Re: [CAD_CAM_EDM_DRO] A "simple" Hexapod project

Graham,

I've built a hexapod, and went through several of the processes you are now
considering.

At 02:47 PM 6/13/2003 +0000, you wrote:

>This is a tentative and probing post just to get an idea of the level
>of interest folks have in making Hexapods. They were mentioed a
>while back and I have seen a couple of designs on the web that have
>been built by amateurs, one using an inverted Stewart platform and a
>router and the other using cables. (Search for hexapod in the archive
>for the links).

I'm one of those with an inverted Stewart Platform.

>I don't have the skills, time or money to make a hexapod capable of
>machining metal but I am an aeromodeller and the ability to shape
>foam, plastic, wax and balsa in 5axis would certainly be of great use
>or at least great interest.

I'm sure you have the skills. I did it with no metalworking skills to
begin with, and had to learn welding, brazing, tapping, and other things to
make it all work mechanically.

>What I propose is a simple Hexapod based on the ideas of the home
>router/engraver constructor rather than the THK and ballscrew set.
>
>Hexapods have 6 actuator legs and as they are legs all designs I see
>tend to be based on telescoping tubes. Personally I can't imagine
>building a very rigid actuator in this manner (i.e. I can't) and
>finding tubes with precision bores makes the task more difficult.

My first model (still the only one in service) used 7/32 hardware store
threaded rod very loosely telescoping inside 1/2" square steel tube. The
nuts were the longish coupler nuts made for connecting threaded rods.

Gradually, there has gotten to be more backlash between the nuts &
screws. The next version (partially constructed, awaiting some time to get
back to building the frame) uses ball nuts and ball screws, but no
precision mating components other than the nuts and screws.

> I
>am thinking of making an actuator based on two silver steel or
>precision ground rods. At the end which would be pillow blocks
>including at one end a stepper motor mount and leadscrew bearing and
>at the other a largeish Delrin bushing. A simple Delrin carriage
>could ride up and down these rails pushing the actuator leg through
>the bush in the pillow block. The part of the leg that moves would
>then have three points of support. The leadscrew would reside within
>the hollow actuator leg and be supported by a derin rider that could
>be a loosish fit in the tube. To stiffen the whole thing up plates
>could be added covering the assembly and locking the pillow blocks to
>prevent twisting.

Stiffness comes from the fact that you have six actuators, which are
sharing the loads, and that each actuator is operating only in compression
or tension. There are no significant twisting loads on the actuators.

>The leadscrew could be cheapo 6mm threaded rod and the nut a long
>nylon job these seem to be pretty good on the backlash and cheap to
>replace.
>
>Pivots for the ends of the actuators might be off the shelf ball
>joints of some kind or maybe gimble types.

For joints, I've gone with gimbals. Universal joints just don't have
enough range of motion. I've found that to get good pitch, roll, and yaw,
the joints where the legs attach to the spindle must have a wide range of
movement. So far, I've done better getting the range with gimbals.

I am jealous, however, of the space that gimbals consume. U-joints could
be more compact, if the range of motion is high enough. From published
U-joint specs, it seems that +- 45 degrees is about the limit. I find it
useful to me closer to +- 80 degrees.

>I am just wondering if anyone else might be tempted, and if my ideas
>are stupid. I suspect if a group of people had a go issues such as
>software and tricky hardware prolems could be ironed out more
>easily. Perhaps just start by making one actuator as I intend to.

I've got kinematics software (scratch written), but have never integrated
with EMC. I have a program that accepts an ad-hoc language of positions,
offsets, include files, and so on, and some tool path generation
programs. One converts a raster to toolpath, others generate toolpath
functions for operations like "cut a hole", and "cut a box".

>Off to Dream,
>
>Graham
>