Measuring Position Without a Model

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

  • Measuring Position Without a Model

    I just made a different post for another question, but I figured I should post this one as a separate post. I will try to give as many details as possible and convey how little I actually know about PCDMIS (I don't know anything!).

    So! I was given a tool and told, "Measure this and find out if it's wrong." I have no idea where to start. There is no model for it. I do have a part with a model, which I have scanned with the Romer arm and checked on the CMM. Both machines are telling me that both holes are about 0.030 out of position. I guess I have two questions:

    1. Is there a way to get position without theoreticals? As in...how do I know where it's "supposed to be" versus where it actually is? I have no print for the tool and there are no markings of any kind referencing theoretical numbers for aligning it. I made a post concerning this before and I didn't get very clear answers (they would probably be clear for someone with more experience, but I don't think I stressed how little I know in that post...).

    2. Can I somehow use the model for the actual part and do something with that after scanning/taking points on the tool?

    Thank you!

    EDIT:

    One more question! I was told before that I need to build the features that I want to check...do this mean to Insert>Feature>Measured? And then run the feature? Then do I need to manually copy the actuals to the theoreticals? I tried this tonight with a cylinder and it gave me an error for being too far off.

    It would be wonderful if anyone has multiple ways to do this to help me understand. I'd prefer to understand the difficult way first...Like maybe using a probe to build features? And I'd love to know how to do this with a scanner.
    Last edited by Alucard; 03-01-2019, 02:22 AM.

  • #2
    Okay Dracula, so many questions.

    I have one of my own, when you say 'tool' do you mean some sort of mold yeah?
    Automettech - Automated Metrology Technology

    Comment


    • #3
      " There is no model for it." " I do have a part with a model" ????? SLOOOOW Down.

      If you do not have a CAD or a print, your part is absolutely perfect. What is the problem ?

      Comment


      • #4
        If you have a model/print for the part the tool goes to then you cal pull nominals for the tool from the part model/print.

        One caveat to that (because I see it on a daily basis) is if its a customer supplied tool and the part is required to fit the tool. We have a few of those parts and we have to pull nominals from the tool for our CMM programs and we supply the dimensions to the CNC programmers so they can write their programs to the tool.
        MAXIMUM EFFORT!!!

        Comment


        • #5
          Question 1: Is there a way to get position without theoreticals?
          Very simple answer, NO. Position is deviation from nominal/theoretical perfect location. No nominal location, no deviation.

          Question 2: Can I somehow use the model for the actual part and do something with that after scanning/taking points on the tool?
          Very simple answer, YES. You can check profile deviation. You can create features from the CAD to compare to the part. You can use the CAD as a visual aid in where to take manual hits. Lot's of things you can do.

          Question 3: I was told before that I need to build the features that I want to check...do this mean to Insert>Feature>Measured? And then run the feature? Then do I need to manually copy the actuals to the theoreticals?
          No simple answer. You can measure a feature on the part by simple taking hits and hitting the end key or pressing the Done button. PC-DMIS will guess what type of feature you measured and does a pretty good job. If it guesses wrong, you can go to Edit | Override guess and select what you meant to be measuring. Then you will want to correct the theoretical values.
          You can also use autofeatures to add features to the program, but there are a lot of settings to make sure are correct. I wouldn't recommend this unless you have time to look at the help files for autofeatures and become familiar with the settings.

          You REALLY need to get some training, but I know that isn't just up to you, so we will try to help as best we can.
          PC-DMIS 2016.0 SP8

          Jeff

          Comment


          • #6
            Originally posted by Schlag View Post
            " There is no model for it." " I do have a part with a model" ????? SLOOOOW Down.

            If you do not have a CAD or a print, your part is absolutely perfect. What is the problem ?
            Hello again everyone! Thanks for the replies! I got pulled in ten different directions at work and kind of forgot about this post. I'll try to give a better explanation of some of the problems I've run into...and use better terminology.

            Okay, so the whole model/no model thing. I have a model for the part, but not the tool which I was trying to measure.

