PDA

View Full Version : Hexapod designs?



chuckknigh
05-30-2003, 12:18 AM
Well, while surfing last night, I found a site that mentioned HEXAPOD designed 5 axis routers. Shortly afterwards I had managed to find a slew of information about these things.

Benefits are said to include no sliding parts, and simplified construction, but with the down side of increased processor power being required.

So, has anyone ever built one of these hexapod systems? They look ingenious. Supposedly they offer 5 or 6 degrees of freedom, and the equivalent to a 5 axis router system, but with the stability of a fixed spindle.

It's an interesting way to think "outside the box."

I've tried to include a picture from one of the sites, but unfortunately it's too big. Here's a URL.

http://www.hexel.com/hexabot.htm

-- Chuck Knight

jlinhart
05-30-2003, 12:38 AM
These are also known as stewart platforms (I guess 'Hexapod' is trademarked by someone). Here's a link to a guy that built one.

Stewart Platform (http://www.foxkid.net/cmm/platform/project-notebook.html)

Does anyone know of off-the-shelf or open source software that would be able to run a system like this?

jason

chuckknigh
05-30-2003, 01:28 AM
It appears that he is using an X10 control system to translate his designs into control signals. Crude, but obviously it works, from looking at his site. I'm guessing he wrote his own control software.

Other than that, the hexapod/Stewart platform design looks pretty straightforward and simple to build. Of course the Devil is in the details...and this design looks to have a lot of details.

-- Chuck Knight

ToyMaker
05-30-2003, 07:24 PM
I one time designed and built a hexapod ...

but it was a six-legged mobile robot :) .

Guess that doesn't count here though :(

robotic regards,

Tom

paulried
05-31-2003, 02:30 AM
I saw one of these last year that the IMTS show in Chicago but it was a whole lot bigger and the cutting tool was mounted from the pod and the table was fixed. It was from one of the Japanese machine companies but I can't remember which. It was fascinating to watch run and was very fast. I believe it was a cool 1/2 million!!

chuckknigh
05-31-2003, 02:47 AM
That's where homebuilding comes in!

Heck, if I had to buy a gantry style CNC router, instead of building one, it'd cost me several thousand dollars! We're homebuilders, on this forum...we don't pay retail.

-----------------------

So...back to my original question. Anyone ever played with one of these things? Built one? Modelled one? Thought about one? Seen software that could run one?

Surely I can't be the only one interested in this kind of thing.

-- Chuck Knight

chuckknigh
05-31-2003, 02:53 AM
The guy whose site is listed above...he's reported some amazing results with his. The accuracy of his is less than it could be, and he mounted his spindle on the Stewart Platform.

He reported, in his sweet zone (his most accurate, centered zone) that he was getting accuracy down to .002". He claimed overall accuracy was down to .005" all across his active area, and that was without antibacklash nuts, precision threads, etc. In fact, it was with allthread and X10 control boxes providing the interface.

He's rebuilding (version 2) with ballscrews and proper parts...I can only imagine what the accuracy of that thing will be.

Improved accuracy was supposed to be one of the benefits of this design...that errors don't add up across multiple planes, and the parallel, triangulated construction provided increased stiffness, further increasing accuracy.

This might be worth playing with!

-- Chuck Knight

EvanRobert
06-01-2003, 11:43 PM
Originally posted by jlinhart
These are also known as stewart platforms (I guess 'Hexapod' is trademarked by someone). Here's a link to a guy that built one.

Stewart Platform (http://www.foxkid.net/cmm/platform/project-notebook.html)

Does anyone know of off-the-shelf or open source software that would be able to run a system like this?

jason

I did a quick search and found a site that might have the information. Check out:

http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html

Good luck. I'd like to see more of the Stewart Platform.

Evan

Fish
06-02-2003, 12:24 PM
This sounds lilke a project that I would love to get involved in. Maybe a joint project to overcome the large R&D side? Anyone interested?

I have a disc from Okuma showing their product line-up. It shows a Hexapod milling machine in it, with a small description. I have never seen any other information from them on it.

A few years ago, there was a company marketing three different sizes of Hexapods with tool changing. I haven't heard anything since. They did claim accuracy that was within the expected range of your average machining centre. Their design used six ball screws, mounted in pairs to a single milling head platform. Each pair had one in tension and one in compression for increased rigidity. I will take a look through all my old stuff to see if I still have any brochures. If you want, I can send you the info on the Okuma version from their CD.

Adam

chuckknigh
06-02-2003, 08:30 PM
I'm not sure there would be a large R&D side to the project. The mechanicals look relatively straightforward, and there is documentation of at least 1 homebuilt unit already, linked at the top of this thread.

And, on a related matter, a linear actuator is a relatively simple thing to build...even considering accuracy and backlash. Since imprecision is not cumulative, like on a gantry machine, the individual members could conceivably be less accurate, and consequently cheaper. (And, easier to homebuild, for someone without a complete metal shop in his garage) For reference, the guy whose site is linked above, got .002" resolution using allthread and hardware store angle iron, on his prototype unit. He's rebuilding it for much higher accuracy, but .002" is pretty darned good!

My concern is control...turning standardized coding into control sequences for one of these machines. Surely, since there are machines available commercially, *someone* has come up with a software library dealing their control. If not, this thing was designed in the slide-rule days...how hard could its control algorithms be to understand? :-)

The only real problem area for a homebuilder would probably be the ball joint. The ball joint requires a rather significant degree of freedom, and no backlash. One common solution I've seen on the attempted homebuilt units, is to use a universal joint. It would provide the level of precision necessary, but has all the grace and elegance of a baseball bat. Surely there is a better solution available. What we would need is the ball joint equivalent to Cranky's rollerblade bearings on gas pipe solution as linear bearings...it's the only really precision part we would need, and with 6 struts requiring one for each end, that's 12 ball joints for a single machine.

I just accessed McMaster-Carr's site, and they're not horribly expensive, but in this type of endeavor, cheaper is almost always better. These may be cheap enough, though I wonder about their accuracy?

http://www.mcmaster.com/catalog/109/html/1043.html

-- Chuck Knight

ToyMaker
06-03-2003, 09:41 AM
Panavise, who build electronics holding fixtures, uses a ball joint style mechanism in their vise bases that looks like it could be reproduced in a home-shop. And it has that "elegant simplicity" you referred to.

robotic regards,

Tom

anoel
06-03-2003, 12:11 PM
Surely to goodness a simple ball joint from an autoparts store. would suffice. Don't know which ones would work but all cars have them in the steering linkage. So, I'm sure it could be cheap.

chuckknigh
06-15-2003, 02:06 AM
http://www.iwf.bepr.ethz.ch/web/de/forschung/wzm/hexa1.shtml

http://www.isw.uni-stuttgart.de/personen/t_franit/modell/index.html

Sorry for the language differences (for most of us) but these sites appear to be excellent sources of information for hexapod style machines. The hexaglide looks quite promising, and the Modell series of machines appears to be based on this principle.

Basically instead of extensible legs that are fixed at each end, and which change in length dynamically, they simplified the design. The legs are of fixed length, and the position of the ends of the legs changes...it's mounted on a glide! Brilliant, and simple...elegant, even.

These look quite home-buildable, to me. So, is anyone else interested in this kind of thing? I'm *building* a traditional 3-axis machine, but I'm *dreaming* of a hexapod.

-- Chuck Knight

P.S. The software end of it is taken care of with EMC, a LINUX based program.

http://www.linuxcnc.org/

Apparently it supports hexapods, directly! Hooray!

cbcnc
06-15-2003, 03:59 AM
From my reading of John Storr's site the Hexpod design was developed thru NIST who also developed EMC.

Chris

Jmtwo
07-18-2003, 03:35 AM
What about turbocnc as software since it controls up to 8 axis?:cool:

jimini
07-29-2003, 02:25 AM
Here's a few more machine sites.. Impressive.

http://www.ifw.uni-hannover.de/robotool/pages/listeenglisch.htm

High Seas
12-16-2003, 09:49 AM
Though I'd post here as most of the links seem to be in this thread.
http://biotsavart.tripod.com/hexapod.htm
Its a paper from the American University in Cairo.
Its a 6 DOF machine (if you take in space AND TIME to consider and build - maybe more? get all 4 dimensions eh?)
:cheers: Jim

Actually these are the basis for all large multiaxis flight simulators - getting rigidity, accuracy, and strength is only a matter of: (man)power, materials, and money.

Klox
12-17-2003, 03:06 AM
This Hexapod / Stewart Platform design is amazing!!! I can think of lots of stuff to cut on this type of machine.......

Klox

Klox
12-18-2003, 09:35 AM
Guys,
I made an enquiery to Hexel. Mr. Michael Fortier said price range is $85000!!!
He said;"They can be programmed using standard G-Code as well as just an ASCII stream of X,Y,Z,I,J,K data. For programming tools any CAM system should suffice and it should be easy to use for anyone with a basic CNC background."

Chucknight i'm with you on this one, i'm also dreaming....LOL!
Thanx for showing us this interesting concept!

Klox

chuckknigh
12-18-2003, 12:25 PM
Let me get this straight... They're asking $85000 for a benchtop machine, and *we're* the dreamers? ;-)

To be fair, no commercial CNC machine is cheap...but the design seems simple enough to build with relative ease. I don't think the software would be a major hurdle, either, once the math is figured out.

-- Chuck Knight

Klox
12-19-2003, 02:47 AM
There's a company near me that uses the same principal to manufacture flight simulation equipment.....
I'm going to pay them a visit January.
I'll see what i can come up with, even if it's only a fact finding mission LOL!

