more visually appealing than one big wheel. The challenge with two wheels was how to somehow couple their rotation together. At first I wanted to just cut a groove in the circumference of the wheels and use an O-ring style belt to drive them both, but I couldn't figure out where to find that kind of belt. The next idea was to just cut gear teeth into the wheels. This is a lot of work, but more visually appealing than a belt. Visual appeal is very important, as the marble machine, after all, has no real purpose. If its not fun to look at, its no fun at all. |
Once I had the two large gears cut, and more holes drilled into them, I jigged everything up with
some pieces of wood roughly cut to guide the marbles between the wheels. Everything worked nicely.
The next thing I needed was an actual drive gear. The drive gear had to be much smaller, and on
account of that would be stressed a bit more, so I decided to inset the teeth into the plywood,
rather than just using the randomly grained rough plywood edges.
I cut a plywood circle to the minor diameter of the gear. then I jigged something up to cut the slots
on my table saw. The whole jig was mounted on my crosscut gadget, which slides in the T-slot.
I used two cheap 7 1/2" skillsaw blades stacked as a poor man's dado blade for this purpose.
Then I cut some hardwood to the exact thickness of the slots, cut it into small squares, and pressed
the individual pieces into the slots of the gar. Then I rounded the edges of the teeth a bit
with a knife, and my drive gear was complete.
Because the teeth all have the grain pointing outwards along them, I was able to make the teeth
thinner without risk of premature wear or breakage. This increases slack in the gear, making it
less troublesome to get running smoothly.
Overall, I am very happy with this approach to making gears. Works much better than cutting the
teeth out with a bandsaw. Now I just need something else to try this style of gear construction on.
I have since built a Gear cutting jig
to help index the angles by precise amounts for cutting wood gears.
Also, on the rising side of the wheel, which has the marbles going up in it, it is necessary to
block off the holes once they are above the axis of rotation of the wheel. This is because past
this point, the holes point slightly down, and the marbles would otherwise start to roll out
of the holes prematurely. This could have been avoided by slanting the holes slightly differently,
but that in turn would have reduced the efficiency with which the marbles roll in and out of the
holes at the top and bottom, so putting a barrier in front of the holes is a more efficient solution. You can see
the barrier at the left of the top wheel, and the right of the bottom wheel. The bottom wheel
turns counterclockwise, the top one clockwise. This also means that the drive shaft and crank
ended up going clockwise (lucky, because I hadn't thought that detail through).
I deliberately chose a direction of rotation for the wheels so that, as they leave the first
wheel, the marbles already have a bit of momentum towards the second wheel. That way, I didn't
need as much slope on the guide between the wheels, and thereby preserving altitude.
The 'spout' at the top of the assembly shown is to guide the marbles onto descent paths that
I had not designed yet at the time I took the photo.
I still used a flipflop to divide the marbles into two streams (see image at left).
This flipflop turns out to be a little overclocked. Nevertheless, it puts roughly the same
number of marbles in either direction, which is all that I really cared about for this machine.
To the left is a round-ish bowl that the marbles roll round and round in, until they gradually
work their way towards the centre, and randomly enter one of the three holes. This mechanism
unfortunately doesn't make very much noise, but at the same time, it creates lot of visual
motion, and uses very little elevation, so i'm very happy with it. When the machine is running
at capacity, and one stops cranking, it takes nearly 10 seconds for all the marbles to clear
the bowl.
The board of nails on the right allows marbles to bounce randomly against the nails. I wasn't sure
if this would create much of a random distribution, but driving some nails into a board and trying
it showed that this approach yielded a lot of randomness. The board is mounted at a 45 degree
angle, so that gravity pulls the marbles to wards the board, as well as down it, and visibility
is optimized.
Another fun gadget is the gadget towards the back (top of image). Marbles
are dropped onto the left of this gadget, and roll towards the right, then stop against
the lip of it, staying on the darker piece of wood. The next marble to fall onto the back
causes the wood to rock, propelling the last marble several centimeters into the air, and
off the piece of wood. The just fallen marble now rolls to the front, ready for the process
to repeat. Its not entirely reliable, but even if three marbles end up stuck against the lip,
the next marble will dislodge at least two of them, thus getting it going again.
The larger block on the right is just for precussion. The top of it is a 4mm thick piece of
maple, with the rest of the block hollow underneath. Marbles drop from one of the holes in the
bowl directly on this, producing a very distinct clack. Also, most of the marbles dropped onto
the soup bowl drum will also bounce onto this block, producing a very satisfying ping-tock
sequence for each marble that falls on the soup bowl drum. Because this is so satisfying, I
slightly rounded the corners of the hole in the bowl leading to the soup bowl drum,
causing that one to get more than just 1/3 of the marbles. I just like that sound so much!
I wanted another mechanism that dumps groups of marbles. The divide by 6 in my last marble
machine was a neat idea, but would not be able to deal with the asynchronous clock nature of
this marble machine. I wanted something that would not jam.
Making the gears
Instead of throwing away my initial experimental wheel, I just drilled more holes into it
so it could elevate more marbles. I then made
a second wheel exactly the same diameter. I didn't have another piece of plywood
of the right thickness around, so I used thinner plywood, and then glued some pieces of wood onto it
to bring it to the right thickness. The pieces I glued I cut from a piece of fire wood.
