Hi everyone, this is my first foray into the community and so I must apologize for any ignorance on my part. I am entering a new period of my life and have decided I would like to look at operating a FDM machine. Unfortunately I'm not bonkers crazy or bankers draft rich so buying one outright isn't an option for me, not an option considering all the points of failure I see in contemporary designs anyway.
Be it Cartesian, Delta, SCARA or Polar, I haven't found a design that doesn't come with baggage. The reality of the matter is that most of us who construct these machines do it without the resources and capabilities of precision that could be afforded in an industrial environment. With each consecutive linkage between stepper and end effector, the tolerances add up and can result in deviations and errors. Given we start off with difficulty ensuring parallelism already, designs that increase our difficulties further just don't sit well with me.
But this got be thinking, and I eventually came across something on youtube that sparked an idea.
This gentleman did something I've not seen anywhere else, he found a way of using a single true surface as datum for all 3 axes. What gets me about Cartesian designs is that not only does each axis have to be perfectly perpendicular to the others, the nesting of axes cause chain amassing of error, and usually on X and Y there is a second parallel rail there for stability. With Deltas all 3 columns must be perfectly parallel, each double linkage between rail and effector has to be perfect and any deviations will cause deviation on all axes.
My discussion here is limited to the horizontal plane as the vertical one is an entirely different issue with its own set of problems (delamination etc)
However the design in the video, though it appears to solve a number of problems, creates many more.
Though all axes maintain parallelism, or radial predictability, the build envelope to footprint ratio is terrible. His choice to move the platform and workpiece ensures that this design could not be operated at speed due to inertia and fragility.
Lowering the work platform from the true surface (hence why I am concentrating solely on the horizontal plane at present) and instead using the radial arms to manipulate the hotend would be much better.
The linear nature of the arms extension creates an even (albeit radial) distribution of addressable coordinates and the footprint has been reduced greatly, however space is wasted allowing the drive rods to retract.
Changing the drive rods to pivoted linear rails provides the same favorable coordinate layout and saves a little more footprint while still keeping the end effector only one linkage away from grounding. This arrangement sets a precedent, a pivoted linear rail can slide flat and true on both sides of a hole in our single, true surface.
Expanding on the theme, what happens when the pivot points are moved closer to work platform? The topmost portion of the build envelope is good and usable however as we approach the bottom of the envelope we run into the issue of the radial axes being less perpendicular and more parallel, resulting in a deadzone pictured, where the majority of the force from the linear rail goes into working against the other linear rail rather than moving the hotend.
The addition of a 3rd radial axis (though absurd since we are talking about 3 axes on a single plane) solves the deadzone problem by ensuring the no direction of travel at any point in the build envelope is more than 60 degrees from a radial axis to propel it. However the mechanism begins to get bulky, each axis would have to cross the work area at a different, non-overlapping height and then drop back down by as much on the other side to reach true. This opens up the possibility of torsion forces twisting the higher rails up off of the true surface.
But 3 axes to do the work of just X and Y (albeit without having to involve challenging fabrication of perfectly parallel rails)? Surely there must be a better way, perhaps going back to 2, and having a single radial arm with the second articulating it at the end rather than the end effector. This creates an unbalanced coordinate system however, in the area marked any movement on the blue axis would require much more time and travel than it should, resulting in a slowing in that area at best and a print failure from alternating speed at worse. At least though, with only one linear rail crossing the work area its easier to keep things true with no overlap.
Having the steppers and linear rails on the outside of the envelope does seem a good idea, lightening what little resistance they added to articulation in previous iterations. Returning once again to the 3 axis concept, at least the radial arms crossing through the work area are now simple rails articulated from the outside, they can be kept thin and close to the true surface, minimizing the danger of torsion. Apart from weight of the hothead spread among them all the forces they experience are horizontal.
That forms the basis of my thinking so far, a true surface being used to keep everything on the horizontal plane parallel through the use of pivots. Unfortunately the solutions I have come up with are all much much more complicated than the less space efficient earlier iterations.
If only there were a way to create the effect of a distant pivot point without the need to actually have it so far away.
Anyhow, I would love to hear your comments and critiques. My aims with this design are simply fast printing speed and ease of attaining effective precision.