Router is only component on z axis, it is mounted using its fixed base, and can be easily removed, the fixed base would be bolted to the bottom (clamp) plate, no top plate would be required. Any router with base, could be simply mounted.
Mounted to bottom plate are the stepper leadscrew nuts and the two uprights with linear bearings. The spools are mounted over cylinders located to side of router, all spools are on single z-axis. I assume the math is doable. Circuit board/fan would be mounted over one of the spools. Less vibration and heat there, also easy to change any one component easily, belts are independent of router z-position.
I think the main problem is that when the sled moves around and the arms swivel the belt will not always stay in alignment with the bit so the direction of force will not be applied directly to the center of the sled, i.e. bit. That would drive accuracy issues.
The original maslow had this problem when it had hard mounted chains before they moved to the ring concept.
With 4 belts and 4 forces it may be solvable through math. But i’m only capable or simple addition so I have no clue.
there are commercial machines that do this with the belts offset, but we haven’t tackled the problem (we are having enough issues with the existing config) I did a quick-n-dirty PR that would add possible offsets, but it was mixed with too much other stuff and didnt get accepted (and given the issues we are fighting with calibration, not the time to tackle that either)
I have a way to have the belts centered, but not moving on the Z axis, but have not had time to build such a machine.
My alternate solution is to extend the spools via a metal extension ear with a ring at the end that rotates around a low cylinder centrally mounted on the sled.
Each extension would have a different vertical offset and the four rings would take up no more than 1 inch in height.
The spool itself would probably have to extend further out or be of smaller diameter to achieve the clearance needed for the angular range required.
The spool itself would probably have to extend further out or be of smaller
diameter to achieve the clearance needed for the angular range required.
note that the current angular range leaves something to be desired. That’s why
my version goes to just two vertical supports (although the router could be
supported on the red cylinder as well)
I extended the width of my frame by ~12" (I think, might be 18") on each corner and the arms don’t interfere now. I think we’ll want to have some different frame/anchor recommendations based on whether people care about cutting a full 4x8 sheet accurately or something smaller.
With spools mounted on offset cylinders, the centers of the cylinders are the vertices of a square centered on the z axis, constraining the sides of the square to be parallel/perpendicular to the x-y axis’, i.e., no rotation of sled allowed, for each desired xy increment of the z axis location each vertex of the square must move the same xy increment, each belt length delta, can be calculated individually the same as if it is the center.
With spools mounted on offset cylinders, the centers of the cylinders are the
vertices of a square centered on the z axis, constraining the sides of the
square to be parallel/perpendicular to the x-y axis’, i.e., no rotation of
sled allowed, for each desired xy increment of the z axis location each vertex
of the square must move the same xy increment, each belt length delta, can be
calculated individually the same as if it is the center.
The problem is that the belts are springs, not rigid, and they are under
different amounts of tension, so this will rotate the sled and the square will
not be exactly where it’s expected to be.
on a small setup, this may not be enough to notice, but the longer the belts,
the more it will show up.
If they are springs and under different tensions, how can you precisely place
the sled now?
We are struggling to fix all the issues from this, but with all the belts
pointing at the center, we don’t have to think about rotation of the sled, which
greatly simplifies the problem.
