My last post described the central configuration options for the 3D printer, and I laid out some basic concepts. Since that time I've rapidly iterated through a large number of design approaches. In general I'm using the idea of combining an outer frame structure with the mechanism mounted inside it. This provides a very rigid outer structure and can be fabricated relatively easily. Initially just the corners of the cube were braced with additional brackets:
This approach had the disadvantage of being difficult to get aligned and tighten down properly. So I quickly evolved to replace the corner braces with full-face sheets. These sheets would be cut on a lasercutter or waterjet and thus can provide additional alignment. This simplifies the assembly significantly:
At this point I started thinking about how to actually mount the X/Y/Z rails inside the cube. In some designs the rails hang from two endpoints similar to several other existing printers. But this approach allowed for too much vibration/error within the motion for my design requirements. I chose to go with fully supported rails for all of the axis except the last one (X). To attach these to the cube I added extra structures inside of the cube:
Although this would have worked, the number of extra (and complicated) components began increasing quickly. Moreover ensuring alignment between all of the components was problematic. At this point another engineer involved in the project suggested a better (and simpler) approach: build the core structure up out of two thick (3/8") sheets and mount the rails directly to those.
As long as the back and top sheets are precisely attached to each other via pins then a precise alignment can be ensured between all three axis. After further discussions we further expanded the angle braces to reduce deflection of the top plate:
Having established the core structural design we began focusing on the details of the axis configurations themselves. I had already decided to go with a CoreXY belt layout for the X/Y axis. This configuration moves the head in X/Y via two interconnected belts attached to the motors mounted at the back:
Normally this configuration would require a twist in the belt where they cross. This adds additional play and can potentially interference between the belts. However I came across an enhanced layout of the belts that avoids this by placing the belts at two different heights:
Another key element evolving in these designs is the z axis configuration. The initial designs used dual motors supporting the build plate. This provided great stiffness, but was difficult to precisely assemble & align. Very quickly we evolved to a cantilevered design using a single lead screw and two rails on the backside (as shown in the later renderings). By focusing on the arm configuration we could provide more than sufficient stiffness for this application while simplifying assembly significantly. Moreover this opens up flexibility in adding automated loading/unloading mechanisms that bring the plate in & out from the front.
Finally by moving to the dual arm configuration in the last rendering we were able to expand the overall functionality to support switching between FDM and SLA modes. This is a key functionality improvement, since for a minimal additional cost we can now support two fundamentally different print technologies:
The final design is a good example of form following function:
The next step is to finish modeling the parts and then start fabricating the prototype.