Poor Parametric calculation within Inventor when involving arcTangent (Sensitive to rounding!)

[Timothyrsands]

Hello!

I am working on designing a parametrically driven truss for a project in college.  I am encountering a bad output value from a parametric equation I have derived.   Since ArcTangent is highly sensitive to rounding, I believe Inventor is doing a poor job of calculating the correct angle by rounding the numbers instead of carrying them through.

The angle I need is off by 0.29 degrees which adds up with you have a truss with many trusses, the gap created is compounded by how many truss members you want!

Please see the attached screenshots and pdf if you want to view it in more detail.  I am looking for a way to increase precision of the angle calculated.  I KNOW my equation works on the T-89+ CE calculator, but maybe I made some mistake on the Syntax with inventor.  But due to the angles being so close in magnitude, I am convinced it is a coding issue like using a float vs a double when running the calculation within Inventor.   Is this just a limitation of the software or has anyone found a work around?

Cheers!

-Timothy

[JDMather]

Can you Attach Inventor file here?

[Timothyrsands]

Hello,

I have attached the inventor file as requested.

[swalton · Accepted solution]

I'm not sure your derivation of the truss angle is correct.  I used the Ti89 angle and the truss spacing calculation in your model and I got a .757mm error in the top corner.  I think you are solving the wrong triangle with your equation.  I have not run through the derivation to find the issue.

Looking at your model, I think the actual problem statement is:
Given an outer rectangular structure that can change in width and height, lay out some number of variable thickness truss sections that will always meet in the inner corners of the structure. 

Trig is not necessary to solve this problem. 

I divided the overall length less 2 times the wall thickness by the desired number of trusses to get the required spacing.  Given that spacing and the truss thickness, I extruded a seed truss at one end of the outer structure.  I then patterned it based on the truss count and spacing.  The angle falls out of the linear measurements and does not need to be explicitly calculated.  I added a driven dimension measuring the truss angle, in case it is needed for later model features.

I used a feature pattern because it is more stable when changing the number of trusses than lines in a sketch.  Normally it is better to use feature patterns than repeated sketch geometry.  I also separated the outer wall extrusion from the truss extrusions.  That way, I can change the truss shapes without affecting the outer wall.  I find it easier to modify a model composed of many simple sketches/features rather than one made from one or two complex sketches.  Check the truss sketch constraints by pressing F8.  I used several different constraints to minimize the number of dimensions needed to fully constrain the geometry.  I normally create the geometry, add sketch constraints, then add critical dimensions.

See the attached Inventor 2022 model. 

[Timothyrsands]

I see what you mean. there is still a small gap somehow when using the TI-89+ CE calculated angle...  I'm not sure what issue there is with the equation as I used Mathematica from Wolfram to solve for Theta, but I will go with you method and forgo the needlessly complicated equation I derived.   

I still can't help but wonder what the issue is... idk why the same equation gets two different answers when using two different calculators, the equations do match one another.  I do plan to create more complicated curves and equations in the future, I loath not knowing why something isn't working.  It usually means I'm doomed to make the same mistake again later on.  I'll mess around with it some more and add a reply if I due figure out what's going wrong.

Thank you for the simple work around Walton, I appreciate your time. Definitely smarter not to do everything in one sketch.