[JBV]

I'm interested in building a Stewart platform (Hexapod). So i started thinking about some technical solutions which would be possible to build accurate without to much expense!

Would be nice with comments and suggestions! I choose to go with brass bushings instead of bearings because of keeping the price down, the adjustable nut also acts as a guide and bushing!

I have not drawn the nuts to secure the threaded rod to the endbushing (left side), the motor mounting plate or the pulleys. The mounting to the frame and the spindle table is a later problem to solve!

[SimonArthur]

I definitely like the idea.

How will you get the smaller tube to fit exactly inside the larger tube? Do you have a big enough lathe to work both pieces?

My current solution for my hexapod isn't especially good; it has a lot of play in the actuator.

I use Acme nuts from Dumpster CNC:
http://gonebowlin.com/dumpsterCNC/

My hexapod software:
http://robofac.sourceforge.net/

CNCZone Thread:
http://www.cnczone.com/forums/showthread.php?t=7467

[JBV]

I have been thinging about Honing the inside of the bigger tube and manually polish the outside of the inner tube. Perhaps a little gap between them and using another bushing at the open end. So the tubes only contact by brass at the nut and the end.

I'm also thinking about building a corner piece with the angles of a octahedron (60 and 70.53 degrees) Then ju just need to make 6 of these and 12 tubes of exactly the same lenght to get a perfect frame!

[JBV]

The corner would be something like this! The rod in the middle of the corner would be for attachment of the actuators (and the worktable on the lower corners!)

It would be hard to close the assembly as the corner is shown now. You would have to cut some of the tubes in half an weld them together later. I have also been thinking about spliting the four "female" connecting tubes in the corner in half and bolt them together over the frame tubes!

Hmm would it be hard to make both the spindle and the table movable? (Software, not mechanical that is!) Would give HUGE envelope, especially angular!

The 2 last images show how you could solve the fit of the tubes a little better!

/Jay

[SimonArthur]

You could just substitute two pieces of threaded rod for the final frame tube. You could then tighten the frame with a turnbuckle.

[JBV]

Check out my simulation of a hexapod with both table and spindle movable
http://www.cnczone.com/forums/showthread.php?t=13530

[miljnor]

you can buy prehoned tubing ID for the outer tubes and get the inner tubes ground. If you don't grind the inner tubes you will probably have some binding.

tubes typicaly aren't made perfectly round and they are usually bowed quite abit. You can get outer tube preground to but it is less common and I doubt you will find a matching outer. (but wouldnt that be cool)

If your using bushings in the outer tube you don't even have to worry about the inside diameter finish. so any straight tube will work.

looks like a good design although abit complicated on the ballnut area. Should work great.

That video at source forge was awsome! There is less metal in that than My mill and its bigger and looks to be more ridgid. I will definitely have to make one of these for myself! Hopefully the software is up to snuff for the 5 axis stuff.

[JBV]

New design! I had little to do so i made this just to try out the principle!
The arm consist only of a big trapezoidal screw with a slot cut in it (to keep it from rotating) and a gimbal bearing at the end (spindle platform).

In the housing, there is a tube with a pulley that is tightly fixed between 2 ballbearings. The tube holds 2 nuts, one is fastened and the other is able to slide in the tube but is not able to rotate, between the nuts there is a strong spring who pushes them apart to take out backlash.

The bearing retainers have a small "ridge" that keeps the screw from rotating.
You would have to make a gimbal bearing on the end of the housing connecting to the frame ("hanging" from the frame i guess).

Think this would work? Would the screw be rigid enough? I was thinking about 20mm (~3/4 inch) diameter or so. You would get great envelope of the actuator (and the hexapod!) as it can retract to the lenght of the housing and extend to whatever lenght of screw you choose! Also it would be easy to put an encoder in the housing for feedback to the software!

/Jay

[SimonArthur]

Really excellent, Jay. I'm using 1/2" (12 mm) threaded rod on my hexapod, and it's OK, but if I had to start again, I think 20 mm would be much better.

Another advantage to this kind of setup is that there is much lower rotational inertia because you're not spinning the heavy lead screw, just the light nuts. This means that you can start and stop movement much more easily.

