A new, smaller version of The Spice Must Flow with servomotors: Arrakis

I’ve started building a smaller version of The Spice Must Flow to use as a coffee table. I haven’t chosen a name for it yet. This one is 120cm long x 83 cm wide, with glass top. The mechanism is being kept near the floor so the top can be higher above it and hopefully not require as frequent cleaning as the original. I am using the same 45 mm square t-slot for the frames.


I have built the top cover and XY frame, and mounted the electronics. I am in the process of remaking the cables to the motors (MUCH shorter now) and redesigning the Y axis bearing/pulley blocks to be similar but a little less fragile than the previous generation. I printed one of them and will print the other tonight. The new bearing block is printed in two pieces without support material and clamps down on the 16 mm square tube X axis. I hope to have the mechanism up and running by this weekend, and then I’ll start on the new sandbox design.

One thing I’m currently stuck on is making the electrical connections to the LED strips in the sandbox. I’m going to try to set it up so that there will be contacts on the frame and on the sandbox and when you set the sandbox on the frame the connections will be made automatically- no loose cables, no connectors. I’m thinking of using magnets mounted on springs to allow some lateral error in the placement and the connections will find each other without any manual intervention.


I think for my next one I am going to try and mount everything in one corner so it could be hidden in a leg or something. The only other options I can up with is a folding tray like you have so it is easy to work on and then fold it up and out of the way 99% of the rest of the time.

Have you seen “spring test probes”? They would work well for making the connection to the lid for the lights. You could still put magnet or two nearby to make sure they align with the pins.

In my design, no local control panel is needed, so the electronics stays under the sandbox, out of sight. I use a Duet wifi enabled controller running as an access point so I can connect to it via my laptop or phone. I can even have it load a macro that calls pattern files on power-up, so if I take it to a public showing, I just set it down and power it up and it will start running patterns.


Yeah, I’m aware of those. I’ll have to do some experimenting…

1 Like

What is that? Like those things that happened in 2019?

1 Like

No motion yet, but I have almost all the mechanical work done on the mechanism. I have to mount the endstops and then I should be able to test it, hopefully tonight or tomorrow.

Business end of the mechanism with electronics and assembled X axis. There are two power supplies, 150 and 200W, a Duet WiFi controller with expansion board, and of course, two servomotors.
Google Photos

Magnet carriage- the magnet will be glued to the top using silicone. The blade(s) will be used to trigger optical endstops.
Google Photos

One of the two Y axis pulley blocks. Both have PTFE bearings that slide in the t-slot - one is screwed to the block, the other is mounted on pins and spring loaded. The blocks are printed in two pieces and the screws that hold them together clamp the X-axis tube between them.
Google Photos

Once it is working to my satisfaction I’ll get started on the sandbox. I’m considering covering the bottom of the sandbox with metal foil and grounding it electrically so the sand won’t accumulate a static charge that will tend to make it stick to the glass top.


How much torsion does the Y axis pulley exert on the slot? Any reason for PTFE instead of UMHW this time?

The Y axis pulley on the other end of the X axis twists the whole thing the other way, so there’s probably not much net torsion. Of course the X axis tube is being twisted, but it doesn’t seem to mind.

I was out of UHMW and happen to have a big brick of PTFE.

It’s all assembled but the motors aren’t spinning yet. I have some troubleshooting to do. One of the endstops is not working properly so I ordered more of them and should have them on Sunday.

1 Like

It’s running, and very quiet at lower speeds. But, or course, it can go super fast, too, with a bit more noise.
I had some problems with the controller board that are mostly sorted out now.

Arrakis test video #1
Arrakis test video #2
A tour of the Arrakis mechanism while it runs very quietly at 200 mm/sec.

Next step is to decide on the LEDs and make the electrical contact pads on the frame so the sandbox can just drop onto them. I may go with some RGB LEDs so I can change the LED colors to suit the mood or season or pattern. They will not be flashy and cycling while the machine runs- just static two colors or white. I’m trying to figure out if I can use some of the PWM fan outputs on the controller board to drive the LEDs.


The sleeper has awakened! Please release your STLs so we may learn the weirding way!

I’ll do better than that- I’ll post a link to the CAD file. The STLs wouldn’t be of much use because they are specific to the hardware I used, but with the CAD file you’ll be able to modify my designs to fit your hardware. I’m using Fusion 360 to model everything.

The t-slot frame is 45 mm square with 10 mm slots (Bosch?), not a common size, I think. The X axis tube is 16 mm square. All that stuff came from a local scrap yard. The springs on one of the Y axis bearings came from a box at the makerspace. I don’t have any part numbers to order that stuff, but if you’re reading forums like this you probably don’t need a BOM and can modify the design to suit whatever hardware you have.

One thing I haven’t tested yet is whether the magnet will interact with the X axis tube and create drag on the mechanism at high speeds/accelerations. It didn’t seem to be a problem in the previous design, but I think the magnet was further from the tube.

