Position Calculation Failed Due to Initialization Error

[bradMFYPZ]

I'm getting extremely frustrated with fusion. I'm deigning a robot. At this point we're looking at around 1200 parts (including duplicates), that I started out doing a top down design because fusion is purported to be great at it. huge mistake. It seemed like every time the model got rebuilt, one thing early on would shift slightly and cause cascading failures. Almost like an excel table that isn't adjusting intra-cell references when changes are made. So I do some research, and find out that fusion is way more stable when you do bottom up design. Well...****. So I go back through and re-model all my components as isolated files. build isolated sub-assemblies. build a top level. it took a day of feverish work but I got it done. Everything seemed really stable for a week. Then yesterday I hit ctrl+B and... EVERY. SINGLE. JOINT. BROKE. compute failed. position calculation failed due to initialization error. What does that even mean. I can roll the timeline back to that joint. edit the joint, everything seems fine on the edit joint dialog, no errors, everything plays nice. hit enter. still red. go back to previous revisions. build is broken. How am I supposed to get anything done if my previous work keeps getting UNDONE, and I have no recourse to repair it. Autodesk; get at me. I'm about to buy a solidworks seat out of pocket, because my organization can't handle these delays.

[ryan.bales]

So there is a lot here to decipher. I'm(someone at Autodesk anyway) is going to need to see the file. 

I don't know that top down or bottom is any better than the other- at least in what i've seen. I'm sure i'll get some flak from others on here saying that as there are camps in either corner. We really could be looking at a scope and complexity issue and that gets complicated. 

Can you share the file? 

[HenryDara]

Yeah, Fusion really sucks for complex designs. I have had a lot of the same problems, and am about to just give in and get solidworks too. Support has been looking at my designs, and crash buckets, and they havent been able to find out whats going wrong. I just tried again, and about a day into redoing my design for the sixth time it happened again. Related to an external file that has ZERO errors. People here will tell you you must have done something wrong, but in the end, FUSION HAS MORE BUGS THAN MAIN IN JUNE. Thats what our problem is, not our workflow, or having not followed some best practice perfectly (of which there are more than one can keep track of in Fuxion360).

When this happens, you cant align components, you cant join components, and sometimes every joint in my timeline will go yellow when I add a joint at the end of the timeline. (oh yeah, always move your timeline slider over cause they havent fixed that bug yet either)

I for one am hope more people start complaining about this problem, which for some is non existant, but for others like you and I it is a regular occurrence. Maybe others designs arent as complex as ours are, I cant say for sure, but I dont think mine is very complex at all. Fusion sure does though. SMHA

[TrippyLighting]

@HenryDara 

Your designs are not overly complex. They are just done incorrectly and that is why you get these error messages and others don't 🙂

You posted some designs in the "Jeff Strater" thread and I've responded there.

[ryan.bales]

@bradMFYPZ 

I want to reply in case this gets off track a bit. @jeff_strater and i looked at your model very thoroughly and we had no immediate answers. Jeff sent it to the team responsible for looking at assembly issues for now. 

My recommendation is likely to use less planer + position capture combos - this is not stating that is the cause, but if related could be something that will mitigate it. 

Peter jumped in here - @TrippyLighting - and has great experience with joints and can provide some insight as well.

I steer clear of planer as it never feels right and has too many degrees of motion to account for me - but joint theory and joint practice vary depending on person. Jeff is a master of joints has stepped in behind the scenes at my request to look at your file. What concerned me most was deleting the broken joints leaves them in the browser- that is scary and is one reason we reported it to that team. 

[bradMFYPZ]

@ryan.bales  Could you recommend an alternate method for forming joints? There's specific reasons why I'm using sheet metal parts and T-slot extrusions in my designs, and to my knowledge there's not a great way to align through holes and T-slots, except for maybe making a rigid joint and measuring all the component dimensions and making a rigid with very specific offsets. This seems like an extremely cumbersome workflow. This is why I'm using planar joints, to locate custom components onto rectangular prism COTS components.  Also, 3 plane datuming is a core concept in GD&T, it seems like something so fundamental should be an inherently stable way to position things in relation to one another. 

