kr200 odd movement behavior

Hello All,

I'm going to do my best to describe this issue as succinctly as possible. If videos would make it easier let me know:

Robot: kr200 sp 2
Controller: krc2 4.1.6

When I jog the robot in axis specific setting all axis move well. A6-A3 move immediately and stop immediately when motion button released (not the deadman but the +/-) without bounce or play. Axis 1 and 2 both move immediately but when movement button is released they continue moving and speed down more than stop motion. This takes place over a few centimeters but is easily perceptible to the human eye. I only mention this because I don't know what "normal" behavior should be.

However, when I switch to any other jogging reference frame (robot coordinates, global coordinates, tcp) I get very significant bouncing of the end effector tool when motion is stopped. Again, I'm not releasing the deadman to stop motion, rather just letting my finger off of the +/- button. Digging a little deeper, I removed the caps from the end of all servos to examine motor motion during this "bounce". On axis 2, it is very noticeable on the exposed nut under the cap, that the nut moves back and forth after the motion button has been released.

I don't hear any audible grinding or see any bouncing while the robot is in motion. Further, when I jog the robot far from home and then run the home command it returns smoothly to its home position.

What I'm wondering is how to determine positively that this is a motor issue and not a gear issue (planetary gear not servo gear). Any suggestions for troubleshooting would be appreciated. My mechanical inkling is that a bad gear would not cause the motor to rock back like it is doing. Is this a braking issue with the motor?

Thank you.

Hey All,

Thanks for the help.

To clarify: when I am running a program holding the green (+) button and let go there is NO jerk. The robot decelerates smoothly and everything looks great. It's only when I'm jogging the robot manually. In axis specific jogging, there is no jerking. When I switch to global coordinates, robot coordinates, or tcp jogging and move the robot it jerks considerably. This is when its visible on the nut of the A2 MOTOR. When the robot stops, the nut wiggles back and forth (mm but noticeable). I'll take some video later. This was a GM line robot. It has 3500 hours on it. No one has tampered with it to my knowledge.

The spindle (tool) that is on the robot flange is only 26.6 kg (this is a kr200). One thing I noticed is my tool loads are declared in R1/system/config.dat. However in robcor.dat (R1/mada) I have this:

seeing as I'm not an expert or I wouldn't be asking, will the config override the default declared settings in my robcor.dat file as I have no load on axis 3 and only 26.6 on the flange. I know, I'm guessing but everything else really looks the same...

$V_ROBCOR[]="V6.5.0/KUKA4.1" ;VERSIONSKENNUNG
CHAR $MODEL_NAME[32]
$MODEL_NAME[]="#KR200_2_TJ H C2 FLR ZH02"
DECL ADAP_ACC $ADAP_ACC=#STEP1 ;BESCHLEUNIGUNGSANPASSUNG ( #NONE, #STEP1, #STEP2 )
DECL ADAP_ACC $OPT_MOVE=#STEP1 ;HOEHERES FAHRPROFIL (#NONE, #STEP1, #STEP2)
BOOL $PROG_TORQ_MON=FALSE ;UEBERWACHUNG DER SOLL-MOMENTE MOTOR UND GETRIEBE
BOOL $ENERGY_MON=FALSE ;UEBERWACHUNG KINETISCHE ENERGIE BEI CRASH
INT $ITER=2 ;ANZAHL DER ITERATIONEN
INT $SYNC=1 ;PHASENANPASSUNG ( 1 = SYNCHRON, 0 = NICHT SYNCHRON )
INT $OPT_APPROX=100 ;REDUZIERUNGSFAKTOR ZUR UEBERSCHLEIFPLANUNG
DECL EKO_MODE $EKO_MODE=#OFF ;EKO-MODUS (#OFF,#ON,#OPT)
REAL $DEF_L_M=210.0 ;DEFAULTMASSE AM FLANSCH
FRAME $DEF_L_CM={x 230.0,y 0.0,z 210.0,a 0.0,b 0.0,c 0.0} ;MASSENSCHWERPUNKT-FRAME
DECL INERTIA $DEF_L_J={X 40.0,Y 40.0,Z 40.0} ;EIGENTRAEGHEITSMOMENTE DER LAST
REAL $DEF_LA3_M=80.0 ;DEFAULTMASSE AUF DER ACHSE 3
FRAME $DEF_LA3_CM={x -505.0,y 0.0,z -1110.0,a 0.0,b 0.0,c 0.0} ;MASSENSCHWERPUNKT-FRAME A3
DECL INERTIA $DEF_LA3_J={X 11.1999998,Y 11.1999998,Z 11.1999998} ;EIGENTRAEGHEITSMOMENTE DER LAST AUF A3

