We received another interesting question from @NicoHanstein who is experimenting with the path recording for deburring with their spindles.

When we programmed with path we realized in certain positions that the robot was moving back and forth (was almost jumping) when we pressed our spindle against the work piece. When we changed the joints of the robot and exerted the same pressure again this behavior was better and almost gone.

At the moment, as you can see in the video, the spindle is not mounted in alignment with the TCP of the UR.

The spindle is mounted in 90° to the Z Axis. Is this causing this specific problem? Is there a best way (axis) to install the spindle to the force sensor?

@NicoHanstein  I don't think there is a orientation to mount the tool better than others. Maybe trying to align the tool such that when you are programming the path you would push and pull in X-Y axis only and not in a combination of 3 axis. I am not 100% sure though if that would improve. @MarcAntoine_Lacasse you have tested this a lot more than I did, can you comment on this?

You can see the robot as a mass-spring-damper system.
If you exerce forces on the robot near its resonance frequency you will probably get buncing.
You can solve this by modifying some parameters.

- decrease or increase robot speed to modify the external force frequency
- change the configuration of the robot to modify its stiffness. The robot is stiffer when the tcp is nearest its base.

Spindle mounted orientation will also affect the stiffness and so the buncing but I think the concern is more about reducing the distance between spindle and robot flange.