TCP wanders around

Hi,

We have a Fanuc R-J3iB with only 4400 hrs. The TCP doesn't stay put when moving around a point. For example, when moving around the Z axis at the origin point of a tool frame, the point moves 4 approximately. We have remastered the robot and recalibrated it but that didn't help so we returned it to the original mastering from FANUC. I have made a new tool frame over the same point as the original tool frame point but that one acts the same way.

Three days ago, after remastering for the nth time, I touched up the tool frame origin and also the user frame origin and points. After 6-8 hours of welding and the robot not being shut down at all, the points are off again about 3 mm or so. I tried reteaching the points but that only made the program farther off track. I thought the program was based off the user frame positions but evidently there's more to it than that.

This problem seems to be getting progressively worse. I thought that welding programs needed touched up pretty often but now they need touched up for every production cycle. In addition, I can see a deterioration during the cycle. The last piece welded now needs a bit of handwork to finish it.

I've read various posts on this forum about the TCP but none seem to quite answer this problem.

I hope I've explained this coherently enough for someone to understand. Any suggestions or advice?

Thanks.

I made a new tool frame based on a pointed rod I welded to the post of the jig. The TCP rotates beautifully around the Z axis. So I did a tool frame offset to one of our welding programs but the robot won't accept the new tool frame. It throws the INTP-251 tool frame number mismatch error.

This is getting worse. The above post mentions a new tool frame and the attempt to shift one program to that tool frame. Not only did that shift not work but now all the programs that use that jig are not working and are throwing the same error, INTP-251 or INTP-253. The programs that I didn't mess with at all won't work, the robot throws an error. I select the tool frame pertaining to the program but the robot still throws the same error. So now the robot is stuck.

Thanks for responding cattmampbell.

I know what the books says so maybe someone can explain what the "error cause code" means.

I did get the robot to fix this error. I run a parent program that calls other programs via just changing registers. I made sure the child program had the code line that told it what tool frame to use even though the parent program had that line already. It fixed it for now, we'll see about the next time.

Could someone respond to my first post that was actually about the TCP wandering around?

Thank you.

No. It's bolted.

Are you thinking that perhaps something is shifting? I can't prove, yet, that something isn't but that would be dependent on a crash and then it would be off in a certain direction. Potentially, a person could do a program shift in the direction it's off and then the program would run again. However, for me, this program I'm touching up right now, I probably could've done a program shift since I'm touching most things up about 2-3 mm in the -Y direction. Other, times however, one end of the jig is off in the -X direction and the other end in the -Y and +Z direction, for example.

Thanks for the suggestion hermann, I'll try it.

Quote from RobcadSim

What is "returned it to the original mastering from FANUC" ? If you remaster the robot ,the "$MATER_COUN" data will change. It can not be the same numbers when the robot bought from FANUC.

You're correct, mastering changes those values.

We ran the robot to its witness lines and remastered it. I did this while on the phone with a tech from the company that helped set our robots up. But we had problems with J4 returning to the witness lines. We remastered it on the other side of the witness line but it wouldn't improve. Remastering also did not seem to improve the TCP tracking around a point. So we went in and manually entered the mastering data from FANUC and now the witness lines all match.

3 mm could be consistent across the jig. Sometimes it's not.

Here's a pic of the orient origin point of my tool frame. I started over the point and rotated it around the +z axis 90 degrees. This was before I mastered it again so it's not that bad now. I'd have to check it again.

Could you explain what you mean by a "bad tcp".

I do have a "zero" program that I run to check the witness marks.

Quote from saberdud

Jog each axis individually and see if there is a noise out of the ordinary on any one joint, or if there is a weird motion in any joint. The jog should be smooth through the whole motion. If there is a mechanical issue, this could help find it and tell you what joint it is. I've seen this work, but I know sometimes noises are hard to rely on in a plant environment.

The robot runs pretty quietly but I can check this.

Thanks for the suggestions everyone.

I'll mark this as resolved for now. Thanks everyone for their suggestions. We retaught the TCP with a new welder neck exactly as the manual describes and we also tightened up the robot mounting anchors. The robot is running well for now. If there are any other issues I'll reach out to this wonderful group of people.