            As for the word theoretical, I know that positions are based off calculations from theoretical points, but I was referring to the the "theo" line in the code. How do I know what the theoretical "should" be? Is that needed or is that based off the actuals I get from my datums (assuming I have datums)? And if so, do I have to manually copy every actual measurement and key it into each datum features theoretical line?

            And another question regarding measuring without a model (slightly related to this post): Does this apply to surface profiles? I have a part I'm trying to check a surface profile on. Again, I can easily do this with auto features and a model. The problem I'm having is that I am lacking a model yet again. It's a simple ABC alignment and then the surface profile of a side related to A and B. I've attached a notepad document with a program I've been working on for few minutes. I basically built planes on each of the three datum surfaces, aligned, then added 20 measured points. Then I ran the program to get actual numbers for my datums and manually keyed those actual numbers into the theoretical line of each datum. All of the points still have a random value for the theoretical...do I need to manually key each of those 20 features actual numbers into the theo line or is there another way? Or am I completely wrong on all this!?

            I don't think I cleared up my confusing questions...I'll stick around this time though. Thanks again!
            Attached Files

            Comment


            • #7
              First of all, if you want good results. you need to do more than 1 alignment. A manual alignment is nothing more than locating the part,(zeroing X,Y and Z to the part), you need a second alignment (usually DCC) with more points, to really lock it down, and these would be the ones to fix all your THEO numbers and Vector hit points. Once you got that, then all your other hits, features, and such will be a little more accurate, and there really is no need to origin PLA2 twice to "Y", I believe PC-DMIS understood it the first time.

              Code:
              LI2       =ALIGNMENT/START,RECALL:ALN_1,LIST=YES
                            ALIGNMENT/LEVEL,ZPLUS,PLA1
                            ALIGNMENT/TRANS,ZAXIS,PLA1
                            ALIGNMENT/ROTATE,YPLUS,TO,PLA2,ABOUT,ZPLUS
                            ALIGNMENT/TRANS,YAXIS,PLA2
                            ALIGNMENT/TRANS,XAXIS,PLA3
                            ALIGNMENT/TRANS,YAXIS,PLA2
                            ALIGNMENT/TRANS,ZAXIS,PLA1
                          ALIGNMENT/END
              (In Memory of my Loving wife, "Ronda" who I lost March 7, 2016. I love you baby.)
              They say "Nobody's Perfect." I must be Nobody.

              Comment


              • #8
                But Kirb? did you believe the Demon understood ALIGNMENT/TRANS,ZAXIS,PLA1 the first time as well?

                Comment


                • KIRBSTER269
                  KIRBSTER269 commented
                  Editing a comment
                  As corny as it sounds, on past experience, I'm not sure it does.

              • #9
                Originally posted by KIRBSTER269 View Post
                First of all, if you want good results. you need to do more than 1 alignment. A manual alignment is nothing more than locating the part,(zeroing X,Y and Z to the part), you need a second alignment (usually DCC) with more points, to really lock it down, and these would be the ones to fix all your THEO numbers and Vector hit points. Once you got that, then all your other hits, features, and such will be a little more accurate, and there really is no need to origin PLA2 twice to "Y", I believe PC-DMIS understood it the first time.

                Code:
                LI2 =ALIGNMENT/START,RECALL:ALN_1,LIST=YES
                ALIGNMENT/LEVEL,ZPLUS,PLA1
                ALIGNMENT/TRANS,ZAXIS,PLA1
                ALIGNMENT/ROTATE,YPLUS,TO,PLA2,ABOUT,ZPLUS
                ALIGNMENT/TRANS,YAXIS,PLA2
                ALIGNMENT/TRANS,XAXIS,PLA3
                ALIGNMENT/TRANS,YAXIS,PLA2
                ALIGNMENT/TRANS,ZAXIS,PLA1
                ALIGNMENT/END
                Okay, okay...I fixed the redundancies...

                So how do I do a second alignment with DCC without a model? That's really stumping me. I activate DCC mode after my manual alignment, then create new planes for my datum surfaces, then do I key the actual values from my manual alignment into the theoretical lines of the these new planes? Or is there a way to export my manual planes to the CAD and then use those surfaces to create auto-planes? I've basically got a first graders knowledge of what I'm doing here. The biggest problem I'm having is understanding how the program knows where to go without a model if I'm not keying in every single theoretical based off my actual "test" points.

