Re: [CAD_CAM_EDM_DRO]floating z axis

Hi Nic,

That's an interesting approach to cutting a board! I'm not quite sure
what you meant by your statement "if you get islands...aviod isolation
problems...reorder".

I get "ordered" Gerber files from my experiments with Eagle, and then I
convert them to Gcode. However, a trace may start at one side of the
board, and go around the board and end up on the other side! The
ordering that you mention (I think) only helps to shorten the G00's
(rapids that don't cut board). Were there some other benefits to the
isolation? I'd like to hear more.

I'm working on a digitizing routine in hopes of being able to remove
surface warps in a similar manner to what you describe. There were
recent conversations on the TurboCNC list a week or so back.
You might also be interested in a new "Home-brew PCB" list, also on
Yahoo.

Alan KM6VV


NCI van deer Walt wrote:

>
> >This will likely drag highly abrasive grit under the pressure foot, and
> >chew up either the foot, the board, or both.
>
> I played with this when I built my first router out of printer steppers,
> round
> bar and 1/4" plate back in school.
>
> I used a dremel in the head, and I connected the routing bit as one side
> of a
> limit switch, and the PC board and bed as the other. That way the
> software could "feel" the
> height of the PCB at the start and end of each cut, and compensate.
>
> The actual motion looked quite cool, it would tap-tap-cut quite
> rhythmically.
>
> The big problem with this is that if you get islands it stops working.
> You either
> need intelligent software to re-order the cuts to avoid isolation
> problems, or do
> what I did and re-order them manually for each board. I never got past
> that though.
>
> Now that I understand the math better I would measure the height of the
> board on a
> .1" grid, and then add Z offsets to the 2D milling path by simple
> interpolation.
> Conductive feedback is the easy way. Capacitive feedback will avoid
> problems with
> oxidation on the PCB, and avoid the grid of dots you'll get otherwise.
> Or you could
> use a digitizing probe.
>
> Regards,
> Nic.