Question about IRCalibration Frame Shift

Frank Yeh · April 6, 2022 at 5:33 AM

Hi everyone,

New to the forum here. I have recently started programming for a Fanuc robot with a R-30iB controller.

It uses a laser to do a frame calibration every time a thermal formed hot tub is loaded onto the fixture.

My question, does the Frame Shift calibration program simply adjust the tilt and shift of a Cartesian frame, or does it also account for the wrapping and shrinkage of the tub as well.

From the controller manual, it seems to indicate some level of compensations for dimensional changes,. I am just trying to understand how it works behind the scene.

Thanks in advance.

Frank

TitusLepic · April 6, 2022 at 5:56 PM

I run CellFinder (later updated to CellCal, then Frame Shift) on my RJ3iB on thermoformed/FRP reinforced bathtubs. So keeping in mind that I'm a few generations of software older than you: It uses some sort of best fit algorithm to match the found points against the mastered points and adjusts the frame XYZWPR, not the individual points of your toolpath. I found that 13 search points with an allowable convergence of 4 keeps me well within my tolerances of +/- 1/8."

Not sure if that answered your question, but hope it helps. I spent a while trying to figure out what's going on behind the scenes once and all I was able to deduce was that Fanuc's engineers are a lot smarter than I am.

Frank Yeh · April 6, 2022 at 6:12 PM

Thank you for sharing the insight. I am currently using 7 points, 4Z, 2Y, and 1X on the lip portion of the tub. It is working reasonably well, but some points on the opposite side of the X are slightly off (1/8). May I ask what do you mean by "allowable convergence of 4"? Do you mind sharing your search point selection strategies?

TitusLepic · April 7, 2022 at 11:33 PM

In the cellfinder menu, go to schedules. There's two options for offset type, "3 Planes" and "Volume." 3 planes works great if you have three flat orthogonal planes you can measure off of, you can get a pretty decent reading using 6 points (3 on one plane, 2 on another, and 1 on the third). Unfortunately we don't make cube shaped product so I'm stuck with using the volume method. The attached picture shows more or less how one of my search programs is set up. My methodology changes depending on tub geometry, but I typically to 4 Z- searches on the deck (as close to the bathing well as possible gives the best results), 6 split between Y+ and Y-, and 4 on X+ and X-. With a tub like the one pictured with no real flat surfaces, I'll need all those points. On a mostly rectangular tub I'll usually do 4 Z-, 3 X+, and 3 Y+ and that will be plenty sufficient.

My schedule is

Frame Type: UFRAME

Offset Type: Volume

Max XYZ Change: 30.000 mm

Max Ort Change: 5.000 deg

Convergence Tol: 4.000 mm

Auto Update: True

Register Number:2

Convergence Tolerance (as I understand it) is a measure of how confident the robot is that it put the new UFrame in the right place. The smaller the number, the more accurate the new frame is.

We're more or less a job shop, so we don't batch anything which means I don't have changeable fixturing. Our fixturing consists of a hold down screw through the drain hole and a pair of 24" C-Clamp vise grips; and my cellfinder search is required on every single tub that we run. The tub's main program will contain a section like this (edited for clarity):

Code

   2:  JMP LBL[1] ;
   3:  !UFrame Data ;
   4:L P[1] 1000mm/sec FINE    ;
   6:  LBL[1] ;
   7:  R[5:DRILL JETS]=AR[1]    ;
   8:  IF R[5:DRILL JETS]=0,JMP LBL[2] ;
   9:  List Menu(5,5) ;
  10:  LBL[2] ;
  11:   ;
  12:  !LOAD USER FRAME ;
  13:  PR[8:UFRAME BUFFER]=P[1]    ;
  17:   ;
  18:  !END LOAD USER FRAME ;
  19:  CALL CELLCALL(43) ;

Display More

Cellcall is my program to call the appropriate search program:

Code

   1:  MONITOR CRASHDET ; 
   2:  MONITOR MON_CBS ;
--condition monitors for recovering from overtravel or contact before search

