CV toolpath smoothing

[hmnijp]

So, not only acceleration, but also feed rate effect corner rounding? Fast feed - more rounding, slow feed - less corner rounding?

The maximum speed when moving depends on the centripetal acceleration and the radius of rounding the corners.
https://www.khanacademy.org/science/physics/centripetal-force-and-gravitation/centripetal-acceleration-tutoria/a/what-is-centripetal-acceleration
V = sqrt(a*r)
If the acceleration of your machine: a= 0.3m/s², max R =< 0.1
then the maximum speed on this arc = 0.328 m/min - Feed will not exceed this value.

But the opposite situation is if you set R=<10, F=1000, a=0.3,
then the machine will pass a trajectory with a radius of ≈3.7mm(not 10), since its acceleration is large enough to maintain the set feed on this trajectory but with a smaller radius.

This explains why a small Corner error makes your machine slow and jerky.
The machine will be faster if you increase smoothing or increase acceleration.

Another screenshot from linuxcnc:

[Doncis]

I also working on complex parts (4axis simultaneus), inverse feed G73 of course. Trajectory planner work same way for rotary axis also?

https://youtu.be/4wU5BIvge8I?feature=shared
https://youtu.be/T1L3NnEx1NM?feature=shared

[hmnijp]

I also working on complex parts (4axis simultaneus), inverse feed G73 of course. Trajectory planner work same way for rotary axis also?

https://youtu.be/4wU5BIvge8I?feature=shared
https://youtu.be/T1L3NnEx1NM?feature=shared

I cannot say this with certainty. I don’t know what smoothing algorithm is used in UCCNC when a rotary axis is added to linear movements - after all, the value of the linear error when using a rotary axis is not very easy to calculate - the larger the radius of rotation, the larger the linear error (not angular) - in order to calculate it you need to do inverse kinematic transformations. Is it possible that uсcnc calculates only the angular error?

For example - the linuxcnc trajectory planner does not use blending with lookahead in this case - it switches to simle TP mode, providing only 1 line look ahead - this makes it quite slow.
I think that CNCDRIVE will give more explanations of the CV algorithm for this case...

[Doncis]

In my case 4th axis is set as linear axis. Or trajectory planner works up to 3 axis?

[hmnijp]

In my case 4th axis is set as linear axis.

In this case, the error will increase with the diameter of the rotating part. I don't see any other way to calculate this)

It's better to wait for comments from CNCDRIVE.

[Doncis]

It could be blind smoothing by angle, because we do not have TCPC. Just hard to imagine how to fit corner error into 4th dimension.
I can corfirm, that uccnc multiaxis simultaneus working quite smooth and fast in some cases, wheare toolpath contains lots of small segments.
By The way, you give very valuable information. I will do more testings in the 3axis world first.

[Doncis]

May cncdrive can comment how uccnc CV trajectory planner works on 4 or 5 axis?

[cncdrive]

It works the same way on 1,2, 3, 4, 5 and 6 axis as described in the users manual.

[Doncis]

You have a plan add some NURBS based toolpath smoothing?

[cncdrive]

Yes. We also have in our plans for a long time now to implement S-curve trajectory planning, unfortunately so farwe always had something more important (customers ideas) to implement.