So, yesterday I messed up good. I figured out a way to reset the origin of the table and continued (more or less) where I left off. The finished blade shape is a lot better than the ones so far. Still that leaves plenty of work:
the warp I experienced before is gone (I made a billet of the right height by laminating a number of strips lengthwise, see photographs), the vertical orientation is perfect. And because there are a number of strips parallel that does not get exchanged for similar trouble in the 'y' direction of the blade. (width).
the whole shape now fits inside the billet, no more pieces poking out. That's mostly due to the new software that allows me to see the shape as it will be within the boundary of the material, that way I can guarantee that I won't overrun the material anywhere.
I'm using a drill with a tip ground to a ball head because my 3" long ball router bit still hasn't come in. The drill bit is not balanced too well (my grinding only goes so far), which means a lot of extra vibration on the router. Very annoying, I'll pester the dealership on Monday once again to get that bit in asap. (or at least before I mess up the bearings in the router)
the glue is MUCH too hard, whenever the router bit hits the glue it bites way more than it should, the glue stays in one piece but the foam around it is compressed, and sometimes ripped out. I need different glue, probably wood glue will be better for this (the one I used for this billet is construction glue), woodglue will hold the billet together long enough for machining but will desintegrate nicely when hit with the bit. Also woodglue won't leave the high spots that this glue does (and since the foam sands much easier than the glue they are pretty hard to get rid of).
the blade tip is still too narrow and too thin for my taste, too hard to get that stiffened up properly without an inlay, and I want to try to avoid that.
registration still isn't perfect. That really sucks, I though I had that down pat a few days ago. It may be that I lost some steps on the first layer, there is a suspicious step at the beginning of the first pass of the backside. I'll add some checks to make sure that I get rid of that in the next round.
the software that drives the mill can be better, the path should be more continuous so that it doesn't ramp up / down between stations but just turns a lazy loop at the end and keep going. That should save a lot of time.
Pictures:
The billet just before the first run:
The last pass on the backside of the blade:
A view from the root of the finished shape:
A view from the tip:
