True Position Again =(

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • True Position Again =(

    I have a part that looks like a can. It has holes on the sides at set angles from a datum hole (C) which is also on the side. I set [A] as the plane, [B] as the OD, create a line from B to C and rotate accordingly. On top of the part there are holes that are TP to A-B-C using MMC on the feature, -B-, and -C-. When I dimension TP using datums with MMC on the feature, -B-, and -C-, i get out of spec conditions. When I don't use -C- for MMC then it comes into spec. I figure I am safe not using -C- for MMC because I rotate basic to these holes on top anyway. Any ideas?
    I used to be high on life but I built up a tolerance.

    Brown & Sharpe Global Advantage
    PCDMIS CAD++ v2011mr2
    PH10MQ/SP600M


    sigpic

  • #2
    Where is the MMC in the FCF? If it is only on the feature, you shouldn't be using MMC on the datums.
    Also, back to my old standby, can you verify it off the CMM (on a plate)? Is it in tolerance there?
    What about using the CMM results with standard dimensioning and calculating TP?
    Just some thoughts.
    When in doubt, post code. A second set of eyes might see something you missed.
    sigpic

    Comment


    • #3
      There are times that I use variables to calculate my true positions and MMC because I get erroneous results otherwise and I can also do a better job with the MMC from the features.
      DeWain Hodge

      Comment


      • #4
        Leave off the MMC and dimension them using the old stand-by, XY locations. Then, do TP RFS, no datums in the TP. THe deviations should be exactly the same (since you should already be aligned to ABC). Then do TP MMC on the feature ONLY, no datums, again, the deviations should be exactly the same. Then do TP~MMC WITH datums and see what you get.

        Myself, I NEVER use datums in the TP since every version of Pcdmis comes with a different 'system' of using the datums in TP, and each time it changes, B&S/Wilcox/Hexagon tells us that the 'new' way it the right way and the 'old' way is the wrong way. Just simple align to teh datums used in the FCF and leave the datums OUT of the TP. If your process is so 'bad' that you need bonus from the datums, maybe you should change the process. Usually (at least, HERE) the available bonus that could be used from a datum is so small, it just isn't worth screwing around with the datums in the TP. I mean, my B/C datums will hardly ever give more than 0.002" bonus, and when the TP tolerance is 0.5mm at the smallest, 10% extra (MAYBE) isn't worth worring about.
        sigpic
        Originally posted by AndersI
        I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

        Comment


        • #5
          I totally agree with Matt here, not that I normally don't, if you need just that bit more of tolerance, you are going to make bad parts. Unless the MMC is really needed on the feature, I don't normally use that either. The reason being, if you make good parts without the bonus, you have more leeway for your process to move, without making ANY bad parts. I am no production guru, but I think Matt has hit the nail on the head here.
          Jim Jewell

          Comment


          • #6
            I'm on board with Matt as well. Especially doing PPAP layouts I never use MMC. I have some "discussions" with our tooling department at times but the way I see it, if a brand new tool needs MMC for a part to be in spec what will the part measure after several thousand hits and a few acknowledged misfeeds mixed in?

            These *%#&!)% toolmakers get paid top money, EARN IT BY BUILDING A QUALITY TOOL. If we want something cobbled together that makes a marginal at best part we'll go off shore to get them.
            Perry
            B&S Mistral
            3.207 Beta on XP

            Older'n dirt

            Comment


            • #7
              Originally posted by Underspec View Post
              I have a part that looks like a can. It has holes on the sides at set angles from a datum hole (C) which is also on the side. I set [A] as the plane, [B] as the OD, create a line from B to C and rotate accordingly. On top of the part there are holes that are TP to A-B-C using MMC on the feature, -B-, and -C-. When I dimension TP using datums with MMC on the feature, -B-, and -C-, i get out of spec conditions. When I don't use -C- for MMC then it comes into spec. I figure I am safe not using -C- for MMC because I rotate basic to these holes on top anyway. Any ideas?
              You know that you can only properly analyze MMC on datums (Virtual Condition or VC) when using the new XactMeasure GD&T? The lagacy can not handle that.

              What you may have is that your C-datum size is out of tolerance. If one of your datums is out of tolerance, PC-DMIS will revert ALL datums back to RFS and calculate as if the MMC was not on C AND on B. Look at the size for C.

              I know it is not popular to disagree with the Holy Hoedeman, but in this case I do. Even though what he professes makes perfect practical sense and I always initially evaluate my features the way he does it (that's the only way to understand what PC-DMIS exactly does), to me it does make sense to also evaluate the feature as it is listed on the print (which means with MMC on the datums). This is the way the engieer wants it, that's the way he gets it. I accept parts based on that call. Not on any of the previous calls.

              Adding MMC to the datums does not always give a "small" bonus. I have had cases where the VC on the datums ligitimately DOUBLES or even tripples the tolerance on the feature to be evaluated. It all depends on how the datums are layed out.