I made both wheels to have exactly 31 inches circumference, then wrapped a non rigid tape measure
around the edge and put pencil marks every 1/4 inch. Then I cut out the space between every other
set of pencil marks with a bandsaw. This has to be done with great precision.
Gears that are inaccurate will easily jam, and not run smoothly. It took me a while with a file
and sandpaper to eliminate spots where the gear teeth would bind against each other.
Marble tracks for the elevating system
Shaping the marble guides to allow the marbles to easily roll into the holes in the wheels took
some optimization. I made the guides curved to follow the path the holes take at the bottom of
the wheel, to maximize the opportunity marbles have to just roll into a hole in the wheel.
Randomized descent path distributors
My first marble machine was based on logic. My thinking as I built it had been to invent lots of
deterministic marble logic elements, and perhaps to show that it would theoretically be possible
to build a marble operated computer.
The approach to this marble machine was more of one of complete chaos. I had done logic, now I
wanted to do chaos. Besides, the wheel based elevating system, aside from skipping the odd
marble, could deliver marbles at such a high 'clock rate' that it would overwhelm most marble
logic elements.
The percussion section
The bowl on the left distributes the marbles into three paths originating from the three holes.
For each of these paths, I created a separate and noisy thing for the marbles to fall on.
The first thing I did was to take an empty soup can, and dropped some marbles on the bottom.
This made a satisfyingly loud noise, so I detached the bottom along with 4 cm of the can, and
mounted it upside down on a piece of wood (bottom left part of image). A channel mounted on
the bottom of the bowl lets the marbles roll to directly over this can, and drops them. This
gadget is the loudest and most distinct noise of the marble machine.
Dump-O-Matic and clanking metal
|
The mechanism works by having a wooden cup on a pivot with a counter weight. This is arranged in such a way that once the cup is full, the cup tips over, and dumps its marbles. The tricky part is balancing it in such a way that the cup, when it dumps, dumps completely, and subsequently rights itself again.
The cup is made of a solid piece of wood. I took a thick hunk of wood, and drilled a big hole in it, which is the inside of the cup. Then I cut away a lot of wood around it, giving it the shape of the cup. Its important to drill the hole first, because the wood, once cut to a small piece, would not withstand the force of a 1.5" drill.
The mechanism had the problem that often marbles would just roll along the edge of the cup, and do this in such a way to fall out of the machine, causing unacceptable rates of marble loss. The coat hanger bow (the white U-shaped wire) cured that problem. The mechanism still has the problem that when marbles fall onto it while it is in the dumping phase, it will sometimes deflect the marbles in such a way that they fall out of the machine. Higher fences around the edges of the machine would solve this, but would reduce visual appeal.
I also had a piece of angle-iron that I noticed made a nice noise when the marbles were dropped onto it. So I used that piece of angle iron as an L-channel to take the rest of the marbles from the nail board. This piece is loosely mounted, allowing it to make more noise. I also had to put a screw in the front end of it, as some marbles would roll off the front end, instead of rolling towards the back.
Hard drive gong and bicycle bell
One of the things I really wanted to integrate was a hard disk platter. Unfortunately, modern hard drives don't make as good a gong as the old 5.25" hard drives did, and I didn't have any of the old platters around. Still, the newer platters are much shinier and prettier than the old ferrite coated 5-meg jobs. This hard drive gong in the machine is strategically placed to catch the marbles from the dump-o-matic, and deflect them inwards to the marble machine, making marble loss less likely. It is mounted in such a way that it is only in contact with felt, and with minimal pressure, as not to dampen the vibrations. It makes a nice 'ting' when the marbles hit it, but I wish it was louder. People keep thinking its a CD-ROM disk. This first surprised me, because I just assumed everyone had taken apart a hard drive before, but I guess most people just throw out computers when they stop working. Pity. At any rate, CD-ROMs have lousy acoustic properties. They look neat when put in a microwave though. For the marbles from the clanking angle iron, I deflect those forward again and let them roll down the angled track to hit a bicycle bell. I wish I'd had a bigger bicycle bell - the one I use isn't loud enough. Still, adds a nice 'ping' when a marble hits it. The bell is placed to bounce the marbles towards the gong, so the marbles hit it just once and let it ring out. |
The dump-o-matic and clanking metal channel didn't use up all the vertical drop I had available,
so I added some more stuff to convert the remaining height (energy) into noises of
various kinds.
Power source
Originally I was planning on motorizing Marble Machine 2, just like Marble Machine 1. Experimenting
with just the big wheels, I realized that with a smaller gear on the drive shaft, it would be just
the right speed for hand cranking. This marble machine is very efficient, so its a pleasure
to hand crank.
Also, a hand crank adds a neat interactive element to the machine. Whenever I show it to people,
they very much enjoy cranking it, and keep on cranking it. Its just so much fun and so fascinating
to watch the chaos of marbles tumbling and falling.
34 second video of marble machine 2 in action: The sound of Marble Machine 2. (160k mp3)
|
