Tool calibration error

Yes. If you just set A2 by eyes it will be a problem.
All position calculations are wrong at this point.
The further away from A2 the worst the error.

Reason robot drops the message about mastering error is that it lost the motor position (transportation caused mechanical movement of the gearbox and motor position does not match any more) Referencing by eye will always be wrong.

Just so we are clear.
Robot math will always be correct. Meaning it will calculate like everything is in perfect order.
But your actual position (mechanical) versus calculated position will be off.

with properly calibrated robot and correct tool load you should have no problem getting error of about 0.1mm or so.

G'day,
Seems like tool calibration issues are the order of the day, so I thought I would tag onto the end of this thread.. I hope that is OK .....

I have just crashed the welding torch on our Kuka Robot into the welding jig. The torch is fitted with a Fronius 'crashbox'. There is no obvious damage to the robot.
When I stepped through the next weld the torch was at least 5mm out of position. I did a tool calibration and came up with an error of 4.2mm(!!!). Sure enough when I tried to use the mouse to rotate the tool around the TCP the tip 'corkscrewed' away from the datum point by about 10mm during a 180 degree rotation.
When the crash occurred I did not get any axis remastering error messages.

I have called our local Kuka agent but he cant get here until later this week.

Can anyone suggest what I should look at next?

Hello,
have you checked A6 axis?
If everything with axes will be ok, then you should be able to make proper tool calibration. (Even if tool is bend or is moved on robot flange). For first, you can check visually, if marks on flange are matching, when A6 axis is in 0.0 deg position. Maybe something happened with A5 and A6 belts, if crash was hard?

Thanks for your reply animisiewaz,
The local agent turned up today. We checked the Fronius crashbox, the belt drives to A6 and A5 (all good) , and we put the dial indicator in and checked zero position on A2, A3, A4, A5 and A6. A4 was 1.4 degrees out. He remastered the A4 axis and then we tried to recalibrate the TCP using the 4 point method. Came up with 13mm error.
He has no ideas and has gone back to check his notes, and I have just sent him a copy of the latest archive.

Any ideas?

Regards,
Rod

how about A1?

Quote from Ozducati

Thanks for your reply animisiewaz,
The local agent turned up today. We checked the Fronius crashbox, the belt drives to A6 and A5 (all good) , and we put the dial indicator in and checked zero position on A2, A3, A4, A5 and A6. A4 was 1.4 degrees out. He remastered the A4 axis and then we tried to recalibrate the TCP using the 4 point method. Came up with 13mm error.
He has no ideas and has gone back to check his notes, and I have just sent him a copy of the latest archive.

Any ideas?

Regards,
Rod

If he did A4 he needs to redo A5 and A6 as well. Wrist axes are coupled together (software wise) (check attachment)
If only check mastering was done and new values were saved for A4 that will explain why A5 and A6 were not done afterwards.
They still needs to be remastered.

General rule is (KUKA Manual) that you have to remaster the robot after the crash. Even though it is not asking for it.
There is mechanical deviation after the crash not detected by motor resolver (no error message on mastering)
So you start with unmastering (deleting) all axes and than remaster all of them.

Thanks for the reply.

We checked the error on A2 to A6.
This robot is different from the kukas that I have previously worked on: this one has the A5 and A6 drives mounted next to the A5 knuckle (see attached) instead of up near the A3 knuckle, so A5 and A6 do not have to compensate when A4 is actuated.
Even though we checked each axis for angular error I think you are right in respect that we should remaster the whole robot.

G'day panic mode,
regarding A1: the tech said that he didn't check the angular error on A1 because that wont affect the tool calibration (?!!)

I did not think this sounded right but he insisted that he knew what he was doing.

I'll keep you posted on any progress.

Regards,
Rod

that was very stupid of him

when teaching TCP it is extremely unlikely to get 4 poses without moving A1. of course it affects the tool calibration. even if A1 was not moved, its position is factor in transform that computed cartesian position of flange and TCP.

you want good results? make sure:
a) robot is mechanically ok (belts, gears etc)
b) robot is using correct MADA
c) robot is calibrated properly

only when all of above is met it makes sense to measure TCP

So what type of robot is that exactly ?

Looks like one of the hollow-wrist MIG-welding specialty models.

Seems like they don't have coupling between 4 and 5,6.
I see only coupling values between 5 and 6.

The tech says its a KR16 L6 HW ARC..... I guess the HW stands for Hollow Wrist

Just to check... What are you using to teach TCP? Are you replacing the tip with a spike or just using a length of weld wire?

If you are using the stick-out/wire; the wire will likely not be straightened perfectly and so will twist/bend/move in the liner as you re-orientate the robot.... resulting in poor TCP.

also if using weld wire - length need to be correctly trimmed before measuring TCP