Re: Digitizer algorithms (mine worked)

Doug,

I can't quite compare my algrithm fairly with Tcnc as I am using a
pretty bad timer and am limited to 1ms between pulses. I also don't
accelerate and have such fast rapids from my program. Anyway despite
this I shall try a test comparing mine and Tcnc on a step object, it
should still be far quicker (visually it is anyway). Oh the first 20
attempts were not scans they were more crazy probe dances and scarey
plunges straight at the bed. My hand was firmly on the parallel port
switch as I was a good 4' from the machine using my proper PC not my
486 laptop. :)

It would not be difficult to change my program to use your algorithm,
I like the idea of taking readings on the break of contact, I had
thought of that but wasn't sure if it would trigger in exactly the
same way. I can't see why not. Your method like mine would not work
well with a ball ended probe, dome ended would be fine as would
pointed and parrallel. This is because you could end up sampling
under cuts.

A word on my offset, the first offset was merely to take the tension
out of the spring and is possibly not needed for hard materials. In
softer ones it may tend to drag a little but then if that happens it
will lift automatically. The second offset was added later after my
first scan. The mill bed had a little double sided tape on it, or
what had been. No stickyness just a generally rough surface. The
algorithm took ages to get over it carefully ramping over the fluff.
I thought it would be better to just have a sensible off set and jump
out of the noise, afterall in my algorithm I don't care what is
between the sampling points..

Doing a scan at the maximum resolution of the machine at least for my
machine would not be practical, I would rarely want 400 samples per
mm. For courser machines it might be perfect. There is also the
question of is it quicker to do a biggish jump in z then in y or to
ramp up in steps following the surface perfectly.

I'll definately give your algorithm a go and I suspect it is best to
have a suite of these things for different applications anyway.

The idea of tracing around the object is also an excellent one!!!
Must have a go at that, there must also be plently of applications
where tracing is all that is required anyway.

Cheers,

Graham