Joints seem to be a challenge for Fusion 360 Ultimate

[TrippyLighting]

While making pogress in my LED lamp assembly (and getting better at recording little demo video's with quicktime to demonstrate odd behavior of Fusion 360 Ultimate) I again came across a problem with joints.

I needed to edit the base sketch of basically the first component that went into the assembly. It was only a small edit that should not affect any of the joint constraints, however, a number of erot messages in respet to joint problems popped up. Interestingly enough, everything in the assembly is still in its place and the one moveable (hinge) joint funktions just fine.  

The good thing is that nothing crashed this time and that already is an improvement! 

Here is a link to a video on dropbox to demonstrate the behavior. I'll gladly share the link to the file if someone wants to look at it. 

[O.Tan]

What does carrier board do?

And I'm willing to look at it 🙂

[TrippyLighting]

Thanks for the offer. At this time there is not much to check. I re-opened the file this morning and interestingly all but two of the joints shown in the video seem to be fine. Interesting!

Here is a little image to compare with the video. It's a lot less yellow:

I've posted several times highlighting problems with joint constraints. The other posts usually refered to ptoblems that crashed Fusion and on one of them seems to serious enough to result in a hot fix.

Let's wait for that an see how things behave afterwards.

[Jon.Dean]

Have you seen the training videos on You Tube?

https://www.youtube.com/playlist?list=PLmA_xUT-8UlJpYbha60oAQMxFlBY371oZ

See the videos on Joints, they are pretty good.

Jon.

[prabakarm]

We would love to get your data and look at the behavior in more detail.  Can you share your model?

Thanks,

Prabakar.

[kgrunawalt]

Hi, I'm a developer of Fusion joints. Thanks for the informative video. I cannot completely diagnose the issue without the model, but I can shed some light on what is probably going on here.

The "conflict" errors arise when joints restrict positioning in conflicting ways. For example, a component that is a box can be rigidly fixed to two grounded components (or bodies in root component which does not move). This can solve as long as the component has certain dimensions. When these are changed, the joints are pulled apart and cannot solve the system. The good news is that this is probably not a Fusion bug. It is a case where a redundantly constrained system was changed by the sketch edit, causing a real conflict. If so, this can be fixed by you.

This circumstance happens when you have a loop of joints. A loop can be needed (example: a four-bar mechanism) but they can be introduced by accident, making the assembly prone to conflicts. Ground components function as a single unmovable component, so these loops can occur unexpectedly when attaching components to ground and each other. Fusion will warn when adding a new joint will cause a conflict, but in your case, the conflict did not happen until you changed geometry later. Removing unneeded loops will make this problem less likely. A good Fusion enhancement would be to help identify these loops.

All CAD systems with assembly constraints will have this problem, but Fusion offers an interesting way to solve it. To resolve this, a good practice is to look at the first joint or snapshot feature in the timeline that has a conflict. Fusion uniquely adds kinematic joints and snapshots in a linear feature history, so you can go back to where the problem started after the geometry change and look at the model before it becomes more complex. Once there is a conflict, later joints and snapshots will also complain of the same conflict. It may be that there is just one conflict to resolve that results in a lot of noise.

It is interesting that the revolute still works even though there are conflicts in other joints. This indicates that your mechanism has multiple independent joint systems which can fail individually without locking up the others (another thing Fusion can do). Grounded components tend to isolate systems. For example, two doors attached to a grounded wall are independent systems sharing the same (grounded) component. When a system has a conflict, it is frozen -- no joints in that system will move. The first failure in your model is in the cylindrical joint cyl6. The edit to the sketch probably caused a kinematic loop of joints to not be able to assemble. The Cyl6 closed the loop and is therefore marked as in conflict.

The best way to fix this is to look at the system of joints that includes Cyl6. You can move the timeline marker back to just before the Cyl6 joint or just after. Is there an unnecessary joint that is overconstraining the system? How does the geometry change affect this system? Editing one of the joints to change its type (degree of freedom) or select different geometry can resolve the problem. If there is an unnecessary loop of joints, deleting one might be the solution. You can temporarily delete a joint by suppressing the feature in the timeline (right-click on feature in timeline and choose Suppress Feature).

Also, joint occurrences can be suppressed individually by right-clicking on the joint in the view or browser -- but I don't think that helps you . A joint definition may have multiple occurrences (e.g. a piston cylindrical joint in a multi-piston engine assembly has multiple occurrences -- each with its own position). These occurrences can be suppressed to help diagnose conflicts. To make a joint be suppressed at a specific point in history, you need to move the timeline marker back to that point in history, suppress the joint occurrence, and save the suppression in the timeline as a snapshot. It is simpler to just suppress the joint feature. [Edit] Another way to avoid loops is to unground a component.

