Skip to main content
J
JonathanHodgson13-Aquamarine
13-Aquamarine
March 27, 2017
Question

Equation curve / sketch projection accuracy

  • March 27, 2017
  • 1 reply
  • 4040 views

Hi all,

I've got a pair of components in which we've modelled an involute spline explicitly, using an equation curve.

The modelling approach is as follows:

Generate one involute curve from the base circle up to a diameter beyond the final tip diameter:

involute_1.png

Extrude this, by Projecting it into a sketch, mirroring it, and adding a root fillet between the two flanks:

involute_2.png

The involute equation includes an offset angle, which generates the clearance between the two parts by effectively rotating the offset about the origin.  If I measure the angle between the two curves at any diameter, I get a constant angle, as I expect:

involute_3.png

However, when I measure between the actual solid geometry (and this is what our analyst noticed when running FEA), the angle is not constant, and the OD is closer than the ID:

involute_4.png

It appears to affect both flanks, regardless of whether it's the 'original' or mirrored side, so I don't think it's caused by the mirroring op.  A curvature analysis also suggests that the mirror is good, although it does show some worrying unevenness:

involute_5.png

Is this an accuracy issue?  I've tried tightening the accuracy (absolute) but without fixing the problem.  Has anyone else experienced anything similar?

Thanks!

    1 reply

    J
    JonathanHodgson13-AquamarineAuthor
    13-Aquamarine
    March 27, 2017

    Also, the curvature for the two parts doesn't seem to match:

    involute_6.png

    P
    psobejko1-Visitor
    1-Visitor
    March 28, 2017

    In Creo 3 (M120), I compare an involute generated using the "curve from equation" (labelled with blue leader lines)

    to the same-sized involute generated by spline->from file sketching tool with the 90 points in the file calculated using the same formula but in an excel worksheet (light green)

    errors_in_curves_from_equations.png

    visually, the two involutes overlap  - but the curvature analysis reveals that there is something fishy about these curves from equations - their curvature graph should be smooth.

    J
    JonathanHodgson13-AquamarineAuthor
    13-Aquamarine
    March 28, 2017

    I suspect, following further discussion with a colleague, that the curvature is suffering from the transformation from equation to NURBS - and your model suggests that there are probably accuracy / interpolation / rounding / evaluation issues with the equation curves too.  That I can, to some extent, accept; but the larger-scale discrepancy in position is more of a problem to me right now.

    Anyway, I've got a call actively under investigation, so we'll see what's reported.

    Terms & ConditionsCookie settingsAccessibility statement