Equation driven curve accuracy for tangency

[jbhill001]

I'm using an explicit equation driven curve in IV 2020.  The curve is rather complex but is displayed as expected.  However, when I want to add another curve, say an arc and make it tangent to the EQ curve, they don't align and trimming the arc is not possible.  If I project the EQ curve in another sketch, I can then add, constrain, and trim as necessary.  But the projected curve is irregularly shifted from the original EQ curve.  The error is small, with the projected curve being slightly larger ~.0001" and no regular, but certainly enough to cause grief. 

It's even possible to see this on the EQ sketch.  If I add a point that is coincident to the EQ curve and zoom in as far as possible, the point is not on the line but is inside of it (in the mathematically correct location).  I don't know the internal math accuracy if IV, but could that be the problem?  I'm not able to share any data but wonder if anyone else has seen this.  Thanks! Jason

[johnsonshiue]

Hi! The curve definition is true but Inventor has geometric tolerance which could prevent the actual geometry from being created. Please share an example that illustrates the behavior. It could be a limitation or a bug.

Many thanks!

[jbhill001]

I haven't had a chance to make a model that I can share, but I can show a bit more about the problem.  The curve as shown on screen is not correct, but not when I snap a point to it, the point is in the correct location.  The top point in the attached image is located at (-.100,-.1886653717), this is mathematically correct and is constrained to the "curve".  The point below it is graphically on the line, which is where the curve is when projected from another sketch.  This point is at (-.100,-.1889515) which is almost .0003 from the correct location.  The offset along the curve is not consistent.

This is the equation: (.5/sqrt(PI))*sqrt(acos(1 ul-((2 ul*x)/OL))-(sin(2 rad*acos(1 ul-((2 ul * x)/OL))))/2 ul)

xmax = OL = .5

Any idea why a projected equation driven curve would be offset?

[johnsonshiue]

Hi! Many thanks for sharing the detail! I will try it later. In the meantime, you may want to try a simple workflow. Extrude the curve as a surface. Then use Thicken/Offset to offset the surface. Then project the offset surface edge to a new sketch.

Thanks again!

[Cris-Ideas]

Hi,

I suppose that most likely reason for that is that at some point (during projection) inventor switches from exact mathematical definition to spline fitting geometry.

This is the case for example when views on drawings are created.

In idw everything is spline fitted approximation.

As to solving a problem with lack of accuracy in your case you could try to create curve 100x larger or even bigger than do all modeling required and than derive result in to second part scaling it down.

This way you can get "precise" model that is actually spline fitted within accuracy you require.

This is the approach I used in the past when had similar problem.

Chris 

[jbhill001]

Thanks for the idea, Chris, I hadn't thought about that.  We'll give it a shot, but unfortunately, this is one case where SW might be the better solution for me.  I'm hoping the next release might work better.  Thanks!