TCP drifting during rotation

Hello,

I was dealing recently with a precision problem concerning our robots. Problem was, when copying a TP program with user frames from one robot to another, the program always had minor differences although it shouldnt. TCP on both robots are the same and based on CAD and all the programs refer to correctly set user frames so in theory the points should always be the same, but they are not. There are always minor differences approximatelly a FINE give or take. Robots are M710ic-50E R30iB.

Robots are fairly new, 2 years old never had a mechanical or electric failure and they are factory mastered. Never remastered. Running a test on a laser pointer, showed that TCP drifts about 0.5mm during rotation around Z axis. Also, when changing CONFIG of a point from NUT to NDB there is also a drift around 0.5mm on the same point. As I saied earlier the TCP is based on the tool's CAD although I tried reteaching with six point and 4point method but the drifting on rotation still exists. The drifting of the point when changing CONFIG from NUT to NDB suggests a wrong mastery, but, can a fanuc factory master be inaccurate ?

Quote from HawkME

Fanuc factory mastering is not perfect. It is usually pretty good at zeroing out the joints, but does not address the accuracy issue of the robot.

So what do you suggest? I have a robot accuracy enhancement technique in the form of Engineering Bulletin published by FANUC but it involves remastering axis J2 J3 J4 J5. Should I give it a try? I am really hesitating to mess with the factory master but especially the minor difference when changing CONFIG from NUT to NDB suggests a minor false in J2-3 mastering. Do you think a 0.5mm drift is acceptable for this kind of robot ? For my application it sure isn't.

Quote from HawkME

How important is it for you to get this perfect?

Every robot is slightly different due to manufacturing tolerances. You can try the method in the bulletin and see if it helps, but first take a backup. The bulletin method will remaster the robot, but not correct for those manufacturing differences. Maybe that's good enough for you.

The only way to correct those differences is to pay Fanuc to do iRCalibration Signature. It is not cheap, about 10% the overall cost of that robot.

Display More

It is very important but not unbareable. It forces me to basically finetune every program I copy from one robot to another. The bulletin simply uses a technique with point config (NUT, NDB etc) to check for any differences and correct them. iRCalibration Signature costs 7.500 Euro ?????

Quote from HawkME

Every robot is slightly different due to manufacturing tolerances.

This is why all programs are referenced to custom User frames. Every point created is in respect to a user frame so in theory, copying a program from one robot to another should be exactly the same. But it isnt. commanding the robot to go to 0 in respect to the user frame, touches the center of a shaft. Copying the same program on another robot with same user frame, robot is away about 0.5mm. User frames are checked, no problem there.

Quote from HawkME

User and tool frames don't correct for differences in the fabrication of the robot arm itself. The signature cost is in that ballpark, I don't know your pricing so you'd need to get a quote from Fanuc.

Quote from Doctor_C

I have utilized the Robot accuracy enhancement technique many times, mostly the J5 portion.

It is much better than Calibration right out of the box.

Second, can you teach the tool frames. Robot is much more accurate than direct entry.

It turns out the accuracy enhancement technique made a much better Mastering result than the factory mastering the robot came with. I was really surprised as I know FANUC uses special tools to calibrate a robot before it is sent to a customer. Both robots were out on J2, J3, J5. I cannot calibrate J4 unfortunately as I am using an offset wrist and the procedure listed doesn't support an offset wrist.

Doctor_C, before I resort to this Mastering technique I tried re-teaching both TCPs with 6-point method but also 4-point method ( I already knew the angle my tool had) which is more precised but the drifting was there.

Quote from Doctor_C

You taught them before Robot accuracy enhancement,

what i meant is after doing the "precision mastering" as we call it, you should redo your toolframes.

Hope this helps

I use 2 TCPs. The cad TCP after the accuracy enhancement technique rotates near perfection. Drifting is about 0.2 to 0.1mm and I am sure this is about as good as I ll get considering M710i is a medium payload robot. I didn't bother teaching the top of the TCP as I am fine with the CAD TCP right now. Thanks for all the info. All this time I was hoping that re-calibrating the user frames together with a good TCP for each robot would correct the "bad" mastering but it seems I was wrong.