Interesting problem with joint tolerances

[fulcrumusa]

Last night I ran into an interesting problem with a joint that I was trying to make.

My design is a symmetric frame with two ballscrews on each side. I am trying to join the ballscrew nut with a body that was designed to fit around it. I designed that body using the ballscew nut on the left hand side of the assembly, then copy-pasted it for the righthand side. I've attached an image for a reference.

However, when I attempted to join the two parts, Fusion360 popped up an error saying that the joint would create a conflict. Upon a bit of investigation, I found that the two parts are off by 1.98E-6mm on the Y axis when they should be perfectly aligned.

Now, with that small of an offset, I think what is happening is an accumulation of rounding operations. 1.98E-6 is an incredibly small offset. All of my parts are in the order of mm so that is 6 orders of magnitude from my scale. Considering my design scale and the alignment offset, I can't see how this is something that I screwed up.

I can easily work around this by changing the position of the body so the offset is gone. However, the question is what caused this offset and how do I prevent such things from happening? How does Fusion360 deal with such rounding accumulation errors? Does Fusion360 have a way to allow some tolerance for joints or do they always have to perfectly align?

[fulcrumusa]

Any guidance on this?

[fulcrumusa]

I tried to use the workaround that I mentioned in my original post and it did not work. Chaning the position of the body got rid of the offset on the Y axis. However, the joint still creates a conflict. At this point, I don't know how to go about finding what the conflict is. Unfortunately, Fusion is also not very helpful in this. In fact, the "More Info" link is useless as it just repeats the error message from the pop up.

I've attached a screen cast showing the problem. I would appreciate any help in figure this out. I am sorry for the delays in the video. My Fusion is extremely slugish with this design. Even though it's not consuming much resources, the performance is almost untolerable. 

[fulcrumusa]

Can somebody please help?! I am stuck and I can't make progress on my deisgn.

Autodesk, anyone? I can provide a link to the design if needed.

[jeff_strater]

Hi @fulcrumusa,

Sorry for the lack of response here, as you can tell, the traffic has increased recently, and sometimes things fall through the cracks.

First, I think I understand your performance problem.  From the video, it looks like you are using Contact Sets.  This is almost certainly causing the performance problem.  So, the first thing I would recommend is just turning those off.  At the risk of being too honest, I almost never recommend using Contact Sets.  It works fairly well for very small designs with very simple geometry, but beyond that, performance is just unacceptable.

I don't think that this will fix your conflict, but it will make debugging it a whole lot easier.

And yes, please share a link to your design, so that we can help you figure out what the conflict is.

Jeff

[fulcrumusa]

Hi Jeff,

Thanks for responding. I am not at home right now so I can't modify the design to turn off contact sets. I'll do that tonight.

I would also prefer to not post it publicly at this time. Could you please give me an email address and I'll invite you to the project?

[jeff_strater]

Absolutely.  If you invite me at [email protected], I'll be happy to take a look.  Alternatively, you can use the private message function from here to share a link to me.  Either way works for me.

Jeff

[fulcrumusa]

Thanks, invitation sent.

I was thinking about this last night and one guess that I have is that sketch planes used to create the mounting bracket was not correctly aligned with the two components to which is connects (hopefully, this will make more sense when you look through the design). I'll have to figure out the best way to check that.

[TrippyLighting]

I am not sure the assembly problem has to do with the contact sets you have enabled in your design but there's a good chance they might be contributing.

What I am certain about is that the preformance problems entirely are a result of using contact sets as I just recently experienced the same thing when helping another user that also had problems assembling is design. He used contact sets to emulate the stops of a linear slide in parallel to having a slider joint in the same axis with the same parts.

Removing the contact set and setting the joint limits properly helped in that case.

[fulcrumusa]

@TrippyLighting, I did try to make that joint after turning off the contact set for that particular ballscrew and Fusion still flaged a conflict. Regardless, I will be turning off the contact sets because the performance drop is increadible. It's so bad that sometimes I sit in front of my computer thinking "I don't really want to deal with Fusion right now". It just completely takes the fun away.