   3:  R[6:CELLFIND INDEX]=AR[1]    ; --each main program is assigned a number for the select case statement below, and passes it to this program as an arguement

   4:  !Set hole size registers ;
   5:  CALL SETSIZES    ; --not relevant to what we're talking about here
   6:  R[10:ALLOWED CONVERG]=4    ; --When calling a search program on the fly like this, it will execute once and update, regardless of what the convergence is. This program will force a recursive search until the convergence is below the value in R[10]
   7:   ;
   8:L PR[2:NEAR HOME] 250mm/sec CNT100    ;
   9:  UTOOL_NUM=6 ;
  10:  !Check operation mode, skip  ;
  11:  !cellfinder if in manual ;
  12:  IF SI[9:CE/CR Select b1]=OFF,JMP LBL[2] ;
  13:   ;
  14:  PR[8:UFRAME BUFFER]=PR[8:UFRAME BUFFER]+PR[13:UF Manual Adj]    ;
  15:  UFRAME[1]=PR[8:UFRAME BUFFER] ;
  16:  PR[13:UF Manual Adj]=PR[13:UF Manual Adj]-PR[13:UF Manual Adj]    ;
  17:  UFRAME_NUM=1 ;
  18:   ;
  19:  !Run cellfinder  ;
  20:  R[2:CELLFINDER CONVE]=500    ; --cellfinder will store the convergence to R[2] (as set in the schedule), this resets R[2] to a ridiculously high number so that if the search fails R[2] will be greater than 4
  21:  OVERRIDE=100% ;
  22:  R[3:SEARCH COUNT]=0    ;
  23:  LBL[3] ;
  24:  IF R[3:SEARCH COUNT]=5,JMP LBL[4] ; -- It's possible for it to never be happy with the convergence (for example in case of a bad bow, warp, or knot) so this will force it to stop trying after 5 search attempts
  25:  R[28:CELLFIND FAILURE]=0    ;
  26:  SELECT R[6:CELLFIND INDEX]=1,CALL SRCZB1 ; --selects the search program to call
  27:         =2,CALL SRCZB2 ;
  28:         =3,CALL SRCZA2 ;
  29:         =4,CALL SRCZA1 ;
  30:         =5,CALL SRCZG0 ;
  31:         =6,CALL SRCYH0 ;
  32:         =7,CALL SRCZH3 ;
  33:         =8,CALL SRCZI3 ;
  34:         =9,CALL SRCZJ3 ;
  35:         =10,CALL SRCZAEX ;
  36:         =11,CALL SRCZC2 ;
  37:         =12,CALL SRCZC1 ;
  38:         =13,CALL SRCZD1 ;
  39:         =14,CALL SRCZD2 ;
  40:         =15,CALL SRCYJ0 ;
  41:         =16,CALL SRCZK1 ;
  42:         =17,CALL SRCZK2 ;
  43:         =18,CALL SRCZE1 ;
  44:         =19,CALL SRCZE2 ;
  45:         =20,CALL SRCZM1 ;
  46:         =21,CALL SRCZM2 ;
  47:         =22,CALL SRCZN1 ;
  48:         =23,CALL SRCZN2 ;
  49:         =24,CALL SRCZO0 ;
  50:         =25,CALL SRCZP0 ;
  51:         =26,CALL SRCZQ0 ;
  52:         =27,CALL SRCZR0 ;
  53:         =28,CALL SRCZS0 ;
  54:         =29,CALL SRCZT0 ;
  55:         =30,CALL SRCZU0 ;
  56:         =31,CALL SRCZV0 ;
  57:         =32,CALL SRCZF1 ;
  58:         =33,CALL SRCZF2 ;
  59:         =34,CALL SRCZY2 ;
  60:         =35,CALL SRCZY1 ;
  61:         =36,CALL SRCZZ2 ;
  62:         =37,CALL SRCZZ1 ;
  63:         =38,CALL SRCZX3 ;
  64:         =39,CALL SRCZW3 ;
  65:         =40,CALL SRCYA1 ;
  66:         =41,CALL SRCZA2 ;
  67:         =42,CALL SRCYB3 ;
  68:         =43,CALL SRCYC3 ;
  69:         =44,CALL SRCYD3 ;
  70:         =45,CALL SRCYE2 ;
  71:         =46,CALL SRCYF0 ;
  72:         =47,CALL SRCYG3 ;
  73:         =48,CALL SRCYI3 ;
  74:         =49,CALL SRCYK3 ;
  75:         ELSE,CALL ABORTIT ;
  76:  R[3:SEARCH COUNT]=R[3:SEARCH COUNT]+1    ;
  77:   ;
  78:  !Check if cellfinder succeded ;
  79:  IF R[2:CELLFINDER CONVE]=500,JMP LBL[4] ;
  80:  IF R[2:CELLFINDER CONVE]>R[10:ALLOWED CONVERG] OR 
    :  R[28:CELLFIND FAILURE]=1,JMP LBL[3] ;
  81:   ;
  82:  JMP LBL[5] ;
  83:  LBL[2] ;
  84:  Prompt Box Msg(2) ;
  85:  UFRAME_NUM=1 ;
  86:   ;
  87:  LBL[5] ;
  88:  UFRAME_NUM=1 ;
  89:J PR[3:TOOL RACK 1 SAFE] 100% CNT100    ;
  90:  END ;
  91:  LBL[6] ;
  92:  Prompt Box Msg(3) ;
  93:  END ;
  94:  LBL[4:CELLFIND FAILED] ;
  95:  CALL TOOLHOME    ;
  96:  UALM[1] ;

