Joint conflict in assembly with multiple occurances of component

[Anonymous]

I am trying to build an assembly with components which repeat over and over again. I built the Components in their own designs and then inserted them into a new design which is the assembly. When joining the parts in the assembly Fusion shows an error for joints which should have no conflicts. While trying to figure out why this happens I built two designs which illustrate this and have joint errors, which I don't know how to deal with.

In one design the geometry fits together but the last joint creates an error. In the other design having multiple occurrences of the same component creates a conflict. Fusion seems to apply each individual joint to all repeating occurrences of the component. I have added two public links to these projects, I hope they work and allow you to see everything relevant.  I am not so familiar with how the sharing function in fusion works. If this doesn't work I will create a screen cast and add it.

https://a360.co/2MbFSrZ

https://a360.co/2kY96y0

thank you in advance for any help or advice, I am stuck and a little frustrated at the moment.

-Markus

[TrippyLighting]

For the pyramid assembly I converted all joints to cylindrical joints, then it works, but:

1. I provided the extrude features with a user parameter for thickness. that will make it easier not to pick the wrong edge when selecting parts.

2. I made sure I joined all side triangles to the center triangles when picking the joint origins.

3. I provided the first three joints with joint limits for rotation to make assembly easier and provide the joint solver with only a limited room to have to solve the joint for. Try to avoid using the position capture feature in assemblies. It's not a problem in this small assembly but can hinder performance in large assemblies.

In general I personally wouldn't build this assembly this way. It can be done with a single rigid group joint if the parts are designed in place.

The other assembly has a few more issues I'll describe in a separate post.

[TrippyLighting]

Tri-leg assembly:

This is not a recommended assembly structure.:

It's an assembly with only one other assembly or component at the same level. Only one level is needed.

I am pretty convinced that the planar and cylindrical join can both be replaced with rigid joints. 

Mirroring components is often not such a great idea as it messes with your BOM (bill of material, oder in Deutsch Stückliste). In the end you want an instance count of 2for a part if it is an identical part and not two entries in that list with a count of 1. In this case you can also do that with a circular pattern as you've designed everything around the origin.

In general, however, your design is generally well structured and easy to follow.

However, having said all the above, the actual problem you are describing here looks like two big F.A.T. bugs in Fusion 360's assembly joints and this have noting to do with the things above. 

When re-opening one of the problem joints the preview shows several ghosted components in the linked assembly out of position.
That's a BUG.  

Then it complains about a missing joint origin reference as also shown in then last screen shot. You can reselect geometry and then it completes the joint, takes a second to think about it and then reports a joint warning. When you edit the joint again it tells you again there is a missing joint origin reference.

That's a BUG!

As can be seen at the end of this short screencast, the preview show the joint in the correct location, however, when the joint then completes it completes with the location flipped

That's a BUG. 

@paul.clauss Please add this to the other joint bug I just reported. I'll tinker around with this, but I believe this also has to do with patterned bodies in the linked assembly.

[Anonymous]

Hi,

Thanks for you reply, there's a lot of info here for me to work through. 

The Pyramid was a test to see whether I made a mistake in the geometry. To see whether the lengths in the actual design would fit together as planned. 

I tried to build this using offset planes or planes through three points or 3D sketches, none of these approaches worked. The lengths in the original sketch are all the information I have. If there is a way to build this without folding the shapes together could you please show how?

Also why does a user parameter make the selection easier?

I have started using joint limits in the larger assembly to stop the parts from getting knotted up in wrong directions, but it didn't help with the joint problem so I didn't focus on this. It is good to know that limits make the calculations easier/faster. I'll take more care to add them. 

It makes sense that cylindrical joints would add a slider freedom to the assembly and avoid a conflict, but shouldn't this assembly be able to work with revolve joints? the lengths that are joined together are all identical so they should't have to slide past each other. Since this was a simplified test object to try and find the root of the joint problem, and I got an error message was the reason I added this in the post. So is this also a bug in fusion? To me logically revolve joints should work and using cylindrical joints is a sort of workaround, or am I seeing this wrong?

