TCP calibration - mean error
jay_c
✭
Hello everyone,
I have a question regarding the Tool-Center-Point calibration using the FlexPendant. Does anyone have any tips or suggestions for how to achieve a small mean-error? We have been having some trouble getting this value to be accurate enough. Ideally this value should be less that 1mm, but we are consistently getting 4mm as a result, which is much too large. This is pretty strange because we are use 2 calibration tips which have a very sharp end, and using very small jogging increments to place the 2 very precisely on each other.
So far the steps I've tried are:
Thanks!
I have a question regarding the Tool-Center-Point calibration using the FlexPendant. Does anyone have any tips or suggestions for how to achieve a small mean-error? We have been having some trouble getting this value to be accurate enough. Ideally this value should be less that 1mm, but we are consistently getting 4mm as a result, which is much too large. This is pretty strange because we are use 2 calibration tips which have a very sharp end, and using very small jogging increments to place the 2 very precisely on each other.
So far the steps I've tried are:
- Using more calibration targets (6-8 instead of 4)
- Jogging the robot in every axis quite significantly, before moving back towards the calibration spike.
- Moving the calibration spike to different locations around the work-cell.
- Confirmed that the "fine calibration" is accurate.
Thanks!
Tagged:
0
Answers
-
Check the robot at zero position, I recently had one which was off only 1.6 degrees on axis 2. Hardly looked off center, but after updating rev counters, it stayed centered in the mark. You said checking fine calibration, on should not fine calibrate, but check the calibration offsets in the parameters vs. the sticker values.Lee Justice0
-
Pretty sure that ABB claims Absolute Accuracy calibration improves TCP accuracy.
0 -
As well as moving well clear of the stationary spike in between points, each of the points should be at different angles around the spike for highest accuracy.
0 -
On the same topic
New 6700 robots, fine calibration matches tag, counters all done.
Super accurate postion, tried up to 8 points, use jog increment of 50us and feeler Guage onto spike.
Mean error 12mm(Max >30!) is about average? It's out by about 5mm compared to CAD?
Ideas?
0 -
Check that the system was created with the right robot type. When you jog linear in world or base, does it make a curved path or veer off in a different direction? What version of RW?Lee Justice0
-
To teach a good TCP:
Mount a spike on the tool or the flange.
Mount a spike on a fixed table.
Angle the tool spike 45 deg
Teach 3 points with a offset of 120 deg
Teach the 4 point vertikale.
Observe that there is a angle higer than 10 deg between axis 4 and axis 5 when teaching the points.
To check after teaching:
Select tool and tool coordinate and rotate over the spike.
Regards
Knud Erik Lindberg
Jorgensen Engineering0
Categories
- All Categories
- 5.5K RobotStudio
- 394 UpFeed
- 18 Tutorials
- 13 RobotApps
- 297 PowerPacs
- 405 RobotStudio S4
- 1.8K Developer Tools
- 249 ScreenMaker
- 2.7K Robot Controller
- 309 IRC5
- 59 OmniCore
- 7 RCS (Realistic Controller Simulation)
- 783 RAPID Programming
- AppStudio
- 3 RobotStudio AR Viewer
- 18 Wizard Easy Programming
- 105 Collaborative Robots
- 4 Job listings