Accuracy of BASE calculations

[jimtyrer]

Hallo All,
Has anyone noticed any inaccuracies when teaching a frame using SETUP->MEASURE->BASE->3-POINT method?
I have a frame that is at a compound angle relative to WORLD, I mean that it is rotated slightly about all 3 axes.
I find that when I do the teach of the frame and when I am done and just before I press SAVE I press Meas. Pt and take the values that the robot is using to calculate the angles and write them down.
Then I go to AutoCad and draw lines with the same coordinates and find that I get different angles than the robot does after SAVE is pressed.
I find that AutoCad's calculations are more accurate that those of the robot.
If I take the AutoCad values and go to SETUP->MEASURE->BASE->NUMERIC INPUT and set X Y and Z to 0 and enter the angles for A B and C from the AutoCad drawing the result is more accurate.
If I write my own frame program and let the robot do the ATAN(deltaX,deltaY) calulations I get very close to what the robot gets but still not as good as the numbers from the drawing.

[Werner Hampel]

Well, i think the Robot is Calculating his own Accurate to the Points.
But i dont really know.
But I am sure, that the Result of the Robot Calculated Base is more Exact, than the theoretic Result from Autocad.
But i never tried this.

[jimtyrer]

Danke Werner,
I think what is happening is that the frame I am trying to teach is difficult for a number of reasons:
The reference points I have to use are close to the extremes of the reach of the robot.
The are obstructions in between the the points.
And it appears that when the robot reduces current to the drives and brings the axis brakes on it changes the points slightly.(when I am slow )

In most cases this slight inaccuracy of the frame would not be a big problem.
If I were trying to palletise with calculated positions on this frame then it would make it difficult.
At the present time I am using the AutoCad values and getting acceptable results.
 

[Werner Hampel]

Good Work 

Thanks for the Feedback