Hbot/CoreXY support in marlin

[iquizzle]

Good news! The patch to support hbot and corexy printers has been accepted and merged into the Marlin main branch. Those of you looking for hbot support can now use marlin main instead of my github repository.

[Pat K]

What is accuracy of corexy compared with regular Cartesian printers. Did any body perform any testing? What are advantages and disadvantages of corexy compared with Cartesian ? 

Just curious. 

Regular Cartesian -- (example: Makerbot, Ultimaker or Mendlemax) -- Each Axis is independent. X, Y and Z are individually controlled. Error on one axis can not be contributed to other axis.Easy to find rootcause and debug. 

Delta printer -- ( example : Rostok, Kossel, Cerberus) - Each axis is resultant of all X, Y and Z.  Error in one axis will contribute to all others. With some fudge factors and complex algorithm  (with sine 60 and cos 60 functions ) Johann was able to make this work with great accuracy. Based on my experience, Delta  is same as that of Cartesian.

Corexy -- The movement of printer on XY gantry depend on combined effect of X and Y motors. Complexity of algorithm is  simpler than Delta. 

[iquizzle]

Hi Pat -- I have both a cartesian (Mendelmax 1.5) and an h-bot (UConduit). I'm pretty happy with the h-bot so far. As far as accuracy, the only disadvantage or loss in accuracy to an h-bot (that I'm aware of) is gantry twisting. Basically, in order to move the carriage back and forth, the force to accelerate the carriage + friction in the carriage bearings causes slight forces to be applied on the x-ends in opposite directions. This means that there is backlash and carriage shifting that depends on the maximum angle that the linear rod can move through the x-end.

That is why I used two LM8UU bearings on each x-end and spaced them as far apart as I reasonably could (actually they could go slightly farther apart if need be). I have attempted to minimize the angle of the linear rod through the x-end, but LM8UUs are not super-precise. I think a reasonable upgrade would be to use high tolerance frelon-lined bushings (Thomson or PBCLinear) and paired with the correctly spec'ed linear rod. At this point, I have not observed problems in the positional accuracy of the h-bot that would warrant an extra $60 in bearings.

Realistically, corexy is no more complex than a cartesian bot. You can make lines by moving a single motor -- the axes for this motion are just rotated by 45 degrees. The reason NOT to use this "natural" coordinate system is that your build platform would be diamond shaped. By performing the vector transformations discussed on www.corexy.com, the h-bot build platform can be matched with the natural coordinate system of the cartesian printers, so things like the slicing platform shape. If you want to make straight lines with single motor motions, rotate them by 45 degrees when you slice.

[-DK Automation]

hi iquizzle,
is it possible for me to get the copy of that h-bot marlin? could you please let me know your github link? thank!

[iquizzle]

It is not a fork of the main marlin branch. Corexy/h-bot support is included.

See this line of Configuration.h:

https://github.com/ErikZalm/Marlin/blob/Marlin_v1/Marlin/Configuration.h#L254

[Rupin Chheda]

I used this very code version and had a few questions.

Why does it use the min endstops? It has a config code which disables max endstops. I atleast need the zmin and zmax to work because I have a z-probe. do you know how to fix this?

[Y VDB]

that's the thermal runaway protection :p

aint it https://github.com/ErikZalm/Marlin/blob/Marlin_v1/Marlin/Configuration.h#L296 ?