Runout problem with JT4, JT5 axes

    Hello!

    I have a Kawasaki RS007L Robot and I encountered a positioning problem. At first I noticed a very strange behavior of the TCP point. To figure out what was wrong, I 3D-printed a calibration pin with clearly defined dimensions and installed it as the TOOL of the robot. Then I rotated the TCP relative to another sharp object and noticed that as long as the robot didn't change orientation, the TCP held its position quite accurately. But after the orientation is changed there is a deviation of about 2mm. And in this orientation this shifted point is already held. (see video1.mp4). I saved this point in one robot orientation (Translation point), changed robot orientation and moved back to this point. And there also was an error. Same XYZOAT coordinates in robot, but different in reality. The first thing I decided to do was to check the zero marks and set the robot to the 0 position on all axes. The marks matched. I noticed that the 5th axis was not completely symmetrical (photo1.jpg) and photo2.jpg). I decided to use a micrometer and found that rotation of the 4th axis had a runout of about 0.8 mm in amplitude (video2.mp4).

    I don't know, is this a normal behaviour, but in K-ROSET 5th axis looks symmetrical. And if it is a normal behavior, then how to fix the positioning error in different orientations.

    Because maximum file size is 12 MB, I uploaded all files on Google Drive: https://drive.google.com/drive/folders/…iAG?usp=sharing

    Thank you in advance!

    I haven't had any experience with new RS007 arms but from what I've seen on your videos it does indeed look as though a mechanical shift has taken place at the wrist section and inline with your rotational runout checks.

    I don't think incorrect zeroing would have any effect as the it's not on the rotational part of JT5 or JT4.

    However it should be as mechanically zero'd as accurately as possible.

    I don't know if there are any precision jigs available for this arm, I don't see any in the documentation.

    Has it had any collisions - big or small?

    Is there any mechanical motion - left or right at the wrist section, with any minimal force?

    I wonder if a bearing has shifted inside the housing or possibly any shims within the assembly?

    What about warranty, does it still lie within warranty?

    I don't know what else to suggest as a test except if you just used joint mode and rotated JT4 whilst JT5 and JT6 remain at mechanical zero with a fine point/needle centrally mounted to the flange and 3 DTI's placed around the tip of the tool, then if the wrist section is not central, then you would see the deviation on one or more of the DTI's as you rotate JT4 between 180 and -180 degrees.

    This would tell me for sure the wrist section should be removed for further investigation into the wrist section, but also the housing each side.

    However, from the outset It does look like a strip down to investigate that further though as if the wrist section has moved laterally, then you will never achieve any sort of accuracy as it's current mechanical datum will definitely be outside of the kinematic modelling tolerances so there would be a mismatch between the physical position and calculated positional values and will very unlikely produce any errors during use.

    Thank you!

    * Yes, it had a couple of collisions, but usually the tool broke (I was printing it out of PLA plastic) and I didn't think it could have affected any mechanical shift. But apparently I was wrong.

    * I did not observe any motion when applying force in the wrist section.

    * I could take the robot body apart, but I couldn't find any instructions. I'm afraid that specialized jigs may be required for reassembly. If not, I will take it apart and see.

    * Alas, the robot is no longer under warranty, and there is no official support in my country.

    I have an access to robot once a week, so I will prepare tools for your test and make it as soon as possibly I can.

    Hello again!

    An update to topic:

    I got access to another RS007 Robot and found that different distance between left side and JT5 axis and right side of JT5 axis is normal.

    I also found that the runout occurs in the direction of travel of the JT5 axis. I used a micrometer and zeroed the 5 axis to get rid of the runout.

    In doing so, the position of the marks did not match. But considering that on RS007 robots, the calibration marks are just glued-on pieces of paper, this is not surprising.

    Now everything is fine! Thank you for your support.

    I personally still have doubts about the integrity of your wrist section, but this is based on your description of the problem and not anything I can conclusively expand on.

    I hate the fact Kawasaki still use this 'gluing method for scribe lines', yet on other arms it is cast in :wallbash:

    Good to hear you have it resolved though.

    I shall place this in my little black book should I come across something similar on RS07 arms in future.

    Many thanks for feeding back your results.......... :top:

Advertising from our partners