Re: Re: Scanner/Digitizer skip function/probe details?

dave e wrote:
<big snip>this is not going to be fast. :-( Post process for to

>compensate for slope, i.e. the unless slope is zero the point of
>contact with a spherical probe end will not be the real point of
>contact( but the solution is straight trig). After compensation
>(correction) this gives a point map in x, y and z which will have to
>be processed again to give you g-code to duplicate the surface.

Dave,

Although speed may not seem too important now, I can assure you that
if the process is not fast ENOUGH (insert your definition here), it
will NOT be used. Seen that too many times not to comment...

Also, The post processing being discussed is nearly never done in
real life situations, and when done,is usually handled in a CAD/CAM
program which has imported the "point cloud". There are SO many
variables, you would need to keep a journal while you were 'getting'
EACH point, with notes like "this was the BOTTOM of the probe. This
was the midpoint of a 45 degree slope", etc.

What the level of digitising we have been talking about does is 'get
us close'. Then we can clean it up in CAD, if necessary. If 'close
enough' IS close enough, you can often get good results by using a
probe tip the same diameter as the ball endmill you are going to use
to cut. This makes the math unnecessary, as the compensation is then
automatic... FUN part here is/are the 'roughing' passes!

>Naturally if you have overhangs this won't work and vertical walls
>will probably need another routine which slides along the wall
>taking points. All in all this is not a simple process. Hope someone
>has a better idea.

There IS a reason CMMs' are so expensive! And why metrology is a
Large science! :-)

Hope this helps.

Ballendo