Revolve feature seems to be using the wrong axis

[adamsd5]

Today I used a feature I've not used before.  It is behaving in a way that surprises me.  In the attached file, I have a center "dial" and an outer "case".  I want a pin in the dial that I can turn 90 degrees.  The pin would slide along a circular groove in the case.  So I modeled the pin, and tried using the revolve feature on one of the pins faces to model the groove.  This mostly works, but as the pictures show, Fusion 360 doesn't use the axis of revolution I selected.  It is as if the axis has been project into the plane containing the face I selected to revolve. 

For comparisons, I copied the pin body and rotated the copy 90 degrees to show where the revolved should have landed.  I also have a picture showing a nearly full revolution circle, which demonstrates the location of the axis.  I have tried selecting the origin axis and also the circles on the dial and the case bodies.  Nothing seems to do what I'd expect.

This picture shows a 90 degree revolve of the bottom face of the square pin.  Note that the revolution goes to the center of the target pin location, not the "first" face, which is what you'd get if it used the proper axis.  Also note that the arc doesn't reach the tip of the second pin.

This is a closer image showing the same thing.  You can see here how the revolution circle/axis is shifted down.  It is as if the axis I selected is projected into the face that I'm revolving.  Is that expected?  It seems surprising to me.  How do I get it to use the actual axis I selected?

I've also attached a .f3d file if that is helpful. 

How would you experts solve this problem?

[TheCADWhisperer]

You have picked the side of a body that does not go through the axis as your profile.

Also, you have posted this question in the wrong location. https://forums.autodesk.com/t5/fusion-360-support/support-board-best-practices/td-p/8445159

[adamsd5]

Thanks for the reply. You are correct. I saw nothing that said the face
and axis had to be coolanar, and I can't see a reason why an axis outside
of the face plane couldn't work. Is it possible? If not, do you have a
solution that does work?

[TheCADWhisperer]

I suspect I would model the geometry very very differently - but I do not know where you are going with this design.

(I would use the BORN Technique.)

[adamsd5]

Great! I am eager to learn. What do you have to suggest? I want a dial
that has that square pin in it that I can turn 90 degrees. I plan to put
an acrylic top on it with a knob on the dial coming out a hole in the
acrylic.

What do you have to suggest?

[adamsd5]

I would also like to reply to your comment about me posting this in the wrong forum, and quote some things from the best practices page you linked.

"If the issue you are seeking help starts with the words “How do I” or “What is the best way to..” then the inquiry may be better directed at the Design, Validate & Deliver board."

My question was not a how-to question.

"If you do have an error/warning message that doesn’t makes sense, spotted a bug or are seeing an unexpected behavior, please report these issues to the technical support team. The Support board can be used for this purpose"

Yep, I thought I found a bug.  In fact, I still think this is a bug.  I see no reason why a face needs to be coplanar with the axis of revolution.  I think it is a weakness at best, and possibly a bug, so I reported it here.

"The most impactful thing you can do on the board is communicating clearly."

If I may, with due respect, bring this one to your attention as well, it might speed things up.  If you think I should post to another forum, state which one it is and why.  Linking best practices makes me guess what your point is.  Also, stating that the face is not coplanar with the revolution axis is not the same as explaining to me that the face must be coplanar with the axis, if that is even the case.  Definitely saying that you would design it differently is not clear communication, and in fact it felt a bit insulting.  If you have a _constructive_ suggestion, I am all ears.

I am working on a different solution that might work, and when I have it, I will post it.

[TheCADWhisperer · Accepted solution]

I didn't add any clearance - understanding that we cannot manufacture perfect parts.

You should adjust for appropriate clearance for the Pin.  (See Attached.)

[adamsd5]

Thank you!  This is indeed a simpler approach.

For those who can't open the file and see the changes, rather than trying to revolve / cut one of the faces of a 3d body, he has used a sketch to get the cutout outline and then stretch / cut from that sketch.

[francescaxiellep]

For those of you simply interested in fixing a problem like this in your project, make sure the plane you sketched the profile on also contains the axis. If you are using the face of a body, sketch a profile of that face in the same plane as the axis (or move the body/reconstruct the face in plane with the axis).

To understand why this happens (and unless there is another use or good reason this approach is taken, why it could be considered a flaw - or bug), here is what is happening and why.

Another way to describe what is happening is that the axis selected is used in two ways, it is a direction and component of displacement measurement:

1. The direction of the axis around which the profile revolves (e.g around the y-direction - not axis!).

2. The perpendicular component of displacement (between profile and selected axis) that has the same plane as the selected profile/face.

What then happens is the axis of rotation is constructed at the component of displacement using the same plane as the profile with the same direction as the selected axis.

Hopefully that all made sense.

Ultimately there are three ways this revolve manouvre could work:

1. When selecting the profile and axis the full displacement and angle is taken into account, from there the rotation would move the profile at whatever angle it may be at around the selected axis (an example of this would be a fan where the blades are often angled or curved, the rotation moves the blade profile at an angle around the axis).

2. The axis is reconstructed in the plane of the profile so the profile moves perpendicular to the 'new' axis (this is what happens and in simpler terms than those above), it makes the most sense for this method to only use a component of the displacement.

3. Instead of the axis in 'option 2', the profile is moved (reconstructed) so that it occupies the same plane as the axis, this would also require using a component of the profiles displacement from the axis.

While option 2 is what occurs and is likely a simple way to compute this problem, I would have to argue that option 1 is by far the most intuitive for what should happen (devs if your reading!). Either way I think if the profile and the axis are not in the same plane (and mainly if option 1 is not used) there should be a message - or warning - indicating this, as there is little to determine the position of the constructed axis/profile that would suggest this to be something done intentionally.