# Tool Calibration

• Hi guys,
recently we've installed an ES165D (DX100 controller) with a tool for milling. Appoximately it weights 40/50 Kg, so I start with the calibration( 5 points procedure), but when I verify the tcp rotation around a point I find an great error: the tcp shifts about 2.5/3mm from the center (-2.5/3mm in the reverse rotation, so the error is around 5/6 mm).
Has anyone had a problem like this? Is a problem the load?
Any suggestion, experience or idea.
Thanks

• Hi
I don't think it matters,but, did you put any angle values?
I don't know the shape of the tool, maybe you need all zeroes,maybe not

Retired but still helping

• Hi,
A shift of 1-3 mm is something that you can expect on any robot.
Also, the repeatability of your robot is 0.2mm which makes it less appropriate for accurate tasks.
I would say that 5-6 mm is something to expect with your setup (innacurate robot + heavy tool)
As Fabian said, try to uniformly change the orientation of each of the 5 points.
If it was a TCP error calculation you would have much bigger errors.

• Thanks Fabian an Nubi86 for your opinion.
Fabian: are you asking if I put the angle in the tool calibration? the answer is yes-> for the X=-180 and Y=-90
nubi86:

As Fabian said, try to uniformly change the orientation of each of the 5 points.
Where does Fabian write this? I'vel lost a part of his message?
Anyway I try to change the orientation in a "simmetry" way around my "calibration tool".

I would say that 5-6 mm is something to expect with your setup (innacurate robot + heavy tool)
This is very bad. There are a couple of application that request "NO ERROR(in another world, of course)" or a negligible shift(around 1 mm).

• I guess (actually, assuming) that the robot is calibrated reasonably.

Francesco: Is it perfectly 180 and 90 ? Are they machined pieces ? When you move the robot to 0,0,0,0,0,0 pulses, does the tool (visually) looks -180 and -90 off or you see another weird angle (like 21 degrees) You are right, I think nubi86 misinterpreted me wrong.
Is the tool a gripper (hand) or some kind of gun ? How do you know is off ? HOw do you do the 5 TCs ?

Retired but still helping

• @ Is it perfectly 180 and 90 ? Are they machined pieces ? When you move the robot to 0,0,0,0,0,0 pulses, does the tool (visually) looks -180 and -90 off or you see another weird angle (like 21 degrees). Is the tool a gripper (hand) or some kind of gun ?

Visually the tool looks -180 and -90 and yes it'made of machined pieces ( a mill tool mounted on an electric motor with a spindle).

• stupid question but you never know. do you use your calibrated tool in jogging or a different tool?

• potis
Yes it's the same tool, of course.
But your question is not stupid, because brings the speech to another point:

usually I've always used a different NUMBER tool for jogging in the calibration(for example the tool number 0 to calibrate a new tool that has a the tool number 2) and I don't know if it was correct or no, but until now it works.
During the calibration of this new tool, I've also tried the calibration of the tool 2 using the same TOOL NUMBER in the calibration(2 in this example).
The results are the same in both tests: not good.

• Try 7 point calibration (standard setting for arc Weld in Europe), little bit more accurate, but im afraid unreachable without Yaskawa code(but not really a paid option). Or try calibration by 25 point using special software (MotoCalcv) . If you aren't able do a proper calibration even after several attempts, your abso data can by also corrupted.

• tray a new calidration of robot and after that a new tcp. after that you can consider a call on Motoman representative.
god luck.