Callibration of robot and position accuracy

Simotion

I have a robot S4C+ where I have a workobject and tool specified.
The tcp of my tool is specified with the 4 point method : x = 2.77,y = -155.14,z = 1162.76, average error = 1.26, maximum error = 4.16
I have placed an object along the y-axis of my workobject and have put the tcp of my tool at the start and end of the object.
So, only the value of the y-axis changes between those two points and the measured distance by the robot is 1207.5 mm.
When I measure the distance myself it is 1200 mm, so an inaccuracy of 7.5 mm on 1200 mm.
Because I suspected an incorrect callibration I executed a moveabsolutejoint to 0A? of all axes. I noticed that the zero positions of axis 3 and 5 are not as accurate as the other axis (see attachment). Could that be an explanation for incorrect distance measurement.
I also noticed when I do a reorientation that the position of my tcp can move about 5 mm.webwiz/4664/robot.zip

graemepaulin

Hi,

 

Robots are repeatable (typicaly to less than 0.5 mm) but not accurate (typically in the whole mm range), unless you have taken the absolute accuracy option where a correction table is used to correct for the inaccuracies.

 

That said if your calibration is out then that will not be helping.

 

Have you checked the resolver offset values are correct to the label (master) values?

If these are correct then you could reset the rev counters (on the 2 axis that are out) and see if that corrects the small offset you are seeing.

It could also be that the scales have moved slightly?

 

Regards

Graeme 

Simotion

The callibration values on the label of axis 2 are the same as in the controller. I already did a revolution counter update. As you see on the picture the offset from 'zero position' on the two 'bad' axes are in that proportion that I don't suspect the scales to have moved over that distance. I know without the option the robot has no complete absolute accuracy, but is 7 mm on a distance of 1 m not a little bit high?

graemepaulin

Hi,

 

This would be at the high end but from what I have heard would not be out of the ball park.

 

The only other thing would be if the motors on these axis have been changed?

If they have the offset would be different to that on the label.

 

Regards

Graeme

graemepaulin2012-04-29 23:29:53

Simotion

Thanks so far

But what do you think about the picturs of the scales after a moveabsolutejoint to 0A??

Aren't the two 'bad' axes far away from the 'zero point', or is this not 'abnormal'?

Mayarol

It's completely normal

maldrax

This robot had a mechanic intervention?

USLuser1

Instead of using the 4 point method I directly change the motor calibration values on the pendant and use jog.  It is time consuming, but I can get very good results.

I use my machinist rule in conjunction with the calibration marks on the respective axis.  The calibration marks and a precision rule create a vernier scale so you can very accurately position each axis.