Dynamic spindle RPM when cornering (future feature discussion)

[Jaknil]

I suggest we break out the discussion of dynamic spindle control to this issue. @JensDyvik

Summary of the idea: We want to use the laser mode in GRBL to dynamically slow down the spindle as the feed speed is decelerated due to cornering and the like. If done right this will give us a more constant chip load and reduce unnecessary heat buildup massively when using high rpm spindles.

Things we will need to look into:

• GRBL default laser mode turns spindle off when jogging. Some workaround seems to exist. If we use bark beetle I assume we could also program jogs as air-cuts.
• Spindle acceleration is slower than laser power change "acceleration"
  Example:
  60000 Rm spindle, 0.1mm chipload single flute = 100mm/s feed.
  If we want to do a 90 deg corner and maintain chipload we should in theory go from 100mm/s and 60000 rpm to 0mm/s and 0 rpm (and then opposite when accelerating), this will take 0,3s if we have a 300mm/s^2 feed acceleration. I doubt the VFD can get the spindle from full rpm to zero, or worse zero to full in 0,3 s.
  So then we can either let the VFD try it's best and adjust the RPM less than all the way during short duration speed changes or we lower the acceleration on the feed axes so that the VFD+spindle acceleration can keep up.
  Practically I guess we can start with setting the spindle to max acceleration and see how high feed acceleration it can keep up with before we notice problems.

I guess low melt-point plastics will be an acid-test to see if we have got it right.

Looking forward to trying it!

[JensDyvik]

I dont think we need to worry to much about spindle speed up delay.

1. With constant spindle speed we already have quite good performance
2. The biggest problem is not corners but cutting jobs with a lower speed over a longer time. For instance small diameter helical entry into metal when using adaptive clearing. Or many small features in a row for instance when milling small teeth. So that makes spindle responsiveness even less of an isssue.

I think the GRBL firmware allows PWM duty cycle to be mapped to a minimum value. If this is correct i suggest we start testing at with a range from 25% to 100%. I dont think we have any need for zero spindle spindle in corners.

I think starting with testing with Bark Beetle is a good idea. If these tests are successful we can dive into modding the GRBL firmware so that dynamic laser mode works with M3 instead of M4 commands and that the spindle will spin during G0 moves. This will allow generic post processors to work with dynamic spindle speed.