Klox

biomed_eng
12-19-2003, 10:29 PM
Like that StarWars ride at Disneyland....???

samualt
12-21-2003, 09:09 PM
Interesting, but those big hydraulic struts would cost a fortune. Precision valves, motors, all kinds of stuff. Hydraulics ain't cheap.

Electric Muscle:
However, I read an article in Scientific American (http://www.sciam.com/article.cfm?articleID=0003FCFC-BB3C-1F5C-905980A84189EEDF&catID=2) about artificial muscle. It seems many types of plastic can be used as 'Electric Muscle'. It might be fun to make a mini hexapod machine using that and a Dremal. The only problem would be it's not really precision so you would have to have some very precision sensors to determine the power you send to them and their positioning.
Perhaps a mini 1' tall hexapod could be done with a minimum of electronics, and no motors, acme screws, messy hydraulics, etc...
However, as far as I know no one is selling the 'Electric Muscle' yet. And, you would have to use allot of them because they don't have a great deal of power.

Air Muscle:
To get even more weird, I know for a fact you can make 'Air Muscles' for super cheap. They run on compressed air and look pretty cool. Of course, you would have to use allot of them because they don't have a great deal of power.
Imagesco (http://www.imagesco.com/catalog/airmuscle/AirMuscle.html)
Shadow Robot Co (http://www.shadow.org.uk/products/airmuscles.shtml)
Chris's Air Muscle page (http://mapage.noos.fr/chrismaker/air-muscle.htm)
Robot Store UK (http://www.robotstoreuk.com/ACTUATORS/AIR-MUSCLE/Shadow-1.htm)

Air Actuators:
Or, why not just use simple 'Air Actuators'. I don't know much about them but they couldn't be too expensive. Although, again, they don't have allot of power.

Satellite Dish Actuators:
The BUDs (Big Ugly Dish - from 4'-15' diameter) have powerful actuators on them. They arn't very fast but they come ready to mount, with an encoder for positioning, have a long throw, and are pretty powerful. You could make a large hexabot! Plug-n-play baby! And, since everyone is moving to cable or the little dish you can sometimes pick up whole systems for nothing.
I think this would be the best idea for a quick, easy to do, hexapod. Get 6 actuators of the same make and your ready to rock!

Just throwing some ideas out there!
:D

chuckknigh
12-23-2003, 12:27 AM
Satellite dish actuators are an EXCELLENT idea. I wonder if they've shown up at BGMicro, yet? (Mail order electronics surplus, that's located in Dallas)

I know they have a "windshield wiper" style system available, right now, and the price is right. I'm not sure if it's just an "all or nothing" type of actuator, or if it's controllable.

As for the air muscles...they're an interesting technology, but they're EXPENSIVE for what they are. They are nothing more than a Chinese finger puzzle, with a balloon inside, and a few pneumatic fittings. That's IT! I had some discussions with their inventors, a few years back, for another project...fascinating and really brilliant idea. They'd managed to increase the pull to several hundred pounds, for a muscle you could hold in the palm of your hand.

-- Chuck Knight

foamcutter
12-31-2003, 10:02 PM
Ok guys,
I've been reading the replies on Hexapods. Doing some searching. Dang now you got me wanting to build one. I have experience with g-code on a three axis router. Have used DeskNC, and a g-code generator, turbocad and some other. Can write or edit g-code when necessary. Think I can copy and improvise design to build the machine. Here's where I get stuck. On a Hexapod we are working with 6 axis at the same time to make the machine do anything. How do you get from a CAD design to g-code? I know on a three axis machine the g-code generator (either me or software) reads the drawing and assigns x and y coordinates. I have been then going in and editing in the z axis coordinates. If I am understanding the Hexapod there would be imputs from all six axis at the same time to create each movement or function. As an example to move right one inch would take a correct input to all six axis together, each having it's own seperate distance and direction it must travel in order to move straight right 1 inch. Is there a g-code generator that understands to take the CAD drawing and convert it to the six axis of control? Any help would be greatly appreciated. Please just tell me where to go guys (maybe I said that wrong) to find the info I need. Thanks Ron

ger21
12-31-2003, 11:34 PM
I think most hexapods you see all use their own custom software. I don't believe there are any standard CAM packages that can generate code for one, but I could be wrong.?

Gerry

samualt
01-01-2004, 01:43 AM
foamcutter:
This site has some kind of software for Hexpods, downloadable. Their machine looks like an upside down hexapod.
LME Hexapod Machine (http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html)

Here is some more code for you:
Sandia Hexapod - Software Utilities (http://ppsc.pme.nthu.edu.tw/~tclee/mirror/Sandia/soft.html)

Try this page also. He seems very nice. I bet if you asked him he could help you:
Stewert Platform Notebook (http://www.foxkid.net/cmm/)

Or, just search Yahoo:
Yahoo - Keywords: Hexapod Software (http://search.yahoo.com/search?fr=fp-pull-web-t&p=hexapod+software)


Also, I would look up all the hexapod pages on the net and email every person involved. You might be suprised how much they help!

chuckknigh
01-01-2004, 02:04 AM
And, as I understand it, EMC (free controller package that runs under LINUX) has some limited support for hexapods.

That's my main concern...I'd LOVE to build one, but controlling it without predesigned software could be a real chore. I've never played with EMC, before...it's supposed to be an industrial type package.

-- Chuck Knight

chuckknigh
01-01-2004, 02:26 AM
On a Hexapod we are working with 6 axis at the same time to make the machine do anything. How do you get from a CAD design to g-code?

In exactly the same way as for any other machine.

Regardless of how many axes the machine has, or what is required to move it, g-code merely describes the *movement* desired...X axis should move +1.00 inches. That's it!

It's up to the controller software, to translate that into the instructions needed to move the machine.


If I am understanding the Hexapod there would be inputs from all six axis at the same time to create each movement or function.

Essentially correct, though I think there are some movements that could require less than 6 movements simultaneously.

This is a simple matter of doing some math (remember translations in geometry and algebra?) and creating some instructions from the resulting coordinates.

Once the equations are described, it's a fairly simple process to change it into the necessary pulses...the programming isn't inherently hard. The difficult part, until recently, was having enough computing horsepower to do it, in real time.


Is there a g-code generator that understands to take the CAD drawing and convert it to the six axis of control?

Nope...g-code just describes the movements. EMC translates them into the control sequences...

http://www.linuxcnc.org is the EMC home page, if memory serves. If I understand the process, correctly, it's just a matter of setup...telling it you have a hexabot, that sort of thing.

Take a look at http://www.hexel.com/rotobot.htm and see if you can figure out the math for this one. It makes my head hurt! They already have full software available, too, but they sell it *with* the machine, and the whole shebang costs US$85K. Yes, eighty-five-thousand dollars...

-- Chuck Knight

ger21
01-01-2004, 02:50 AM
In exactly the same way as for any other machine.

Regardless of how many axes the machine has, or what is required to move it, g-code merely describes the *movement* desired...X axis should move +1.00 inches. That's it!

It's up to the controller software, to translate that into the instructions needed to move the machine.


Not exactly. On a standard 3-axis machine, the X-axis (1 motor) only moves 1". On a hexapod, all 6 would have to move vs just the 1. If you look at the Sandia link above, their software translates what seems to be 3-axis g-code into the proprietary code that their custom controller runs on. I don't know how EMC handles the hexapod, but I'm guessing that for TurboCNC or Mach2, you'd use 6 linear axis'. Since the majority of CAM software writes g-code for 3, 4 or 5 axis machines, you could convert this code into what would essentially be just another g-code file, but it wouldn't really be readable like a standard g-code file is. Follow me? I bet if you ignored the 4th and 5th axis to start with, the math wouldn't be that difficult to translate x-y-z moves into the 6 linear moves required. It is, just a bunch of triangles, right?

Gerry

foamcutter
01-01-2004, 03:12 PM
Ok,
As I read your replies I get the idea that there may be actually two ways to control.
1. standard 3 axis g-code x,y,z would be fed to a controller software that would translate it into impulses to the six motors to make it move. This way your CAD drawing is just converted into g-code same as a standard 3 axis router. The controller software takes the g-code and translates it to the seperate pulses to each motor.

2. The mathmatical equasions are figured out to make each arm move the specified distance per rotation of each motor and these factors are used to create six axis g-code. ie: CAD design would be converted to g-code using the factors giving you six axis of g-code.
With my limited knowledge I'm thinking plan #1 would be the way to go if possible. Any thoughts?

Thanks for the sites to go see and input. I can see I'm going to be doing a lot of emailing, asking questions, and asking for help. Thanks guys, if you have any other ideas please let me know and I will keep you posted. Thanks Ron

bb99
01-01-2004, 05:06 PM
Any 3- to 6-axis CAM package will do, for example TurboCAD/CAM.

OK, now you ask yourself ďis this guy correct or cracked?Ē Well, g-code does NOT control your CNC machine. Itís the software that interprets the g-code and converts it into electrical pulses that controls the machine. So, itís the software that interprets the g-code that will see the command to move ďright 1 inchĒ and properly move the pods legs so that the point of contact between the material and tool moves 1 inch in a horizontal plane. Think about Mach2. It interprets the g-code and moves the CNC machine for you. The ďtrickĒ is to think of g-code as describing ďwhatĒ you want the machine to do while the software figures out ďhowĒ to do it.

OK, so the next question is ďwhat software to use?Ē I donít believe Mach2 can control a pod. Iím not aware of any general package that does. If you were to buy a pod, the s/w that comes with it will properly interpret the g-code. For the DIY CNCer, I think weíll have to write our own s/w. If done correctly, it should be able to handle a wide variation of pods built using similar design principles.