I went through machine.dat and robcor.dat...here's what I found.

Machine.dat:

1) $RED_ACC_OV (acceleration reduction override) :

INT $RED_ACC_DYN=100
INT $RED_VEL_CPC=2 ;REDUZIERFAKTOR FUER BAHN-UND ORIENTIERUNGSGESCHWINDIGKEIT BEI KARTESISCHEM HANDVERFAHREN UND KOMMANDOBETRIEB [CP] [%]
INT $RED_ACC_CPC=7 ;REDUZIERFAKTOR FUER BAHN-UND ORIENTIERUNGSBESCHLEUNIGUNGEN BEI KARTESISCHEM HANDVERFAHREN UND KOMMANDOBETRIEB [CP] [%]
REAL $VEL_CP_T1=0.100000001 ;BAHNGESCHWINDIGKEIT IN T1 [M/S]
REAL $VEL_CP_COM=0.150000006 ;REDUZIERUNG DER FLANSCHGESCHWINDIGKEIT IN [M/S]
REAL $RED_JUS_UEB=100.0 ;REDUZIERFAKTOR FUER UEBERNAHMEFAHRT [%]
INT $RED_ACC_OV[12] ;AXIALE REDUZIERUNG DER BESCHLEUNIGUNG FUER OVERRIDE ACHSE[I] (I=1:A1,I=7:E1) [%]
$RED_ACC_OV[1]=100
$RED_ACC_OV[2]=100
$RED_ACC_OV[3]=100
$RED_ACC_OV[4]=100
$RED_ACC_OV[5]=100
$RED_ACC_OV[6]=100

2) $RED_ACC_EMX (emergency stop ramp)

FRAME $ACC_CAR_TOOL={x 0.0,y 0.0,z 0.0,a 0.0,b 0.0,c 0.0} ;FRAME (ACCORDING TO FLANGE) FOR CARTESIAN ACCELERATION MONITORING
DECL ACC_CAR $ACC_CAR_LIMIT={X 0.0,Y 0.0,Z 0.0,A 0.0,B 0.0,C 0.0,ABS 0.0} ;LIMITS FOR THE CARTESIAN ACCELERATION $ACC_CAR_ACT
BOOL $ACC_CAR_STOP=FALSE ;ENABLE (TRUE) OR DISABLE (FALSE) CARTESIAN ACCELERATION MONITORING
INT $RED_ACC_EMX[12] ;REDUZIERFAKTOR FUER BAHNTREUE NOT-AUS-RAMPE [ % ]
$RED_ACC_EMX[1]=140
$RED_ACC_EMX[2]=220
$RED_ACC_EMX[3]=370
$RED_ACC_EMX[4]=200
$RED_ACC_EMX[5]=200
$RED_ACC_EMX[6]=200

3) $TIME_POS (positioning time)

every axis is 512

4) $AXIS_JERK

Don't have this variable.

I have attached my machine.dat and robcor.dat.

Thanks Frenchman!!

Nothing seems notably different from your parameters, do you agree?...

