Does this count as a build?

I saw the Sisyphus – The Kinetic Art Table by Bruce Shapiro,

I want one…so I got out some of the prints I couldn’t sell and went to work. Good sunday project right? I hope Mr. Shapiro and his team don’t get irked. Come on though quick and dirty, ugly as heck but it works (kinda).

[attachment file=36712]

I think I need to try a different shape magnet and this super fine sand is kinda crunchy so maybe I’ll dump it and go to baking powder instead. It was fun, glad I did it. Goes to show the Mostly Printed CNC is pretty versatile, right?
[attachment file=36713]

So any suggestions for a single line art type program? Then I can just drop it into EstlCAM and do a engrave, or carve. It would keep the sand super level if all the art went from out to in then back out again. Pretty self leveling.

[attachment file=36714]


Uhh, we need more pictures of the underside!! I’ve been wanting to make something like this since I saw the first prototype table a couple years ago!

Under the skirt…you dirty dog, you.

[attachment file=36718]

This is the center assembly, both parts. The rest is the current build, no feet or locks, just corner bottoms pop riveted to the frame.

It is almost a standard build. I have a set of the second series of z guides in there (the ones right before the “better middle Z” parts were released. and a big fat magnet sitting on the bolt head. I’m thinking a smaller diameter stack might be better, or maybe a doughnut shape. Right now the ball kind off moves in little spurts, even on just a smooth glass surface.

Its ugly and doesn’t work well enough yet. But If I can iron out some bugs, I would love to make some new parts and do a full sized dinner table. How cool would that be, If I can get a good font, Have it draw each person’s name in front of their seat! While it keeps drawing in the “sand” though a meal.

Wow, cool! Damn. I wish I would have bought more clear plastic.

Hmmm. Might be a fun gcode simuator. I wonder if you’d like the design if you made it just 3D print your parts, drawing one layer at a time, but scaled up to fill the area. On second thought, you probably see that enough.

You could certainly do some fun things in estlcam, but you’re going to be limited to peel and parallel pockets for anything interesting. I guess you could try doing some engraving, but it won’t be able to lift the ball up, so at some point, estlcam is going to run travel moves all over your things.

Maybe there’s a good way to do the name thing by milling out everything but the name, with a parallel pocket. Do that for a rectangle in front of each person’s place. Oh, but it would do that from the outside in, and then get stuck. You might have some manual gcodes to get it from rectangle to rectangle. Even with the travel moves, I’m pretty sure 100% of the people would be impressed (and if not, then why are they eating with you?).

You could go to the bare metal and write up some gcode in python. It would be pretty easy if there was some math to just make interesting patterns, like a spirograph. There are some gcode python libraries. Even ones that will let you preview it, IIRC. It’s also not terribly hard to write the gcode after you figure out where the x, and y should be. If you have some mathy patterns you want drawn in gcode, send me an email and I’ll write it up if you’re too scared.

What speeds do you recommend? What size should I make it if I want to push aluminum sand around?

Any chance you’ll make a vid soon? I think that would help me convince my wife that I should make it, and that it should be ahead of fixing the sprinklers.

HAHAHA slow down, I read that like you didn’t take a breath between sentences…

I made another path in solidworks, I laid out a font and used the spline tool to make a single path that started and ended in the same corner in cursive. Exported as a DXF. Came out okay, but it did not let me cross the path so it wasn’t perfect. But it is pretty fun, I think using solidworks I could come up with some decent ones but nothing like the stuff for the Sisyphus table.

I haven’t done more than 10 minutes of research into designs, but I think they would need to be generated straight to gcode, not having to go through estlcam.

For speeds I was running slow…like always, then I tried 12mm/s, then I ran it again at 200% no issues. I want to try and get the ball to roll more smooth before I put too much coding time into it though. Right now it is chunky. I’ll order a new magnet stack and see if it helps now that I have an idea of load put on the ball.

I threw together a janky “will it work” build using duct tape and everything. turns out it is pretty cool. I would love to make some low profile parts for this. My chef buddy just came over for a beer and saw it, and just about lost his mind with ideas. Even if you didn’t have it running while you were around it but once a day it turned on and kicked out a new random pattern it would be so awesome. Like, come home from work and the dinner table is full of new randomly generated patterns. Hope they don’t have a patent on it cuz this is too cool for school.

Very nice, It would be therapeutic to watch it. I love kinetic art.

I don’t think they were the original. I remember seeing one a really long time ago.

And yeah, we got a new coffee machine, and I haven’t figured out the dosage yet.

Jim, It is pretty cool but a bit to chunky right now, if it were smooth and silent I can see myself zoning out for a few hours while it does its thing.

Hefe, That happens to me all the. A little extra in the filter and I am off the hook for a few hours.