Iím a software engineer and have worked on hardware control systems (nothing to do with CNC, but the principles are the same). Iím willing to donate some of my time working on the controller s/w (g-code interpreter) if someone is willing to build the h/w.

--bb99

foamcutter
01-02-2004, 10:16 PM
bb99,
Thanks for the offer to create the interpit code. I spent most of newyears looking up pods on the net. There seems to be a couple places that talk about controller code. One is NIST who developed the EMC code using linux as a base. Another is LME Hexapod Machine, the Sandia site listed earlier in this thread, and the The Stewart Platform Notebook also listed earlier. I emailed each asking questions, haven't heard back yet but didn't expect a reply for a couple days because of the holiday.

By h/w I assume you are talking about the motor controllers (drivers). Are you thinking of something using a PIC chip? I saw a guy on ebay selling some controllers that he sells as a kit with PIC's either pre-programed or not. I will look again at them, try to get the specs and get an auction number for you to look up. I myself have a little knowledge about them, very little actually, but just enough to think maybe it would work. I can assemble electronics, have some basic knowledge but don't know how to design. Let me know what you think. Or would something like gecko drives or any of the numerous other ones out there work? I guess what I'm asking is do you think off the shelf drivers would work or do we need a custom driver? Ron

bb99
01-02-2004, 11:24 PM
foamcutter,

Iím purely a s/w guy. In part thatís why Iíve decided to build a CNC machine (I want to learn more about the h/w side). With my ignorance about h/w firmly in place, Iíll make a guess and say any 6-axis linear motion system will do. The pod is not linear motion, rather its rotary motion (at least the design presented at the beginning of this thread). However, that is not an issue since most of the motion control systems described here have no intrinsic linear knowledge. All they do is say ďspin the motorís shaft clockwise/counterclockwise X number of stepsĒ. For linear motion, that translates into Y number of linear inches moved. While for rotary motion, that will give you Y number of degrees of motion for a leg. The trick is to have the s/w produce the correct movements on the legs to produce the desired movement. As long as the motion control h/w can handle moving up to all 6 axes simultaneously, I donít see any reason not to use off-the-shelf h/w.

Itís all in the math!

--bb99

chuckknigh
01-03-2004, 12:59 AM
Precisely...the g-code says to move the X axis 1 inch, and the controller software interprets that, and outputs the step and direction pulses over the pin/s specified in the setup, for however many steps that requires.

It's all a matter of setup.

Once the hexapod motions are understood (that's where the math comes in) then writing a program to interpret the g-code, and output the pulses, is not that difficult.

But, like I said a while back, EMC has *some* support for hexapods, already. Why reinvent the wheel? It even has source code available...maybe a new "module" at most?

http://www.linuxcnc.org

-- Chuck Knight

bb99
01-03-2004, 04:44 PM
I agree that EMC is the most promising platform for the home CNCer. My point is that the motion control h/w can and should be off-the-shelf. The issues that first need to be addressed revolve around the pod itself. For example, since the pod is symmetrical, limit and home switches can be associated with each leg rather then with the base. This would allow full 360 rotation of the pod table without any of the legs inadvertently tripping a limit switch.

As you said, itís all in the setup. But, that canít be defined until the geometry of the pod has been worked out.

--bb99

chuckknigh
01-05-2004, 01:25 AM
What level of detail do you need, from the hardware side of the project?

The basic geometry is simple...it's a standard octahedron, with some variable length struts. We have 2 possible configurations. (imagine triangles) One has 2 variable length legs per triangle, and a fixed base...this is a traditional hexabot. The other has a fixed leg length, and a variable base length...this is the rotobot.

With a traditional hexabot, "home" is fairly simple to define, as a return of all legs to some standardized length...the home switch could be within each leg assembly. I don't know if that's how they do it in industry, but it makes sense to me...

With a rotobot, it becomes a little more complicated...each leg has to return to a predefined position, probably equidistant, at 60 degrees apart. Same disclaimer as above.

Do you need any other information? I'd think anything machine specific, like the exact leg length, would be a variable...something to set up in a configuration file.

Seriously, think about it, and let us know the specifics of what you need... I've included a picture of a traditionally designed hexapod.

-- Chuck Knight

chuckknigh
01-05-2004, 02:01 AM
Something just occured to me. This is going to be hard to explain...forgive my ramblings.

Most of us have no real use for 5 or 6 DOF machines. The majority of our work is done in either 2 1/2 or 3D space.

So, it might be possible to simplify the design of the hexapod to a tetrapod.

Thinking about it, it's a fairly simple matter to take something like a tetrahedron and distort it within 3D...it wouldn't give control over such things as yaw, but how many of us would use that?

It might be a simpler initial project, and would require only 3 stepper controllers (instead of 6), if I have it figured right...the only thing I can't figure out is how to hold the cutter rigidly at the vertex of the tetrahedron. Maybe something like making each arm a parallelogram linkage, so that the end of the strut is always aligned as vertical? I know that'd work for something with a fixed strut length.

Am I missing something? I WELCOME input...

Let's see..basic regular Platonic solids are tetrahedron (4 sides), hexahedron (6 sides, cube), octahedron (8 sides), dodecahedron (12 sides), and icosahedron (15 sides). The hexahedron is not inherently stable...the 2 lowest order units, that are inherently stable, are the tetrahedron and the octahedron.

I might have to play with some models, to see what kind of movements I can get out of these things. If a tetrahedron based model would work, that'd be a MUCH simpler homebuilt project...not to mention a heck of a lot cheaper!

I'm virtually positive a "hexaglide" tetrahedron would work just fine...not sure if a tetrahedron with a fixed base (traditional design, like the picture above) could be done with ease in a home shop, though...and that's my goal.

-- Chuck Knight

bb99
01-05-2004, 06:05 PM
I Iíve been thinking of something like the rotobot but that isnít a requirement. Iíll begin to play with the math and let you know what I come up with.

--bb99

chuckknigh
01-05-2004, 10:16 PM
OK, this is the geometry I'm playing with, now. I'm tentatively calling it a tetrabot, after its base solid, the tetrahedron.

Compare it to the hexabot, above -- same triangular base, and the "tip" should be able to move by varying the lengths of the struts. The hard part will be keeping the router bit "vertical" but I think I've already solved that problem by using a parallelogram linkage.

I'll build a model, soon, and test out what is, at the moment, purely theoretical. It works in my mind's eye, though... ;-)

Hmmm...now that I've mentioned it publically, I wonder if I could still get a patent on it? Wouldn't mind being the patent holder for something simple, like this...

-- Chuck Knight

bb99
01-06-2004, 12:21 AM
Ahhh, and the software is working flawlessly in my mine!

OK, now down to the nuts and bolts... Iíve got a few questions for you.

1. I believe there are 6 edges in your drawing. Will all 6 be able to elongate/shrink in length?
2. If so, then how are the legs mounted to the base?
3. How is the table mounted to the legs?
4. Could you put a horizon line in your drawings and use some color scheme to differentiate between moving and non-moving parts?
5. The original design you shared had the rotation about the z-axis as being an integral, however with this design, Iím not seeing it. Can you explain?

--bb99

jimbo
01-06-2004, 12:44 AM
...

bb99
01-06-2004, 01:17 AM
Thanks for the links! Lot's of stuff to go through.

--bb99

chuckknigh
01-06-2004, 01:39 AM
Ahhh, and the software is working flawlessly in my mine!

:-) Exactly.

Before I build anything, I've found that it's always a good idea to create a dynamic model in my mind. Once I *understand* how it's supposed to work, and how the parts all interact, I can draw it, and then build it. It's just the way I design things...

On a related note, I'm guessing you have a basic framework for the software, already...it might not be formalized, but it's something you're working out in your mind.


OK, now down to the nuts and bolts... Iíve got a few questions for you.

1. I believe there are 6 edges in your drawing. Will all 6 be able to elongate/shrink in length?

There are 3 ways to do it...

First, keep the base triangle a fixed size. Make the upright legs of variable length, like one the hexabot platform a few messages back.

Second, keep the upright legs a fixed length, and vary the geometry of the base.

Third, make everything variable.

I think, for the purposes of my thought experiment, I'll focus on the first two. And, the problem of keeping the router orientation fixed is easier to solve on the variable base/fixed uprights version of the design.

I'm calling it the router end -- it is just the articulated "tip" of the machine. Conceivably, you could also mount a workpiece on it...and move it under a fixed spindle.

Have you seen the "hexaglide" design? It's an ingenious approach that uses fixed length struts, and a variable geometry "base" of sorts.

http://www.isw-uni-stuttgart.de/personen/t_franit/modell/e_primod1.html

I think it looks promising, even if it looks like a bit of a kludge. Very home-buildable, and a reasonable model for a 3-axis machine. I think the *6* drive screws are "in the way" in this particular implementation...3 screws won't be, though. And, it could be further modified by mounting the screws horizontally, such that the base of the tetrahedron would move in and out, rather than up and down.

It shows one of countless variations on the construction, though...and a particularly simple one, at that. The picture below is of this type of unit...they call it Modell, and it's based on the hexaglide designs.


2. If so, then how are the legs mounted to the base?
3. How is the table mounted to the legs?

These are some of those details that have to be ironed out, in physical models.

My initial thoughts are twofold, depending on which design works better.

In design 1, the base would be fixed, and the upright legs would be of variable length. They would be fixed to the base with a ball joint, or possibly a u-joint...same difference, in practical terms.