INT $RED_ACC_AXC[12] ;REDUZIERFAKTOR FUER AXIALE BESCHLEUNIGUNG BEI ACHSSPEZ. HANDVERFAHREN UND KOMMANDOBETRIEB (PTP) ACHSE[I] (I=1:A1,I=7:E1) [%]
$RED_ACC_AXC[1]=10
$RED_ACC_AXC[2]=20
$RED_ACC_AXC[3]=20
$RED_ACC_AXC[4]=20
$RED_ACC_AXC[5]=20
$RED_ACC_AXC[6]=20
$RED_ACC_AXC[7]=0
$RED_ACC_AXC[8]=0
$RED_ACC_AXC[9]=0
$RED_ACC_AXC[10]=0
$RED_ACC_AXC[11]=0
$RED_ACC_AXC[12]=0
INT $RED_ACC_DYN=100
INT $RED_VEL_CPC=2 ;REDUZIERFAKTOR FUER BAHN-UND ORIENTIERUNGSGESCHWINDIGKEIT BEI KARTESISCHEM HANDVERFAHREN UND KOMMANDOBETRIEB [CP] [%]
INT $RED_ACC_CPC=7 ;REDUZIERFAKTOR FUER BAHN-UND ORIENTIERUNGSBESCHLEUNIGUNGEN BEI KARTESISCHEM HANDVERFAHREN UND KOMMANDOBETRIEB [CP] [%]

and...

; general MOVEMENT - parameters:
;----------------------------------
INT DEF_OV_PRO=100
INT DEF_ADVANCE=3
; PTP - MOVEMENTS
;----------------------------------
INT DEF_VEL_PTP=100
INT DEF_ACC_PTP=100
; CP - MOVEMENTS
;----------------------------------
DECL CIRC_TYPE DEF_CIRC_TYP=#BASE
REAL DEF_VEL_CP=2.0
REAL DEF_VEL_ORI1=200.0
REAL DEF_VEL_ORI2=200.0
REAL DEF_ACC_CP=2.29999995
REAL DEF_ACC_ORI1=100.0
REAL DEF_ACC_ORI2=100.0
REAL DEF_VEL_FACT=1.0
; APO - parameters
;--------------------------------
INT DEF_APO_CPTP=100
INT DEF_APO_CVEL=100
REAL DEF_APO_CDIS=3.0
REAL DEF_APO_CORI=5.0

I'll spend some time playing around with numbers tomorrow. There is nothing glaringly wrong though. Anything making you think this isn't just a simple motor going bad/internal damage to it?

Thanks again.

Four part video:

1. Axis specific jogging
2. Global coordinate jogging (some parts of the work envelope have more dramatic bounce than others...I would call this example mild)
3. Servo motor axis 2 nut movement (this is a slow motion video!)
4. Program to home motion

So, an update.

I've started milling some really soft foam and am seeing all kinds of bouncing when executing the program...as noted before in this thread (same person, same robot, same controller). Nothing I can't sand out in foam, but looking to move on to hardwoods and this would not pass. The bouncing is occurring (yes at a1 change of direction, but that's been minimized) while the robot is moving in arced lines...it completes the line straight, but the tcp is bouncing up and down. This generally is happening at the top of a curve (change in tool z). There are no noticeable glitches in jogging each axis independently. Load data has been set for all tools (it's only 30kg on a 200kg rated robot).

I wrote a warm up program for the robot. So it runs some continuous (c_ptp) motions on loop to warm up servos. I start slow and work up to faster speeds. Some other folks with the same robot/cabinet set up say that it will get better with time after running the robot for longer more consistent durations. I have my personal doubts.

I have attached a video of the bounce along with pictures of the surface + my postprocessor and sample code.

I have tried changing my C_DIS values along with exporting in both polylines (different point distributions) and arcs (while also changing milling path tolerances and removal of redundant points. I have also changed the feed rates significantly and this does not really help either.

Nothing seems to really help.

I guess here's the list of questions (and some of this is fishing so feel free to disregard and not have to explain):

1. is this within expected robot (considering date, model, and controller) tolerances? --> 2003 kr200 gmkrc2 controller 4.1.6
2. could this be a mastering issue? are different axes trying to go different places at the same time?
3. could this also be linked to the servos...should I try swapping?

Any place to start (other than a new arm) would be welcomed advice. Thanks for your time.

Hours: off the top of my head 3-4K when I checked (cabinet and rdw)

Maintenance: gm welding robot. Hummer line (they didn't sell very many). Talked to robot engineer from the plant and he said they were well maintained. I couldn't say whether this is true or not.