              Be aware that V4.1 does NOT properly evaluate multiple VCs on datums. If your datum has more than 1 VC, 4.1 adds them up. 4.2 has been corrected and it uses the most restrictive one. So if you have multiple VCs on your datums, you MUST upgrade to V42. No way around it. If you have a single VC on your datum, you're all set with 4.1.


              Jan.
              ***************************
              PC-DMIS/NC 2010MR3; 15 December 2010; running on 18 machine tools.
              Romer Infinite; PC-DMIS 2010 MR3; 15 December 2010.

              Comment


              • #8
                Originally posted by Jan d. View Post
                You know that you can only properly analyze MMC on datums (Virtual Condition or VC) when using the new XactMeasure GD&T? The lagacy can not handle that.

                What you may have is that your C-datum size is out of tolerance. If one of your datums is out of tolerance, PC-DMIS will revert ALL datums back to RFS and calculate as if the MMC was not on C AND on B. Look at the size for C.

                I know it is not popular to disagree with the Holy Hoedeman, but in this case I do. Even though what he professes makes perfect practical sense and I always initially evaluate my features the way he does it (that's the only way to understand what PC-DMIS exactly does), to me it does make sense to also evaluate the feature as it is listed on the print (which means with MMC on the datums). This is the way the engieer wants it, that's the way he gets it. I accept parts based on that call. Not on any of the previous calls.

                Adding MMC to the datums does not always give a "small" bonus. I have had cases where the VC on the datums ligitimately DOUBLES or even tripples the tolerance on the feature to be evaluated. It all depends on how the datums are layed out.

                Be aware that V4.1 does NOT properly evaluate multiple VCs on datums. If your datum has more than 1 VC, 4.1 adds them up. 4.2 has been corrected and it uses the most restrictive one. So if you have multiple VCs on your datums, you MUST upgrade to V42. No way around it. If you have a single VC on your datum, you're all set with 4.1.


                Jan.
                First of all, I LIKE that one!

                Second, I stated that the most bonus that I would ever see is a couple thou (0.002") on my B&C datums. Automotive sheet metal does NOT give much tolerance on locator sizes. Typical tolerances are:
                +0.05mm/-0.00mm
                +0.025mm/-0.010mm
                +/-0.050mm
                These are the 3 that I see. Rarely do I see anything OTHER than those 3 toelrances, and I have NEVER seen one with as much as a total of 0.25mm. So, in MY case, the MMC on a datum is useless. Add to that the fact that 90%+ of the time, my B&C holes are on solid pins in a fixture with no way to measure them. 9% of the time I get a stab pin where I MIGHT be able to measure the hole, depending on stock thickness/net pad clearance. The other <1% is an iterative alignment. BUT, again, the size tolerance for these locators is RARELY over +/-0.05mm.

                But, when trying to figure out WHY the TP is giving you the answers it is giving you, you go back to the basics, eliminate ALL the fancy extra's and see what you get, then add them back in, one at a time, until you see something different.

                Also, another question (or two!)
                1) Are you using surface sample hits? If so, HOW many?
                2) Are you reporting TP PERP to CENTER LINE?

                PERP to CENTER line will eliminate the SURFACE deviation when used with a surface sample hit and will give you a 'different' TP result than just printing the XYZ deviations for a hole that is NOT square to an axis. I always use PERP to CENTER LINE since I have to simulate a 'stab pin' being put in the hole.

                Surface sample hits will change the 'major' axis deviation by including surface deviation in the result. A single sample hit will set the 'depth' for the circle touches, 3 or more will also change the 'nominal' vector of the touches to match the actual surface vector that the hole is in. When done with 3 or more, I do NOT know what it uses for PERP to CENTER LINE calcualtion, does it use the nominal vector OR the actual vector of the surface.
                sigpic
                Originally posted by AndersI
                I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

                Comment


                • #9
                  90% of my parts have a very flat plane for datum A and then 2 pin holes (B and C) that must be measured as cylinders since they are deep (compared to the diameter). I use the XactMeasure GD&T to calculate the perp from the cylinderical features to A. Most of my other features are also cylinders and flats that must be evaluated to the pin holes. I take regular vector points and circle features (construct a cylinder).

                  I tried many data sets and evaluted them in many ways. I am convinced that V4.1+ takes the VC on the datums into account properly.

                  Jan.
                  ***************************
                  PC-DMIS/NC 2010MR3; 15 December 2010; running on 18 machine tools.
                  Romer Infinite; PC-DMIS 2010 MR3; 15 December 2010.

                  Comment


                  • #10
                    Looks like I am just going to remove the MMC bonuses for now and just use RFS. I just upgraded from v4.1 and I am sure v4.2 is calculating differently. I don't remember seeing this problem with v4.1 when I wrote it.
                    I used to be high on life but I built up a tolerance.

                    Brown & Sharpe Global Advantage
                    PCDMIS CAD++ v2011mr2
                    PH10MQ/SP600M


                    sigpic

                    Comment

                    Related Topics

                    Collapse

                    Working...
                    X