In design 2, the base would be variable, and the uprights would be of fixed length. They, too, would be connected to the framework with ball joints.

In both models, I see the machine "suspended" over a fixed workpiece, and holding a router.


4. Could you put a horizon line in your drawings and use some color scheme to differentiate between moving and non-moving parts?

Once I start doing my own drawings, I'd be delighted to do that. The diagrams, above, are just "clip art" off the web. I think the irregular tetrahedra were from the Math Forum.


5. The original design you shared had the rotation about the z-axis as being an integral, however with this design, Iím not seeing it. Can you explain?

Sure, I posted a rambling message last night, about a different idea I'd had...a simplification of the hexapod concept. By basing a machine on a tetrahedron, rather than an octahedron, we could create a 3 DOF machine with a space frame construction (which means averaged errors, rather than cumulative errors like on a traditional machine)...and do it with only 3 motors, 3 controllers, and 3 actuators...half the complexity, and half the cost of a hexabot. A hexapod is a 5 or 6 DOF machine (depending on whether it's a fixed base or a rotobot) and its capabilities would likely never be used by the home hobbiest.

It's just a different, but related idea that occurred to me, and now I'm trying to figure out the dynamics. A home-buildable, simplified, "tetrabot." At least, that's what I'm calling it, for the time being.

-- Chuck Knight

chuckknigh
01-06-2004, 01:49 AM
This is something closer to what I have envisioned...but this one uses 2 tetrahedra to allow the other degrees of freedom. Just imagine it, but with only 3 legs hanging down, and the router head held rigidly vertical.

In fact, this might be a good design on which to base it, as it could add other degrees of freedom by adding another tetrahedral module on top.

BTW, this is a commercial unit designed by Turin, and marketed by Motorola.

-- Chuck Knight

foamcutter
01-06-2004, 03:10 PM
Chuck,
Just a thought. Do you think that the "tetrabot" will be as rigid as a hexapod? Also how about using the tetrabot with a verticle rotating work surface for doing 3d? I have been also looking at the different designs and believe the fixed leg length with variable base would be easier and more rigid for the home builder like myself. I'm also thinking it might require less powerful motors to run it. Also wouldn't any backlash in the control threads running the variable base have less effect on the spindle than if the threads were directly in the legs to vary leg length?
You lose me on the parrallel mechanism to keep the spindle straight. Can you please clarify.

I have downloaded the EMC Manuel and am reading thru it. It does appear to have a part about 6 axis control.

I am a little slow to pick some of this up sometimes, but then all of a sudden it will dawn on me so please hang in there with my questions.
Ron

chuckknigh
01-06-2004, 07:03 PM
There are only a few basic platonic solids...the tetrahedron, with 4 faces; the hexahedron (cube) with 6 faces; the octahedron with 8 faces; the dodecahedron with 12 faces; and the icosahedron with 15 faces.

The tetrahedron and the octahedron are the two lowest order forms, which exhibit inherent stability.

Now, in English, that means that all their faces are triangulated...triangles make things stiff. A cube is a platonic solid...but none of the sides are triangulated, so it's not inherently stable.

The tetrabot would be at least as rigid as the hexabot, and possibly more rigid, if built to the same tolerances. The higher the number of "edges," the less severe the angle between the faces...less severe angles can result in play in the system.

This is an issue with geodesic domes, which when they get bigger and more complicated (more edges) allows them to sometimes "dimple" inwards. There's just not enough difference between adjacent faces, to keep them apart. The tetrahedron hs the most severe angle between the faces (dihedral angle) and is the strongest and most rigid of the potential structures.

As for the parallelogram linkage, it's a standard engineering solution. Ever seen a "Luxo" lamp? It uses 2 parallelograms to make sure the lamp "head" remains in the same position, regardless of how the arm is moved. Same idea...I want to keep the router in the same position (vertical) regardless of how the arm is moved.

I'll draw up some pictures, etc, and post them, when I get some time.

-- Chuck Knight

turmite
01-07-2004, 11:17 AM
Guys I am going to state my ignorance righ here right now, but for the life of me I can't see a real world need for something like the hexapod when it looks to me like anything you can machine with a pod cal already be machined with a 5 axis router that has plenty of software already available. Can someone please explain to me in layman's terms what the "real" advantage of a pod is over an good 5 or 6 axis std machine. Now on the other hand if this quest is because " I want to do it" I fully understand and wholeheartedly cheer you on. I just need a little help understanding the why!

Mike:confused:

buscht
01-07-2004, 12:56 PM
Here's. what they say at http://www.hexapods.net/hexapod.htm

Cost might be an advantage in the right application. I heard the number $80,000 being quoted for a hexapod to sit under a brideport type mill. Compare this to a comparable HAAS 4 or 5 axis mill and it may look attractive in the right circumstance.

I did some playing around trying to see how a person would make a G code translator and noticed something interesting. A 1" move in X, relates to much smaller moves in a hexapod. All 6 axis must move, but each one in a smaller amount than the 1" X move.

It seems to me that this would equate to a faster overall run time.

The hexapod is a very rigid platform as Chuck has mentioned throughout his postings. It might not matter when cutting wood, but in the brain surgery application shown on the hexapod website, it might be vital.

I would say that for 99% of existing applications, a hexpod is overkill, but its very cool and if more developers work on the design, overall costs will drop, applications be discovered, and it may become a mainstream mechanism for CNC work.

balsaman
01-07-2004, 02:10 PM
Chuck,

Wouldn't a hex with 3 movable axis (arms) in a triangle make just regular 3 axis motion (x, y, z)? If so then it's way more complex than a conventional 3 axis machine but offers no advantage. The 6 legged one creates the 4th and 5th axis and is the reason hex's exist in the first place....no?

Eric

Bloy2004
01-07-2004, 05:21 PM
but it's so elegant!

chuckknigh
01-07-2004, 07:51 PM
Well, of course it offers some advantages.

To start with, it's inherently stiff. We go to such pains to ensure rigidity in the gantry or moving table, on traditional machines...then we have to increase the weight, which increases the cost of the bearings and linear slides, which in turn requires us to increase the size of the machine...there has to be a better way? This might be it. After all, there are NO linear slides or bearings, on this type of machine...just ball joints.

A major advantage is that, with a parallel system, error is averaged out, rather than added together like in a tradtional machine. What that means, in practical terms, is that hardware store allthread could deliver the same accuracy level as a good quality ACME, and ACME could deliver accuracy rivalling ballscrews.

A tetrahedron based machine should be of similar cost and complexity to a "traditional gantry machine," since it uses the same number of actuators...hexabots use double the number. It might even be cheaper, since there is less "structure" involved, for larger volumes, and the parts can be of lower quality. Being "forgiving" is a major advantage for a homebuilt machine.

A possible advantage is a much greater working volume, without a tremendous increase in weight and cost, or a loss of rigidity. See above. Take a look at the Z-axis heights of the hexapod and othe rmachines that I linked to...it's HUGE!

It would also provide an "entry level" model into parallel kinematics...possibly even upgradable by adding a second tetrahedral module on top, like in one of my more recent pictures. That machine could have provided 5DOF, with ease.

And, of course, it's just so darned cool! :-) Hey, the geek factor is always a major player, in my world!

-- Chuck Knight

balsaman
01-07-2004, 08:08 PM
Well, it is cool.

I am not so sure it's more rigid. It also has a MUCH larger footprint for the amount of cutting area it has.

Don't let me discourage you. It's interesting to be sure!

Eric

chuckknigh
01-08-2004, 02:05 AM
Anyone got any ideas for how to turn 2 screws with 1 motor? I know I can do with gears, belts, etc...but these two screws would be changing angles relative to one another...possibly distances from one another.

I could do it with a "flex shaft" type coupling, but that wouldn't be a particularly good solution...lots of backlash, I would imagine.

BTW, I think I've solved two problems tonight...

First is that I figured out how to hold the router, rigidly vertical, while not compromising the rigidity of the tetrahedron.

Second, I figured out how to provide for upgradability to 5DOF...the 3DOF variation is just a "special case" of the 5DOF machine. It should make a good "Kleinbauer-approach" entry level design.

Of course, in the process of solving these two problems, I've created about 10 more. Engineering can be fun!

-- Chuck Knight

foamcutter
01-08-2004, 02:24 PM
Chuckknigh,
Two shafts one motor. How about using splined shaft into a socket for the change in distance with a universal joint to the splined shaft to take care of the angle change. Use belts or gears to run both shafts from motor. If you could engineer it right maybe the splined shaft could run through a bearing it could slide in and out of while rotating and the splined shaft would be the gear for the shaft running against the motor gear. Similiar to the way the drive shaft and velocity joint goes into the transmission on front wheel drive cars. Don't know if this makes sense to you if not let me know and I will try to draw a crude diagram. Just a thought Ron

chuckknigh
01-08-2004, 05:31 PM
That's a very good idea...and one I may have to borrow. If I understand it right, make a flexible joint that will telescope in and out, but which is held "together" by use of a spline. Makes sense.

Thank you. I have about a dozen possible ways to do it...all have tradeoffs, of course, including this one.

The ideal would be if I could turn the drive nuts, rather than the drive screws...that way they'd stay in basically the same place, but change only in angle. That, I could accomplish with any number of flexible joints...next time I'm in the hardware store, I'll have to play in the connectors aisle.

As to basic overall design, I think I'm going to start with the arrangement used by Turin...stacked tetrahedra. That should provide both ease of construction, and easy upgade to 5DOF at a later date. It's a good starting point.

As to the limited working volume that Eric mentioned, I think the limitation comes, not from the size of the machine, but the limited range of travel of the big linear actuators. I have an idea of how to address that limitation, on the cheap.

