Revolute Joint - Odd behavior.

[Beyondforce]

Hi Guys,

Since the last update, the Revolute joint is behaving really odd. I'm explaining in detail in the screencast. This post also experiencing a problem with the Revolute joint. Maybe they are connected, I don't know!

Cheers / Ben
---------------------------------------------------------------------------------------------------------------------------
Did you find this reply helpful ? If so please use the Accept as Solution or Kudos button below.

Check out my YouTube channel: Fusion 360: Newbies+

[jeff_strater]

I need to investigate a bit, but the thing I notice is that your pin components are not grounded.  So, when you add the Motion Link, what is happening is that the pins are rotating.  The problem is that, because both components are free to rotate, the solver can choose to rotate either.  You will most likely get the expected behavior if the pins are grounded.

Jeff

[Beyondforce]

@jeff_strater, something doesn't make sense here. 

1. Why the Motion Link (Solver) is moving only one gear!?

2. Why the solver doesn't take into account that the gears have been chosen first on creation?

3. In the last example, the pins are connected to a bridge with a Revolute joint, which means they are free to move from both sides. When I animate the gears everything works fine, even though nothing is grounded?

As far as I know, the grounding is for the model not to move, when manually testing the joint animation, and that makes sense!

I was working on another model, everything is grounded as it should, but I still have problem with the Revolute. Which is why I created this test file.

Ben.

[jeff_strater]

Hi @Beyondforce,

I think I can answer some of your questions.

1. Why the Motion Link (Solver) is moving only one gear!?

[jeff]  It's not only moving one of the joints.  The right-hand side is also moving, you can see the joint symbol move.  It's just that the solver chooses to move the pin component, not the gear.  Which, because the joint system is underconstrained, is a valid solution for the system.  Remember that the solver doesn't know that the gear is the "important" component, it only sees two components related with a revolute joint, so it moves one of them

2. Why the solver doesn't take into account that the gears have been chosen first on creation?

[jeff]  This is a fair question.  The "which component is picked first" behavior is, I admit, somewhat confusing.  This behavior, though, is only used in the command preview animation.  That preview animation is not a full solve, it is only an animation of the two components selected for this joint.  That's why the rest of the mechanism doesn't follow.  We did this for performance reasons.  As you know, the performance of a large assembly can be slow, and we didn't want to bog down the command.  But, for the full solve, all components are equal.  This is why we recommend more completely constraining your system, and that includes grounding at least one component.  In this case, you would need two grounded components (or one grounded and a rigid group created)

3. In the last example, the pins are connected to a bridge with a Revolute joint, which means they are free to move from both sides. When I animate the gears everything works fine, even though nothing is grounded?

[jeff]  Most likely you just got lucky.  That can happen - it's pretty random in underconstrained cases

As far as I know, the grounding is for the model not to move, when manually testing the joint animation, and that makes sense!

[jeff]  Yes, and also it more fully defines the behavior.

Jeff

[Beyondforce]

Thanks @jeff_strater,

If you could send out a documentation on how the joints work, that would be great.
1. Right now I'm very confused whether it matters the order you choose the components or not!
2. The "solver" is not following the basic principles of the Fusion 360 Joints!
Joints are allowing freedom of movements. In my example, the freedom of rotation should ONLY apply to the Gear and Not the Pin.
In order to create a joint, you need 2 components, but it doesn't mean that the joint should be apply to both of them.

I'm working on a complex model with 2 gears, one is normal and the second gear is triangular. I'm trying to join them together, and I was very close. But for some reason the Revolute command doesn't work as it should.

If you are 100% sure, that there is nothing wrong with the Revolute joint, then I'll have to rethink the technique.

Ben.

[jeff_strater · Accepted solution]

Hi @Beyondforce,

I don't really know if these lower-level aspects of the joint solve are documented anywhere.  Maybe they should be, I agree.

Regarding:

1. Right now I'm very confused whether it matters the order you choose the components or not!

The answer is:  It only matters during the command preview animation, that is when you see the Animate control in the dialog:

Once the command has been OK'd, the order doesn't matter any more.  If both components are free to move, the solver can move whichever its algorithms pick.  To control this, the best way is to "lock down" the one you don't want to move.  That makes it unambiguous.  

Regarding:

2. The "solver" is not following the basic principles of the Fusion 360 Joints!
Joints are allowing freedom of movements. In my example, the freedom of rotation should ONLY apply to the Gear and Not the Pin.
In order to create a joint, you need 2 components, but it doesn't mean that the joint should be apply to both of them.

The Fusion joint solver is what we internally call a pure "variational" solver.  Meaning that there are no directional relationships.  A revolute joint is not "from one component to another", it is an equal relationship:  Those two components can rotate with respect to each other.  A good analogy is to think of the sketch solver (which, by the way, in Fusion is the same solver).  If you have two lines with a parallel constraint between them, either line can move, and the other will follow.  You wouldn't want the sketch solver to restrict you to only move the "first" line, and have the "second" line react, and not be able to move the second line.

I would not say that I am 100% confident that there is nothing wrong with the Revolute joint, or that what we looked at here is what is going on in the other case you are looking at.  However, I am probably 95% sure that in this case, what I said is the reason for the behavior you see.  If you just ground those pins, I'll bet you get the behavior you want.

This is a good discussion.  This is not obvious stuff, and as you say, it is not well documented, and further, not obvious at first.

Jeff

[Beyondforce]

Maybe something is wrong with the model as well. I'll try to use a different technique and we'll see.

Thanks for your feedback.

Ben.