PR 249136 The computation of error Profile seems wrong with the scan

huangyhg

The computation of error Profile seems wrong with the scan
<< Domenico Varacalli  --  07/12/07  19:22:58>>
The examples are a scanning of a line and arc of a circle.
The program SCAN_04_LINE measure SET1 element as DATUM A.
The LIN1 is a element constructed with the hits of SET1 and it becomes the DATUM B.
The output is on the PDF file. In the 1st file there are the results of the scannings SET1 and SET2
These two scannings are identical as path of points and the 1st run of program (SCAN_04_LINE_1.PDF) should give identical results for both the scannings.
During the 2nd run of the program (SCAN_04_LINE_2.PDF), I produce an error on the hits of SET2 (putting a piece of paper on the part).
The error of STRAIGHTNESS is evident (0,1302) but I do not understand the error of profile (0,0038). The example of the circular scanning is similar (program SCAN_04_CIRCLE).
In the 1st run (SCAN_04_CIRCLE_1.PDF) I do not understand the error (1.4876) since the two scannings are identical and the CIRCULARITY is 0.000.
In the 2nd run (SCAN_04_CIRCLE_2.PDF, putting a piece of paper on the part), the error of Profile seems excessive.
There are particular shrewdness, for the tolerances in profile and the scanning?
<<END>>


<< Domenico Varacalli  --  07/16/07  08:56:46>>
Ok Don, thanks so much.
<<END>>
<< Changes made by Domenico Varacalli -- 07/16/07  08:57:34>>
Status:  RESOLVED to CLOSED
<<END>>
<< Don Turcotte  --  07/13/07  09:43:15>>
The profiles in these programs are all of SET2 relative to either datum  A or datum B.  Profile is a calculation which compares the nominal hits of a feature or scan to the measured hits of the same feature or scan in the datum alignment.   The datums constrain degrees of freedom which means that unconstrained degrees of freedom can be used by the datum fitter to reduce the profile deviation (when using an FCF profile).  So a profile of SET2 to datum A is not comparing the hits of SET2 to the hits of SET1.  The datum feature is use to construct an alignment based on the type of feature.  Generally datums are prismatic features so an alignment level, rotate and origin can be constructed.  Since a scan is not a prismatic feature, the datum reference frame treats it as a point feature that constrains only the X,Y,Z origin.  This is not very useful so scans should generally not be used as datums.  Profile and form (straightness or circularity) are not the same calculation since form is calculated independently of datums.  In the case of straightness, form is the deviation of the measure hits from the least squares solution.
<<END>>
<< Changes made by Don Turcotte -- 07/13/07  09:43:25>>
Action:  Don Turcotte to Domenico Varacalli, Status:  OPEN to RESOLVED
<<END>>
<< Changes made by Tim Wernicke -- 07/12/07  17:49:17>>
Priority:  to High
<<END>>
<< Don Turcotte  --  07/12/07  17:30:30>>
Part of the problem here is that Datum B is a LINE (even though it is derived from the SET1 Scan) which means it is treated as an axial datum and contrains the level.  Datum A which is created directely from the SET1 Scan is treated as a Point type datum since a Scan is not a recognizable prismatic feature and so constrains only translation in X,Y,Z.  The X,Y,Z position derived from the Scan as a Point type datum A is different than the Line X,Y,Z of Datum B.
<<END>>