Getting the workflow right

[kjellhar]

Hi,

After using the tool for a few weeks now, and getting through with a somewhat larger design, I have made a few observations that I would like to discuss with someone more experienced then me. This will be a long one, but I hope at least some of you will stay with me. Maybe there are simple solutions to my problems, but if not, they may be worth a discussion. I'm not looking for point by point answers here, but a few viewpoints, long or short would be nice.

There is no doubt that F360 is a powerful tool. It is promoted not only as a design tool, but also a tool for doing top-down design in teams with build in (or soon to be) support for version control (partial now), data sharing through A360 (should be there, but I still haven't got it to work), and I'm sure there is or planned to be some sort of review functionality, annotation of drawings and models and other management goodies.

What I don't see, and this may be good or bad depending on who you ask, is a build in work flow. When you work on models and assemblies, the tool don't guide you through a specific way of organizing your work. This is good if you or your team manager if you're part of a team, is well organized, and the flow is defined and followed. It needs both a plan and discipline, but you get the choice of how you want it. The bad thing is that it needs both a plan and discipline.

So, let's get a bit more specific. I started a design, which in the end would involve around 40-50 objects + hardware. There would be 3 moving joints.

Needless to say, I started out without a plan. Or at least not a plan for how I would organize my design. Everything went well until I was about 80% complete and I discovered a minor skew in the symmetry of a component. Some bolt holes that were shifted a few fractions of a mm. This puzzled me since I was sure I had placed them exactly where I wanted them. Well, no problem I thought, I'll just go back in time, fix some sketches, and everything is good...... wrong.

It turned out that since I had no good plan. the relevant sketches was all over the place. They usually was in some component up the hieararchy, and since I had moved around the components after drawing them to align them with the bolt holes, the sketch did not necessarily match with the part. Once I started to look, there were small skews all over the place. I have no idea how they got there, but I tried to walk back through the time line to figure out where it all started, which I actually managed to do. The plan was to fix the features in turn, but now the time line failed me. When I started to fix features, I broke down stream features all over my design. Since the driving sketches was partly shared accross several components, and things had been moved around and modified afterwards, everything fell apart. Then went to the end of the time line and tried to delete the offending features in order to create them again. This also failed since a delete operation apparently goes back in time to delete the feature when it was created. If something depends on the feature, it's getting difficult. 

Long story short, there was a tangle of dependencies criscrossing my design the prevented me from doing anything to fix this.

Right now, I can see to options:

1. Revert to DM mode to ditch the timeline, and then delete and recreate the features.

2. Start with a new file, and just do the drawing again, this time with a plan. Shouldn't take too long since I know how everythinkg should look, but very boring.

Before I found my current job, which involves a weird mix of mechanical and electronics design, I used to be a chip designer, making integrated circuits in a teams as large a 30 designers. These designs were usually performed in a manner we called "top-down, bottom-up" which meant we did the specification from the top, detailing out as we went down through the hierarchy. Then,when we had all the specifications ready, we started to design the components from the bottom up.

In my design, and in the training videos I have watched, the flow is more like you start to design from the top down doing specification at the same time. This could work, but given the state F360 is in at the moment, you need a fairly good idea of where you want to end before you do this. Why is that. Well, a couple of examples just of the top of my head.

• Well, there are limitations to when and where you can move a sketch in the hieararchy. If a sketch is created in one component to create another, you cannot necessarily bring it into the new one.
• Even though there is supposes to be a context menu choice for breaking dependencies, I rarely find that I am in that context.
• If you join stuff, then work on, and then you need to break a joint in order to move something a little bit, there is a good chance that many parts will fly apart, reverting to some old position.
• It is also difficult to know what is depending on what. There is nothing that presents the links between a feature, and the entity that drives that feature in a good way. At least not when things are getting complicated.

All in all, once your timeline starts to get long, you got more and more limits to how much you can clean up your design. I would like to isolate each component as much as possible, and keep the interfaces as clean as possible, but if you mess with a couple of sketches early on, you will stay in that mess.