The main downside I see is that it has a lot of parts and you'd have to make 12 of these assemblies to make your final design work.

I'm curious: what is your background? What kinds of things to you hope to build with your CNC machine?

[JBV]

Okay! I'll have to build one to do some testing on later when i have bought and converted an ordinary mill & lathe!

Well my background is a little odd! I'm 25 years old, flunked out of gymnasium (something like senior high school?) But in some strange way got work at the countrys largest villa manufacturer, doing house plans and garden planning. And most recently i worked for Microsoft personal support! I'm also educated welder. I don't have the dicipline to work at the same place for to long! I think i'm ment to start a business

Actually i'm planning on starting a company with a friend in a year or so. We will mostly be developing products in biocomposites and -plastics! (He is a civil engineer in chemistry) But we will also research and develop some other things, we have got hold of a patent of a very interesting turbine (Some tests in the early 80s at a local university showed very promising results)

Further i have lots of other ideas i would love to try out! But as now i don't even have tools for wood! I'm also planning on building an vacuum induction furnace to be able to make and cast exotic alloys and metals! Also its nice to be able to reuse the metal of finished experiments and from other sources!

What do you use your equipment for?

/Jay

[SimonArthur]

Very interesting. I use my home shop mostly for building fighting robots.

[Evodyne]

JBV,

Awesome diagrams.

Very interesting. I use my home shop mostly for building fighting robots.

Also cool. My whole interest in CNC is to be able to build kit robotic stuff to support AI software I'd like to sell. That's where I'd really like to see the combat robotics go-autonomous bots with survival instincts. No human intervention. There are online simulations of such, but the real thing would be oh so much better.

Had anyone considered (back to the hexapod) using a rack and pinion setup? Something along the lines of a piece of linear gear rack bolted to an extendable beam or slide. It cound be driven by a worm gear (to eliminate backdriving). The worm could be made into two, spring loaded, sections to eliminate lash. Again, a toothed belt and pulley arrangement would connect motor to worm and give extra reduction as well.

Lance

[JBV]

Thnx! Yes it would be awesome with fighting AI robots! But i guess it would be hard to solve the interpretation the input (feedback) from the sensors! Would you use neural networks? Very interesting stuff!

Wouldn't a rack and pinion be very slow and heavy? I have no experience of them but i just imagine that

I just began a new design! (Much better looking, yep i want a really cool machine to show off with ) This one would have a encoder on the rotating nut assembly to get closed loop and greater control in the software. I replaced the belt and pulleys with a bevel gear and pinion (1:2 ratio) Really just for the looks Think this would cause any problems with backlash? Even if i used an encoder of 400 ppr or so? I wouldn't let the software "chase" the exact point like a open loop system.

/Jay

[miljnor]

desing looks kool!

But.....
Bevel gears, like that of a car differential, are inefficient and transmitt a lot less power than just using a direct connection. I don't know what the exact loss is but its pretty significant (like >30%).

The only reason most machines use something like that is to change directions on rotating shafts. I believe every 90 degrees you go thru has the potential for droping 30-50% of you torque.

as I said, not sure on the percentage but its very significant.

Although for coolness I would go with that desing

[skippy]

Lance, I don't want to sidetrack the thread but regarding combat robots check out this one: http://www.mecsoft.com/Mec/Products/Case/kondo.html
Phil

[JBV]

Yeah i know its less efficient and has more backlash than belt and pulleys. But i like the idea of a closed unit You could always get bigger steppers

[smarbaga]

are these attachments drawn with unigraph or some other program,
if other , are the dimensions included in the drawing ?
thank you

[Xerxes]

Looks relatively complex to me. There are simpler linear actuators like these in attachments. Might be hard to find though.

Photos by Jiri Kuukasjärvi

[JBV]

Whats the fun with simple? Commercial units with 0.75 meter stroke or so would be very expensive i belive!

The dimensions is just approximate until i have bought real components!

/Jay

[Xerxes]

Then how about hydraulic hexapod?

Build 12 identical cylinders. Six of them are in the machine and the another six are driven by steppers \w screws to produce a hydraulic positioning. It might have some backlash caused by rubber seals and oil compression but maybe that is not something to worry about. I might try to build one some day