-- Chuck Knight

anoel
01-08-2004, 05:41 PM
Wouldn't you have to have one Rod LH threaded and the other RH threaded to make the gear thing work with one motor and two Screws??? Not a problem with belts but gears counter rotate.

High Seas
01-08-2004, 06:50 PM
Sounds like you could use something like a turnbuckle. RH and LH threads (have to work out the ball-joint ends), but you'd spin the body of the turnbuckle and get the in and out motion. The ball joints though - that might build in backlash.
Example: As you spin the body of a turnbuckle, the ends rotate too until they hit their stops and then continue in (or out) as the main body is turned. You could probably "capture" the movement like the heim joints on a suspension system though.
Just thinking this out as I'm typing here - not intentionlly mudding the effort. Seems like a rapid prototype could be built up to prove concept using timber and some turnbuckles - and no motors to check your movement theories for tetra, quadra, et all versions.
:cheers: Jim

chuckknigh
01-08-2004, 07:56 PM
Sounds like you could use something like a turnbuckle. RH and LH threads (have to work out the ball-joint ends), but you'd spin the body of the turnbuckle and get the in and out motion.

A good idea, but no. These two screw drives would not be in line, like a turnbuckle.

Here, take a look at this basic geometry -- two stacked tetrahedra. This image shows a 5DOF machine (used for LASER inteferometry...that tells you how accurate it is), and if you think about it, moving the upper and lower corresponding struts in unison, would keep the tool vertical. 3DOF is just a "special case" within this 5DOF machine.

Imagine if the top and bottom struts were just threaded rods, joined at the center with a small triangular plate...making 2 complete tetrahedra, sharing a "base." Now, just let the linear actuators be nothing more than threaded rods, and the nuts could be mounted at the side supports. We'd have to build in a little bit of tilt and swivel at the 3 corners, but that should give a tremendous amount of linear motion for each strut, at minimal cost.

Down side is simple...the "unused" portion of the threaded rod will "stick out" the side of the machine. But, it should allow *all* of the interior volume of the machine to be accessed, if it works the way I think it's going to work.



Just thinking this out as I'm typing here - not intentionlly mudding the effort.

Read some of my recent messages -- stream of consciousness is the polite term -- rambling nonsense is more accurate.


Seems like a rapid prototype could be built up to prove concept using timber and some turnbuckles - and no motors to check your movement theories for tetra, quadra, et all versions.

Which is why I'm going to the hardware store, tonight, to get some small allthread. :-) Thinking about it is a very good idea -- playing with the geometry is a better idea! This isn't exact intuitive, to me.

-- Chuck Knight

foamcutter
01-08-2004, 11:28 PM
chuckknigh,
Just a thought, have you considered using rack and pinion instead of allthread and nuts? That way the pinion on the two same side shafts could be driven together off of one motor using some type of belt or gear drive with the splined unversal joint idea. I can see it in my head but man it's tough to get the idea on the computer screen in words. Ron

Scott_bob
01-08-2004, 11:47 PM
I don't know guys, this is a lot harder than you think. I think...

If anyone can do the control logic it would be:

http://www.numeryx.com/cnc/index.htm
Unbelievable smooth, motion control...
You just can't believe it till you see it.

Scott_bob

chuckknigh
01-08-2004, 11:51 PM
I think I can see that idea, in my head. The only down side to it, is that I live in a small town, and unless I want to build it the size of a tractor, from actual tractor parts... Let's just say that gears, racks, pinions, etc are just not available out here...unless you want them to literally be 10" diameter for the pinion.

I think you're right, though...that would be a very simple arrangement. I see two possible down sides...

1) The racks will "stick out" at a virtually common point, on each of the 3 corners. There is the potential for interference...but that exists with allthread, too. The two corners will have to be *slightly* askew, in order to not interfere. Not sure how that'll affect the geometry, but I'm hoping it'll be negligible.

Worst case scenario would be when they're fully retracted "out" of the machine, at which point they'd be almost horizontal. Move them back in, and you have a "scissor" arrangement, on the exterior of the machine.

2) The angles of the struts will change, across a 60 degree arc...if the motor and pinion are mounted at the corner, and the whole corner "pivots," this might solve the problem. I don't think it's a problem, in reality.

I wonder if I could do something similar, with parts I can get locally? Hmmm...

-- Chuck Knight

foamcutter
01-09-2004, 12:12 AM
Just think how solid it would be the size of a tractor, no vibration!

Ok go back to the allthread and nuts. The nut is captured in the gear part of a worm gear assembly and the "worm" gear is driven by the motor through the splined universal joint. And the whole worm gear assembly floats with the nut on a bearing that would swivel as the arm changed angles. Ron

sbrpollock
01-09-2004, 01:19 AM
I wouldlike someone to clarify some of this for me because I may be loking at this all wrong.

It seemed as though a leap was made in this thread that the hexapod has six axises or six degrees of freedom because it has six diferent devises that cause motion in six different directions. (I'll call these actuators)

To me, the axises are the driections that the tool can move in regardless of how many actuators it takes to cause that motion. If a tool can move in three perpendicular directions (X, Y, Z) while remaining verticle the whole time, this would be a three axis machine. If you throw in the ability to change the tool angle from verticle, this would be a 4 axis machine. And so on.

A good example of what I'm talking about would be the gantry type machines that have been built that use two seperate motors and lead screws (actuators) to move the gantry along its axis. Just because there are two actuators here, there are not two axises here. It is only one axis controled by two actuators.

Now as far as a hexapod is concerned, there is no doubt that it is capable of moving a tool along many axises. But it would seem to me that the best way of approaching developement would be to build it and develope it to opperate only as a three axis machine to start out with. Then, once developement of this mode of opperation is satisfactory more axises can be added.

The construction of the machine itself would be the same regardless. The difference between 3, 4, or five axises would be in the controlling software. (A software upgradeable machine that could grow along with it's opperator?)

It also seems that by doing it as a three axis machine first, existing code could be run on it without modification since it would be the controllers task to translate the three axises in the program into the six proper signals to move the machines actuators.

To put it another way: If you only move the tool in those three axises then those are the only axises that need to be in the program. (The G-code should remain the same)

The last benefit would be in the developement of the controller software. The algorythms for the three axises could be written and put into opperation before the remaining algorythms are completed. This allows us to have working machine sooner. I think the bussiness types call this time to market?

I could be completely wrong about all of this though.

chuckknigh
01-10-2004, 01:34 AM
It seemed as though a leap was made in this thread that the hexapod has six axises or six degrees of freedom because it has six diferent devises that cause motion in six different directions. (I'll call these actuators)

Actually, a hexapod had 5 DOF (degrees of freedom) which are X, Y, Z, pitch, and yaw. A special type, called the rotobot, adds a 6th DOF...rotation.


To me, the axises are the driections that the tool can move in regardless of how many actuators it takes to cause that motion. If a tool can move in three perpendicular directions (X, Y, Z) while remaining verticle the whole time, this would be a three axis machine. If you throw in the ability to change the tool angle from verticle, this would be a 4 axis machine. And so on.

Yes, though that isn't the only, possible arrangement. A CNC lathe might include X, Y, and tilt...no need for Z.


Now as far as a hexapod is concerned, there is no doubt that it is capable of moving a tool along many axises. But it would seem to me that the best way of approaching developement would be to build it and develope it to opperate only as a three axis machine to start out with. Then, once developement of this mode of opperation is satisfactory more axises can be added.

We're in perfect harmony, so far. Take a look at the image I included, just above. You'll notice that there is an upper "tripod" and a lower "tripod."

Either tripod is capable of motion in 3 dimensions...by linking them (an easy upgrade, in my mind) we can gain additional degrees of freedom.

Thinking about it, keeping the router vertical with this geometry simply requires a "special case" condition, in which the upper and lower tripods have identical geometry. That's why I want to move 2 screws with a single motor...one motor driving both the upper and lower corresponding strut.

I could do it with separate motors, and separate controllers, but at that point I'd have built a 5DOF machine...greater cost, greater complexity, and to be fair, greater capability. But NOT an entry level unit.


The difference between 3, 4, or five axises would be in the controlling software. (A software upgradeable machine that could grow along with it's opperator?)

Assuming the geometry supports the additional movements... In the example, above, it would. It would, however, require a hardware upgrade as well...an additional set of steppers and driver electronics, and disconnection of the link between the upper and lower struts. Not a big deal...


It also seems that by doing it as a three axis machine first, existing code could be run on it without modification since it would be the controllers task to translate the three axises in the program into the six proper signals to move the machines actuators.

Actually, there are standards for any number of axes, within G-code. You just tell your CAM program to generate code for a 3 axis, 4 axis, 5 axis, 6 axis...whatever machine.


To put it another way: If you only move the tool in those three axises then those are the only axises that need to be in the program. (The G-code should remain the same)

Assuming you're talking about the g-code interpreter, that's technically correct. However, since most software is designed in a modular way, it might not be as big a deal as you think to include support for multiple configurations.


I could be completely wrong about all of this though.

Sounds completely reasonable to me...

-- Chuck Knight

Graham S
01-11-2004, 10:51 AM
Hexapods DO have 6DOF the only thing is that when it is the spindle that is moved it is a pointless one. i.e. It allows the spindle as a whole to be rotated.

A tripod can produce 3DOF movement of it's hinge assembly but as it has only three legs the other axis are not "limited". By that I mean that although you certainly only need 3 actuators for 3 axis the structure still needs to support the tool in the other axes.