Display More

Hopefully something here is helpful to you.

Frank Yeh · April 12, 2022 at 1:37 AM

Thank you so much for the additional details sir. I will have to study it a little bit. I am in the process of convincing my client to digitize their hand made thermal forming molds as well. Do you have any experience with 3D scanners? or are your molds are digitally designed to begin with?

TitusLepic · April 12, 2022 at 5:49 PM

Our molds are all hand made; I used a cheap i-pad based 3d scanner to create a surface, imported the surface to Inventor, and essentially manually traced the surface in 3d and took measurements of critical dimensions from the physical molds to create my parametric model. It was/is a real pain; if you go this route don't buy the cheapest scanner that you can find.

Frank Yeh · April 13, 2022 at 10:57 PM

Thank you Titus, that does sound like a real pain. I have experience using photogrammetry for smaller parts. Maybe I will try the same route for the tubs and mount the camera onto the robotic arm. Will need to get some dotted stickers for the tub.

I check the frame shift schedule, and it seems like there is no longer an option to set the offset type. I am going to have to do some trials to see if adding additional points would compensate for minor dimensional changes. I greatly appreciate your help TitusLepic

schittampalli · August 4, 2023 at 11:56 PM

Hi Everyone! I am using iRCalibration Frame Shift feature to find not a perfect rectangle part that doesn't have a locating mechanism. I was able to set this feature up with 6 points (3in Z, 2 in X, and 1 in Y - using 321 example in FANUC's manual. But since the part is so large (1.5 meters width and 2 meters length), 3 points in Z aren't enough to capture the roll and pitch part has, so I am trying to teach more points in Z essentially just one more point in z to cover all the foUr corners. But I keep getting this error
ACAL-055 Too many parallel searches

The error manual shows remedy of either adding more searches than 7 or change the search direction. I added 2 more search points, but no luck. Does anyone have the same issue, if so what did you do get through this and how were you able to add more than 3 points in one direction.

Frank Yeh · August 10, 2023 at 6:10 PM

Hi schittampalli ,

1. As long as you set the search points as far apart as possible, you may not need to add more points to capture the roll and pitch of the part.

2. You can mix different search frames. For example, you can mix UTOOL[-Z] with UFRAME[X,Y,Z] for additional search points.

Hope it helps.

Question about IRCalibration Frame Shift

Advertising from our partners

How to see your Job Posting (search or recruit) here in Robot-Forum.com

Tags