Quick Mastering to a fixture in a cell (Fanuc)

    I cannot master to my whitness marks inside the cells, however, I can master to a common fixture point. If I wanted to do this would I jog the robot in joint mode to that point, and register it as a quick master ref position, then monthly or whenever, I can jog back to that position and perform a quick mastering and calibrate? Am I understanding that correctly, or how would i re-teach that master location if I have an issue with the arm? Thanks a bunch

    Talking with Fanuc tech support, they recommended this as a solution to our issues, we have noticed that our prgrammed points are not repeatable. We see about 1mm in error after about 2 months. We have checked fixturing androbot mounting and they believe it is a backlashing issue. Our robot witness marks were aligned then drifted after 2 months

    Quote from M3WIN

    Talking with Fanuc tech support, they recommended this as a solution to our issues, we have noticed that our prgrammed points are not repeatable. We see about 1mm in error after about 2 months. We have checked fixturing androbot mounting and they believe it is a backlashing issue. Our robot witness marks were aligned then drifted after 2 months

    1mm drift is A LOT ! What type of robot is it ? Check your robot by hand, try moving the J3 Joint by applying some leverage force on the wrist. Push the wrist up and observe J3. Does it have plays ? Do the same by applying some force to the end effector left or right. Observe J1, does it have plays?

    In anyway, when the robot drifts by that much (depending the model) it is wiser to identify the faulty reducer(s) and replace them, not perform mastering every 2 months. It most propably is indeed a backlashing issue but the cure is not to remaster every month.

    They are M-20ib/25, when pulling on the arm / end effector there is some flex but it goes back to the original position, I am not sure if I would call it slop tho, it feels solid not like there is slop between the gears or something, but I am able to flex the arm and move the TCP it quite a bit, maybe 5mm or so before it "springs" back to its original position. Should the arm be rock solid when ir pull on the end effector?


    So you think there maybe a faulty reducer in one or more of the joints causing the "drift"? I was able to observe "drift" in the whitness marks in 2 identical robots and cells in joints 2 and 3, both after about 2 months per robot/cell, would this be enough evidence to look into replacing those reducers, these robots and cells were only installed 2 years ago and have a payload of only 2 lbs, also the programs never exceed 50% speed, so it's troubling to think backlash or other mechanical issues could happed with this type of workload.


    The help is greatly appricited I hope to someday give back to the boards and help others, this has been a reacurring theme across our 10 cells and has us in a never ending "touchup", re-programming loop!

    When you say you are seeing drift in the witness marks, are you actually taking the robot back to zero degrees on that joint and the witness marks do not line up? How far off are they?


    That is not normal behavior and is an indicator of something majorly wrong with those robots just as ps0f0r said, maybe a bad gear reducer.

    Checking drift by looking at the witness marks sound like A dumb idea to me.

    We have found most of the time that the fixturing is moving around when things seem off.

    The only way to be sure is to have the arm and tooling shot by A laser scanner, recorded then checked again when things seem off. That would tell you what is moving and how far.

    Unless your arm and tooling are both mounted to pileings buried in the ground below you floor,

    there is A chance that the floor can move do to thermal expansion and contraction.

    Ours are mounted to the cement floor, we see movement in both when the season changes from hot or cold weather.

    Starting with 8.3 there is a "single axis quick master" that can be helpful in tight spaces.


    It would be surprising if two year old, 25kg payload robots operating at 1kg payload and 50% override have much reducer wear. Is the TCP a great distance from the faceplate? Have your robots had preventative maintenance performed and if so did you verify it was performed correctly? I've had robot service vendors come out and make mistakes, turnover for PM techs is high. If the witness marks are actually off after a couple months that would be surprising, as the robots are only 2 years old and not overloaded.


    Common sources for positional errors (all assume witness marks on axes are not off):


    Failed floor anchors. Perhaps too small or insufficient in number. Wedge anchors can often be used, but not always, sometimes epoxy anchors are required, so of course Fanuc specifies them.


