Before i even think about milling something i need to be sure the spoil board is
not crooked, so my first step is to level the board.
The working area of our milling machine is about 1 meter by 1.30 and you have to use a large endmill if you don't want to wait for ages!
Large endmills can be a problem with harder materials but our spoil board is made out of MDF and so i can use pretty aggressive parameters.
I modeled a simple board on fusion360 and generated a 2D toolpath using the largest
endmill we have in our lab: 25mm.
I exported the toolpath as gcode using "grbl" postprocessor.
Our milling machine is controlled by Mach3, i don't like its interface but it
After homing all axis i zeroed Z axis "by hand" lowering the endmill by 0.5mm steps until it touched the board.
I repeated the leveling going down 0.5 each step and after a lot dust the spoil board was nice and flat.
I milled a test model to try to measure runout errors, compare up-cutting vs
down-cutting and try different spindle speeds.
I designed the model on Fusion360, it looks like a skull just by chance :)
It took a lot of time to generate the toolpaths changing all those parameters, maybe it was a smarter choice to generate a gcode by hand or with some python script.
I used a 3mm 1-flute endmill on 18mm birch plywood.
I calculated the runout measuring the space between the "teeth", it's under 0.1mm
but don't trust too much wooden surfaces against my hardened steel caliper.
Scrap wood inside the "eye" milled using up-cut toolpath was ripped off but the surface finish is smoother than the other "eye".
On the "forehead" i tested various spindle speed and feeds, at 2000mm/min the best results are at 15000rpm or above.
After testing our milling machine i realized that the console was a bit inconvenient.
So i designed and builded an articulated arm tha can be rotated o shifted at will.
This design use a lot of joints, to compensate for endmill radius i used an external script for Fusion360 that generate dog-bones "automagically".
I manually splitted M0 commands (rapid movements) in single axis commands because of a problem of the controller board. Hopefully we will repair it soon!
The stock was 18mm birch plywood fastened with wood screws on the spoil board.
The assembly was very laborious and i needed to use a lot of "elbow grease" to
sand down every single piece.
Joints was very tight (i used the rubber hammer!) but i also used a bit od wood glue.
We are updating air milling machine using servo motors instead of steppers and we are mounting the arm on the wall! :)
We just bought some chinese 3D printers that unfortunately have a very weak
acrylic frames, so I designed a "package" of reinforcements including a new
aluminum base plate using SolidWorks.
Anyway i like the CAM integrated in Fusion360 so exported and reimported the model.
Unlike wood milling parameters for aluminum need to be chosen wisely using
shallow stepdown and ramps.
After the initial holes are milled i planned to add some screws to block the aluminum stock more firmly so i paid a lot of attention to safety heights.