Again, thanks for your response and thanks for taking the time to answer my questions.

[Anonymous]

And there is even more info here to work through. Thank you for a very thorough analysis of the design. 

After hitting a wall building the assembly I have done so much tinkering that my project folder has gotten pretty cluttered. and doing things like pushing the parts of the Kugelleager (ball joint/tri leg) into a new component are results of this tinkering.

I didn't know that difference in a BOM between mirror and patterns. I use mirror quite often, I guess I'll be adding more patterns. 

The Planer/Cylindrical joints are definitely overkill, I thought I had replaced all of those with revolves, but your're right most of them could just be rigid joints. I have also started using rigid groups, is there any reason not to turn the whole central light fixture into a rigid group?

I try not to use motion captures in my assemblies, and when I do, I delete them when they are no longer needed. In this case though, the parts would not come together on their own, so I had to manually push things into place and then join them.

I will make some changes to the design including to the patterned parts in the sub assembly. I don't have as much time on the weekend to work on these so I might not be able to upload before Monday. When I can, I'll share another link or two to show what I mean and where these assemblies are supposed to be going. Maybe there is a workaround or something so I can still build what I am try to do.

oh and I also sent this problem as a question to Autodesk, I guess I'll send them a link to your response. You're way faster! 😉

[TrippyLighting]

It's not the user parameter that makes the selection easier, I phrased that incorrectly. Thickening he plates makes selecting the correct edges easier. Thickening all plates at the same time is being made easier by the user parameter.

I'd agree that in a pure analytical math environment this should work with revolute joints only, but my guess would be that the joint solver does not use analytical but numerical math. where rounding errors can be a problem. Pure speculation on my part.

[TrippyLighting]

The reason you hit a wall is simply because of bugs.

I've not looked at a workaround but I believe I know how this can be solved.

The reason your design caught my attention and why I went through it in more detail is because I was tracking down a bug with patterning that I reported just yesterday. @paul.clauss whom I tagged in this thread is a Autodesk Support Engineer that looked at my case and I invited him because I believe patterning the components in the central hub and then using a rigid group joint is what is creating this buggy behavior in this case as well.

When joining components, as a general rule use the joint type with the least amount of freedom as it creates less of a burden for the joint solver.

[Anonymous]

OK, so I spent some time with the design this morning and changed a few things. Now everything is working much better.

-First off I got rid of the circular pattern in the central hubs.

-There are joint limits on all the central joints.

-The Lighting fixtures have had all their as built joints removed, then I turned them into rigid groups.

I still have to move everything around by hand to get the joints to line up, I think the Form is to complicated for Fusion to solve automatically. But now There are no more error messages.

Thanks for you help here! I am guessing that the key change was taking out the circular patterns in the central hub, since that was what you pointed at as being the root of the bug.

I will attach a link to my new assembly in case you want to see the changes, but in the next few days I will delete all my tests that are cluttering my Project folder and break any links into it.

https://a360.co/2l0fPYp

[pawel.potyrala]

Hello @Anonymous,

thank you for the updates, bringing this issue to our attention and for sharing your designs! @TrippyLighting thanks a lot for help finding a workaround.

Using the feature "Circular pattern" for one of the components of the assembly seems to cause further joint conflicts. As I mentioned in a private e-mail to Markus, I logged this issue in with our development team for further investigation with the number FUS-40731.

Best regards,

[Anonymous]

Yesterday things were working better. Today there was an Update to Fusion, since then I have been having troubles again. Maybe it's related, maybe it's a coincident. I rebuilt all the central hubs with "Plane at an Angle" instead of circular patterns. but the individual instances of the joining arms seem to be linked to each other. I made a screencast to show what happens when I try to push two surfaces together that should be joined. I manipulate one ball jointed arm and another one moves with it.  This is what keeps happening to the joints and why I get conflicting movement error messages.

