Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Fri Dec 03, 2021 12:51 pm
by cnckeith
parameter 228 is used to turn off S curve when you are running G64. If you aren't running short vector program and G64 changing 228 will make no difference in motion.
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Fri Dec 03, 2021 2:41 pm
by cnckeith
scope shot of Acorn S curve accel decell. 3" line at 100 ipm, acell set to 1 second
G1x3F100 (ipm)
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Fri Dec 03, 2021 3:00 pm
by eng199
drdennis wrote: ↑Thu Dec 02, 2021 12:43 pm
Indeed there is an S-curve. Very interestingly (at least to me) it is implemented through some pretty course dithering. I suspect the dithering is fine, given the inertia in the system, but I would love to hear some reasoning behind it. Thanks.
The spikes in the plot are due to the low steps per revolution setting. If you are only studying the waveform (not running stepper motors), increase the steps per revolution and step rate as high as possible. Monitor the DB25 step outputs to avoid rate limits on the open collectors. The plot will smooth out significantly.
There is no way to output a partial pulse. The calculation of pulses per time interval always gives a whole number. The fundamental unit is pulses, so the artifacts show up even when you change units to inches or millimeters. Steppers use more microsteps and servos use higher encoder counts per revolution for smoother motion. Your plots show why low resolution systems have rougher and noisier motion than high resolution systems.
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Fri Dec 03, 2021 4:27 pm
by Cncninja
Wow, thanks to everyone
I just tracked it and it should be here tomorrow!!
@eng199 Makes perfect sense thanks
@CNCKeith Thanks for the info, For the build its uses C5 1604 4mm for better resolution with the scale of the application, THK20mm rails, a Jagger ATC spindle 100k-RPM, the machine is made up of a Granit base, Epoxy granite/steel columns Its a small work area only 200X200mm buit its going to be moving fast. I have TMC drive but have some clear paths on the way Would you recommend using the RAS in the clearpaths or with the 2 S curves mess each other up?
Here is a video of the RAS its an option built into the drives.
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Fri Dec 03, 2021 5:21 pm
by drdennis
The spikes in the plot are due to the low steps per revolution setting.
The spikes are there because the beagle motion controller generated them. It may be that the motion controller introduces such noise only for lower resolutions microstepping (if you want to call 1000ppr "low resolution"). My point is that the noise is not implied by, nor is it a result of the stepper resolution, rather it is due to the motion controller algorithm.
Your plots show why low resolution systems have rougher and noisier motion than high resolution systems.
My plots show what kind of frequency noise one may expect to see while running Acorn under the condition of my test. By definition, lower resolution systems exhibit higher noise than those with higher resolution. But my plots show that Acorn run in those specific conditions will produce much more noise than what a 1000 pulse microstepping would suggest.
I don't know that the jitter I showed will manifest in any way on a running machine producing chips. If it does not, then designing something "better" would not make any sense regardless of the fact that one could easily make a controller that would not produce such noise. I am guessing from the comments that Acorn will have easier time controlling steppers with much higher resolution; I will test that simply to satisfy my intellectual curiosity -- I have designed precision motion controllers, but never worked on a mill or lathe controller, so this is an educational exercise for me.
Re: Does Centroid have S curves or jerk control in motion planning?
@CNCKeith Thanks for the info, For the build its uses C5 1604 4mm for better resolution with the scale of the application, THK20mm rails, a Jagger ATC spindle 100k-RPM, the machine is made up of a Granit base, Epoxy granite/steel columns Its a small work area only 200X200mm buit its going to be moving fast. I have TMC drive but have some clear paths on the way Would you recommend using the RAS in the clearpaths or with the 2 S curves mess each other up?
Here is a video of the RAS its an option built into the drives.
the clearpaths will work well, they perform better than steppers partially due to the buffering and internal pid controls Tecnic has in them, several machines that i've seen performed very nicely with them sized right and tuned, nice smooth action.
next step up would be Estun's, Delta's or Yaskawa's etc.
Re: Does Centroid have S curves or jerk control in motion planning?
The spikes in the plot are due to the low steps per revolution setting.
The spikes are there because the beagle motion controller generated them. It may be that the motion controller introduces such noise only for lower resolutions microstepping (if you want to call 1000ppr "low resolution"). My point is that the noise is not implied by, nor is it a result of the stepper resolution, rather it is due to the motion controller algorithm.
Your plots show why low resolution systems have rougher and noisier motion than high resolution systems.
My plots show what kind of frequency noise one may expect to see while running Acorn under the condition of my test. By definition, lower resolution systems exhibit higher noise than those with higher resolution. But my plots show that Acorn run in those specific conditions will produce much more noise than what a 1000 pulse microstepping would suggest.
I don't know that the jitter I showed will manifest in any way on a running machine producing chips. If it does not, then designing something "better" would not make any sense regardless of the fact that one could easily make a controller that would not produce such noise. I am guessing from the comments that Acorn will have easier time controlling steppers with much higher resolution; I will test that simply to satisfy my intellectual curiosity -- I have designed precision motion controllers, but never worked on a mill or lathe controller, so this is an educational exercise for me.
one reason why we don't recommend a step per rev value below 1600.
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Sat Dec 04, 2021 2:14 pm
by cnckeith
Cncninja wrote: ↑Fri Dec 03, 2021 4:27 pm
Wow, thanks to everyone
I just tracked it and it should be here tomorrow!!
@eng199 Makes perfect sense thanks
@CNCKeith Thanks for the info, For the build its uses C5 1604 4mm for better resolution with the scale of the application, THK20mm rails, a Jagger ATC spindle 100k-RPM, the machine is made up of a Granit base, Epoxy granite/steel columns Its a small work area only 200X200mm buit its going to be moving fast. I have TMC drive but have some clear paths on the way Would you recommend using the RAS in the clearpaths or with the 2 S curves mess each other up?
Here is a video of the RAS its an option built into the drives.
great. be sure to go with the native 6400 pulses per rev Clearpath units, they make two different types the less expensive ones are 800 steps per rev, avoid those for this type of application, get the 6400 units.
Re: Does Centroid have S curves or jerk control in motion planning?
Posted: Mon Dec 06, 2021 11:44 am
by Cncninja
Look what i got today!!
Re: Does Centroid have S curves or jerk control in motion planning?