What I think I'm going to do is to regard this first messed up model as my top-down specification, and just look to it while doing my bottom-up design. That way I can keep the separation between my components, and only maintain a smaller amount of inter-component dependencies. The important thing is to prevent things from exploding if you fix one little thing inside a component.

Back in my chip designing days, we spend a considerable effort inventing our work flow. Quite a lot of work was put into how changes on sub component should be done to avoid breaking the rest of the design. Clean and simple interfaces was imperative. Same in software design. I know all of this is not directly applicable, but much is.

I don't know if the Autodesk tema had any particular work flow in mind when they made this tool, but I'm guessing they went for flexibility to create your own flow. I also guess that a large majority don't have a particular flow. Many because they don't need it (simple designs with few parts and only one designer), while a few will get themselves in a mess. Others, especially professional teams, will maybe fail the first time, but they will invent a flow.

Maybe what F360 needs is some sort of best practice manual. Or at least a suggested work flow for those who don't have the resources to invent their own.

Ok, I think I will stop here. Hope this makes sense to someone out there. 

Sorry for typos and bad English, it's getting late here.

Take care,

Kjell

[Anonymous]

Hi, really good and interesting post.  Here are a few things I follow.

1) Only one body creation action per sketch.  So a hub/root/parent object sketch for the rectangle object in below image.  Then a spoke/branch/child sketch that is defined on a face of the parent object, in this case the parent sketch face.  This a) insures the position of the child sketch/object properly updates, via dependency on projected geometry, if the parent sketch/object is altered, b) errors do not occur when the multiple objects are attempted to be joined as a single body, c) ease of understanding, etc.  **Another important benefit of this workflow, going along defining new sketches on existing body faces (even if sketch is for new component, just be sure new component is activated so new sketch goes into it), is no need for subsequent body or component moves, which avoids as you mentioned sketch location difficulty (reverting to original location upon editing).  There is a specialized workflow that deals with this as discussed here ( http://forums.autodesk.com/t5/general-fusion-360-questions/my-workflow-and-future/m-p/5559908/highli... ) but is not in general recommended except when needing specific capabilities, like the sketch of one component driving the sketch of another component.  

Note in the below image, the single child sketch involved two profiles (the tapered profiles), one driving the other via mirror/symmetry constraint, yet the subsequent revolve action was a single action. 

2) if a parent sketch has part of its geometry projected into a child sketch, rename the parent sketch to specify this.  If other subsequent sketch(s) needs to project same geometry, use the same parent sketch (which is now named accordingly as the master projector sketch of that specific geometry).

Regarding fixing broken events/actions in the timeline, if you can share the exported f3d file here, I can try to make a short demonstration of the best ways to fix stuff like that, without actually too much pain. 

some other thoughts at the moment:
3) Turn on color swatch at the right of the timeline, which color codes timeline actions to components.  In combination with this, try to work as much as possible on one component at time, resulting in nice separated component groups in the timeline.  It's also possible to even keep timeline very tidy by deleting old move actions (that are superseded by new move action or snapshot), event reordering...

In near future update, it sounds like will have more freedom to move sketches into different components, to allow more logical grouping.  For now have to remember to activate component that would like subsequently created sketch to go into.

Regarding joints, there are several things to consider to insure good position update and what not.  I made an instructional video for this stuff as I was learning it, located on this forum page (playing video as html5 allows playback speed increase):  https://www.youtube.com/watch?v=XyFJscakKs8, which was posted and discussion begins on page 4 and 5 of a thread called "having-trouble-with-fusion-in-general" (for some reason posted link not working). 

As you get used to following good procedures, the editability of your models will improve drastically, as you mentioned.

You've probably also seen good pages like this for best practices : http://forums.autodesk.com/t5/design-differently/fusion-360-modeling-best-practices/ba-p/5502252

.

