[bedo]

Hi to all members !
I'm new to this forum and I'd like to present my ongoing project of an hexabot
machine. All is completely DIY by me, attached is cad design and a frame from a test cut video on a block of resin.
Machine frame is composed of two ( 3 with the base ) exagonal wood base joined by columns and stiffened by diagonal chains. I'll post some other photos soon.
Mill ( at the moment a plain drill... ) is placed on a round platform joined by six variable lenght legs with stepper motors.
Machine will be driven by selfbuilt control board that interface a standard pc to the six driver boards via RS232 and custom protocol.
Software will convert standard Gcode ( from any cam software ) to drive commands to set the correct lenghts of the legs.

Anyone interest in sharing comments and ideas is welcome !

ciao,
Bedo.

[bedo]

It's me again, I'll post some more details for my machine :
here attached the joint shoots. There are 2 kind of homemade ball-joint : one for base attachment and one for platform attachment. The one for the base has a plate for attaching the stepper motor, while the other one has a square metal part that fits on end of leg.
These have been made using spherical ball handles I found in a furniture spare parts shop. This handles have been put inside a resin part ( bicomponent epoxy ) so that it fits exactly around the sphere, keeping it locked for just a couple of mm.
In this way I can get an inclination from vertical of about 60°.
This have been one of the hardest part to build, I made really a lot of test before reaching these results.
I think this part will be remade in metal ( if ) when the machine will be working.

[bedo]

Here You can see one of the legs during assembly.
In "leg1" You can see the end of the fixed part of the leg. Here I fitted a round resin part with a square hole, where the other part can move without rotating.
In "leg2" the other side of the fixed part, with base for fitting the stepper motor.
In "leg3" the top of the movable part, with inside the precision screw. The round part on top of the screw keep it centered inside the square part while rotating ( otherwise I'll get weird noise when moving...).
In "leg4" the other side of the movable part, with round bae with o-ring that
exactly fits the outside tube. This part has been molded with a screw inside so I get an exact copy of the screw to drive to movement. At the base of the driving screw the joint for the stepper motor. This resin parts have been pretty hard to get them right and working.
In "leg5" the stepper motor assembled on screw and finally in "leg6" the complete assembly with the ball joints.

[bedo]

here You can see the structure of the machine assemble as in the first cad drawing I sent.
The base are 19mm MDF which is the only material I found at a reasonable price to keep vibrations low.
The orange columns are made with standard pipes nicely cutted all at the same lenght with a precision saw. Then I filled this with cement with nuts "glued" inside, to give some weight and stiffness.
At center of the base at moment I mounted a standard clamp for some testing.
On this shoot the legs are not assembled because I'm doing some reworking at the moment but I will soon post a complete assembly.
The chains keep the structure in tension so it cannot bend while working.
Below the base, for each column, a nice stand with screw to keep the whole machine perfectly vertical.

[greybeard]

Hi Bedo.
Many thanks for posting your pictures with your thread.
I know very little about hexabots, though I think they are interesting because of being so different from 3-axis cnc.
Makes my brain work in a different way, which has to be good.
Given the dimensions of the one you are building, what will be the possible "angle" downwards from the top of the work piece that you will be able to reach ?
For example, will you be able to come in horizontally on the "equator" if your workpiece was a sphere ? Or even lower ?

Regards
John

[bedo]

Actual figures cannot be evaluated exactly because of odd maths behind it...
Near the origin I guess the platform can withstand angles of around 45-50°
from vertical.
This is in strict relation with the maxim angle the ball joint can do. With the
one I've built the angle is still pretty limited, mainly because the base of the joint are vertical and not oriented to the base one, and so for the opposite one. If the machine will ever work I already plan to machine better ball joint in metal with the base oriented lets say 45° so it will be able to have more travel in both direction. This is the main limit at the moment: when at origin the joint have a lot of travel in one direction and limited travel to the opposite. I don't know if its clear, its not easy to explain in words...
Other post and photos will follow in a few moments...

[bedo]

this evening I finished assembly the legs and reassemble the movable platform, so here the photos.
As in other post here in the platform images, You can see a position near "home" position ( all legs completely retracted ). As You can see the ball joint position is already around 30° from their vertical, so possible movement will be almost 90° in one direction and only 30° in the opposite. But You must take in account that the opposite ball-joint will limit the movement on the other side, so the maximum vertical angle for the platform is limited by this at the moment. Same problem is on the base ball-joint.
Solution for this would be to have the base of the joint angled of around 45° in the direction of their leg, so that movement will be limited of roughly the same amount in both directions.