So with this in mind you would have to build both top and bottom halves of the structure. But in order to ensure they move in unison yo will need to have 6 actuators.

So your suggestion is to drive two screws with a single motor, but you have to make sure the screws have the opposite action, one leg shortens while the other lengthens. Coupling two leadscrews that can change there angles constantly is difficult enough but this would make it even more difficult. I might say impossible if I wasn't so against the word

The other option is to have 6 motors but drive them in pairs with one of each pair having a wire reversed to give oposite motion.


Graham

chuckknigh
01-11-2004, 02:28 PM
A tripod can produce 3DOF movement of it's hinge assembly but as it has only three legs the other axis are not "limited". By that I mean that although you certainly only need 3 actuators for 3 axis the structure still needs to support the tool in the other axes.

True. To this end, I'm investigating two possible arrangements...one is the stacked tetrahedra model that I posted above, and the other is a tripod made up of parallelogram linkages. Either should restrain the motion in the other axes.


So with this in mind you would have to build both top and bottom halves of the structure. But in order to ensure they move in unison yo will need to have 6 actuators.

Almost correct...you have to have 6 active actuators. That doesn't necessarily mean 6 motors.


So your suggestion is to drive two screws with a single motor, but you have to make sure the screws have the opposite action, one leg shortens while the other lengthens.

No...to keep a router vertical, you need only extend the upper and lower struts the same distance. That means that they can be driven the same direction, if the screw thread is the same handedness.


Coupling two leadscrews that can change there angles constantly is difficult enough but this would make it even more difficult. I might say impossible if I wasn't so against the word

Ah, a man after my own heart.

Here are a few thoughts. Assume the stacked tripod arrangement shown above.

Drive the screws at the "common" point, located at the 3 corners of the stand. Their distance from one anothr is constant, but their angle is variable.

Two possibilities spring to mind. One is a gear/pulley system that would drive both screws from the same motor. Remember...their distance is fixed, only their angle changes. With the right geometry, some gears could be made to mesh over a reasonable set of angles.

A simpler and more obvious way to do it might also be a "flex shaft" feeding both screws. I think there'd be a lot of backlash in a system like this, but nevertheless it would drive both screws at once, with a minimum of fuss.

Hardly impossible...doing it well is a challenge, though.


The other option is to have 6 motors but drive them in pairs with one of each pair having a wire reversed to give oposite motion.

Increased complexity and expense for a 3DOF machine...why not just build a 6DOF machine and be done with it, if you already have 6 motors and driver boards?

-- Chuck Knight

P.S. Update: I was at my local flea market, yesterday, and had a brainstorm about cheap ball joints. A vendor had some very nice "ball" drawer pulls, which are a very good ball form, and which come pre-tapped. Now, the thread is tiny...but it's a starting point. Some nylon washers, or maybe PVC pipe fittings, and I might have a cheap and workable ball joint.

-- CK

Graham S
01-11-2004, 03:26 PM
>Almost correct...you have to have 6 active actuators. That doesn't necessarily mean 6 motors.

That is what I meant.

>No...to keep a router vertical, you need only extend the upper and lower struts the same distance. That means that they can be driven the same direction, if the screw thread is the same handedness.

If you want to lower the spindle for example then for the lower set of legs they must be lengthened. For the upper set of legs they must be shortened by the same amount. Or is the upper tripod not fixed firmly to the shaft between the two tripods but allowed to slide up and down the pole so that it is only it's X and Y position that is important. This is what is done on a Tetrapod where I think the shaft it help vertical bya conventional x,y stage.

>Two possibilities spring to mind. One is a gear/pulley system that would drive both screws from the same motor. Remember...their distance is fixed, only their angle changes. With the right geometry, some gears could be made to mesh over a reasonable set of angles.

They would have to be plain pulleys rather than toothed not a good CNC solution. As for meshing gears the angle would be far more than reasonable.

>A simpler and more obvious way to do it might also be a "flex shaft" feeding both screws. I think there'd be a lot of backlash in a system like this, but nevertheless it would drive both screws at once, with a minimum of fuss.

Then you need a gearbox to produce two flexible shafts.

Replace impossible with impractical.

Graham

keithorr
01-11-2004, 03:54 PM
Hadn't been reading this thread until today. Instead of moving the workpiece, consider this approach.

http://www.mmsonline.com/hsm/hsmevent/dm2bmach.html

sbrpollock
01-13-2004, 02:15 AM
Chuck:

Concerning the problem of driving two struts from the same motor:

This seems like the perfect application for the alternate geometry you mentioned earlier in this thread, I think you called it the "HexaGlide" in post #44 of this thread.

Leave the strut length fixed (and equal) and attatch the ends of each pair of upper and lower struts to a common actuator. You will have three acuators in a horizontal plane, each pointing toward the center of the machine. On the end of each of these actuators will be two struts (an upper strut and a lower strut). All three of the upper struts will meet in the center above the horizontal plane and all three of the lower struts will meet in the center below the horizontal plane.

Just imagine the machine in the picture in post #59 of this thread with solid, equal length struts. Each pair of struts connected to a single actuator sticking horzontaly out of each of the three corners of the machine.

This method might solve a couple of problems such as the need for fancy ball joints for the struts to go through. I would think some simple heim joints would work well.

Pat, the sober pollock

foamcutter
01-13-2004, 11:17 PM
sbrpollock,
Actually that may be a very good idea. I have been leaning toward trying the Hexaglide because of the idea I think it would be easier to construct. Would transmit screw error less than actuators that shorten and lengthen directly I think. Build the trolly rails out of C channel with the screw down the center and roller blade bearings to run on. Also use square aluminum stock to machine out nice little universal joints using skate bearings to swivel on. I was also thinking about running the trolly rails out from the center on a triangle or hexigon configuration instead of parrallel to each other to make calculations for control software easier. One could actually build it with six trollys and rails and use belt or gear drive to run two trollys the same if you wanted to have just the 3 axis machine and be able to expand to the five axis later with six motors and controllers.
Not sure about getting software to interpet g-code to run the actuators though. Maybe bb99 or anyone actually might have a thought or two about that. Any thoughts from anybody? Ron

BrianS
05-13-2004, 05:50 PM
Although I am not certain I've heard that Linux EMC can control a hexapod. At least that's what I am hoping cause I too would like to build one of these instead of a big gantry.

Brian

foamcutter
05-13-2004, 09:13 PM
BrianS,
Yes the guys over at the EMC site says there is a basic setup in EMC for Hexapod. I have not tried it yet cause I don't have a computer to didicate to Linux and at the present work has been cutting into my play time way too much. If you have read this whole thread you will find how to find the EMC site and you can join it. If you don't find it, I think if you do a search on yahoo it will come up, I think, been a while since I joined it. Someone here will surely be able to tell you where to go. Keep us posted Ron

Dee
11-12-2004, 09:56 AM
Hi all. Newbie here.
I think this might be a simpler approach to a 3-axis design. It uses a single parallelogram and doesn't need any additional drives to keep the router upright.
The double tetra is a great idea for more degrees of freedom. For now tho I don't know about you but I'd rather start with the simple design.
Any thoughts?

Hope the attachment works....

Red dots are ball joints with linear slide, black dots are plain ball joints. It works. The parallelogram needs to be kept vertical though, and not allowed to twist, so it'd have to be beefy.
Oh, the vertical links shown are fixed, one on the base and one on the head.

chuckknigh
11-12-2004, 09:09 PM
I don't see the attachment, and I'm curious now. Do you have a direct URL?

-- Chuck Knight

Dee
11-12-2004, 10:13 PM
Ok, I can't seem to get attachments working. No URL unfortunately, but if anyone wants I can email a bmp or slasm file.

I'll explain the thing. 1000 words, LOL.

Basic parallelogram on one strut only. The strut is longitudinally extendible. Lets call it a pneumatic piston for sake of visualization. Oh, and lets pretend that the piston rod can't rotate relative to the cylinder. Bear with me..!
Might be easiest if you try to sketch it.

Base and head are triangles.
The piston is attached by ball joint to the base.
The end of the piston rod is attached by bearing to the head, allowing rotation in one plane only (vertical).
The base has a fixed vertical extension that makes up one end of the parallelogram.
The head also has a fixed vertical extension, makes up the other end.
Stack another piston (but this one is undriven) on top of the first to make up the top of the parallelogram. It's mounted the same as the first, with ball joint to the base (top of base extension) and bearing to the head (top of extension).

So far we have a basic parallelogram, which can extend in and out. But the head can still rotate away from vertical around the axis of the strut bearing.

Here comes the trick.......Put a fixed length link (same length as the other two vertical extensions) between the top and bottom piston rods. Mounts by bearings, not ball joints. The ends of the link are fixed to the rods, but free to rotate.
Now the head is fixed vertical. If you've sketched this assembly, try moving the driving piston rod in and out. The link will always stay vertical. No proof is required.....the vertical extensions are the same length as the link, so it too will always be vertical as its length is fixed. It will push the top piston rod in and out the same distance as the bottom piston rod, no matter what the angle. Awesome.

Now the other two struts are connected as usual, to control the thing.

Okay, not too concise, but hopefully you get the gist.

Maybe someone can help me get a picture up, cos I can't work it out. I can upload a file with the "Manage Attachments" button, but it doesn't stickto my post. I can click the "Insert Image" button, but I don't know the URL for the uploaded files.

chuckknigh
11-12-2004, 10:50 PM
Just send the picture to me, and I'll post it for you. I have very little trouble with attaching pictures, as this thread proves! :-)

Here's your picture...

-- Chuck Knight

chuckknigh
11-14-2004, 01:48 AM
OK, your picture is available in the post, above. Let's open the floor for comments.

