The z-axis of my new Cetus 3D printer just plummets down when the power goes off. I want to fix that.
Well, colour me embarrassed. I have recently learned that my Cetus Mk2 is supposed to have a system in place to stop the arm from falling down when the power is off. All it took for me was to tighten a screw. I guess I can shelf this project for now.
There are already a few designs out there that deal with this problem, but they usually just lock the z-axis completely. This is done either manually or when the z-axis reaches a certain height. When the z-axis is locked, it needs to be released manually when one wants to start a new print. I would like to avoid any manual locking or unlocking. Also it is much more fun to design your own solution.
My first idea was to print some sort of spring mechanism that would catch the y-axis when it comes crashing down. In normal operation, the z-axis would push the spring down to build the first centimetre of the prints. The spring would probably not be able to stop the thing completely before the nozzle touches the build plate if the z-axis fell from the top position, but a decelerated impact would be better than a full speed one.
So I designed a plastic spring in Fusion 360 and got printing:
The first iteration already seemed promising, but it was too high. The nozzle could no longer reach the build plate. It took me until iteration 4 to get the height just about right. While printing nr 4, it also became obvious why this parameter had to be tuned:
I forgot to remove nr 3 before starting the print of nr 4. The z-axis could not reach the position it wanted to and the first layer was extruded at a too high position. This reduced the adhesion and the raft peeled off on the right side of the print. Fortunately it stuck on the left side and the print was finished and produced a usable (albeit lightly skewed) spring.
This is when I noticed that the spring approach might not work after all. When calibrating the nozzle height with the spring installed, I got considerably larger values for the z-offset than without it. And it is easy to see why that is happening. The z-axis belt seems to be slightly flexible. At least the z-axis moves a little bit up and down when pushing or pulling on it. This is not a problem if the load on the belt is constant, but when the load changes, the height will not follow the steps of the motor one-to-one any more. This is what happened with the calibration. When the motor has to work against the spring to push the nozzle down, the slight flexibility in the belt means that it has to do more steps to get to the build plate. Since the software only knows about the steps and thinks they correspond one-to-one to a distance change, it tells me that the z-offset is larger that it actually is.
Back to the drawing board.