[Vince]

Hi,

I have some questions about the TCP calibration by XYZ plane method. I usually use 4 point method with teaching spikes and the controller's builtin calibration method. In order to improve the calibration I am now using a probe and RoboDK however the I am surprised to see the result is worse.

Should the same point on the plane be touched or should the different postions be as far apart as possible?
When using an ABB robot with AbsAcc should poses or joints be used to feed the calibration data to RoboDK? 

Thank you in advance.
Best

[Albert]

When teaching the TCP by plane I recommend you to pay attention to these details:

1. Change the orientation of the TCP as much as possible (at least 90 deg change of orientation in all directions).
   
2. Use 8 points or more to get a good understanding of errors.
   
3. Use joint values (same joint values ready by the robot), make sure you use at least 6 decimal positions or use the driver to retrieve the joint values.
   

The point found is the center of the probe or the tip of the tool if you use a conical shaped tool.

The constrain is on a plane, so you can use the same point or different points as long as you remain on the same plane.

If you can share your RoboDK project we can better advise.