This motion starts off small from the furthest side from the bent 8mm rod, and increases in amplitude as the carriage approaches the side nearest the bent rod. For instance, if an 8mm rod is bent, as that rod spins to move the axis, the axis is forced up and down (and side to side, but that is much less visible) with the motion of the smooth rod. The sine wave error on Ultimakers comes from the 8mm smooth rods that are used for linear guides AND belt transmission.

For example the droop due to gravity in large robotic arms is characterized and counteracted to achieve precision at larger scales. Also, while ringing is basically a structural issue that needs to be solved physically in this case, known offsets could be calibrated out if sensing is relatively cheap. I’m interested to know why a sinewave offset would occur, do you have any resources on that? I would have expected a different error. I’m sure the k2 printer could be improved to lose quality at a higher speed, there are probably some ratios between belt/rail rigidity and moving mass that could be optimized further (at an increased cost).

However, you do have to start somewhere and some of these projects lead to more polished versions in the future after the creators fill in some of their knowledge gaps. A lot of the fundamental physics behind these systems is kind of glossed over by hobbyist grade efforts. People have tried to use the same belts, motors, and rails from small printers on square meter gantries with predictable results. Thanks for the insight, scalability is definitely a huge issue. And that is the setup doesn’t scale, increasing the size causes all kinds of problems (including not enough stiffness, and making it stiffer increases the weight a lot more then other setups), scaling it down causes a different set of problems (including cost)

Small gain, for a lot of increased cost (more motors, expensive rails)īut, it doesn’t solve one of the bigger drawbacks of the Ultimaker gantry setup. Now, the linear rails should solve one of the Ultimaker issues, which is X/Y/Z offset depending on the head position, due to the axes not being a 100% straight you get sinwave offsets in all directions. Acceleration/jerk settings are slowing it down a lot. You can also see it’s not moving at 400mm/s, far from it. Sure, you can go crazy fast, at a high cost of quality.Īnd putting a heavy print head in there just makes it worse. Balancing the mount of ghosting/ringing vs speed is something we where struggling with all the time. And, we wouldn’t have called this print “beautiful”, far from it. We called this “ringing” instead of ghosting at Ultimaker. Posted in 3d Printer hacks Tagged 3d printer, croxy, k2 Post navigation For more design patterns contributions, have a look at Kinematically coupled bed design. With commercial 3D printer manufacturers spending the last few years in a race to the bottom, it’s exciting to still see new design pattern contributions that push for quality and performance. When finished, the probe simply slots back into its holster, and the print can begin.īoth CroXY and a variant of K2 are up on Github complete with bills of materials if you’re curious to poke into the finer details. Once mounted, this probe becomes the lowest point on the carriage, allowing the carriage to travel around the bed probing points. When the machine needs to level the bed, it travels to a corner to “quickdraw” a magnetically attached limit switch from a holster.

Both gantries also feature a unique take on a removable Z probe. However the overall belt length has been reduced tremendously, resulting in a much stiffer setup.īut the innovation doesn’t stop there. Like the CoreXY design, this setup is tailored for clean prints at high speeds since the motion-related motors have been removed from the moving mass. Each crossed rail controls a degree of freedom with vanilla Cartesian kinematics, but each degree of freedom also has a redundant motor for added torque. Borrowing concepts from Ultimaker’s crossed gantry using rods, the Hypercube Overkill project, and perhaps even each other, the results are two compact machine frames capable of beautiful prints at extremely high speeds–upwards of 400 mm/sec in case!īoth gantry designs take a rotated MGN12 rail (a la the Railcore) and cross two of them, mounting the carriage at the intersection point much like an Ultimaker. In the last few months, both and have been piloting a rail-derived crossed gantry architecture, a “CroXY” as it’s come to be known. Simply having a few go-to 3D printer motion system designs is no reason to stop exploring them, as even small iterations on an existing architecture can yield some tremendous improvements.

Hypercube 3d printer