Re: Where should I head on a probe? (Was:Digitizer probe progress)

Tim and Fred, OT perhaps? ;)


Jeff,

> 1) Where do you find a probe or what major components for a home
> built one are redily available? Is there something used for

another

> funtion (joystick turned upside down? ;-) that could provide parts?

Buy or build, if you wanted you could make a much better version of
mine. All you need is a lathe a mill and a rotary table. You don't
even need CNC. Materials required:

1. Some metal assuming a more robust and accurate body assembly
assembly (not that mine isn't you understand :) ),
2. Some PCB material, this would be excellent for sitting the ball
bearings in as it is an insulator (when copper is removed) an is also
dimensionally stable.
3.More pcb for the ball clamp
4. Some ground hardened drill blank (for the rods that sit on the
balls)
5. A spring.

If you you care to spend the time you could make something very
concentric and will excellent repeatability. Mine was built on
a "toy" mill and it is doing VERY well.

> 2) OK, lets say a probe is obtained/built. Grahams process to
> digitize an object is clever and works, but an easier way with less
> steps would be more useful for mear mortals like me. The ability
>to find the boundries of a piece of material (for use as a
>reference point) would be helpful most CNC programs (MACH2 for
>example). I may be wrong, but it should remove the requirement to
>hand measure everything when beginning a new project.

In TurboCNC If you pop a spherical or cylindrical probe in a hole in
jog mode and then zero the axis and press C, it finds the centre of
the hole. If you then enter the probe diameter it will tell you the
diameter. You can also use it as a wobbler. A general "job mapper"
would seem tricky but perhaps a conversational style could be used.

E.g. <map edge left>, probe moves to the left, finds the edge and
then maps it by moving along it doing little sideways pecks.

<find hole centre> finds centre of hole

All the while this info could be added to a file, either 2D, 3D or
whatever.

>
> 3) While I understand that tool paths will not be optomized, and
> tools/feeds/speeds will have to be hand entered, it would offer the
> oportunity for a CNC program to act as a duplicator. Once the
> boundries of the original probed, a marix of points (of a user
> selected resolution) could be collected. The program could then
> suggest an appropriate size/shape tool based on "best fit" of three
> tool botton shapes, of 3 user selected sizes each to create a
> duplicate of the work with a simple tool back and forth (don't know
> technical term for it) path.

Horrendously slow. Say you had a plate with just a few holes in it,
you would still need to digitize the whole thing to a high accuracy
assuming you wanted the copy to be accurate. Although the turbo CNC
digitizing algorithm can be improved upon it would still be slow.

In a similar way that it is better to use 2 1/2D machining for simple
jobs of facing and pockting rather than parrallel milling the surface
it would be better to take the route in 2, where specific features
were defined, like holes, pocket, edges, faces etc.

Just MHOs I am new to this digitizing lark, and CNC for that matter.

Cheers,

Graham