    Gripper design. Some of the companies doing robot integration are not experts in that field, maybe PLC experts but not so good with robots. I found no dowel pins in a gripper design from 2010 at a tier 1 auto supplier. Bent, loose, sticking parts of the gripper are something to look for.


    Fixturing, as a previous poster mentioned, is a much more likely source of positional error than backlash.


    TCP/Uframe manipulations. If the TCP/Uframe is being changed programmatically and someone made an error in the program, then this could make it look like there is positional error.


    Good luck and hope to hear what resolved the issue(s).

    I appretiate all the help. We have considered laser scanning to determine the source of movement but have not pulled the trigger on that just yet. We are hoping the in cell quick mastering every 2 months will hold us over while we can look into other potential solutions. Fanuc tech support mentioned that some users will mount a calibration grid in the cell and use iRVision mounted on the robot to auto calibrate after X amount of cycles.


    The robots are mounted on a large metal skid, along with everything else in the cell, so if there was expansion or changes on the concrete floor I would think it would be uniform change. Also our shop is temperature controlled to stay within 65-70* all year.


    The end of arm tooling is 1.1FT off the faceplate, it is an aluminum cylinder with a Shunk end effector. The grip of the parts being minipulated is very consitant, it does not slide or rotate in the jaws.


    The cells have not had any PM done to them just yet, I am looking into having that done, according to the fanuc manual it seems like some basic checks, the only "bigger" item would be greasing the joints.


    Both times the drift was about at the tcp was 1mm or so and we mapped out the drift on points around the cell and it did not appear to come from the base of the robot, also the end effector does not feel lose, however there was no great way to measure it.


    We re-mastered the cell to align the joint witness marks, took a photo, then after the cell fell out of precision again, compared the witness marks and they were off around half to 3/4 of a witness mark thikness on J2, J3. The witness marks were aligned in the zero program, so after there was drift noticed we would re-run te zero program and see that the robot thinks it is at 0 however, the witness marks are no longer the same.

    Can you please describle your application? What is the robot's model? What kind of parts do you grip and how do you grip them? are you gripping inside or outside a hole? what kind of manipulation are you doing ? is it a pick and place ? do you use special tools to process your workpiece ? give some more info. There is no way a Nabtesco RV reducer develop backlash measurable by month. If it is a backlash issue develeped every month you will be able to feel it by hand. Apply some force to your robot by hand.

    We have 8 identical cells and programs with R-30iB TP and Fanuc M-20iB 25. They are loading small blocks of material into a CNC then removing the finished peice, deburring the part edges, serializing the parts, visually inspecting the serialization, then placing the finished parts into a bin. We are trying to hold around .025" at our tightest tolerance. "Drift" has been observed in 2 of the cells after 2 months, however, the other cells "fall out" of program quality in a similar timeline.


    The finished parts are being gripped in the center and are thin curved pieces, the end effector has 3D printed fingers that match the curve of the part. After being removed from the machine, the robot brings the part to a re-grip stand. re-grips, then begins deburring the edges on a number of different deburring wheels. The whole robotic process takes about 10 minutes per part.


    The arm will flex when pressure is applied to it, but it does not feel "sloopy" or "loose" after pressure is applied it will move back to it's orignal location

    Quote from M3WIN

    We have 8 identical cells and programs with R-30iB TP and Fanuc M-20iB 25. They are loading small blocks of material into a CNC then removing the finished peice, deburring the part edges, serializing the parts, visually inspecting the serialization, then placing the finished parts into a bin. We are trying to hold around .025" at our tightest tolerance. "Drift" has been observed in 2 of the cells after 2 months, however, the other cells "fall out" of program quality in a similar timeline.


    The finished parts are being gripped in the center and are thin curved pieces, the end effector has 3D printed fingers that match the curve of the part. After being removed from the machine, the robot brings the part to a re-grip stand. re-grips, then begins deburring the edges on a number of different deburring wheels. The whole robotic process takes about 10 minutes per part.


