Welding Circular Path

    Hello all,

    I'm integrating a FANUC robot, model R2000iC-165F for a laser welding application and I'm having some problems with circular trajectories.

    When we try to weld small circles or trajectories with small circular radius... the result is not as expected. It's like an oval.

    And this result is with fast... and slow speed.

    Robot has the payload calculated (with robot), we installed Adv-Ctrl Speed and Adv-Ctrl Path, also we tried with more points in the trajectory but at the end it's equal.

    I attach a picture with the programmed points and the result.

    Any advice for programming circles with FANUC?

    I'm by no means an expert and still learning but how confident are you in your TCP? That would be the first thing I would check for this.

    If the TCP seems perfect then I would try finding a similar sized circle and manually teach points going around it. Does this problem still occur if you run the taught program?

    My third attempt to trouble shoot this would be to actually try it with less points. In my job im working with bigger circles (>=12" pipe) but most of the time we are doing significantly less points than this. I wonder if the problem would occur using just 2 circular moves to complete the weld.

    This is a strange one for sure, never seen circle movements be this wavy.just some tips I would try to troubleshoot this with, take it or leave it. Let me know if I've said something wrong, I'm trying to learn more being on here

    Small circles are problematic with articulated arm robots. Generally you shouldn't expect them to be great. It's caused by compounding backlash in the joints.

    Show your code of what you have attempted. Have you tried using the actual circle or Arc movement type.

    There is also a paid option you can install called "Small Circle Accuracy" that can improve results.

    Hi

    KyleAR: measure the tcp of a laser head is quiet complicate, I got it by Roboguide like an approximated tcp. Then I measured another tcp in a spike. Created a point on the working table with a measured frame and then compared both tcps going to a point. After it, i made some corrections to achieve a better tcp.

    Also in my opinion and after some tests, i think it's better with less points.

    HawkME: yes, as you say backlash is a problem, and more in this kind of tasks, also a big robot is not the best choice to achieve great circles.

    Small Circle Accuracy is not supported in this robot.

    I post the code:

    So, to make circles the start and end points must be fines. :thinking_face:

    Something like this:

Advertising from our partners