[bedo]

Some figures I did not post previously to give an idea of size of machine.
Distance between the two base is 950mm. Witdh of base hexagon is 1000mm, platform is 300mm. Legs can extend from center of joints from 550mm to 950mm.

[bedo]

Here the electronics box. On the left side, from top, mains filters and transformer and power supply board, with high voltage for pwm stepper ( around 35-36V ), 12V and 5V.
On the right side mainboard with pic16f877a controller. This will receive command from PC by serial port and convert it to dir and step commands for all the driving boards.
In the center there will be the stepper driver boards ( 8 in total ), only one
built at the moment...
On the top right space is for mill driver board, with digital variable speed controller, still to be designed and built.

[bedo]

A screnshoot of the alfa version of the controller program. It reads a standard Gcode program from any cam program.
In the left window there are X,Y,Z position and A,B and C angles ( or other option will be vector direction, but is not standard in gcode ), feed speed, spindle speed and tool number.
The program is parametrized with the real machine size and a model in directx is built in the right windows. The platform is then moved in the correct X,Y,Z and angled as specified. Then it simply measures the distance from the ball-joints centers ( You can see in L1...L6 ). Movements are interpolated from one gcode position to the next in delta-T time steps ( around lets 1/10 of second ) and number of steps for motors in the correct direction is sent to the mainboard circuit throught serial port.

[Milo]

Hi Bedo
Great project. Keep up the good work. Could you tell us
about the math involved?

[bedo]

Math is really simple, I do no calculation at all ! as I tried to explain in the software post, I made a CAD replication of the machine with the same size of the real one with directX 9. In the software You can see it moving as simulation.
I simply mode the the platform object in the correct position given by X,Y,Z of gcode and orient by the A and B angles. Then when You have the platform in the position You need, You simply take the distance from the ball-joint centers and You have the actuators lenghts. Some checking is done at this point not to reach the end of movement for length of legs or ball-joint angles.
If this test is not passed the position is not reacheable with current machine setup and program stops.
Then the program simply sends the correct number of steps for that delta-T of motion to the control board.
Starting from a know platform position ( all actuators retracted ) You will move the platform in the position You need simply setting the correct lenghts of the legs taken from the directX simulation !

[greybeard]

Bedo, someone has to ask the stupid question, so it might as well be me
Why have you got the lower section ?
Is it just to give you a convenient work table height with a "stand-alone" machine ?

John

[bedo]

simply not to work on the floor !
Also under the machine there will be space for PC, electronics and dust aspiration...
First version of machine was only top part, but just after doing some tests I decided to raise it at a "table" height.

[Milo]

Bedo
Thanks for the reply. I'll have to look into directX. I have watched videos on hexabots and always wonder how they generated the motion profiles for each axis. I researched Inverse Kinematics on the web and got lost in the math. Your solution seems within the reach of my math skills.

Thanks
Milo

[jgro]

Great work! Keep going. I've always wanted to build one of these.

jgro

[Milo]

Bedo

Do you have updates on your progress?

Milo

[bedo]

hi all, here a small video of a test cut I did some time ago. At the moment
machine only has one driver board ( I still have to etch and built the other 5... ) so only one leg was moving. I managed to find a good position for the test cut, the tool tip described a complex arc in space, but it was enough to check for structural stability. I got almost no vibration and the cut was nice and clear.
I'm now writing the firmware of the pic board and the communication software, but my "real" work took almost all of my time, so progress are really slow down at the moment.
Next part, after first beta of software, is to have all electronics and driving board, so I will be able to do the first real test cut of hopefully something usefull !

[samco]

Very cool.. Have you looked at emc? It has kins for a hexapod machine.
I know you have done a lot of work so far - but emc would have all the trajectory planning and such for segment blending..

http://www.linuxcnc.org/
http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl

Here is a hexapod running emc2 in spain

http://www.imac.unavarra.es/parallel

sam

[bedo]

yes I know about this, but when I started my project ( more than one year ago ) I didn't knew about it and started with my own plans. So I decide to built all circuits and wrote all software on my own so I will be able to fit perfectly to my needs.
Surely their structure of machine is much more steady than mine, but I don't have the ability to built such a structure-
Also writing all software and building all by my self will be much more fun for me !