What is wrong with the PTP speed in T1 on the KSS8.7

  • Hi,


    The PTP speed in T1 mode on the KSS 8.7 is unbearably slow.

    How slow? Well let me tell you with the speed test example I run on two systems.

    From Y -300 to Y 400 so a 700 mm running distance and measured the time in T1 mode while at 100% speed on the Teach Pendent.


    System 1:
    KSS 8.6.4

    KR C4 Compact

    KR10 R1100 Agilus 2 robot (A six axis one)


    System 2:

    KSS 8.7.4

    KR C5 Micro

    KR6 R700 Z200 (A four axis scara robot)



    The results.


    System 1 (KSS 8.6.4)
    PTP time: 9,684 sec

    SPTP time: 3,852 sec

    LIN time: 5760 sec (Path was close to singularities so slower then necessary)


    System 2 (KSS 8.7.4)

    PTP time: 47,508 sec

    SPTP time: 3,744 sec

    LIN time: 5,964 sec


    This is nor normal and causes the programmer to test the programs in other modes which is not a safe way to avoid collisions.

    Its over ten times slower than SPTP. Why?
    Considering this was a 700mm Point to point distance... the speed was 14,7 mm/sec ...

    The speed in T1 should be limited to 250mm/s


    What is going on?

  • The differnce between ptp and sptp is usually:



    To investigate further it would be easiest to have a krcdiag of 8.6 and 8.7 to lookup what differences in relevant machine data might be.


    Fubini

  • I do not understand how a link to this thread gives an answer on how to solve this problem.

    This thread has another link to German version of Robot Forum.


    Fubini wrote there:

    Quote

    The quick translation is as follows:

    This is due to reductions like $red_t1. This is added to the override in T1. If you have programmed slow speeds in a PTP movement, it will be even slower. By the way, this is no longer the case with Spline. There is always limited to 250 mm/s no matter if LIN or PTP, because there are no more factors like $red_t1.

    Such general factors usually have historical reasons. In the past, PTP could not be limited to Cartesian 250 mm/s as a pure movement in the axis space. In the meantime this is possible via $speed_limit_teach_mode. But for compatibility you won't get rid of the old method as default. But you can change it somewhere ($custom.dat?). I don't remember the name of the variable, $Red_t1_ov or something like that.


    I am not able to find any parameter called $red_t1


    Can we solve this problem, as this can be safety hazard because nobody will test a program in T1 if it is not working correctly and happens on all new robots straight out of the box.
    It's not like I did something to the robot to act like that... it's how they come out of the box.

  • To investigate further it would be easiest to have a krcdiag of 8.6 and 8.7 to lookup what differences in relevant machine data might be.

    And the difference in machine data between the "fast" and "slow" robot is? If you do not know what you need to look at create a complete archive or krcdiag, open of both machines, unzip them and compare using e.g. kdiff3 or post both archives otherwise its difficult to guess whats configured. Morst important files to diff are r1/$robcor.dat, r1/$machine.dat and everything in the STEU folder.


    There are too many possibilities in configuration to trigger your "faulty" behavior.


    Fubini

  • There is a quote from Panic mode in this thread:
    BAS function speed limit is cummulative in T1 mode and much slower than 250mm/s

    Quote

    PTP is the fastest motion... and this is why in T1 mode PTP is throttled more aggressively than other types of motion.


    but in some cases there is a bug and safety controller reduces speed even further. in such case in T1 mode PTP becomes really really slow, specially when approximated with CP motions... by doing BCO you can skip that super slow part and get straight to CP motion such as LIN and robot will move normally. solution is to contact KUKA. they will tell you how to fix this (for example by changing limits in safety controller or whatever).


    I think I am also hit with this bug.


    PTP motions are limited by a reduction factor ($RED_T1*max. axis velocity).


    If the maximum axis velocity configured for T1 in the safety controller is lower than the maximum possible axis velocity $RED_T1*max., the reduction factor is automatically reduced to prevent the velocity monitoring from being triggered.


    In my STEU > MADA > MACHINE.DAT file there are.


    $RED_VEL_AXC[1]=11

    $RED_VEL_AXC[2]=8

    $RED_VEL_AXC[3]=8

    $RED_VEL_AXC[4]=10


    These values are percentages of the maximum related to these. (values are in RPM)


    $VEL_AXIS_MA[1]=5600.00

    $VEL_AXIS_MA[2]=6000.00

    $VEL_AXIS_MA[3]=5224.99707

    $VEL_AXIS_MA[4]=4975.84521


    So for example the limit for the first axis would be 11% x 5600rpm = 616 rpm.
    I can divide this by 60 seconds = 616/60 = 10,26 rounds per second.

    I can multiply this by 360° = 10,26*360 = 3696 °/sec

    In my limits in the safety configuration are:


    A1 = 30°/sec

    A2 = 30°/sec

    A3 = 250mm/sec

    A4 = 30°/sec

    Basically what I understand I need my $RED_VEL_AXC[1] * $VEL_AXIS_MA[1] to be under the 30°/sec value and if it is not then the system reduces the speeds to ridiculous levels.


    The numbers are close to a 100 times out, so I don't know if there is a massive 100:1 planetary gearbox in there somewhere or what?

  • you are chasing this for more than a month. why not simply contact KUKA as suggested?

    and how come A3 limit is linear when that axis is rotary?

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

  • What I wrote above was pretty much the interaction I had with KUKA.

    Quote
    If the maximum axis velocity configured for T1 in the safety controller is lower than the maximum possible axis velocity $RED_T1*max., the reduction factor is automatically reduced to prevent the velocity monitoring from being triggered.

    This explanation is hinting why sometimes the T1 speedlimits are very very slow.
    I sent them my files and they pointed out the relevant parameters and gave links to explanations of these parameters. (some xpert.kuka.com page that expires in 14 days)


    But when I manually calculated and compared the parameters (3696 °/sec vs 30°/sec) they seem way off.
    I didn't want to adjust anything just yet, rather to understand beforehand what all of that means. Because currently the maximum velocity 30°/sec is way way lower than the $RED_VEL_AXC * $VEL_AXIS_MA.

  • that is a hint alright but that is not a solution. whoever told you this should be moved to a different position or retrained


    you are dealing with unexpected and defective behavior of their product, so they are supposed to tell you explicitly what to do, such as "change value of A to new value B" whatever those A and B may be.


    this is not the first time this happened, they have it documented. yes you could play around with things and mess with values yourself but


    a) that is not something you are expected or supposed to do

    b) changed value is kinematics dependent (robot with longer reach have different reduction)

    c) this is something that can be a safety hazard and exceed T1 limit of 250mm/s.

    d) that need to work equally well with both motion instruction sets. manipulating value of wrong parameter may look ok for PTP bur also affect SPTP and you do not have insight into inner working of KSS.

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

  • KUKA help concluded that:

    Quote

    As mentioned, this is something that probably will be fixed in the next major KSS release.

    At the moment it's recommended to use the SPTP motion. Thre is no "fix" for this at the moment.

Advertising from our partners