A final suggestion for now: you can use the "as-built" joint type when the joint's components are already in assembled positions due to earlier features in the history. This type of joint is much more ammenable to geometry changes earlier in history because it does not try to match up two separate geometries that might be pulled-apart by a geometry change. The "as-built" joint accepts the joined components as they are and defines the joint. Use as-built joints as much as possible when components don't need to be moved into an assembled position by the joint.

If this helps, please let us know. If not, please provide the model shown in the video and I can give you more specifics.

[TrippyLighting]

Hi Jon,

Yes, I've watched some of those and agree thet these tutorials are good. Also I don't mean to suggest that there is a conceptional  problem with Joints in general. I apologize as I can see that the title of my post can be interpreted that way. I do like the joint concept of Fusion 360 as I can see advantages compared with other Software Packages I am familiar with.

I have worked as an Engineer fro 25 years and have approximately 10 years of experience with Solid Works and have used Alibre/Geomagic Desing Pro for the last four years for my private endeavours, such as the LED lamp design I am experiimenting with in Fuision 360 Ultimate. I have also used Blender for about 8 years and hope to get to give the Sculpt part of Fusion 360 Ultimate a spin soon.

What I do mean to say, however, is that there are some bugs that need to be ironed out. I have made a number of posts that all have to do with joint constraints in one way or another.

This thread describes a problem that when restructuring an assembly with joined subassemblies Fusion 360 Ultimate crashed.

This thread describes a problem that when I created a pattern of a assembly containing some joint constraints that broke some of the constraints.

[TrippyLighting]

I can certainly share the the model!

Please provide your email and whether you prefer to receive an archived assembly or want me to share the project. I am not at my home computer right now but I believe I exported an .f3d archive before continuing to work on the model. 

[kgrunawalt]

Great. When you get the chance, send the archive to katrin.grunawalt@autodesk.com.

[TrippyLighting]

Hi Katrin,

I created another screencast - this time using the Autodesk screencast utility - and hope this helps explaining the situation. I've shared the video with you.

I also hve shared the project with you so you can look at the data directly. I really don't remember what version I saved when creating the archive but the online version of the "LED lamp v20" is the version with the problems and perhaps one or two versions prior to that show the problem as well.

Thank you also for the excellent explanations in you previous post. That belongs into a user manual or into a tips and tricks section.

Also, as this gets all too easily overlloked when bugs or unepected behavior are reported. Autodesk Fusion 360 Ultimate certainly has a lot of potential and really, there is no vaible alternative on the Mac platform and I commend Autodesk for tackling this. That was the one and only Software that I was truly missing in my toolchan that previously I was not able to find a solution for that runs natively on a Mac.

So a big Thanks! form here for all the hard work!

[kgrunawalt · Accepted solution]

Hi,

I haven't seen the screencast. I signed up for screencasts after your reply, so maybe I missed the notification. However, I did take a close look at the model and found out some things.

First, regarding the failing joints, there is an issue that I found that makes the joint solve fail when it should not fail. This is something I've not seen before, so it is probably rare. I have a fix that I'll test thoroughly. If that goes well, it should be in the next release or hotfix. You've probably noticed that these failures are not predictable. This is likely due to tolerance issues. The final two joints in version 21 (the cylindrical joints) complete fairly complex loops. The good news is that I think the fix will resolve this issue. The bad news is that you have to wait for it.

A workaround for this model is to do the following (based on version 21):

1. Modify->Compute All to re-compute all features. You should see two workplane failures, a joint origin failure, and the final two joints and snapshot failures.
2. Move the timeline marker to just after the second-to-last joint "Heatsink Adapter <-> Gator board - left"
3. Delete the second to last joint
4. Create a new cylindrical joint, but use the Lamp Arm assembly:1/Ring aperture v1:1 JointOrigin1 instead of the Heatsink Adapter joint origin that the deleted joint used. In my testing, this will solve correctly.
5. Move timeline marker to end of timeline. The final joint should solve.

In my testing, this resolved the joint issues. You might see a different result since this issue is a bit unpredictable.

