I started with fairly typical settings of 80mm/s and 2,500mm/s2. Most printers operate fine in this range, although I should note that many would display ringing at this speed. My MX80L stages should have absolutely no overshoot or ringing unless something has gone very wrong with their configuration.
I chose to try 500mm/s and 5,000mm/s2 for the next test. Maximum velocity has increased over sixfold to 500mm/s, but that is mostly irrelevant. At typical accelerations the axis will never have time to reach its maximum speed - by the time it has built up some speed, the axis is already halfway to its destination.
You can see that although I doubled the acceleration for this test, the axis is severely acceleration limited. It looks very smooth, but the short moves are quite slow.
The third test was at 10,000mm/s2. I cannot raise the speed any higher than 500mm/s because the base step frequency of the Smoothieboard is 100kHz which produces a maximum of 500mm/s at 5um step size. This is never going to be a problem real-world use, but does prevent us from testing the servos to their absolute limit.
You will notice that in practice print speed is mostly limited by acceleration because slicer paths have many short movements. This test seemed to hit a pretty good balance between speed and smoothness but is still somewhat acceleration limited in the short sections.
I performed one final test at 30,000mm/s2 simply to satisfy my curiosity.
My camera could not properly capture the axis at this speed. Individual frames are so blurred that quick moves look more like wild shaking than a set of controlled direction changes. If you watch the video in slow motion you can confirm that the axis fully completes the test program.
To produce these accelerations, the stage must generate 1.5kg (3.3lb) of force at each direction change, and that force produces an equal and opposite reaction in the printer frame and table it is sitting on. If you are touching the table you will feel an impulse every time the stage changes direction almost as if someone is hitting the printer with a mallet.
This was as far as I was willing to push the stage, since going further would probably require bypassing the Smoothieboard and carefully tuning servo gains.
The MX80L stages can actually go much faster than I have demonstrated here - the datasheet claims speeds of 2,000mm/s and accelerations of 50,000m/s2. Its likely that the stages could complete the test program twice as fast as my best run when properly configured. That being said, I don't expect to be able to print any faster than 200mm/s and 10,000mm/s2 so in practice the speeds I've demonstrated are impractical.
My last video is of the homing sequence, which is much more boring than the speed tests but shows all the axes in motion.
The servo homing routine moves until a limit switch is hit, then reverses direction and searches for the encoder index signal. This is guaranteed to repeat to 0.5um under all conditions. It doesn't look very special but you can pick out a little wiggle on the Z axis at the end of the procedure.
No comments:
Post a Comment