-- Chuck Knight

andy55
04-13-2005, 11:51 AM
I don't quite follow the idea in posts 76 and 77. If the head is to remain vertical, wouldn't the upper link of the parallellogram also have to shorten and lengthen in concert with the lower one ?

any other recent developments or bright ideas about tetrapods or simplified 3D hexapods ??

Dee
04-14-2005, 05:40 AM
Andy, that's right, this is what the link between upper and lower struts is for.
I'm still thinking and planning on this idea. Brain strain on the geometry of an elegant solution. Maybe there isn't one. Might post something further soon.

Jonathon
04-14-2005, 10:48 AM
Not totally on topic but while on a robot course for work I discovered these, very alien like flexipicker robots made by ABB. Very cool video on their page.

IRB parrallel robots (http://library.abb.com/GLOBAL/SCOT/scot241.nsf/VerityDisplay/1CB8C5A0F81F684BC1256B4B00450EC6/$File/Overview%20line%20Mildred.mpg)

Xerxes
04-14-2005, 02:46 PM
Has anybody tried using hydraulic hexapod for milling?

My idea is to get 12 identical hydraulic cylinders where 6 of them are in the pod and the rest are elsewhere producing the hydraulic control for the pod. So there would be six pairs of cylinders connected with pipes. When one cylinder is moved with stepper and screw, the another will follow.

JavaDog
04-14-2005, 03:11 PM
Not totally on topic but while on a robot course for work I discovered these, very alien like flexipicker robots made by ABB. Very cool video on their page.

IRB parrallel robots (http://library.abb.com/GLOBAL/SCOT/scot241.nsf/VerityDisplay/1CB8C5A0F81F684BC1256B4B00450EC6/$File/Overview%20line%20Mildred.mpg)

Those are kinda creepy...but oh so cool!! Damn things are FAST too!!

chuckknigh
04-15-2005, 02:07 AM
WOW! I think I want one!

-- Chuck Knight

andy55
04-15-2005, 03:11 AM
Has anybody tried using hydraulic hexapod for milling?
My idea is to get 12 identical hydraulic cylinders where 6 of them are in the pod and the rest are elsewhere producing the hydraulic control for the pod. So there would be six pairs of cylinders connected with pipes. When one cylinder is moved with stepper and screw, the another will follow.

can you use hydraulic cylinders in open loop control with sufficient accuracy ?

or, in this kind of setup, do you need a continuous position feedback ?

12 actuators instead of 6 seems to complicate things a lot although I can imagine that hydraulic rams create _lots_ of force.

pogcarr
04-20-2005, 11:56 AM
I want one! This is sooo cool! :)

Xerxes
04-20-2005, 02:09 PM
can you use hydraulic cylinders in open loop control with sufficient accuracy ?

or, in this kind of setup, do you need a continuous position feedback ?

12 actuators instead of 6 seems to complicate things a lot although I can imagine that hydraulic rams create _lots_ of force.
It may work just fine in open loop. Oil is hardly compressible and minimizing rubber pipe lengths would probably make stiff enough positioning.

12 simple hydraulic cylinders would be much easier to do than servo/ballscrew/whatever struts that are seen in most designs. Also joints at end of struts would be simpler to implement (ball joints will do..).

Graham S
04-21-2005, 06:20 PM
There was a flex picker in the technology museum in Stockholm when I visited, but it wasn't working!!!!! I was gutted.

Notice they use half round ball joints and just pull the pairs together with springs. The video on the website showing lubrication shows what I mean. Basically you can get huge angular movement and only need pretty basic joints

Graham

TimKoene
04-24-2005, 08:19 PM
Wow, this is really an amazing idea. I really like the hexapod setup. If done correctly it will make for a very cheap and fast and accurate machine with 5 axis. Now we need a good and cheap design for the actuators...

pogcarr
04-25-2005, 01:10 PM
Now I am curious... What kind of accuracy could we get if we took a cheap ball screw satellite dish actuator like the one from http://www.burractuators.com/hd_series.htm (I'm thinking the 55:1 gear model) fitted them with cheap stepper motors, automotive type ball joints on both ends, and gave it a shot? How bad a system could this be? Even keeping the out-tilt angles reasonable with a 30" base it would have a workspace over 48 x 48 x 17. For some applications 1/16" accuracy and repeatability in this sort of volume would be more than sufficient. I wonder just how cheaply it could be built and still have use?

TimKoene
04-25-2005, 01:14 PM
How expensive would those actuators be? I would guess very pricey! If you used the parts you were summing up I would wager you would spend 500+ dollars and have an amazing machine capable of doing true 3d work at amazing accuracy.

JavaDog
04-25-2005, 01:24 PM
How expensive would those actuators be? I would guess very pricey! If you used the parts you were summing up I would wager you would spend 500+ dollars and have an amazing machine capable of doing true 3d work at amazing accuracy.

It is very cheap and easy to get salvage actuators from big dishes...of course, getting 3 matched ones might be tough...

pogcarr
04-29-2005, 12:38 PM
I'd say buy new ones... less wear in them so better accuracy, but if you can find a bunch of matched used ones, so be it. If I knew where to come up with the controller hardware so it could be run from standard CAM packages, I would throw this much money at the attempt. Heck, it is cheaper than any comparable 48"x48" XYZ router machine!

Anyone want to handle the electronics side of things if I put together the mechanics?

mattijs
04-30-2005, 01:14 PM
Hi all!

This is my first post on cnczone...i'm an electronics engineer and interested in building my own router/mill for quite some time now.

I learned about the HEXAPOD design two days ago browsing the web and i fell in love with it immediately. I was quite pleased to see a thread about building one over here and i really hope this will be start of a new hobby.


anyway...about the hexapod:

I've read this thread and saw that many people question the availibilty for the translation from standard x y z(possibly more) coordinates to the correct geometry for the hexapod.

Wouldn't it be possibly to solve these rather basic triangle equations with a microcontroller, possibly one for every actuator to keep the program straight forward.

ANother solution could be to have an old computer buffer the input and convert it all together.

As of yet i don't realy know exaclty what kinda signals normally go into the actuators controller but i quess these are simple bit trains acting as clicks with every click being one microstep...is that true?????

in the end i'm just trying to say that i too really want one and i'd like to contribute....possibly in the form of programming a microcontroller.
I guess that 'll be about the only thing i can help with cuz i haven't got a clue to what technology is available for slides, actuators etc....

ger21
04-30-2005, 02:05 PM
I saw a site somwhere that had software that took standard 3 axis g-code, and converted it to 6 axis gcode to run a Hexapod. I think it was a university project, but don't recall. This would be relatively easy to do for 3 axis work. 5 axis work would be much more difficult.

mattijs
04-30-2005, 02:50 PM
why would you want to convert the G code? It shouldn't be machine specific...right?

ger21
04-30-2005, 06:45 PM
why would you want to convert the G code?

Because it would be a lot easier to write a converter than to write a program that could create g-code for a hexapod.

pogcarr
05-02-2005, 01:26 PM
My understanding is that the g-code says things like go 6.2 inches in the x direction. This code then needs to be translated into the "language" of the machine that tells the motors how many microsteps to go. A hexapod needs to move all of it's actuators to accomplish even a one dimensional translation.

Someone mentioned that there is some code available from NIST that does this (using Linux I think). The electronics and software to build the machine driver is beyond my skills I admit, but the machine side of things is doable.

FritzX
02-18-2006, 06:21 PM
Hi,

maybe somebody is interested in the latest design from Germany, the Pentapod (Hexapod=six actuators, Pentapod=five actuators).

You can find a video with a Pentapod, directly milling of molds from sand here (http://www.metrom.com/kurzvideo_mpeg2.mpg),
and another one showing the Pentapod cutting steel here. (http://www.metrom.com/video_d.html)

BTW.: I'm in no way involved with the manufacturer Metrom (http://metrom.com/).

Regards,
Fritz

derekj308
02-18-2006, 09:58 PM
Hi Guys,
This is a very interesting thread, there are some unbelievably cool videos and pics. When I saw the Prsco (http://www.prsco.com/index.html) R series Rotopod working I couldn't stop laughing in amazement of how ingenious the design is.

I thought I'd put my spin in about 5a machines vs. n-pods. One thing I have noticed about hexapods/rotopods is that they have 3 great translation axis, maybe 1 great rotary axis depending on design and 1 not so great rotary axis. XYZ and a 360 rotary C-axis equivalent are fine (like the R series above) but the A/B-axis equivalent doesn't have a great range of movement. We are designing a 5a at the moment ('we' being a consortium of nerds with a little spare cash lol) to reduce the tooling required for the stuff we want to do. For example if you want to face all of the sides of a 100mm high box it is better to do it with a short face mill parallel to the face being machined rather than a long end mill cutting on the side. You can also do drafted sides without ball nosing down the face or having to buy/make a tapered cutter. Its all about shorter cutters and reduced inventory of tooling. We can do what we want only if we have 360 degrees of movement on both of our 2 rotary axis. The n-pod designs are obviously more rigid but do not seem to (I could be wrong, I haven't spent more than a couple of hours researching) have a practical amount of rotary axis equivalent movement. Its what I see anyhow.

