Freesteel Blog » First drop point results
First drop point results
Tuesday, February 13th, 2007 at 11:25 am
Beginning with the easiest, I’ve done cutter location against a set of points picked from a model. I’ve learnt my lesson and done it for the uncommon tapered cutters from the start, rather than going for the easier cylindrical type of cutter, and then having to fit this feature into the code later.
I’m also short of simple machining algorithms to test it on, so this is what it looks like when used by the const scallop routine. The complex shape of the toolsurface gives the algorithm such a good work-out it’s exposed a couple of glitches which I am determined to deal with before I carry on. Unfortunately it requires running in debug mode, which is rather slow. I need patience.
2 Comments
1. Neel D replies at 13th February 2007, 11:54 am :
Is this algorithm different from one you have mentioned in
http://www.freesteel.co.uk/wpblog/?p=59
2. Julian replies at 13th February 2007, 7:12 pm :
That’s exactly the one.
The reason it needs rewriting is that what we’re using is too slow, relative to the competitive software. The main way to speed it up is to reduce wasted calculation. The nature of the
cl(S) = max(cl(tr) | tr in S)
function is that if I was able to choose just the right triangle trm in S, then cl(S) would equal cl(trm). The calculation on all other triangles apart from trm are wasted.
Unfortunately, an algorithm to choose just the right triangle might wind up taking as long as the full calculation. But we might do well with a reduced subset S’ of S where cl(S) = cl(S’).
The obvious first candidate for S’ is just the set of triangles which overlap the circular outline of the cutter. This is what’s being used at the moment. But it’s got to be made a lot more sophisticated to be competitive.
Leave a comment