                Thanks again!

                Oh and for reference of how little I know: I worked in a record store up until a year ago. I graduated college with an English degree. I have no idea what I'm doing. I applied for a QA job and a boss said, "HEY! YOU'RE PRETTY GOOD ON A COMPUTER! WANNA RUN THE CMM!? WE'LL SEND YOU TO SEATTLE FOR TRAINING!" To which I exclaimed, "SURE!"

                Then training got swept under the rug after I was able to figure out how to measure a simple part by watching some videos online. Now there are expectations of me knowing what I'm doing...and I have no idea what I'm doing. So apologies for my absolutely base level questions.
                Last edited by Alucard; 03-18-2019, 10:04 PM.

                Comment


                • #10
                  Ok,Ok, calm down In simple terms yes. you can take manual hits for your DCC alignment, and fix your THEO's on each one or even create auto features and place them in the right place by punching in the correct numbers and vector hits. Let me first start by saying, you know when you have to Home Your machine, and it goes up in the top left corner. The machine is establishing 0,0,0 X,Y,and Z. This does Not included the Knuckle, or the Module, or the Probe. So then you create a new program, the first thing it does is ask what Probe to use. Lets just say you don't have any probe built. So you add the knuckle(rotating head or straight head) then at the bottom of that becomes 0,0,0, but you still need a probe end, so you continue on. Then you add the module and then the probe, and the new 0,0,0 becomes the end of the Probe. Now even though, in the virtual world, your build is perfect. Your knuckle is not perfect, your module is not perfect, and your probe is not perfect. So you take one manual hit on top of your Calibration Ball(manual alignment) So then PC-DMIS has it's own built in DCC alignment and it starts taking the perfect hits, at the perfect vectors, even your ball is not perfect, and so it calibrates and PC-DMIS calculates the errors and creates a new 0,0,0. Then you take manual hits on a part, the part becomes the virtual new 0,0,0. Just like the calibration ball. Then you Calibrate the part in with DCC alignment more hits, right vector settings. Then DMIS will do all of the calibration for the imperfections of the part, that's where you see the difference between THEO and ACTL. A little more complex than your Calibration ball, all in all, the difference between being close or being spot on.
                  Last edited by KIRBSTER269; 03-19-2019, 07:10 AM.
                  (In Memory of my Loving wife, "Ronda" who I lost March 7, 2016. I love you baby.)
                  They say "Nobody's Perfect." I must be Nobody.

                  Comment


                  • #11
                    Originally posted by KIRBSTER269 View Post
                    Ok,Ok, calm down In simple terms yes. you can take manual hits for your DCC alignment, and fix your THEO's on each one or even create auto features and place them in the right place by punching in the correct numbers and vector hits. Let me first start by saying, you know when you have to Home Your machine, and it goes up in the top left corner. The machine is establishing 0,0,0 X,Y,and Z. This does Not included the Knuckle, or the Module, or the Probe. So then you create a new program, the first thing it does is ask what Probe to use. Lets just say you don't have any probe built. So you add the knuckle(rotating head or straight head) then at the bottom of that becomes 0,0,0, but you still need a probe end, so you continue on. Then you add the module and then the probe, and the new 0,0,0 becomes the end of the Probe. Now even though, in the virtual world, your build is perfect. Your knuckle is not perfect, your module is not perfect, and your probe is not perfect. So you take one manual hit on top of your Calibration Ball(manual alignment) So then PC-DMIS has it's own built in DCC alignment and it starts taking the perfect hits, at the perfect vectors, even your ball is not perfect, and so it calibrates and PC-DMIS calculates the errors and creates a new 0,0,0. Then you take manual hits on a part, the part becomes the virtual new 0,0,0. Just like the calibration ball. Then you Calibrate the part in with DCC alignment more hits, right vector settings. Then DMIS will do all of the calibration for the imperfections of the part, that's where you see the difference between THEO and ACTL. A little more complex than your Calibration ball, all in all, the difference between being close or being spot on.
                    Okay, thank you! I sort of understand...A bit. But I was able to tweak my program and get a reading that was quite a bit better. I'll explain what I did and maybe you can tell me if I cheated and everything is wrong (and I put a question in there as well):

