The center mounting is a coincident and is only a constrait. The encoder are
doing most of the work.
having the belts point to the center is a key simplification. If they point
anywhere else on the sled, then there is torque applied to the sled and it will
rotate, the expected length of the belt (end of arm to anchor + end of arm to
bit) will be slightly shorter (for some belts) as they will be two sides of a
triangle instead of a line.
this is why the original maslow went from two solid anchors on the sled to
people building links, to the ring with the belts always pointing at the center.
the error can add up faster than you think.
if those belt setups are allowed to rotate rather than being fixed, they can do the same thing as the current arm rotation.
At one point I thought it was as simple as adding offsets to the belt locations, but in checking the math, found it was far worse (and we don’t know and can’t calculate the torque on the sled as that will also include the force of the bit against the wood)
I explored that avenue first, to determeine that the belt exits start to rotate to retraction direction on all corners, off setting the center position aswell. I also created rotation end stops to see if that would help… no luck
As per @dlang, it’ll be interesting to see what the accuracy is away from the centre, without any software compensation (unless you have some?).
From stuff I’ve experimented with, I’d also keep an eye on:
Flex in the router cradle position when there’s load (linear rails on a solid backbone maybe?)
Rollers snapping under load (unless you are printing solid ones which are stronger under shear?).
It seems like a good test of whether the received wisdom still holds true - I tend to think it probably does, but it’s never a bad thing to test it periodically
Related to all of this - I have pondered a bit of we could route the belts through a…loop (this is going to be a bad description as I am on a phone and it needs a description), to give you the effect of a moving exit point that keeps the centre off effect, not a million miles off what you get from core-xy printers.
But given it’s an arc I couldn’t work out how you would do it, and kept coming back to what @ian_ab has tried being the simplest way to achieve the same effect?
