PR 238753 Inconrrect 2D/3D angles displayed....

huangyhg

Inconrrect 2D/3D angles displayed....
<< Tim Wernicke  --  02/28/06  09:34:04>>
Inconrrect 2D/3D angles displayed.
Steps to reproduce:
-Open the attached part program and look at Item #25 angles in the program. The first one says 2D Angle from LINE 25_A TO LINE_25A. If you F9 the dimension you'll see I've selected feature LINE_25A relative to the X axis.
1.) I measured 2 surface points from cad then constructed a 3d line between.
        ( Points 25_A1 & 25_A2 )
2.) I constructed a 3d line between these 2 points.
3.) The nominal angle should be 0.0 degrees as the X nominals are the same with a 1,0,0 vector.
4.) The measured deviations between the 2 points are 0.03mm over a distance of 1.30 mm ( Z = 11.0 ?9.7 ). This trig? out to be 1.322 degrees.
5.) Note that if I dimension the line 2D to the  Z axis the measured angle says 0.000. This is wrong based on the measured vector for the line which K=.9997239 and the trig? out to be 1.346 degrees.
6.) Note that if I dimension the line 2D to the  X axis the measured angle says
?8.654 which is the complement of 1.346 degrees. The  I=.0234971 trig? out to  be 1.346 degrees.
7.) Now when I did the same dimensioning but using 3D instead of 2D I get the same basic results but the dimension to the X axis went from a ?8.364 to a +88.364 which is a completely different angle based on the 4 quadrants of a circle. Which is correct ? It can? be + & - !
8.) As you know they removed the ability to use the compliment of an angle when dimensioning. The software in 3.7 displayed the word (True) and the operator could simply F7 on the word True to get the compliment. Currently there is no was to get the measured angle to match the drawing as you are stuck with the PC-DMIS default guess.
<<END>>

<< Changes made by Tim Wernicke -- 06/18/07  09:58:32>>
Status:  RESOLVED to CLOSED
<<END>>
<< Jerry Naylor  --  06/15/07  11:50:50>>
wasaddedtoreadme40R
<<END>>
<< Tim Wernicke  --  03/02/06  09:08:29>>
Correct.
<<END>>
<< Bret Naylor  --  03/01/06  15:25:35>>
Reviewed.  I don't believe we will be releasing 3.7 again.
<<END>>
<< Changes made by Bret Naylor -- 03/01/06  15:25:43>>
Action:  Bret Naylor to Tim Wernicke, Status:  REVIEW to RESOLVED
<<END>>
<< Paola Pallo  --  03/01/06  09:42:52>>
Reviewed.
Being the first issue related to an oversight from 235967, that was inserted also in v37, why not to fix it also there ?
<<END>>
<< Changes made by Paola Pallo -- 03/01/06  09:43:14>>
Action:  Paola Pallo to Bret Naylor
<<END>>
<< Changes made by Tim Wernicke -- 02/28/06  18:28:51>>
Priority:  Critical to Stop Rel.
<<END>>
<< Bret Naylor  --  02/28/06  17:09:48>>
There are two issues here.
The first issue with 0 for the measured value is related to an oversight from PR#235967.  There is a section of code that checks for close to parallel lines in order to avoid math errors (such as divide by zero for example).  When the last set of changes were put in, the line that calculates the test value was removed in the case of measured values and so we were still using the nominal values for this test.  The whole test itself may be obsolete since I could not find any divide by 0 issues or invalid parameters to the ACOS function.  This test has been commented out for the measured case (since we now have up to four different solutions that we test against).  In any case, that issue should be solved.
The second issue has to do with the question of why do we get a negative nominal with a 2d angle and a positive nominal with a 3D angle.  This reason for this has to do with the transformation that is necessary into Major Minor Third coordinates for the 2D case.  We need to project the vectors into the workplane and so the transformation becomes necessary.  To determine the sign of the answer, we take the cross product of the first vector into the second vector.  The problem is that we are also doing this in the Major Minor Third transformation.  So in 2D we are crossing +Y into +X which gives us a result of -Z.  In 3D, we are crossing +Z into +X which is gives us a result of +Y.  The result of -Z in the 2d case is causing the nominal to be calculated with a negative value.  The result of +Y in the 3d case is causing the nominal to be calculated with a positive value.  I have changed this so we do the cross product in the original transformation (not Major Minor Third) for 2D angles and this solves this problem.
Note:  A work around is just to type in the desired Nominal. The measured value will update the next time the dimension is executed.
The first fix has been approved for V40R, V40, and V41.  The second fix has been approved for V40 and V41.
Files inserted to server
------------------------
V40\DIMENS\DIM_2D_A.CPP
V40\DIMENS\DIM_3D_A.CPP
V40R\DIMENS\DIM_2D_A.CPP
V40R\DIMENS\DIM_3D_A.CPP
V41\DIMENS\DIM_2D_A.CPP
V41\DIMENS\DIM_3D_A.CPP
<<END>>
<< Changes made by Bret Naylor -- 02/28/06  17:10:12>>
Assigned:  Paola Pallo to Bret Naylor, Status:  OPEN to REVIEW
<<END>>