Do you guys thing that supressing the contact sets is enough or do I need to completely remove them?

[jeff_strater]

Right, I saw the same thing - suppressing the contact set did not get rid of the conflict (I didn't expect it to), but will make your life much more bearable.  It shouldn't matter how you turn off contacts, as long as they are off.  Although, if it were me, I would just delete the contact sets totally, because you won't ever want to re-enable them

I have not yet figured out the source of the problem, but I suspect that you are correct that the mis-alignment is at least part of the problem.  I have asked a couple of colleagues who worked on joints to take a look at the problem with me, but it may take a bit of time for us to get to the bottom of it.  I have made a copy of the design, so that I can edit it without affecting your design.  My first step will be to cut down the amount of components significantly, to simplify the problem.  

Jeff

[fulcrumusa]

@jeff_strater, thanks for helping me.

I checked the alignement of the construction planes and they seem aligned correctly. I also positions all the components as they would be if the joints were working and ran an interference computation. Fusion did not find any interference. I am no closer to finding an answer.

I did also remove the contact sets. I'll post on the affect of Fusion's performance after I've had some more time with the design.

[jeff_strater]

I have not forgotten you, @fulcrumusa.  I also have not had much luck finding the source of the problem.  I suspect that you are right that it has something to do with the alignment - a conflict between the sliders along the drive screw and trying to join the two components together.  Some very small differences.  I will continue to investigate as I have time, and I've also enlisted our joint experts.  But, your design is pretty complex, so it's taking me some time to unravel it.  I can't promise I will fix your design, but we'll keep trying.

Jeff

[fulcrumusa]

Hi @jeff_strater, thanks for helping me. I also noticed these increadibly small offsets that seem to have appeared in the design. I am not exactly sure how that's happening. As I said, in my original post, my suspecion is that it's an accumulation of division remainders or rounding operations. A lot of dimentions and/or points on my design have been calculated using formulas.

Please, let me know if and when you find something? Thank you.

[jeff_strater]

I have a solution, if perhaps not the ideal solution, since we still do not understand the root cause.  I suspect you may be correct that it is just accumulated error.  Long term we would like to add "sloppy joints" (not the technical term) to Fusion, which would allow for some play in joints, rather than rigid mathematical correctness.

The solution is more of a workaround, but a valid one.  I finally realized that what we are trying to do here is to take two components, each of which has a slider joint to the same component, and join them rigidly together.  Which makes one of the sliders extraneous.  So, I just deleted one of them, and was able to create the rigid joint.

Hope this gets you "unstuck" with your design.

Here is a screencast:

Jeff

[fulcrumusa]

Hi @jeff_strater, that seems like a good (and obvious) workaround. I didn't think of it because in my head I always thought "Fusion's joints are supposed to mimic real-world joints" and in the real world, the two slider joints is exactly what's happening. However, removing one of the sliders will not only workaround the problem but also reduce the computational load on Fusion.

In any case, I am still interested in finding the real problem. In that vain, I looked at the other side of the assembly, where the same joint seems to be working. As it turns out, it only works in certain cases. Despite the fact that on the other side of the joint, the bracket was designed around the elements it was joining (using those profiles as starting points), Fusion refuses to make the joint if I use a different joint origin. Below is a screencast to illustrate what I am saying:

I'll keep messing around to see if I can find other things that may be helpful in debugging this. Let me know if you want me to try something specific?

P.S. As you can see, my performance problems are still there despite me removing the contact sets. If this is due to the number of components in the design, I would have to say that Fusion is not ready for primetime yet. I shouldn't be coming down to a crawl if it is to be used for complex designs. But that's a separate problem

[Anonymous]

Hi guys,

I have the same problem. In top of that I am using imperial and metric components in the same time. I cannot make a joint because the components are off by 0.000000010168414 mm I have find out this was a problem after many many hours. In real world, there should not be more than 2 digits (metric) difference when making joints. 

Any update on the "sloppy joints" ? I think the proper term should be "modifiable joint offsets" and everybody should be able to set the valuer as they wish.

Please let me know when this is going to be implemented.

Thanks,

Cristian