This is what the magnet carriage looks like inside:

In the photo, the printed part on the X axis tube is the top and the other printed part is the bottom. I quickly realized that was going to be an assembly nightmare, so I changed it so the screws come up through the bottom part and screw into the top part (which means they want to fall out of the holes while you’re trying to assemble it…). I switched to a single endstop flag and moved it to the top piece as you can see in the videos. There are four 13 x 30 x 5 mm PTFE bearings fit into slots in the top and bottom pieces, and I control the pressure they apply to the X axis tube by shims cut from a beer can. Even with the changes, assembling the carriage is a little tricky because of the way the belts have to be looped around the screws. Having 3 hands would help…


I recently started working on a Sandtable similar to yours, although I’m using 4040 extrusions:

( I don’t know yet if i will keep the hairpin legs, but for now they will work.)
Next step would be the design of motor mounts and pulley blocks etc. Since you wanted to publish your files, I would like to use these as a basis. Do you have a preliminary version of those files I could work with?

Here’s a link to the fusion360 archive: https://drive.google.com/file/d/1g-tXNyelhquofqvxE29muj0ySXJvfP3-/view?usp=sharing

Here’s a link to the STEP file: https://drive.google.com/file/d/1HBe5G30wzYMbKCpXduIP7puPRvg7COm0/view?usp=sharing

Fusion360 no longer allows sharing files on line like it used to, so the above links are current snapshots of the design. They won’t update automatically as I change or add to the design, so I’ll try to remember to refresh those files as changes are made.

The motor mounts and corner pulley blocks should be useable without too much effort- you’ll have to change the size of the tangs that fit into the t-slot to match your t-slot. Likewise the Y axis pulley blocks and PTFE bearings. The files show both bearings as spring loaded, but only one of them is- the other is screwed down tight.

I tried to print you design, but tangs canno’t fit inside my rails, tried to change it with freecad but can’t figure how to do it neither.

As said on youtube, really nice design :slight_smile:

Progress on the Arrakis table has slowed recently due to some issues going on at home, but I’m back to working on the table again.

If you’ve been following my work you know I’ve been trying to minimize noise even as I maximize speed. I recently built (but not finished) the shell of the sand box (baltic birch plywood, 5/8" sides and 1/4" bottom) and ran some tests to see what the noise level is like. There is an air gap between the magnet and the bottom of the sand box, so I can report that that has significantly reduced the noise level. The mere fact that the sandbox covers the mechanism also reduces noise. The sand box is resting on dense foam rubber pads on the corners of the mechanism’s frame to try to minimize coupling vibrations from the mechanism to the sand box.

I ran a test with a ball and found that the steel ball rolling on the plywood is quite noisy, especially when it goes across the grain of the wood. Of course, in normal use the ball will be plowing through the sand which will reduce or, at least, change the noise, but if the sand layer is thin, you’re going to hear the ball. So I got the idea to try covering the bottom of the sand box with some sort of rubber sheet. But I didn’t have any rubber sheet in stock, so I tried the next best thing- a rubber covered steel ball from a very old computer mouse.

The result is impressive. The rubber coated ball is much quieter than the steel ball. Video here- rubber coated mouse ball first, then the steel ball. The machine is programmed to run at 500 mm/sec, but acceleration is set to 2500 mm/sec^2 so actual ball speed in this pattern is about 200-300 mm/sec. My camera’s focus servo is a bit noisy…

A quick web search finds many sources for rubber coated steel balls in different sizes.

It remains to be seen whether a rubber coated ball will generate static electricity that causes the sand to stick to or be repelled by the ball. I am looking at rubber coatings for the bottom of the table, too. EPDM rubber pond liner is readily available and pretty cheap.


There was another thread talking about lining the sandbox with fabric. I believe the Sisyphus tables use it.

I’ve been experimenting with more noise reduction. I noticed that the ball rolling on the plywood bottom of the sandbox was pretty noisy at 500+ mm/sec, so first I tested a mouse ball (rubber coated steel) and found it much quieter than using a plain steel ball. Video here: https://www.youtube.com/watch?v=RL6Qxa-Tfco

Then I got a piece of 1.4mm thick EPDM rubber roofing membrane and laid it in the bottom of the sandbox. It quiets things down nicely. Video here: https://www.youtube.com/watch?v=_Lw-PUZ-V0I

The rubber sheet has a few waves in it that should flatten out when I glue the rubber down to the bottom of the sandbox.

1 Like

Heffe did a test run and I was shocked and the difference in sound between the sand and the baking soda. BS is nearly silent, sand was loud and almost amplified…to me at least.

1 Like

Good information- I started with baking soda after seeing some reviews of white beach sand being full of bugs. I guess I got lucky on that.

I removed the plywood from the sand box to glue in the rubber sheet and found that as the glue set up, the plywood has acquired a distinct upward arch. When it goes back in the box the screws should hold it down flat.

I am planning to paint the inside of the box black in the next couple days, then reinstall the plywood, and then test it with some baking soda.