kr200 odd movement behavior

    Skyefire,

    So, I have my load set for the tool that I'm using in my milling program. It's as follows for tool 2 : LOAD_DATA[2] = {M 32.0,CM {x 3.533,y 0.05,z 77.77,a 0.0,b 0.0,c 0.0},J {X 0.3703,Y 0.5434,Z 0.3407}}

    When I monitor $LOAD.M $LOAD.CM AND $LOAD.J everything appears as it should.

    When my program is running and I monitor $ADAP_ACC it reads as #STEP1. In checking system variables step1 says without kinetic model while step2 says with kinetic model. Is this indeed correct? ie should I be #step1 or should I set $ADAP_ACC to #step2?

    In robcor both $adap_acc (acceleration adjustment) and $opt_move (energy planning) are set to #step1.

    step1

    $adap_acc : dynamic model without kinetic energy
    $opt_move: higher motion profile without energy planning activated

    step2

    $adap_acc: dynamic model with kinetic energy
    $opt_move: higher motion profile with energy planning activated

    Edited once, last by uberdoom (December 15, 2016 at 9:33 PM).

    I'm only asking because I've started doing more curved surfaces with greater changes in z and there's definitely some vibration back. Motion planning does great over a constant z surface. However, and I'll post a picture tomorrow, there is a pattern to the bounces now that I've expanded the surface area of my millings...like you can almost see the pattern of each axis. It looks cool, but not what I'm looking for exactly :(. That is the bouncing doesn't seem to be random in a toolpath, but definitely has a pattern over x, y, and z.

    Can't say as I've ever had to pay attention to $ADAP. But from what I read, it sounds to me as if it's related to Collision Monitoring more than path accuracy. That said, I suppose it might be worth a try to set it to #STEP2 and see if that has any affect.

    I also found a mention of $OPT_MOVE connected to $ADAP, which relates to the activation of a "higher motion profile." Not anything I've ever dealt with, but it's another option might be worth experimenting with (carefully, of course! -- I'd suggest milling on some nice soft material while you're playing with these settings). But beyond that, maybe Fubini will chime in with some actual knowledge, as opposed to my guesswork. :uglyhammer2:

    Well Skyefire, lets try:

    Basically do not mess with these two switches, because depending on these switches you will need different machine data for the robot to work properly. Unfortunatly I can not go to far into this because there are company secrets involved. KUKA delivers fitting machine data for a certain setting of OPT_MOVE and ADAP_ACC for a certain robot type. Moreover usually there only exits one set of machine data for a certain robot type. So in theory you could create fitting machine data for any switch setting, but since the determination procedure for each setting is very expensive and time consuming this is not done. Additionally there is a historic component to this. Newly constructed machines (Quantec, Agilus) will always get setting OPT_MOVE and ADAP_ACC at #STEP2 which supposedly is the overall best setting. Older machines (parts of the KR2000 Family) might only have #STEP1 because the necessry algorithms (higher motion profile) for #STEP2 where not yet included in the controller when the machine data layout was done.

    Last but not least: These switches only affect how fast the robot can run but have nothing to do with path accuracy, collision detection or stuff like that. They are basically switches for different trajectory planning modes.

    I hope this answer is not to vague.

    Fubini

    Edited once, last by Fubini (December 17, 2016 at 9:45 AM).

    Fubini,

    Not vague in the least. I was just trying to make sure that my input load data was in fact being integrated into the dynamic model by the controller. Because for the life of me I cannot get a smooth surface. I was looking through system variables and a way to monitor the dynamic model (in the system variables for kss 4.1.6) and saw that there were 2 settings for it (step1/2 for adap and opt move). Straight lines (lin moves with no change in z) are fine...which is odd to me because it seems the robot is doing just as much work to maintain a straight line as moving the tcp in the x y AND z. So, I figured it was something controller side.

    I attached some rather dramatic lighting photos of my surfaces and a video to give scale to the changes in z. In looking at that video just now it appears to show the velocity being jerky. That's a product of video compression, it is a very constant speed. Also note that when lights are on you can't see the grooves at all and would come out with a sander (probably starting with 60 grit though and definitely changing the tolerance of the surface to the model). This path was composed of polylines and arcs. I have also tried points separated by uniform spacing (5mm, 10mm, 25mm). I haven't tried all fit arcs yet, so that's next. $apo.cvel is set to 100 and $advance to 5. Any suggestions or is this what I should be expecting? This piece is being finishing with a raster toolpath working in straight cuts back and forth and 1 mm stepovers.

    And now I really want to know what would happen if I set $ADAP_ACC and $OPT_MOVE to step2...must practice self control :fine:

    External Content youtu.be
    Content embedded from external sources will not be displayed without your consent.
    Through the activation of external content, you agree that personal data may be transferred to third party platforms. We have provided more information on this in our privacy policy.

Advertising from our partners