If they weren’t the first ones I will put in some more time on this thing. It is worth it. I just don’t want to piss anyone off making a printed one… The key to this will be the design program though. It would need to handle different sized builds, speeds, (later add a time of day to change?), then some sort of randomness. Even if it was just a random seed/s (point or points to build to or from) then work out and only do travel moves at the borders. Do you think I could program this in c++, or what would you recommend I look into that might have an easily configurable GUI? It sounds doable to me, I would have to write out an outline or something to be sure.

Almost any language can do almost anything. C++ can definitely handle it, but if you’re not familiar with it, it can bite. I don’t know what kind of interface you’re thinking, but web would be the most flexible. Then you could attach with your phone or computer. I have a raspberry pi running my sprinklers and it’s just a Python program hosting a web page, with some jQuery to do the dynamic stuff. I also made a thermostat in a similar fashion:

That ran on a Yun, which has an Arduino with a serial connection to an embedded Linux computer (like a pi, but smaller, running openwrt). I ran the Python in Linux.

If you did it in C++, you’d make it simple and probably connect it to a web server with some protocol. The web server would be where the smart GUI went and how you’d connect would depend on how you made it.

Another choice is using the esp8266. There are some neat web servers running on them. I think there are even projects for 3D printers controlled by them, so it might be a good place to look.

Personally, I would do Python. There is so much web support, and text support, and you can easy make plots to debug your algorithms.

I am hijacking a little here. I love the project, and it reminded me of an Axidraw video I saw.

They draw in the sand with a bamboo skewer, and I though it would make a great sign for a beach cafe or something. Is there any way to ‘freeze’ the sand for this type of project? I imagine if you just sprayed epoxy on it it would be a fail.

1 Like

My wife had a great idea. She wants this, but with two nobs like on a large etch a sketch.

1 Like

OMG and etch-a-sketch would be epic let it run a “program” until manual intervention. We need to make this happen. Estlcam has the gamepad function, that would work immediately I am sure you could mess with the pinouts to hook up some knobs if you wanted them instead of the sticks. Time to solder up my protoneer shield.

James, maybe with some slow set epoxy it could work. There has got to be a few additives that would hold the design. MAybe encase it in a thin epoxy after the fact, thin enough to penetrate.

1 Like

I need this code

1 Like

Maybe the joystick to move the “cursor” and then a keyboard to write notes. Shopping lists, reminders, instructions to Maxwell Smart…

You definitely need that code. Since it says it’s using inkscape, I wonder if they literally drew the flower shape, then did some operation to rotate and shrink it 50 times? Actually, that would be interesting too, to make some code to put in a simple polygon, and have the code rotate and shrink (or enlarge) it over and over. That would be a decent prototype algorithm.

I wonder if you could mix the sand with some slow slow setting glue, and then draw on it before it sets up. I think something like the stuff you use for making jigsaw puzzles set up would work. It reminds me of water-based polyurethane, but they sell it specifically for puzzles. I bet you could water down poly and spray it on. It would really have to soak in. Another alternative is to just have the cafe make an MPCNC, and rewrite it every day when it gets ruined (the mandala approach) or you could carve the sign out of wood or small cell foam and then attach sand to it with spray adhesive.

1 Like

(((Responding to myself, I’ve gone infinite!)))

I could make a python script that would take in gcode for a polygon, and then turn it into an infinite spiral. I think that would be <100 lines.

1 Like

Try some kinetic sand. Easy enough to make with household supplies. That might hold it enough but you’d have to find a way to “erase” it.

Cool project!

The part that takes some math would be how to act when it hits a max/min dimension. So if it was doing a square in a rectangle it knows to limit the coordinates at 450 and 0 in the x axis and let the Y keep going until all 4 max and mins have been reached…I started to get serious about this last night. Using a single seed point for circles, 2 for rectangles, 3 for triangles is within my immediate coding abilities. Also having it start at the outside and work in, and then do another random from the inside and work out. That would keep all the sand even and always park the ball in the home corner.

Adding patterns like that video would just put this over the top.

Now I need to build a full sized one with all low profile parts.

My kid’s school makes play sand with: “4 cups whole wheat flour and ½ cup vegetable oil.”

When it hits the outside, it either stops (pattern over) or it can clip it to that extent.

next_coord = compute_something(…)

next_coord.x = min(MAX_X, max(MIN_X, next_coord.x))


if next_coord.x < MIN_X or next_coord.x > MAX_X:

Dang it, I am going to be obsessed with this now…there goes any free time I had this week.

Hefe is there something like the size calculators I use that would make a web based one easy for me? I’m not too scared (just lazy) to learn a new language but I think it would be more fun to get it up and running faster.