[Arm-netbook] $150 taobao knock-off 3d printer doing 200mm/sec

[Luke Kenneth Casson Leighton]

On Mon, May 29, 2017 at 6:58 PM, mike.valk at gmail.com
<mike.valk at gmail.com> wrote:

> I remembered an post from Hackaday.
> http://www.doublejumpelectric.com/projects/core_xy/2014-07-15-core_xy/

 oh wooow _that_ post, yeahh i remember reading that aages ago.
learned so much since then.  not least, i *really* do not like the
non-coplanar aspect of that original experimental corexy design: i've
seen so many people be misled into thinking that you can misalign the
belts and everything will be hunky-dory.

 in a corexy design the belts coming off the pulleys at the ends of
the moving beam *must* be in-plane and *must* be at exactly
right-angles.  anything other than that is just f*****g stupid.  not
only do you end up with non-linear motion but the offset corner idlers
(out of plane) means that the belts "ride".

 the simplest way to achieve good corexy kinematics is to have the X
and Y belts stacked one above the other.  that also has the advantage
that the belt termination points can be in the *middle* of the
carriage, which, if you have dual rods (or a single linear rail)
results in zero twisting of the carriage, even under high
acceleration.

in the diagram shown in that post, under high acceleration one end of
the belt will become slightly slack whilst the other end is under
increased tension: that in turn torques the carriage which in turn
places torquing (side-loading) on the rail (or rods).

the best carriage arrangement i saw was the fusebox.  everything
_else_ about the fusebox was questionable but the belt arrangement at
least was superb.

 anyway the double pulley arrangement i am going to try out will have
a 4x reduction in the amount of force on the belt (traded for a 4x
increase in speed, which i will have to think through the consequences
of).

l.