Straight Part Tangent to Curve in Assembly - Part can be flipped with drag

[Matthew_Policelli]

With two tangent constraints to get the two ends of a straight part tangent to a curved part, plus a mate constraint to set one end's height above the bottom plane (see screenshot), the part is tangent to the inside of a curve properly, and will adjust itself to any changes in the curve or in the length of the straight part:

However, when dragged, the part can flip quadrants:

The way I know of to keep it in the correct quadrant involves deleting one of the tangent constraints and constraining an angle to the bottom plane with Predict Offset and Orientation checked. However, this removes the adaptiveness to changes in the length of the straight part or the radius of the curve.

I have tried at one point to keep both tangent constraints and use an angle with limits so that it would stay in the correct quadrant but still be adaptable, but if I recall correctly I had to pretty narrowly limit the angle otherwise it for whatever reason caused inventor to crash when I tried dragging the part, so to me this is not an ideal solution.

Is there a better way to constrain the part within the desired quadrant?

[kacper.suchomski]

Hi

If the corners are not rounded, tangent mates are not needed.

Constraints that connect an edge to a surface are sufficient.

Which does not change the fact that such a set of constraints is not sufficient to take away all degrees of freedom; and from a logical point of view, both solutions are correct.

You can create a surface offset inward and constrain it to one corner to exclude logically unrelated scenarios.

Just remember that this distance is not the thickness of the board - it must be calculated or plotted because the thickness of the board is not radially directed.

[johnsonshiue]

Hi! I suspect Transitional constraint (use surface-to-surface to limit the scope) may work better in this case. Please share the files. The forum experts can help take a closer look.

Many thanks!

[3D4Play]

Is it an optical illusion, or is the offset distance different in the two screenshots? I can clearly see they are different scales, but it seems obvious the offset distance is not the same, so it begs the question: what do you call the "bottom" plane? I tried recreating this using an extruded surface, with two PL1/4"x4"x4", and couldn't recreate the issue, so I think we're missing some details. Generally speaking, tangent constraints have multiple solutions, so it isn't surprising to see this type of behavior.

See the video, where I've placed an extruded surface - my "curved" surface is cylindrical, not spherical- into an assembly with two plates, and partially constrained them both before recording. I highlight the two tangent constraints (which I believe replicates your assembly), then drag to see if it flips (it doesn't). I then constrain the bottom edge of the plate to the bottom plane with an offset as you also indicated, and drag again. The plates go wonky because they aren't constrained in one axis or to each other, so I add a flush constraint between the two plate edge faces. I went so far as to ground the bottom plate, and I could drag the tangent plate all the way around the cylinder face without a flip, so I wonder if one of your tangent constraints is getting suppressed in a model state, or something like that (because that would certainly allow the flip). Is this similar to your conditions? BTW, tangent constraints are necessary in this case, because Inventor doesn't allow mating an edge to a curved surface, and will force the edge to mate with the (cylindrical) surface's axis. I hope this is helpful.

[Matthew_Policelli]

It's not exactly the same situation as my original post, but this is perhaps an even better example of the situation I am trying to find a better way to address.

I have a curved piece and I want to place some square(ish) pieces on the curve. A good real-life example this would represent is a curved railing. This assembly would be welded together so I don't want the railings sticking out, I want them to smoothly follow the curved piece on both the inside and outside of the curve, and I want to be able to place them by arc length from one end of the curve (if possible). I think there's gotta be some kind of solution involving work planes normal to the curve at a specified point (and using sketches to place those points along the curve), which is something I can do in an .ipt (as shown below), but I need to do it in the assembly, not the .ipt.

I've attached the sample assembly.

Thank you!

[kacper.suchomski]

It is good practice to include a hand-drawn sketch with the intended effect/goal.

[Matthew_Policelli]

Good point.

Here you go

[SBix26]

Here are two of the three files revised to show how this might work.  I created a pattern of surfaces in EX1 following its shape (which is completely unconstrained, by the way...).  They are evenly spaced along the curve, though they could also be done as a sketch-driven pattern for irregular spacing.  The first square post is constrained flush to the first surface and tangent to the inside of the curve.  This post is then patterned following the feature pattern from EX1, so it maintains orientation relative to the shape of the base.

The problem with tangent constraints is the large number of possible solutions (as you have discovered).  The trick is always to find other constraints to reduce the number of solutions to one.

The attached files are Inventor 2022 format.

---

Sam B

Inventor Pro 2025.1.1 | Windows 11 Home 23H2

[johnsonshiue]

Hi! This seemingly simple motion is a bit tricky to animate. Transitional Constraint is probably your best bet. The key here is to narrow the scope of the transition using surfaces. Please take a look at the attached assembly.

Many thanks!