                    So! I tried entering numbers manually for a few hours to no avail (theos kept changing or the program would just immediately error out after trying to run and it would say the program was out of parameter or something). Then I loaded up my original program, made a few tweaks to the feature alignment (changed one plane to a line and one to a point since the locations of the datums are extremely awkward). After this I pressed F9 on every single feature and checked 'Learn Hit During Execution' <Should I not have done this? I've never noticed this option...but it helped a lot>. After this I deleted my DCC command I ran the program entirely manually in order: I aligned, then aligned again (for what would become the DCC alignment), then took my points. Then I hit F9 on every single feature (points and alignment features) and copied their actuals into their theo boxes. Finally, I inserted some move points and put the DCC command back in. After this I got a deviation on my surface profile of 0.012" to AB. A lot better than my 18 inches that I was consistently getting before.

                    Now I'm just wondering if this is actually how you're supposed to do this. I think you explained already in your previous post, but much like the clutch in my first car, I think this will take me longer than it should to grasp the concept. I fully understand how it works with a something that is preprogrammed or to a model, but how does the program know what something should be when everything is entered off values gathered from a finished part that could be completely different from what the perfect model is? It's theoretical points are gathered from it's current state, but what if the current state is horrible? Or how can you base theoretical numbers off something that could theoretically be wrong...? Does it not matter what the theoretical points are? I feel like I'm over-thinking this and believing that models matter a lot more than they actually do. I think I need to study some GD&T...

                    I've attached my latest program as well if you would be so kind as to glance at it.

                    EDIT: Oh one more question! Will my program update itself if I moved the part? I'm worried about the way I did this. Without a model, do I need to put the part in relatively the same spot every time? I had to enter every theo manually for all the DCC features so it doesn't appear they update based on my manual alignment. Is there a way to make this work if so?

                    EDIT AGAIN: So I just moved the part by 0.020" and the program crashed because it was too far from nominal (first point on the line for plane B). This makes me think the whole program is wrong and I basically just created a program that will check itself to itself and pass as long as it's exactly where it was the first time I checked it. Is there a way to make the DCC features update to match the manual alignments theoreticals? The manual alignment and DCC alignment don't seem to be talking. Sorry for all the questions.

                    EDITING PART THREE: Would this have something to do with the 'Recall' option in the Alignment Utilities? Right now all my alignments are set to "Startup".

                    EDITING PART FOUR: IT'S ALIVE:
                    I consolidated my most questionable of questions.

                    1. Did I do the correct thing by turning on "Learn Hit During Execution" for each feature so that I could initially run through it to get the theos?

                    2. How does the program know what everything "should be" when theos are entered off values gathered from a finished part that could be completely different from what a perfect model would be...or simpler, how does it know these theoretical points are correct to begin with? Does it not matter since everything is being measured from datums? What if there aren't datums?

                    3. Is there a way to make the program update itself to the manual alignment? When I program to a model, I use auto features, align, and then enter DCC mode where it seems to know where the part is based off my manual alignment. But if I move this part, manually align, then enter DCC, the probe just runs into the part because I moved it. The DCC alignment isn't recalling anything from the manual alignment. Shouldn't the DCC mode adapt to my manual alignment? Does this have to do with recall? Okay question number 3 is about ten questions in one.
                    Attached Files
                    Last edited by Alucard; 03-20-2019, 12:06 AM.

                    Comment


                    • #12

                      ALI2 =ALIGNMENT/START,RECALL:STARTUP,LIST=YES ALIGNMENT/ROTATE,XPLUS,TO,LINB,ABOUT,ZPLUS
                      ALIGNMENT/TRANS,YAXIS,LINB
                      ALIGNMENT/TRANS,XAXIS,PNTC
                      ALIGNMENT/TRANS,YAXIS,LINB
                      ALIGNMENT/TRANS,ZAXIS,PLA1
                      ALIGNMENT/END

                      You should recall the previous alignment, ALI1

                      You have no LEVEL in ALI2

                      You have translated YAXIS twice.

