Okay, so a little update.
I checked my marlin settings and in the EEPROM, there was an imbalance between the max feed rates and accelerations for X/Y, and Z. Now, this shouldn't make a difference as the firmware should, when set to delta mode, use these to control effector kinematics and not carriage kinematics. As each axis does not operate as it does in a cartesian printer. But setting them all to the same values has solved the problem of being limited when a move required a fast change in Z carriage position.
So that's 1 down.
Next, I replaced the top idler pulleys with toothed idlers and in the process tightened my belts.
I recalibrated and achieved a standard deviation of 0.02 via the auto-calibration routine. I am using a Precision Piezo under bed sensor system, which is so sensitive it can trigger if I blow on the print bed.
I also printed a 50x50x50 cube using a segments value of 30, based on my results above. The lower half of the cube is 0.202 layer height and the upper half 0.161.
Its a cube with a smoothness some order of magnitude smoother than before, which is a great result. Interestingly you can see the tree ring pattern in the upper half printed with the 0.161 layer height, but not the lower half printed with the 0.202 layer height.
So a fair few issues solved here. Now to try the S-curve acceleration control.
Simon Khoury
Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions