Datapage

[VONDY]

I Am Checking Three Variables Exactly The Same On One Part And Reporting Out Each, D1, D2, & D3 And Running 30 Parts. What My Boss Wants To See In Datapage Is All Three Combined Together. Instead Of Having D1, D2, & D3 Showing On The Statistics Report, He Just Wants To See One Report. Is There A Way To Combine Them All Together In Datapage? The Samples Have Already Been Ran And The Data Is There, Its Just Not What He Wants To See.

[Matthew D. Hoedeman]

Nope, you can not 'add' different 'variables' together to get a combined, one value report in Datapage. If they are 3 different variables, then you need to tell your boss that they HAVE to be reported as seperate variables since they ARE NOT THE SAME.

[VONDY]

Nope, you can not 'add' different 'variables' together to get a combined, one value report in Datapage. If they are 3 different variables, then you need to tell your boss that they HAVE to be reported as seperate variables since they ARE NOT THE SAME.

I'M NOT ADDING DIFFERENT VARIABLES TOGETHER IN THE SENCE THAT D1 PLUS D2 EQUALS THERE SUM, WHAT I HAVE IS 3 HOLES IN EACH PART THAT HAS A 91 DEGREE ANGLE IN EACH HOLE THAT WE REPORT OUT. THEY JUST WANT THEM ALL TOGETHER IN ONE MULTICHART INSTEAD OF THREE SEPERATE ONES. (ACTUALLY THERE ARE TWENTY HOLES WITH TWENTY ANGLES AND TEWNTY CHARTS, I JUST SIMPLIFIED IT HERE)

[JamesMannes]

If you want to add the 3 variables together, open the database the 3 are located in, then open editor. From there choose the pull down menu view>data values>data values, choose the appropriate part name,variables, and transactions. Then go to edit>new>variable. Create a new variable. Then go to edit>new>transaction. Make as many new transactions as you need in order to fill the new variable with the 3 variables data. After making the new transactions area's, highlight the data in D1 and hit cntrl-c now move your cursor over to your new variable and hit cntl-v and do the same with the other two variables and you now have a new variable with all transactions in 1 variable.


Thanks for the exercise, I learned something from this too!!!

[Matthew D. Hoedeman]

If it is NOT the same hole, then you can not 'add' them together, they are seperate, individual entities and MUST be reported as such. I did not say you were adding D1+D2+D3, I said you do NOT have a 90 piece study, but still a 30 piece study with 3 unique variables and Datapage will NOT combine them into a 90 piece study ('cause it's NOT).

[JamesMannes]

Note, I would agree with Matt that this is not a good practice. I didn't know what you were really doing I just gave an answer to the question, "Can such and such be done?"

[JamesMannes]

Why not a Bar and whisker chart? That should give him all the data on a few pages.

[TRD_racing]

Vondy, Just save you file as an Excel file. (in datapage) file, save as *.xls. You can open it up in excel and format as you like. If you do this a lot you can write a macro to do it for you everytime. If you don't know much about macro's and you do this often PM me and I can help you w/ this.

[craiger_ny]

I Am Checking Three Variables Exactly The Same On One Part And Reporting Out Each, D1, D2, & D3 And Running 30 Parts. What My Boss Wants To See In Datapage Is All Three Combined Together. Instead Of Having D1, D2, & D3 Showing On The Statistics Report, He Just Wants To See One Report. Is There A Way To Combine Them All Together In Datapage? The Samples Have Already Been Ran And The Data Is There, Its Just Not What He Wants To See.

I will have to side with the others in that Datapage does not have an option to merge data simpler than what James has posted. I will however detract from others that tell you what you are doing is not statistically sound.

Let us suppose one has a drilling machine that drills a hole then indexes 91º then drills another hole, indexes 91º drills etc. Now let us suppose we want to learn a little about how well the machine indexes 91º. Why would it not be sound to evaluate each 91º index on the part as the same feature even if it is multiple features? The answer is that it is perfectly valid in some cases to do just that(never say never, especially with statistics). Too many people have fallen into believing statistics is limited to nothing more than 30 piece capability studies handed down as a customer requirement to ship product. The most value I have gotten from statistics is not from 30 piece capability studies of key characteristics but studies just as the one you may be trying to do Vondy. I have used statistics to evaluate machine capability, tolerance feasibility, and for process trouble shooting in general. Anyone that tells you statistical analysis must be limited to a single dimension for a single feature that is generated not multiple times on one part by the same process but only from part to part is selling themselves short.

Here is how you can take a 91º angle between multiple features and send them to Datapage as a dimension that treats them as one feature:

CIR1       =FEAT/CIRCLE,POLR,OUT,LEAST_SQR
            THEO/2,91,0,0,0,1,1
            ACTL/2,91,0,0,0,1,1
            MEAS/CIRCLE,4,WORKPLANE
            HIT/BASIC,NORMAL,0.4651,1.9997,0,1,0,0,0.5349,1.9997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,2.4997,0,0,1,0,0.0349,2.4997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.5349,1.9997,0,-1,0,0,-0.4651,1.9997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,1.4997,0,0,-1,0,0.0349,1.4997,0,USE THEO = YES
            ENDMEAS/
            ASSIGN/ARR_ANG[1] = CIR1.PA  [color=green]'set an array index (1) here to capture the PA of CIR1[/color]
LIN1       =FEAT/LINE,RECT,UNBND
            THEO/0,0,0,-0.0174524,0.9998477,0
            ACTL/0,0,0,-0.0174524,0.9998477,0
            CONSTR/LINE,BF,2D,CIR_ORIGIN,CIR1,,
            OUTLIER_REMOVAL/OFF,3
            FILTER/OFF,WAVELENGTH=0
AL_CIR1    =ALIGNMENT/START,RECALL:, LIST= YES  [color=green]'create an alignment so your next circle's PA is 91º from the last[/color]
            ALIGNMENT/ROTATE,XPLUS,TO,LIN1,ABOUT,ZPLUS
            ALIGNMENT/END
CIR2       =FEAT/CIRCLE,POLR,OUT,LEAST_SQR
            THEO/2,91,0,0,0,1,1
            ACTL/2,91,0,0,0,1,1
            MEAS/CIRCLE,4,WORKPLANE
            HIT/BASIC,NORMAL,0.4651,1.9997,0,1,0,0,0.5,2,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,2.4997,0,0,1,0,0,2.5,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.5349,1.9997,0,-1,0,0,-0.5,2,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,1.4997,0,0,-1,0,0,1.5,0,USE THEO = YES
            ENDMEAS/
            ASSIGN/ARR_ANG[2] = CIR2.PA  [color=green]'set an array index (2) here to capture the PA of CIR2[/color]
LIN2       =FEAT/LINE,RECT,UNBND
            THEO/0,0,0,-0.0174524,0.9998477,0
            ACTL/0,0,0,-0.0174524,0.9998477,0
            CONSTR/LINE,BF,2D,CIR_ORIGIN,CIR2,,
            OUTLIER_REMOVAL/OFF,3
            FILTER/OFF,WAVELENGTH=0
AL_CIR2    =ALIGNMENT/START,RECALL:AL_CIR1, LIST= YES  [color=green]'create an alignment so your next circle's PA is 91º from the last[/color]
            ALIGNMENT/ROTATE,XPLUS,TO,LIN2,ABOUT,ZPLUS
            ALIGNMENT/END
CIR3       =FEAT/CIRCLE,POLR,OUT,LEAST_SQR
            THEO/2,91,0,0,0,1,1
            ACTL/2,91,0,0,0,1,1
            MEAS/CIRCLE,4,WORKPLANE
            HIT/BASIC,NORMAL,0.465,2.0084,0,0.9998477,0.0174524,0,0.465,2.0084,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0436,2.4996,0,-0.0174524,0.9998477,0,-0.0436,2.4996,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.5348,1.991,0,-0.9998477,-0.0174524,0,-0.5348,1.991,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0262,1.4998,0,0.0174524,-0.9998477,0,-0.0262,1.4998,0,USE THEO = YES
            ENDMEAS/
            ASSIGN/ARR_ANG[3] = CIR3.PA  [color=green]'set an array index (3) here to capture the PA of CIR3[/color]
            ASSIGN/LOOP_INDEX = 1  [color=green]'this variable will count your loops and move to the next index of your array in the loop[/color]
            ASSIGN/LOOP_COUNT = LEN(ARR_ANG)  [color=green]'get the length of the array[/color]
V1         =LOOP/START, ID = YES, NUMBER = LOOP_COUNT, START = 1, SKIP = ,  [color=green]'note that the loop NUMBER property is the length of the loop[/color]
               OFFSET: XAXIS = 0, YAXIS = 0, ZAXIS = 0, ANGLE = 0
FEATURE_TO_DATAPAGE=GENERIC/POINT,DEPENDENT,POLR,$  [color=green]create a generic point whose PA is set from the array created earlier[/color]
            NOM/XYZ,2,91,0,$
            MEAS/XYZ,2,ARR_ANG[LOOP_INDEX],0,$  [color=green]'notice that LOOP_INDEX is the index number of the array to use[/color]
            NOM/IJK,0,0,1,$
            MEAS/IJK,0,0,1
DIM DIMENSION_TO_DATAPAGE= LOCATION OF POINT FEATURE_TO_DATAPAGE  UNITS=IN ,$
GRAPH=OFF  TEXT=OFF  MULT=10.00  OUTPUT=BOTH
AX   NOMINAL     +TOL       -TOL       MEAS        DEV      OUTTOL   
PA    91.0000     0.0000     0.0000    91.0000     0.0000     0.0000 ----#----
END OF DIMENSION DIMENSION_TO_DATAPAGE
            ASSIGN/LOOP_INDEX = LOOP_INDEX + 1  [color=green]'add one to the index so the next loop will be the next array index value[/color]
            LOOP/END

You can also, instead of measuring 3 (or however many) features, loop the feature. Code your loop to offset it at whatever angle you need and dimension it in the loop. Be careful there though to code it so your alignment/dimension scheme tracks the angle from hole to hole. I sugest coding the offset yourself, I can post what that looks like if you'd like to see it. The dimension will go over as one feature. I give you these suggestions not knowing how you have programmed it. My first suggestion that I posted code for can be retrofitted to a program you may have written already. If you are going to re-write it then looping the feature about a certain angle might be the cleanest solution. Like I said I can offer nothing now that your data is databased but if you want to do this again in the future you can program the CMM accordingly.

Craig

[VONDY]

You can also, instead of measuring 3 (or however many) features, loop the feature. Code your loop to offset it at whatever angle you need and dimension it in the loop. Be careful there though to code it so your alignment/dimension scheme tracks the angle from hole to hole. I sugest coding the offset yourself, I can post what that looks like if you'd like to see it. The dimension will go over as one feature. I give you these suggestions not knowing how you have programmed it. My first suggestion that I posted code for can be retrofitted to a program you may have written already. If you are going to re-write it then looping the feature about a certain angle might be the cleanest solution. Like I said I can offer nothing now that your data is databased but if you want to do this again in the future you can program the CMM accordingly.

Craig[/quote]

Craig, I would like to see what that would like like also if you could post it.
Thanks,
Vondy

[craiger_ny]

Here is how I do it. If you put 91º in the ANGLE property of the loop it will revolve your features at 91º increments regardless of their actual location. What I am doing is measuring a feature at a 91º nominal location. Then I take my generic point "OFFSET_FEATURE" and move it to the same location of the last measured feature. Then I costruct a line that I rotate to. Doing it this way makes the nominal location of your next feature 91º from the last even if the last feature was located at 89.2º from the feature previous to it. Doing it this way will give you data that is purely feature-to-feature so you can evaluate the feature-to-feature variation. That is what I am guessing the intent is. Good luck PM me if you have any questions. I could even take your code and retro mine on it to make it more understandable.

