Busy couple of days... After the last - very successful - test run I wanted to see if we could speed up the cutting operation by automatically computing a more efficient toolpath. It took a couple of days, but we ended up with a fully three dimensional simulation of the routing operation, to show us how the cutter would move through the blank. After the simulation has run you end up with a numerical representation of the surface cut out which then gets fed into a 'roughout' generator that produces a toolpath that will approximate the shape with the fewest possible (in theory) machining operations. All this software was written and tested in a few days, and seems to work ok.
It took some tries to get it right, but finally we thought we had something that we could feed to the cutter to check that it worked as expected.
When testing a radical departure from your previous cnc run it's always good practice to check your design in a playground environment, so wood gets replaced by foam again to see if we've missed anything. Sure enough, halfway into the next-to-last pass of the roughing out the cutter decided to take a shortcut through 2" of material left standing... oops! Since this was just foam, it was not a big deal and the cutter actually made it through without incident. Had it been wood that would have looked a little different. The run made it to completion and the the blade was actually quite nice.
When powering down the router I heard a nasty sound, and I decided to pull the router and inspect it (this was thursday evening). After disassembling the router we found that the top bearing had roughly 1 mm of play in the housing. Now, get this right, at this point after the last rebuild - by a 'competent' shop this router had less than an hour of work done, and that in FOAM ! Not exactly the most demanding application.
I went back to the repair shop the next day and gave them a piece of my mind, asked them go get me another housing so I could at least put the thing back together again.
I cruised around the local town (Sault Ste. Marie) some more, where the only thing they usually stock are catalogs and found a very heavy 3 1/4 Maktia router mounted in a table, and got a good deal on the whole assembly.
I had looked at this type earlier and decided that it was too cumbersome to try to mount it, this thing is meant to be mounted on a table or some kind of plunge base, definitely not on the Z axis of a cnc router.
The machine had run only 5 minutes in its whole life so far, I decided to open it up and have a look at the bearing housings (call me suspicious) and sure enough, the beginnings of a scoring pattern where already visible.
To stop this from getting worse I used a couple of drops of locktite to make sure that the bearing would not walk around the housing without permission. (The BLUE stuff, not the RED stuff which is impossible te remove without heat, not a good idea on a machine with a plastic housing).
It's my theory that these bearing housings don't work too well because the aluminum expands more than the steel of the bearing, so essentially the housing expands away from the bearing, not a good idea with a loose pressfit. Suddenly it's just 'loose' and starts rotating with the shaft. Proabably a bit slower, but it won't take very long for a hardened steel bearing casing assisted by a good bit of vibration to smash an aluminum casting.
We'll see how it holds up, time will tell.
Getting it mounted was quite a chore, we chopped up the housing wherever it intersected the machine and built a VUB(1) for it. This took us the better part of Friday, and at the end of the evening we had the beast mounted. This router is so much heavier than the previous one that the slightest tap on the housing caused the whole works to come crashing down, clearly the steppers didn't have enough torque to hold it, let alone move it.
Some fiddling around with dipswitches cured that problem, and at 3 x the current of the previous setup the unit now holds and moves beautifully, in spite of the extra weight.
As soon as I have the blades done and the table restored to its real function (plasma cutting) I'll make a nice bracket to replace the VUB. The whole setup is quite an ugly hack job, it's less than satisfactory, but it's the best I could come up with on short notice.
On Saturday we ran another foam test to verify that all was well (I had fixed the bug that caused the machine to eat 2" of foam two days earlier)>, it ran fine, really quickly (47 minutes for the entire blade) and made a good shape.
We've just glued up a new blank, and we'll be running a full sized blade again tomorrow.
The simulator is an invaluable tool for this 3d milling stuff, it saves a lot of time and messed up blanks, by the time the simulator runs a piece well you are almost assured of a good outcome on the table.
Tomorrow I'll try to do a good screenshot of the simulator in action.
Pictures follow:
Roughing out the foam blade (3' long, tsr=8):
Final pass as seen from the top:
Final pass as seen from the left (router and blank have been switched in this picture relative to the previous picture, it looks the same at first glance but there are two days between these images):
The Makita in the VUB, notice the tight clearance of the housing on the left (the router is in the 'down' position here, when it comes up the left hand side is actually rubbing the dust shroud, on the right there is about 3/8 of clearance). Also notice the excellent quality of the scrap the VUB is made of, and the 1/2 threaded rods sticking out of the top of the router housing, wonder if they will honour the warranty on this one 
:
A view of the root of the blade:
A view of the top, showing the taper:
The blank for tomorrows run in the glue clamps:
(1)VUB: a VUB is a very ugly bracket. Normally I make stuff quite neat, but I was sorely pissed at having this happen in mid stride - again ! - and did not want to spend a lot of time so I used whatever scrap I had on hand and welded a bracket that holds the Makita securely in place. It works, but it's butt ugly, hence the VUB
