XZC Lathe Rotational feed rate - Kinematics?

cts1085 wrote:

Is there a way to adjust/configure the kinematics to take into consideration the current X value and compute the arc feed rate during rotation to be the same as if there was no rotation?

This isn't really a job for kinematics (and no, nothing has changed in newer releases)

The motion control system has no real way to know how long the path defined by a combined linear and radial move is. There is no way to tell if your X value starts at the origin or the material surface, for example. You might know, and it might always be the same way, but there is no means to pass that to the motion controller, and making assumptions would be wrong as often as right. Your lathe will always have X = 0 at the rotation axis, but an engraver engraving onto cylindrical parts would always have Z=0 at the material surface.

The solution is to work out the feed rates at the G-code level (or earlier) and to then use the "inverse time" feed mode to tell the motion controller how long the move should take in the time domain.

Inverse time feed is a bit strange because higher numbers move faster.
www.linuxcnc.org/docs/html/gcode/gcode.h...sec:G93-G94-G95-Mode

It is a bit painful, but would be amenable to an input filter if your system is always set up with X=0 at the rotation axis.
www.linuxcnc.org/docs/html/config/ini_co...l#sub:FILTER-Section

cts1085 wrote:

is this a situation where the move is so short that the acceleration / deceleration exceeds the feed rate?

I don't think that there should be such a situation. Short moves without time to accelerate should be slower, not faster.

In a pure rotary move the current F-rate is interpreted to be degrees per minute so will typically be slow.
If there is a linear move then the F-word is interpreted as mm/min or in/min and the rotary axis moves at the speed necessary to reach the end point at the same time as the linear move.

I think that that is all that you are seeing.

cts1085 wrote:

Is this true with inverse time as well?

Yes, but it won't look that way.

If you are in inverse time mode and program F30 then the move will take 2 seconds no matter how long it is.
So, as you reduce the length of the linear move the linear feed rate will reduce to still take 2 seconds.

The rotary move will still be set at a rate to finish at the same time as the linear move, but in inverse-time mode you won't see tiny linear moves making very fast rotary moves happen.