Collisions: no noticeable/visible damage on exterior of the robot. I've attached videos of robot running at speed. No grinding or out of the ordinary sounds.

Function generators: this ones beyond me. ??

Panic Mode: I hear that and I'm not trying to kid myself that this robot has been around the block. I can't know more than the robot will tell me for run time. 3179.4 h to be exact, and yes I know that can be reset. However, its physical appearance is decent and it sounds fine when running, even at speed. There is no weird movement when it's jogging in axis specific...tcp stays nice and smooth through each axis arc. This is also true when running continuous ptp movements. There is no shaking or oscillations. Then there's this bouncing described above in toolpath execution for milling.

When jogging in tcp workplane mode the bounce is most prevalent when starting and stopping movement. Tiny bounces can be seen while the tool is in motion.

What I'm trying to get at is if this software or hardware side or just impossible to really know without tearing it down and swapping parts. Could this have something to do with the resolver?

I have loaded all factory defaults for the rdw and machine data on the controller. Beyond that I guess it's time to try and switch motors. I was just looking for some advice on if anyone has seen something like this before, and if so, what were the culprits...have any arms you want to get rid of?!

SkyeFire

Thanks for the advice. I will definitely try that tomorrow. Titanium! I wouldn't even know what to do with that!

I did however do another test. I set up a face milling toolpath over a large workplane. So, one LIN code is defining a tcp movement of 1600mm in the x, 5mm stepover in y, back the other way with a small linking path in between...no change in z. The thought was while the robot is doing the same amount of work (or close to it) to keep the tcp at a constant level, it's interpreting a considerable less amount of motion code to do so. Low and behold, the bounce is gone. Yes, there is still slight movement, but expected robot movement, not the bit skipping off of a surface. The finish isn't perfect. But I'll chalk that up to my mastering and tool tip alignment calculations. I will definitely hammer those out more exact very very soon.

So, I'm thinking that yes, this is an old bot with some wear, but it seems to be more of a motion code interpretation. Tomorrow I will set up evenly distributed points for my toolpath and see where that gets me.

Thanks!

Here's my code (well, start of it):

$VEL.CP = 0.2
PTP {X 523.550,Y 2353.020,Z 10.000,A 32.1262,B 0.0000,C 0.0000,S'B110',T'B010011'}
; First Toolpath Point
$VEL.CP = 0.2
LIN {X 523.548,Y 2353.019,Z 5.000,A 32.1262,B 0.0000,C 0.0000} C_DIS
; Plunge Move Starts
$VEL.CP = 0.06
LIN {X 523.548,Y 2353.019,Z 0.000,A 32.1262,B 0.0000,C 0.0000} C_DIS
; Cutting Move Starts
$VEL.CP = 0.12
LIN {X 523.548,Y -13.019,Z 0.000,A -29.4052,B 0.0000,C 0.0000} C_DIS
; Cutting Move Ends

Okay...I recalculated my toolpath using evenly distributed points 5mm apart. I set C_DIS to 5 and ADVANCE to 3. No bounce and extremely smooth cutting even on very soft polystyrene--couldn't ask more from my gantry based cnc. Thanks for the advice Skyefire.

In this case, am I using C_DIS to my benefit or am I negating it's use altogether? I've read up on what it does but am a bit confused to it's use when putting points in an evenly distributed manner. Any other advice?

Program size goes through the roof with evenly distributed points, but that's something I can work around. What was throwing me off is how incredibly bouncy the TCP is when jogging the robot. When I saw this transfer to running programs I was incredibly worried something was completely off with the robot.

Thank you.

As far as load. I did try load determination but found out quickly my load was too small for it to determine (was loaded on the bot as a package). I modeled the spindle and mounting plate in rhino. Multiplied the volume by the mass to get density and then multiplied that my rhino's calculations of volume moments of inertia (and yes all in kg and m...except for CoG). My incorrect assumption is that the mass is evenly distributed through the mounting and spindle, but it will have to do. I think that's the best I'm going to get to accurate load approximation.