Sketch Dimension not updating with references

Angles have always been a twitchy thing in Creo and most CAD systems I've used. Notice, for instance that if you are defining a plane at an angle from another plane, entering a negative angle will move the plane to the desired position, but the angle will not be -40.0, for example, but just 40.0. The "sense" of the angle changes, but the dimension is forced to be a positive number. Even a simple sketch of an arrow, a line with a triangular "head", defined in the X,Y positive quadrant, exhibits odd behavior if you change the angle of the arrow from the "X = 0" plane to something like 170 degrees.

How to avoid this nonsense? You could define a Datum->Curve->Curve from Equation. Make the curve Cylindrical, then if you have parameters  for the ANGLE and RADIUS, you could have something like the following equations:

R = 3.0 + t * 6.0
Z = 0.0
theta = angle

This curve will behave correctly, Setting an angle of something like 227 degrees gives the correct result. It seems to work because the handling of cylindrical coordinates is more mathematically "rigid" than the typical cartesian coordinates.

Try using a polar grid in sketcher to create a sketched curve used to "clock" the angle. I am not certain about this but by setting the grid to polar it may alter how the intent manager treats the clocking angle dimension. In Creo 7.09 using this technique supports full 360 degree rotation of the clocking angle without failure. See the video below and the construct of the sketched curve in the picture.

Hi @VV_10659302 

This would be a good example for PTC to review, because it does highlight the deficiency here in how the software produces strange results when the angle dimensions "lose" their "sense" and switch from CW to CCW clocking.

However, I do think this particular problem is more related to the sketch scheme and the use of the perimeter+variable angular dimension and the mathematical reality that there are multiple solutions that satisfy the perimeter constraint.  For example, 2 solutions that produce 1500 perimeter at angle 200degrees:

As you demonstrated, if  you vary the angular dimension and "approach" the intendent solution, you get the "correct" result.  But if you jump directly to the  "multiple solutions" zone, the system picks the "wrong" result for you.

It would be great if the software could somehow remember and stay with your "intent".  Seems like a reasonable request.  I recall a product idea that asks for similar thing, but can't locate it right now.