    The arm will flex when pressure is applied to it, but it does not feel "sloopy" or "loose" after pressure is applied it will move back to it's orignal location

    Can you verify in which step of the cycle the "drifting" occurs ? From what I read I can seperate your cycle in those phases:

    1) pick

    1) load

    2) unlock

    3) place

    4)re-pick

    5)surface treatment wheel 1

    6)surface treatment wheel 2

    7) surface treatment wheel 3

    Does the drift occur on all of the above phases? Can you seperate the drift by axis in reference to the tool frame ? Is it for example -1mm on x axis on all phases ? Even better is there any phase NOT affected by the "drift" virus ?

    The virus I like that lol, affects all points in the cell and the dift is a similar distance but changes direction depending on what userframe the program is calling.


    The latest idea from Fanuc support is that some of the joints on this robot can skip teeth by the style "non-wearing" syle of gearing they use, since we have frequent crashes when loading the raw material into the cnc it could cause this. They claim that a strong push or pull on the EOAT can cause a gear to skip

    Quote from M3WIN

    The virus I like that lol, affects all points in the cell and the dift is a similar distance but changes direction depending on what userframe the program is calling.


    The latest idea from Fanuc support is that some of the joints on this robot can skip teeth by the style "non-wearing" syle of gearing they use, since we have frequent crashes when loading the raw material into the cnc it could cause this. They claim that a strong push or pull on the EOAT can cause a gear to skip

    From all this info gathered, I may have to agree on the FANUC support for your problem. I myself specialise in robotic material handling applications involving deburring, grinding, linishing, polishing and I can tell you for sure that M20iB/25 is not a suitable robot for those kinds of applications. It is not rigid enough and I think the joints are harmonic drives and not RV Reducers but this other members should confirm. I use M710i series robots and the payload requirement is not even 20% of the payload capacity of my applications. The problem with these applications is that the wheels used for grinding/deburring/polishing put a lot of strain and vibration on the joints of the robot. Maybe installing a Mastering Fixture to your cells is the only viable option. You should check your deburring wheels. Try to minimize the vibrations by tuning the rpm, use as low rpm as possible to do the job, check the straightness of the shafts and the pressure applied to the robot when it contacts the wheels.

    I wanted to share some updates on my robot "drift" issue, after 2 more months of running parts in the cell the robot in fact moved again, this time 1.5mm in X+ and Y+ when compated to a programmed point in the cell that was being used to check for this. Also the robots joints, J1, J2 and it appears J3 have moved away from the original pre-calibrated whitness mark locations.


    So I re-masterd the robot the the quick master ref position fixture we created, as recomended by tech support, however, that did not work. I believe it was becuase in order for quick master to work the encoders need to be outside of 0.2* degrees, so it was not accurate enough for our applications. After re-calling tech support the do not think that re-master is the way becuase it does not address the root cuase, it was simply something to try that could maybe work as a bandaid untill the real issue is fixed.


    With that being said tech support also did not think using iRvision to re-teach frames or to re-master is a good fix (not sure if iRvision re-masters, only calibrates?) . They said if we were to re-master the robot to the Fanuc fixture every 2 months that would probalby work by resetting all the points, however, it would not be easy and makes almost no sense to do. They believe there is a mechanicle issue, as you guys were thinking.


    They instructed me to check the crash logs and do some other things to look for damage. I was not able to feel vibrations or notice backlash, however, each one of our robots logged on average 625 crashes since they were installed about 2 years ago. Also only one payload schedule when we should have more, and our robot sensitivity is at 100% when they recommend trying to get it to 150-200% to reduce damage when a crash occurs.


    I am planning on having a service engineer come out ot do PM on these robots, they have not been done in 3 years. While they are here, I will ask them to check if there are mechanicle issues, I was told they may see metal chips when re-greasing the joints.


    With all this being said, I am working on touching up all of our robot programs (8 Cells). If Fanuc does deem a gearbox reducer issue or something like that and we do have then change out the components, will I have to re-program all of the cells or would they simply copy and re-load the encoder values.


    Thanks again this forum has such a great help, you guys are wonderful

Advertising from our partners