In the X-Y plane, you can get this effect from micro-stepping of the extruder. Its commonly called moire effect, where a repeated 'waviness' happens as the lines are extruded. This is caused by the systems extrusion being too close to the available torque of the extruder. If you look at the extrusion from top-down, the extruded material would not be constantant, but look more like an hourglass shape wave. This is due to the torque of the motor during the micro-steps. Ways to fix this is higher current on the driver, gearing your extruder, higher steps/rev motor, etc.
Now your upward/downward curve is not the same thing. The moire effect I explained above is extrusion related, and even though cumulatively it looks like a pattern, it is still random in nature. This curve you show that follows in the Z axis has to be something else. My best guess is that it is either something not mechanically true, or it may be caused by the line segmentation of the firmware. By this I mean, the firmware does a trick to make things 'straight' in the Z axis. The firmware still receives g-code movement commands in the traditional cartesian format (go right 1 mm, go up 1 mm). Obviously what makes the delta style printers so cool is the arms move in a coordinated way to accomplish the task. The firmware will take the cartesian coordinate and calculate the distance each motor needs to go to get from point A to point B. The problem is that while the arms move at the same time, the direct path thy would take would not keep the nozzle confined to the Z plane. Think about a position where one arm is almost flat and the other two arms are almost straight up/down. Then command the nozzle manually to move to the opposite side of the printer but not have the Z distance change. The linear movement calculated would end up with the nozzle moving in an arc, dipping down in the middle of the move. Now, there is a trick in the firmware to counter this. It takes each g-code move command and breaks it into a bunch of tiny moves. This way the stepper pulse conversion ends up with a bunch of tiny 'dipping' moves that start and stop in the same Z plane to the point where it is no longer noticeable. This trick is limited by how many segments per second as a setting in the firmware. So a bunch of tiny moves in cartesian format would not be turned into a million micro moves in delta movement. Since you are printing cylinder, arc movement is already broken down. So maybe what you are seeing is an effect caused by the cartesian conversion coupled with constant arc moves.
I have no idea but its a guess.
If this was the case, people would see this effect with any type of printer.
Now your upward/downward curve is not the same thing. The moire effect I explained above is extrusion related, and even though cumulatively it looks like a pattern, it is still random in nature. This curve you show that follows in the Z axis has to be something else. My best guess is that it is either something not mechanically true, or it may be caused by the line segmentation of the firmware. By this I mean, the firmware does a trick to make things 'straight' in the Z axis. The firmware still receives g-code movement commands in the traditional cartesian format (go right 1 mm, go up 1 mm). Obviously what makes the delta style printers so cool is the arms move in a coordinated way to accomplish the task. The firmware will take the cartesian coordinate and calculate the distance each motor needs to go to get from point A to point B. The problem is that while the arms move at the same time, the direct path thy would take would not keep the nozzle confined to the Z plane. Think about a position where one arm is almost flat and the other two arms are almost straight up/down. Then command the nozzle manually to move to the opposite side of the printer but not have the Z distance change. The linear movement calculated would end up with the nozzle moving in an arc, dipping down in the middle of the move. Now, there is a trick in the firmware to counter this. It takes each g-code move command and breaks it into a bunch of tiny moves. This way the stepper pulse conversion ends up with a bunch of tiny 'dipping' moves that start and stop in the same Z plane to the point where it is no longer noticeable. This trick is limited by how many segments per second as a setting in the firmware. So a bunch of tiny moves in cartesian format would not be turned into a million micro moves in delta movement. Since you are printing cylinder, arc movement is already broken down. So maybe what you are seeing is an effect caused by the cartesian conversion coupled with constant arc moves.
I have no idea but its a guess.
Quote
The pattern is due to positional accuracy limits, bound by your stepper motors' degrees per step.
If this was the case, people would see this effect with any type of printer.