My Motion Toy - yet another Stewart platform piston design

Another 6DOF motion platform design with cost being a notable design goal. But also actuator range and speed.

I don't have welding skills, but I have a 3d printer and a CNC that can cut wood. (And without the CNC, one could use Send Cut Send or OSH Cut.)

Still working on refining my piston design before building all of them.

Rough tour:

• AASD-type servos, see this thread for more info.
• Cut plywood and all thread to make the base. These takes the strong motion forces and routes them around the coupling and servo.
• 1616 ball screw. My calculations indicate I need to use 1616 or 2020 to reach my target speeds and forces.
• Angular contact bearings with 3d printed alignment/base plates transfer the ball screw forces to the plywood.
• 2040 aluminum extrusion and 1½" EMT tubing for the sliding parts. Strut channel doesn't have good enough tolerances.
• The 3d printed piece on the ball nut has a ton of heat set inserts for every screw and holds the guide wheels that engage with the aluminum extrusion. The EMT presses directly into the part for positive forces and two screws through the EMT hold it on for negative forces. (Ignore the rectangular cuts on the side of the EMT.)
• At the top I have three polyurethane-coated bearings to hold the EMT in alignment and another plywood plate.
• An anti-wobble part clamps onto a bearing at the end of the ball screw. It had little arms that were meant to flex, but they broke off easily. I probably should just reprint it without them at a larger radius.

Thoughts:

• Still need to figure out my u-joints. Pictured here with a stub and something to hook weight to for testing.
• Getting the right distances on the all thread to get the plywood plates parallel is a bit fiddly, but not awful.
• Thrust bearings (the kind that comes in 3 parts) are garbage. I'm having much better results with angular contact bearings.
• Heat set inserts are nice.
• The guide wheels between the ball nut and extrusion are presently a bit loose, in theory I can adjust it by rotating the off-center inserts inside the bearing, but in practice it's not so easy and maybe doesn't need to be perfect.
• I don't like how the bearings at the top work. It's fiddly to put together and adjust. Then, the print is weak enough that the bearings can bend away enough that things still rub. A simple plastic bushing didn't appear to be enough though. Ideas welcome.

I like the idea, but unfortunately not the result.

The lateral play of your guide determines the noise and stability of your actuator, and unfortunately I see some serious design flaws there. I don't even want to think about the noise at the moment.


Ideally, the printed parts should be pressed onto your tube to ensure proper force transmission. What is the wall thickness of the tube? The wall thicknesses of your printed parts are clearly too small, and from a design perspective, the bearings can move far too much because the plastic will always flex. The bearings must be guided on both sides; fastening on only one side, and then only with a heat insert, will fail within a short time! Please note the forces acting on it when the actuator is almost fully extended! I would design the printed parts significantly larger with a closed body on the outside so that the bearings cannot flex outward under lateral stress.

And I very much doubt that the wood on the motor side can withstand the stress. Please take this as constructive criticism; I certainly don't want to discredit the whole thing. My motion rig also consists of many self-designed printed parts, but you shouldn't underestimate the load.


For Addition, please think about the design, small radiuses will tend to brake more and always think about how you have to print them at the same time. A printed part has the weakest properties when it comes to layer adhesion, unfortunately this is what is most stressed in almost all your printed parts.

Have you tried cutting plastic with your CNC machine? 3D printing is ideal for prototyping and trying things out. But as Misanthrop said, the parts are not very robust. Parts machined from solid PA (Nylon) or PC ("bullet proof glass") don't break as quickly as 3D printed parts.

Thanks for challenging my assumptions; not having more eyes on a project is a disadvantage of working alone.
I plan to put it through load testing next which should help show which things need a tweak or redesign!

Well, I did some testing. I'm kind of surprised at the results. (TL;DR: The piston did better than my testing rig.)

Here's my Sketch-o-meter:



Called so not because it measures how sketchy things are, but because it is a sketchy thing that measures stuff. It uses pulleys and wire rope to transfer the piston's pushing forces to a hanging hook I can apply weight to.

For reference, my design goals include the ability to accelerate the user upward at 1G and downward at -1G. According to my math I need about .67kN/piston for a 1G upward acceleration with the Stewart platform in a neutral position. (300lb payload minus losses and geometry across 6 pistons.)

In compression:

• With the piston stationary, no issues with ~285lbs, even with some bouncing. Roughly 1.2-1.3kN. (Servo disabled and brake applied, didn't see it rotate at all.)
• During testing, I started to tear off the handles on my static weight, oops.
• The upper bearings design needs a rework. The EMT conduit makes contact with the surroundings instead of always riding on the bearings. And it squeaks.
• Seeing a lot of wobble at the bottom end of the piston as the servo turns. I wonder if my ball screw or servo shaft is bent. In theory, the coupler should be eating any minor misalignment issues...
• Running with a slow sine wave for motion I got to around 500N force before my test rig bound up and I stopped the test.
• Found this after, probably from the test rig binding up:
  
  My math says the piston with this servo should stall somewhere around 1.4-1.8kN.

Then I rearranged things for testing the piston in tension. Aside: the piston weighs about 20lbs.

• To my surprise, it didn't have issues holding 1.3kN force while stationary. The servo brake did start to slip a bit, though (that's fine). I figured the core piece around the ball nut that holds the EMT conduit on with only two screws had a decent chance to delaminate on the printed layer lines.
  
  
• With a slow sine wave, it worked quite well up to 1.2kN. Just picked me and the added weight up.
  
  
• The plywood plates held fine. There is a little flex (the gap on the servo coupler changes when under load), so I may switch to thicker plywood.
• The little bearings and wheels held on with M4 screws and heat-set inserts are holding up mostly fine?! I figure I'd have issues with the wheels getting pushed into the adjacent wall and binding up, but they appear to rotate fine under load.

Also a grab bag of thoughts/responses:

• There is more lateral play that I'd like, but I'm also limiting travel to not use the last 6 or so inches right now, so it's not as bad as it should be.
• The EMT conduit wall thickness is a bit under 2mm. The measurements I'm using for the EMT are 44.5mm OD, 41mm ID.
• The heat set inserts I'm using for the guide wheel bearings are longer than average, but I'm kinda surprised they do as well as they do.
• I could cut plastic on my CNC, or even print fancier materials than ABS. There's also this sort of method to help a 3d print hold together better.
• If I did redesign the ball nut part, I'd probably cut something from wood to hold the bearings better. Having tested it and not seeing it woefully insufficient, I'm tempted to leave it as-is for now so I can work on other aspects of my simulator. (And probably have to tear things apart and retrofit a redesign later, but, that, likely after I have a functioning sim with some hours on it.) As they say, the greatest performance improvement is the transition from a non-working state to a working state...

At your Test you have only a static force and no sideforces which will you get, when your rig will move the first time and the main forces will be in compression, not in tensile.

Yes, but 3D-printed parts usually break much easier under tensile load than under pressure. So it makes sense to test the worst case by pulling instead of pushing.

What he tested is not the worst case! He should worry about lateral forces. The fact that 3D printed parts can withstand the stresses and strains has been proven here more than once, even if you don't want to admit it.

Well, ideally I would test it harder in compression than tension, my testing rig is just insufficient.

You're right, I haven't tested side loads on the piston. I was expecting them to be relatively negligible, just the weight and inertia of the piston itself, since the main loads are always axial on a Stewart platform. Do you have some quantitative suggestions for how much side force it should be able to withstand?

Not possible, because the design and weight of your rig and yourself are the main factors.