I have no idea where this came from, I just only now ran across it…but I thought it was interesting:
It’s a Delta/Cartesian Hybrid. X/Z axis delta, Y axis Cartesian.
That’s a great way to make sure your bed is level! I’ll have to try that.
Worst of both worlds.
Why did it need a round table?
Can I just say that I’m really tired of people calling every parallel robot with a parallelogram somewhere in it a “Delta”? This is not a Delta. There are a few different things you could call it (most accurate probably being a 2-DOF planar parallel robot, with a separate Y stage bolted on) but Delta definitely isn’t right.
I came here basically to say what the previous three comments did, so I feel my work here is done.
At least it’s not a drone.
Some people like deltas partly because the bed is stationary, and some else said “nuts to that!”
@Ryan_Carlyle I always thought that it had to do with the need to calculate the triangles between effector, arm, and linear rail, thus delta.
@Ryan_Carlyle Yeah, 2-DOF planar parallel robot with a separate Y stage just rolls off the tongue better than stupid delta.
@ThantiK No, “Delta Robot” was originally a proper name for a specific design of parallel robot invented by Raymond Clavel. It was originally a ROTARY actuator Delta like the Firepick, but with a fourth gripper rotation DOF implemented via a telescoping double U-joint shaft to the center of the end-effector. That’s why we capitalize the word “Delta” – it’s a given name like RepRap or CoreXY.
Over time, the term “Delta” has become genericized to refer to the particular spatial 3-DOF arm arrangement of three push/pull parallelograms, which was the brilliantly clever part of Clavel’s original design. What’s special about it is that the three parallelograms reinforce each other: each parallelogram only stays “in plane” and thus able to provide rigid rotational constraints because the other two arm parallelograms are also in-plane and thus providing their own rotational constraints. Each of the three kinematic chains is badly under-constrained until final assembly – and then it all magically becomes a highly-rigid 3D translation robot with fairly simple kinematics math and minimal parts-count construction. That’s very special and unusual behavior for a parallel robot design. So the parallelogram arm arrangement is really what we should be using to decide whether something is a “Delta” or not.
Compare to the “Deltesian” we have here. It’s badly overconstrained and the parallelograms aren’t being used the same way. Note the mid-span rod clamps designed to keep the arms from swaying out of the XZ plane. That’s a weird design choice. Why does it have ball joints at all? This kinematic arrangement is truthfully a linear actuator version of a parallel-SCARA robot (ie like a Morgan). But it was built out of Delta drivetrain parts like ball joints and rods for no good reason, which made it needlessly more complex.
If you simplified the weird “Deltesian” construction to its basic kinematic chain arrangement, you’d only need three arm links: one SCARA-style rotary arm with hinge joints on either side, and one push-pull “parallelogram” rod on ONE side to keep the end-effector parallel to the build plate. Like this 2-DOF parallel robot, but with column linear actuators instead of rotary arms: https://i.pinimg.com/736x/41/22/56/41225658582a239bee86bcd1f1fbf0c9--a-robot-d-print.jpg
@John_Best It doesn’t have a good classification name, because it’s fairly novel. But it’s explicitly not a Delta, which makes the given “Deltesian” name a poor choice.
If I build a polar printer out of threaded rods, should I call it a “RotaMendel”? No, that would be a dumb name, because threaded rod construction isn’t the defining characteristic of a Mendel. Likewise, building a 2-DOF parallel robot out of a weird mishmash of column Delta parts doesn’t mean you should put “Delta” in the name. Using ball joints and column linear stages isn’t the defining characteristic of a Delta.
Also… Here’s the same basic mechanism arrangement, circa 2006. http://computationalnonlinear.asmedigitalcollection.asme.org/article.aspx?articleid=1394698
Name used: “two degree of freedom planar parallel manipulator, which is a subpart of a hybrid machine tool” Catchy, right?
Is there any benefit at all to this kinematics style over just cartesian or just “delta”? It seems like it has all the issues of “delta” calibration plus the print volume inefficiency of cartesian with a moving bed.
@Adam_Steinmark looks cooler.
Hello
How are you doing
@Ryan_Carlyle I’m the guy working on this monstrosity! I’ve been looking for any research into this style of machine and am excited to review that ASME Paper.
I never intended to turn this into a project. I had a cheap Kit Delta that I got tired of tweaking. I purchased a Cartesian and no longer needed the Delta. It sat in the corner for a while until I decided to disassemble it and while thinking about what I could build with the parts.
I was planning on a full Cartesian conversion, but before I ordered the linear rails, extrusions, belts, stepper etc. this dumb idea came to mind. I honestly didn’t expect it to work. Here’s the tech demo I did for myself zip-ties and all: https://gfycat.com/HiddenJealousGuernseycow.
I was already planning going to full Cartesian route when it failed. And then it didn’t.
The “Deltesian” name is as much about the kinematics as it is, the fact the build got off track halfway between a Delta and a Cartesian.
As far as the kinematic equations go: it has much more in common with a Delta than a SCARA. In Marlin, I set tower 1 & 2 angles to 180°, zero’d out all the values for the 3rd tower in the dot product calculations and substituted Cartesian code for the Y-Axis. Yes, I could have rewritten the XZ moment as 2D dot product vector calculations, but I just wanted to see if it would work. So it’s currently running with a phantom 3rd tower, generating the dot product vector results, outputting to nowhere.
Ultimately, a lot of the design is non-optimal. However, I do see value in a design which can take a $300 Delta kit, 1kg of plastic and (2) 220x220 build plates and produce a printer that has a 220x440x250mm build capacity.
I’m very curious to hear your feedback!
More gifs if you’re interested: https://gfycat.com/@bornity/albums/deltesian_project
@Rob_Stuart_bornity thanks for posting. Really interesting way of doing the kinematics!
@Ryan_Carlyle Thanks!
That paper is awesome, btw. It will save me a bunch of time, not having to derive the equations for the optimal part dimensions.
I found it interesting the Chinese worked on the 2-DOF design with a 2-DOF rotational machine head and then found it worked well for large blades and guide vanes for hydraulic turbines. Well enough they built at least one more. 8feet between the towers just massive.
Honestly I’m really curious why this design, modified for the 3D printing space, isn’t already a thing.
IEEE link if anyone can’t get it from ASME: http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6076213