PNT_ORIGIN =FEAT/POINT,RECT
            THEO/0,0,0,0,0,1
            ACTL/0,0,0,0,0,1
            MEAS/POINT,1
            HIT/BASIC,NORMAL,0,0,0,0,0,1,0,0,0,USE THEO = YES
            ENDMEAS/
A2         =ALIGNMENT/START,RECALL:STARTUP, LIST= YES  [color=green]'this is just a generic alignment scenario yours will be different[/color]
            ALIGNMENT/TRANS,XAXIS,PNT_ORIGIN
            ALIGNMENT/TRANS,YAXIS,PNT_ORIGIN
            ALIGNMENT/TRANS,ZAXIS,PNT_ORIGIN
            ALIGNMENT/END
            ASSIGN/POLAR_ANG = 0  [color=green]'this variable will hold the angle of rotation from feature to feature[/color]
V1         =LOOP/START, ID = YES, NUMBER = 3, START = 1, SKIP = ,
               OFFSET: XAXIS = 0, YAXIS = 0, ZAXIS = 0, ANGLE = 0  [color=green]'notice I do not offset the angle here. If I did it would not truely be from feature to feature[/color]
OFFSET_FEATURE=GENERIC/POINT,DEPENDENT,POLR,$  [color=green]'this is for alignment purposes. This feature acts as an origin of sorts to rotate to[/color]
            NOM/XYZ,2,POLAR_ANG,0,$
            MEAS/XYZ,2,POLAR_ANG,0,$
            NOM/IJK,0,0,1,$
            MEAS/IJK,0,0,1
CL         =FEAT/LINE,RECT,UNBND  [color=green]'line from the origin to our offset feature for rotation to[/color]
            THEO/0,0,0,1,0,0
            ACTL/0,0,0,-0.9993908,-0.0348995,0
            CONSTR/LINE,BF,2D,PNT_ORIGIN,OFFSET_FEATURE,,
            OUTLIER_REMOVAL/OFF,3
            FILTER/OFF,WAVELENGTH=0
AL_OFFSET  =ALIGNMENT/START,RECALL:A2, LIST= YES  [color=green]'this alignment will be how we rotate from feature to feature[/color]
            ALIGNMENT/ROTATE,XPLUS,TO,CL,ABOUT,ZPLUS
            ALIGNMENT/END
FEATURE_TO_DATAPAGE=FEAT/CIRCLE,POLR,OUT,LEAST_SQR  [color=green]'this is the feature you want to pattern around the origin[/color]
            THEO/2,91,0,0,0,1,1
            ACTL/2,91,0,0,0,1,1
            MEAS/CIRCLE,4,WORKPLANE
            HIT/BASIC,NORMAL,0.4651,1.9997,0,1,0,0,0.4651,1.9997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,2.4997,0,0,1,0,-0.0349,2.4997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.5349,1.9997,0,-1,0,0,-0.5349,1.9997,0,USE THEO = YES
            MOVE/CIRCULAR
            HIT/BASIC,NORMAL,-0.0349,1.4997,0,0,-1,0,-0.0349,1.4997,0,USE THEO = YES
            ENDMEAS/
DIM DIMENSION_TO_DATAPAGE= LOCATION OF CIRCLE FEATURE_TO_DATAPAGE  UNITS=IN ,$  [color=green]'this dimension goes to datapage[/color]
GRAPH=OFF  TEXT=OFF  MULT=10.00  OUTPUT=BOTH
AX   NOMINAL     +TOL       -TOL       MEAS        DEV      OUTTOL   
PA    91.0000     0.0000     0.0000    91.0000     0.0000     0.0000 ----#----
END OF DIMENSION DIMENSION_TO_DATAPAGE
            ASSIGN/POLAR_ANG = POLAR_ANG + FEATURE_TO_DATAPAGE.PA  [color=green]'now we add to the rotation the actual location of the last measured feature[/color]
            LOOP/END

By the way instead of this:

*
*
*
OFFSET_FEATURE=GENERIC/POINT,DEPENDENT,POLR,$
            NOM/XYZ,2,POLAR_ANG,0,$
            MEAS/XYZ,2,POLAR_ANG,0,$
            NOM/IJK,0,0,1,$
            MEAS/IJK,0,0,1
CL         =FEAT/LINE,RECT,UNBND
            THEO/0,0,0,1,0,0
            ACTL/0,0,0,-0.9993908,-0.0348995,0
            CONSTR/LINE,BF,2D,PNT_ORIGIN,OFFSET_FEATURE,,
            OUTLIER_REMOVAL/OFF,3
            FILTER/OFF,WAVELENGTH=0
AL_OFFSET  =ALIGNMENT/START,RECALL:A2, LIST= YES
            ALIGNMENT/ROTATE,XPLUS,TO,CL,ABOUT,ZPLUS
            ALIGNMENT/END
*
*
*

I tried this:

*
*
*
AL_OFFSET  =ALIGNMENT/START,RECALL:A2, LIST= YES
            ALIGNMENT/ROTATE_OFFSET,POLAR_ANG,ABOUT,ZPLUS
            ALIGNMENT/END
*
*
*

But it kept changing my nominal location of the feature to be measured (thank you very little PCDMIS). As clean as that looks it does not work, be nice if it did.

Craig

[VONDY]

Craig,
I have two parts in mind right now that I will be trying this out on.

Thanks,
Vondy

[craiger_ny]

I did not put move points in that loop so you'll need to program accordingly or use clear plane. Play with it, if it makes sense you should have no trouble applying it. Like I said good luck. Statistics can be fun when you get away from the hum drum of Cpk et al.

Craig