A couple notes about this model

1. This is a very "bottom-up" design where all joints are assemblying joints that use joint origins. With Fusion, if parts are created in-place, then the "as-built" joints are better. This is a more top-down approach where the history constructs parts in place and joints are added without the need to identify two joint origins. The bottom-up approach is still valid and necessary of course, but top-down designs can be much simpler to construct and modify. A simple example is block body that is split into two bodies which are turned into components (via browser). These components an be joined by an "as-built" joint without any need to assemble them. Then the original block can be modified and the whole assembly will update nicely. When possible, consider building things in place.
2. This model inserts other models as part of the bottom-up workflow. When Fusion supports external references (xrefs) as planned for early next year, this workflow will be greatly improved. The inserts will simply reference external models which can be edited to affect referencing models. A complex assembly can be a mix of top-down created assemblies and parts that are inserted into other assemblies and assembled without copying all the features into the history. Stay tuned.
3. This model has unnecessary snapshots created to temporarily move a part to make it easier to assemble. This workflow is pretty common. The assemblying joints don't need these prepositioned parts, but the designer finds the prepositioning convenient. The resulting snapshots will not harm anything, but they don't contribute anything and they clutter up the timeline. This is something we'll work on minimizing.

I hope this helps a bit. The fix for the troublesome joint solves is in the pipeline.

[TrippyLighting]

Hi Kathrin,

Yes, I realize now that I have a little more experience with Fusion 360 that I shoud use the In-place Joints and not rely so heavily and unnecessarily on joint origins.

The concept of joining parts in Fusion is quite different from Solid Works or Alibre/Geomaic Desing of which I have a combined 14 years os experience with. This reqires some re-thinking and breaking of old habits. Making mistakes is how we learn 🙂

The reason that I inserted all of the components from external files is that I had to basically toss my first model due to constant crashes (also joint related. Differnt thread ;). I could not export all the parts that I had desigend, because I had not realized that when I don't activate the component first the Sketch for e.g. the base extrusion of that component is created at top level instead of it being in the component substructure. On export I would have lost the sketch and unfortunately at the moment you cannot drag a referecned sketch into a component. Catch 22.

So for backup reasons I decided to model all the separate pieces in single files and assemble them later. The initial design was done in Geomagic Design and actually I have most of the parts in form of machined aluminum pieces in my hands.

While Xres are nice I actually very much like the top-down ability of Fusion 360. It allows for much deeper parametrisation of the model without having to touch more than one part file and I am going to test that next with this model.

Also thanks for the workaround . As long as noting crashes I can wait for the fix.

Lastly, you guys Rock!!! I'ts great to be able to reach developers directly with a problem and is greatly appreciated!

[kgrunawalt]

Thanks for the encouragement and patience! Fusion has always been supportive of the top-down assembly modeling approach back when it was a direct-only non-history-based modeler. When we added a parametric history, we did something that is pretty unique and powerful, but takes some getting used to. We combined top-down modeling with a single parametric timeline that includes joints and positioning operations.

The assembly-based parametric history means that parts can be built in place in a deterministic way. This allows us to create as-built joints in history that adapt when earlier history is modified. Typical CAD systems will solve assembly constraints in a bottom-up way based on the component hierarchy. Fusion's timeline will solve them as joints and snapshots are encountered in the timeline. Snapshots can be used to position parts for later modeling operations.

Because Fusion has not supported external references, the traditional bottom-up method using separate models has been hampered. This will be rectified soon. It is unfortunate that you have been using separate models in a defensive way due to instabilities. Ideally, separate models will be strictly for re-use and collaboration. This should become much easier in the coming year.

[O.Tan]

Yeah! I've always wondered what is As-Built Joint for and realized that it's only available in Parametric mode. So for Direct mode, is there anything similar?

[Mike.Zhang · Accepted solution]

You can use Joint in Direct Modeling, the As-built Joint is only designed for Parametric Modeling.

[kgrunawalt]

As-built joint features in the timeline use the parametrically computed positions of components to define How the joint relates the two components. For example, a timeline creates a box, splits it, and then converts the split bodies into components. Then a revolute as-built joint is added by selecting the two components and the midpoint of an edge created by the split for the axis. The split can be edited by moving the split construction plane. The as-built joint will adjust because the as-built feature in the timeline re-computes after the split is edited.

In a direct model, an as-built joint is feasible, but it lacks the history context needed for it to adjust to changes. A *regular* direct joint can adjust to changes without the history context because it is created with more explicit selections of defining geometry on both of the joined components (we call these geometries "joint origins" if the are created separately). This extra information allows a regular joint to assemble the two components and adjust when the geometry changes in both direct and parametric models.

We've thought about having direct as-built joints, despite their inability to adjust to changes. The rigid group is effectively an as-built direct joint. We think the direct rigid group is a special case that is warranted. We understand that regular joint creation can be trickier than as-built joints when a model is already assembled. It would be nice to have the simplicity of as-built joints in direct models, but they are more dependent on parametric history for smartness.