[bradMFYPZ]

I'll be watching that thread. To my knowledge, I'm doing everything in the most stable manner possible.

[TrippyLighting]

---

@bradMFYPZ wrote:

I'll be watching that thread. To my knowledge, I'm doing everything in the most stable manner possible.

---

The "knowledge" part tends to be the problem. If your timeline has position capture feature for the purpose of assembling components then that is a problem.

But I do have a couple of suggestions for the fine folks at autos to improve the behavior of assemblies in Fusion 360 my magnitudes. Maybe after this thread they listen 😉

[bradMFYPZ]

The first time I built this assembly I made gratuitous use of position feature and rigid group. Realizing that was a big part of the instability, I rebuilt it using only joints, and deleting all the position features as I went along.

[TrippyLighting]

I am going through the 3D printer design and if you are interested in learning then I might be tempted to explain a few things.

There are lessons to be learned here for the users, but also for the Fusion 360 team!

You repeatedly and unfortunately consequently move components into position using the move or align tool, then you use the position capture feature to capture their position and then you use a rigid group to "assemble" them. That is entirely the wrong technique (for a timeline based design) and I have no clue where you picked that up. No Autodesk Expert Elite here on the Forum would recommend that!

If you've seen that in a tutorial, please provide a link.

Another technique is that you use component mirror to mirror not only individual components but subassemblies. This was also subject of a recent thread, so I went and asked the 5 senior mechanical design engineers in our company if they use that technique. Bear in mind that we work with Solid Works and design manufacturing automation systems with thousands of components. "Senior" means that 4 of those guys have 30 years of engineering experience and each of them 15-20 years in Solid Works alone. It is not a recommended technique!

In Fusion 360 it creates an incorrect BOM for obvious reasons. Interestingly enough the feedback was that in the past it did similar things in SW.

Just avoiding these 2 things alone will make a lot of difference.

[TrippyLighting]

@ryan.bales 

I'll share the design later.

The sketch is missing its reference plane. Why do I not get a selection for that in the menu ?

Selecting stuff in the timeline has become a real hassle lately and @HughesTooling has reported some of these things as well. Time to get moving on that!

[TrippyLighting]

@ryan.bales never mind on the "redefine sketch plane" issue. I simply had to roll back the timeline to right after the sketch, then I could select it.

[bradMFYPZ]

I don't know how you know what I'm doing in my designs. I only sent them to @ryan.bales . I initially was using that method with select items, but upon realizing the instability of the position feature, I stopped and rebuild my model using only joints. The issue in question is with an assembly built using joints, at every level of the assembly/sub-assembly, not using any position features or rigid groups. 

Also, I explicitly don't use the "mirror" feature on sub assemblies because of fusion's inability to create new sub assemblies without mirroring every single part. I don't know what you're getting at about the "senior" engineers at your company, but Solidworks is perfectly capable of creating mirrored assemblies/sub assemblies without creating new parts, as well as mirroring the mates. I do sometimes use the mirror feature with parts that need a left/right handedness. 

[TrippyLighting]

My last post I simply posted in the wrong thread. I've not seen your assemblies!

As it pertains to mirroring components, I am fully aware that this can be done in Solid Works. I've worked with is since 1998 😉 It simply isn't a recommended technique. That doesn't mean that sometimes it can be helpful.

[HenryDara]

Thank you Peter for being of so much help. I picked apart what you did bit for bit, and started applying these changes throughout my design, and magically things are just getting better. 🙂

You put a bit of work into my problem, and I just wanted you to know I appreciated it.

Thanks again...HD

[bradMFYPZ]

@ryan.bales any movement on this from the "team responsible for looking at assembly issues"?

[ryan.bales]

As far as i know, no there is no change.