[pawel.potyrala]

Hello @Anonymous,

Thank you for the update! Would you mind sharing the link to the latest version of the design? I would like to see if there are any links between the components that might cause this behaviour. Thanks in advance!

Best regards,

[Anonymous]

Here is the link to the test design from the screencast.

https://a360.co/2MkxXbQ

[TrippyLighting]

@pawel.potyrala You might want to record this behavior in the viewport as a bug. 

If you move the timeline marker back to before the 2 grounding operations and drag the 4er Kontenpunkt Reduzier around in the viewport, the revolute joint symbols, which belong to the linked Kugelgelenk component, remain in place.

If you grind the part (and capture the position) and move drag around on the parts the symbols jump back not position and it seems the whole assembly still works, but this is still a bug!

[pawel.potyrala]

Hello @Anonymous,

thanks for sending over the design. @TrippyLighting - thank you a lot for the help and for noticing the bug. I am logging this one in for further investigation.

Best regards,

[TrippyLighting]

@Anonymous When working in assemblies with such large degrees of freedom, you have to be very, very, V.E.R.Y careful with what is joined to what, so what you consider to be the stationary component, when creating the joint, should actually be the second component selects in the joint creation dialogue.

Then it probably helps not only to define joint limits, but also rest positions somewhat close to the final orientation of the part. This will avoid having to use the position capture feature.

I believe when using the position capture feature I see another bug or two. Holy Crap!

[TrippyLighting]

@markus.hueppauf You had asked how to design this pyramid in 3D from the sketch you ha with the unfolded sides.

This is more or less a 2D pencil and paper approach transferred to 3D CAD.

That way you can design everything in place and can avoid the laborious problem with having to create tons of joints. 

[Anonymous]

Thanks for that, brilliant way to solve the position of the last point of the pyramid. The only thing I had to watch a couple of times, because I didn't know the feature, was when you did the intersect of the circle with the sketch plane. Glad I know that feature now. Saw you Draw the circle and had to face palm, why didn't I think of that. So obvious once I saw it, I was going about the problem entirely the wrong way, trying to project or sketch the lengths, trying to figure out the angles.

I'm still frustrated with the actual problem from this thread, but I'm learning along the way.

-I tried to rebuild the Assembly differently using lots of joint limits.

-If I try to use a rest in the joint limit it doesn't work. It just doesn't stick there. I haven't had much luck finding google results for joint position rest.

-There are some things that I always thought were strange about Fusion, just never thought of them as a bug. Like the weird floating disassociated joint symbols.

 -At the end of the screen cast you can also see how the sketches of the components show up at the places where the components were inserted into the design. I'm not sure if that is supposed to happen.

Here is a link to the new setup of the assembly, maybe you can take a look at it.

https://a360.co/2JEjRoj

[TrippyLighting]

I think this exceeds what the joint solver in Fusion can solve as there are too many interconnected joint loops that it would have to evaluate. I created a more simplified version and could not get it to work either.

[Anonymous]

I think I will have to take out the solved tag from this thread. I thought I had a solution but I keep simplifying the design and keep getting the same conflicts. Maybe this is too complicated for the joint solver, if so I guess that could get passed on to the development team. I have another version where the last joint should work but fusion keeps giving a conflict. Here is my last attempt for now.

Last attempt:

https://a360.co/2JXpugo

The Strange thing is the day I marked this as solved the assembly worked. I have a rendering of everything fitting together, no conflicts. Now I can't get it to work no matter what. I think I'm going to drop this for a while. I managed to produce one rendering, that will have to be enough for the client. We will just start building this and hope there are no surprises along the way. 

[pawel.potyrala]

Hello @Anonymous

I am submitting the latest version of the design to our developers and will inform you here on the progress. Feel free to write a private message as well. Thanks a lot!

Best regards,