Review - PR 247981 Lower tier of composite true position shows large deviatio

huangyhg

<< Don Turcotte  --  05/07/07  16:44:33>>
From Spirit -
Lower tier of composite tp fcf reports large deviations.

Steps to reproduce:
Load attached program and look at FCFLOC1, etc.

Expected Results:
Composite deviations for lower tier should be no greater than upper tier deviations (and usually less since the lower tier has a tighter tolerance).
<<END>>

<< Changes made by Neil Kay -- 07/23/08  16:29:09>>
Action:  Paola Pallo to Yanhua Huang
<<END>>
<< Changes made by Neil Kay -- 07/23/08  11:57:22>>
Action:  David Petrizze to Paola Pallo
<<END>>
<< Jerry Naylor  --  06/29/07  07:41:11>>
wasaddedtoreadme42R
<<END>>
<< Don Turcotte  --  05/10/07  09:52:51>>
An additional problem here is that the THEO vector of the circle set is dependent on the global_workplane.  The Tp dimension should not be dependent on user workplanes.  I have fixed this so that in the case of a circle set, PC-DMIS will use the vector of the first (non-set) feature of the set.
Fixed in V42R, V42, V43B.
Files inserted to server
------------------------
V42\DIMENS\DATUMREF.CPP
V42\INCLUDE\DATUMREF.H
V42R\DIMENS\DATUMREF.CPP
V42R\INCLUDE\DATUMREF.H
V43B\DIMENS\DATUMREF.CPP
V43B\INCLUDE\DATUMREF.H
<<END>>
<< Don Turcotte  --  05/09/07  17:06:51>>
I have found a problem with the fix that was made yesterday for this PR 247981.  The basic fix was OK, but didn't work correctly when executing from the NC journal file (or any execution for that matter, since this is not specifically an NC problem) because the THEO vector of the internal set constructed for the considered features did not have a correct vector at execution time.  As a result the wrong degrees of freedom were being constrained for the lower segment of the composite true position so the lower segments reported large deviations.  I have fixed this and tested against Tony's journal file.  All the FCFs in the report for Composite True position now are correct.  The lower segments are all showing very small deviations as would be expected since there are no datums on the lower segments.
I have uploaded my fix against PR 247981 for V42R, V42, and V43B since this was a Stop Release PR.
Files inserted to server
------------------------
V42\DIMENS\DATUMREF.CPP
V42R\DIMENS\DATUMREF.CPP
V43B\DIMENS\DATUMREF.CPP
<<END>>
<< Don Turcotte  --  05/08/07  11:16:13>>
Merged into V42R.
Files inserted to server
------------------------
V42R\DIMENS\DATUMREF.CPP
V42R\DIMENS\FEATCTRLFRM.CPP
V42R\INCLUDE\DATUMREF.H

<<END>>
<< Changes made by Don Turcotte -- 05/08/07  11:16:38>>
Action:  Don Turcotte to David Petrizze, Status:  OPEN to REVIEW
<<END>>
<< Tim Wernicke  --  05/08/07  09:44:53>>
Hi Don, please add to V42R.  Thanks.
<<END>>
<< Changes made by Tim Wernicke -- 05/08/07  09:44:56>>
Action:  David Petrizze to Don Turcotte, Status:  REVIEW to OPEN, Priority:  Critical to Stop Rel.
<<END>>
<< Don Turcotte  --  05/08/07  10:16:35>>
The primary issue here is that the lower segment of the composite has no datums (this is legal) so there are six degrees of freedom allowed.  Since the considered features are circles, the fitter has very little data to work with when solving a problem in six degrees.  I have changed the code so that on the lower segment when the considered features are circles, PC-DMIS will restrain translation along the circle axis and also tilt of the circle axis.    This reduces the problem to three degrees of freedom which the fitter has no problem solving.  The restrained degrees of freedom would not contribute to an improved solution anyways so this is a valid solution.
I also fixed a minor problem in that the size tolerance of the datum feature was not reporting a negative value when PC-DMIS was configured to report minus tols as negative.  This also affected the linear graphic for the datum size report.
Fixed in V42 beta and V43B.
Files inserted to server
------------------------
V42\DIMENS\DATUMREF.CPP
V42\DIMENS\FEATCTRLFRM.CPP
V42\INCLUDE\DATUMREF.H
V43B\DIMENS\DATUMREF.CPP
V43B\DIMENS\FEATCTRLFRM.CPP
V43B\INCLUDE\DATUMREF.H
<<END>>
<< Changes made by Don Turcotte -- 05/08/07  10:29:27>>
Action:  Don Turcotte to David Petrizze, Status:  OPEN to REVIEW
<<END>>
<< Don Turcotte  --  05/07/07  18:01:33>>
I found a problem in Datumref with how degrees of freedom are constrained for the lower segment of a composite.  Fixing this has corrected many of the issues seen in the report with composite lower tier.  There are still some other reporting issues, however, which cause large deviations to be reported in the lower tier segment of some composite tp's, as well as incorrect datum shift values which I believe are related to the same basic reporting problem.  I am still looking at these issues...
<<END>>