                      That’s as far as I looked at your code.

                      B&S CHAMELEON/PCDMIS CAD++ V2011

                      There are no bugs, only "UNDOCUMENTED ENHANCEMENTS!"

                      sigpic

                      Comment


                      • #13
                        Originally posted by dph51 View Post
                        ALI2 =ALIGNMENT/START,RECALL:STARTUP,LIST=YES ALIGNMENT/ROTATE,XPLUS,TO,LINB,ABOUT,ZPLUS
                        ALIGNMENT/TRANS,YAXIS,LINB
                        ALIGNMENT/TRANS,XAXIS,PNTC
                        ALIGNMENT/TRANS,YAXIS,LINB
                        ALIGNMENT/TRANS,ZAXIS,PLA1
                        ALIGNMENT/END

                        You should recall the previous alignment, ALI1

                        You have no LEVEL in ALI2

                        You have translated YAXIS twice.

                        That’s as far as I looked at your code.
                        Okay, thank you. I keep doing that to the YAXIS...

                        Any chance you could check out my last few questions? I'm having a problem with getting my DCC alignment to communicate with my manual alignment. Is this even possible without a model?

                        Comment


                        • #14
                          You can program without models. I have never had cad models for any of the programs I have written.

                          Follow steps like this.

                          MANUAL MODE

                          Manual plane1

                          ALI1 level and translate to plane1

                          Manual line1

                          ALI2 Recall ALI1 rotate to line1 and translate to line1

                          Manual point1

                          ALI3 Recall ALI2 translate to point1

                          DCC MODE

                          Measure plane2

                          ALI4 Recall ALI3 level and translate plane2, rotate manual line1 and translate line1, translate point1

                          Measure line2

                          ALI5 Recall ALI4 level and translate plane2, rotate and translate line2, translate point1

                          Measure point2

                          ALI6 Recall ALI5 level and translate plane2, rotate and translate line2, translate point2

                          Hope this helps.


                          B&S CHAMELEON/PCDMIS CAD++ V2011

                          There are no bugs, only "UNDOCUMENTED ENHANCEMENTS!"

                          sigpic

                          Comment


                          • 2013_junkie
                            2013_junkie commented
                            Editing a comment
                            Nice to see that there's still someone else out there that doesn't use cad models for everything .

                        • #15
                          Originally posted by dph51 View Post
                          You can program without models. I have never had cad models for any of the programs I have written.

                          Follow steps like this.

                          MANUAL MODE

                          Manual plane1

                          ALI1 level and translate to plane1

                          Manual line1

                          ALI2 Recall ALI1 rotate to line1 and translate to line1

                          Manual point1

                          ALI3 Recall ALI2 translate to point1

                          DCC MODE

                          Measure plane2

                          ALI4 Recall ALI3 level and translate plane2, rotate manual line1 and translate line1, translate point1

                          Measure line2

                          ALI5 Recall ALI4 level and translate plane2, rotate and translate line2, translate point1

                          Measure point2

                          ALI6 Recall ALI5 level and translate plane2, rotate and translate line2, translate point2

                          Hope this helps.
                          Thank you for the simple clear instructions! This is exactly how my brain works...I need to see exactly what to do rather than just vague random answers that go over my head. This will help tremendously.

                          EDIT: Okay, back to my redundant basics questions. I followed your steps in order without altering anything, just to see how it would run. Once DCC mode kicked in the probe just started moving off in a random direction and then said something about being out of limits. I did not alter any theos or anything. All features were built with 'measured' features (Insert>Feature>Measured) so all theoreticals are at zero. Should I run the manual alignment, then key those actual values into the theoretical values of all the DCC features? Also, should I choose "Yes" for "USE THEO" during DCC? Also, does measuring without a model update values of features after aligning or do you always need to manually update every single feature each time you move the part to a different spot on the table? I can't seem to wrap my head around this. Everything I've tried seems to leave the alignment useless because none of my features update beyond the alignment. Should I not be using measured features? Am I not activating a setting?
                          Last edited by Alucard; 03-22-2019, 01:06 AM.

                          Comment

                          Related Topics

                          Collapse

                          Working...
                          X