On the subject of controller translators for a hexa-ma-pod (Simpsons reference lol) I think it is as hard (definitely not simple) as creating translation and rotation matrices that would output the lengths of the 3 sides and one angle of each of the 3 triangles (since the other angle is inferred) that form a hexa-ma-pod. There is also another angle which needs to be defined which is the angle of the plane from vertical/horizontal that the triangle lies within. Sounding a bit more difficult than a couple of scaling factors. Hmmm. For the R-series Rotapod the only side of the triangle that changes length is the 'bottom' side which lies on the circular track. I would assume this would be easier to code for (and less processor hungry) than a hexa-ma-pod that has two variable and one fixed side of each of the 3 triangles. I believe this would be a more rigid design than two variable sides. Once you have the required lengths you equate that to pulses generated by the controller to the drives. The translator would be part of the controller software which would generate required pulses to each of the 6 actuators (not axis) which define the lengths of the side(s), the internal angles of the 3 triangles and also the angles of the planes in which the 3 triangle lie. The R-Series rotapod has another variable which is the relationship of each of the three triangles to each other.

From what I can gather the three points under the table define the desired xyz and two rotary axis position. Once you know where you want these three points in space you construct triangles back to your 6 pivot points on the machine.

Please be kind lol. This is all based on 10 minutes of assumption lol.

Cheers
Derek

Graham S
02-19-2006, 08:31 AM
I have written a little document on using Mach3 to control parallel kinematic machines. It only really applies to 2-4 axis machines as you need twice as many axes in Mach as the machine has.

www.indoor.flyer.co.uk/kinematic.pdf

I like the pentapod makes sense and the kinematics are simplified, a pentaglide would be easier to make for the amateur the rotabots are also sweet.

chuckknigh
02-19-2006, 03:32 PM
The pentapod and the rotobot are my favorite designs, yet. Simple, elegantly designed, and actually buildable!

I still worry about software, especially on these non-traditional geometries, but software is comparatively easy to take care of. If I understand MACH (and EMC, the program on which it's based) then you can program in any type of geometry.

-- Chuck Knight

P.S. I'm glad to see I'm not the only one interested in these little machines. This thread has gone on longer than I *ever* expected!

Graham S
02-19-2006, 03:38 PM
Mach does not support kinematics in the same way that EMC does I am using a feature only seen in Mach to do it in a less generalized way.

The pentapod would seem to have many of the same building challenges as the hexapod as is except for the clever joints on the tool but even they require some rather large ID bearings. I looked at building a rotobot but the large ring gear is a problem to get hold of unless you want to use friction.

Graham

gotis
02-19-2006, 06:18 PM
we had a robot at work a couple of years ago called Neos Tricept,
drilled some holes and milled an oval in the volvo subframes.

FritzX
02-19-2006, 06:48 PM
I looked at building a rotobot but the large ring gear is a problem to get hold of unless you want to use friction.

For those who don't know the rotobot (rotopod ?): Parallel Robotic System Corporation (http://prsco.com/) have some animated images on their website.

Only a suggestion: The large gear ring, needed for a rotopod design can be found at your local scrap dealer...

The flywheel of a truck diesel engine contains of two parts: The flywheel, usually of cast iron and a gearring for the starter...

The 800mm diameter (approx. 31.5") should be enough for a small machine...

Regards,
Fritz

chuckknigh
02-19-2006, 10:37 PM
Given the diameter of 800mm, I wonder what the working envelope would be? Surely there's a way to calculate it.

-- Chuck Knight

FritzX
02-23-2006, 08:50 PM
I did a quick search and found a site that might have the information. Check out:

http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html

Good luck. I'd like to see more of the Stewart Platform.

Evan

John's address in the www has changed.

http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html is not valid any more.

Check out for:
Computer Craftmanship (http://www.laboratoryformicroenterprise.org/lme/ComputerCraftsmanship.html) instead.

Regards,
Fritz

FritzX
03-10-2006, 07:04 PM
Just found a video from an hexapod design from Poland:

CAMT - Centrum Zaawansowanych Systemow Produkcyjnych (http://www.camt.pl/dodatki/Felix.avi).

Regards,
Fritz

unterhaus
03-10-2006, 09:40 PM
Someone mentioned that there is some code available from NIST that does this (using Linux I think). The electronics and software to build the machine driver is beyond my skills I admit, but the machine side of things is doable.

EMC originated from NIST. The kinematics module in EMC can be replaced. Nist built a hexapod that they ran with EMC. They used it to demonstrate the software, back when they still admitted they had something to do with it.

You could use hydraulic cylinders for a hexapod. I've built a machine with 3 Parker hydraulic cylinders that came with linear feedback devices. They work pretty well, but each cylinder was thousands of $. I'd go with the newer electronic cylinders that have servomotors on them. Much better design for this application. Of course, we can't afford those either.

bryanrabb
03-13-2006, 04:15 PM
Check out the "free floating milling machine" video. I think this is the same thing yall are discussing.
http://www.cuttingedgecnc.com/movies.htm

sandos
03-15-2006, 09:23 AM
This looks like a pretty simple design.. I think:

http://suvalleynews.com/modules.php?name=News&file=article&sid=520

I dont know exactly how the thin "dual-bars" down to the platform behave.. but I guess they must atleast be retractable/expandable?

Graham S
03-15-2006, 09:29 AM
Its a 3-axis delta robot just like the flex picker. Dual arms act as parallelograms to keep the "tool" pointing downwards (it never tilts). It is operated by what looks like brushless servo motors running through harmonic drives.

Building a machine like this is possible for the amateur but not at this scale, as soon as you try to define angles accurately you need very high ratio gear boxes. Even 1cm resolution on a machine like this would need a high ratio gearbox.

Graham

Graham S
03-15-2006, 09:31 AM
correction it is 4-axis but it will only be able to rotate around the tool's axis which is perfect for pick and place.

Graham

sandos
03-15-2006, 03:53 PM
correction it is 4-axis but it will only be able to rotate around the tool's axis which is perfect for pick and place.

Graham

Ah, yes, thats correct of course, and not that useful. Not as useful as a tilt-able machine anyway ;)

I thought it seemed a bit too easy.

unterhaus
03-15-2006, 05:18 PM
some professors here at Penn State had a platform they built with 3 linear stages on a plane at 60 degrees from each other. The only disadvantage is that it is large compared to its envelope. It had limited tilt range, but it was good enough for most milling. A full 6 actuator stewart platform can gain some stiffness from the interaction of the actuators since they can push against each other.

I think this design has obvious advantages for home built machines.

FritzX
03-16-2006, 05:55 PM
A nice photo album of Carl Mikkelsen's big home built hexapod, made without expensive bearings, joints, etc.:
Stewart Platform (http://www.foxkid.net/cgi-bin/tn4-page.pl?home_album=&basedir=./Processed/Carl%27s%20Projects/Stewart%20Platform&album_name=Stewart%20Platform&contains=P1280453.JPG)

hma
03-20-2006, 07:45 AM
Here another and interesting design of a stewart platform:

http://www.lewetz.de/ > Bilder > Very low on the page

Regards,
Hugo

sandos
04-06-2006, 05:38 AM
Heres a very simple thing to build (I think?), but also very "weak"

http://www1.atwiki.com/gingery_machines/pages/robocrane

Not sure if this could be useful for anything ;) possibly for carving in foam or something? Im pretty sure its "useless" for even wood and definitely for metalwork ;)

JBV
04-06-2006, 08:15 AM
Should work with welding and plasma cutting ;) Then you could use your entire garage floor as worktable :D

phantomcow2
04-17-2006, 09:55 PM
Hmm...interesting.
Even more interesting is that Hexel Corp is about 10 miles from my house. I may pay a visit

gotis
10-11-2007, 03:43 AM
They re still in business.
http://www.pkmtricept.com/video.htm

Jamison
09-23-2010, 09:46 PM
I've been working on hexapods and delta robots for a while...
YouTube - Delta Tau's Brick controlled Hexapod

the video above is a hexapod with magnetic joints, and I made it for @$15k. I made it for Delta Tau, and they travel with it to shows including IMTS etc... along with my little mini deltas that people like so much.
YouTube - Mini Delta Robot with background noise to verify actual speed

These are some very capable machines, but you will find some gotchas in controls, beyond complex kinematics with multiple solutions, some of my biggest challenges were in precision calibration required due to mechanical imperfections. Try tracking a small error due to math or mechanics.... not fun. Also remember that the hexapod is rigidly coupled so good, stiff, powerful actuators must be perfectly coordinated or they will crash the hell out of the machine... so start off with mush tuning or suffer the expensive consequences.... or use magnetic joints... (God bless Sina at Delta Tau for that idea!) Actuator error from the upside down hanging, magnetically joined hexapod in the video only resulted in a dropped end effector or actuator.

Fun stuff tho!
YouTube - Fast 140ppm mid sized delta robot

Jamison
09-23-2010, 09:51 PM
http://www.ecuriechotis.com/pkmgallery.pdf

I've stopped accumulating these... they are too many...

Jamison

PKM
08-31-2012, 11:51 AM
Just to revive the topic.
LinuxCNC (EMC2) hexapod machine tool - YouTube
LCNC isn't flawless, but definitely fits to control any DIY parallel machine. I've done a few kinematics already.

Jamison, I admire your machines. Making a delta too, but with cheaper servos it probably won't be so fast.

samco
08-31-2012, 11:57 AM
I have watched your video's many times. Very cool - Nice work!

I know linuxcnc isn't perfect when controlling non-trival machines. They keep working on it though. (an there really isn't anything out there)

sam


Just to revive the topic.
LinuxCNC (EMC2) hexapod machine tool - YouTube (http://www.youtube.com/watch?v=G_UmhUjZhNo)
LCNC isn't flawless, but definitely fits to control any DIY parallel machine. I've done a few kinematics already.

Jamison, I admire your machines. Making a delta too, but with cheaper servos it probably won't be so fast.