Jesse

[fredsi]

kjell,

Indeed, as you have discovered, Autodesk has not put forth a series of best practices for creating fairly complex mechanisms (though there are many excellent videos dealing with various aspects of the software). It has been left to individuals - such as Jesse and TrippyLighting to contribute successful workflow descriptions. Here is a video from the screencast library by TrippyLighting that also describes needed elements of a good workflow (hope I have the link correct):

https://screencast.autodesk.com/Main/Details/ce351e26-a8b5-4c90-821f-dd06936524b9

Over the last six months I have had several false starts looking for a robust workflow (to handle part, assembly design, CAM, etc all in one file); have not quite found it yet. Certain elements still don't conform to the one central file concept (CAM in my case). Further, in the next major update, external references are due to be added - this may alter or invalidate some current best practices. 

Lastly, should you find a methodology that results in solid, editable models and assemblies I hope you can find the time to detail those methods here as well. Thanks.

Fred

[prabakarm]

Kjell, thanks for a thoughtful post.  Jesse and Fred, thanks for engaging with your thoughts.

One of the tenants of Fusion 360 was to provide more flexiblity than current systems offered in being able to explore and iterate on ideas before committing to a particular design, especially when employing top down design workflows.  Our intent (maybe a naive one) was to have 101 training material (link below) and based on user needs move to higher level of training material.  

http://help.autodesk.com/cloudhelp/ENU/Fusion-GetStarted/files/GUID-CC9A8E39-356C-43EA-9EDD-0E9C1321...

Fred, am I to assume that the Fusion 101 training does not give you the information you need to be successful in having a good workflow?

In addition to the learning content, I am also sure we have bugs which might come in the way and workflow gaps that we are not aware of that needs to be addressed.  We are going to puts our heads together to provide a set of best practises but we will also need your help in seeing how you have gone about leveraging tools so that we learn from you guys and provide the best solution possible.

Thanks,

Prabakar.

[kjellhar]

jjurban55,

Thanks for you suggestions. I haven't read everything in your link yet, but I will sit down with it once I find some time to focus. You obviously have thought a lot on this subject.

fredsi,

Thanks for the link. And yes, if I manage to figure something out, I'll share it.

prabakarm,

I like the flexibility, and the ability of the software to explore. The only problem, as I see it, is that exploration and iterations tends to result in messy designs. F360 is not good at keeping things tidy, and there are still a few bugs / missing featurs that prevents the users from cleaning up. I know there are features in the pipeline that would help. For instance, I just watched a preview video of branch and merge, which is a feature that would impact any work flow greatly. So, at the moment I think F360 need a more strict workflow in order to work good for big designs. As soon as the issues that are mentioned are straightned out, the flow can loosen up, allowing more agility in the design process.

One temporary solution to this could be to keep a scratchpad model that you mess around with, trying out concept, and make things fit. And then transfer the ideas to the final design file. 

And last, a few thoughts on what to consider in a flow.

1. Naming convention, both for components, bodies, sketches and construction elements. This can be used to tie these elements toghether.

2. Component hierarchy. How it is structured. What is the condition for creating a new level. Only bodies in the lowest levels, while the higher levels only contains sub components.

3. Placing Origin for each component. Should you make all Origins coincide when the components are placed right, or should you releate each origin to a major feature in its component.

4. Drawing sketches in the component it belongs.

5. Well defined interfaces, like mating surfaces and bolt holes. Keeping the sketches for these features separate from other features that does not directly affect other components.

6. Putting joints at the right hieararchy level. Maybe they should be kept in the component that also keeps the sub components that are joined.

7. If you need to create some concept sketches to see how things will fit together, don't create other entities from it. Just use it as a guideline to create the real driving sketch before making the part. Make sure that deleting or changing the concept sketch will do nothing with the actual design.

Not a complete list, but at least some things to consider.

Kjell

[fredsi]

Hi Prabakar,

I went through the Fusion 101 training documents and exercises when they were first released. Once again, they provide a good examination of the trees (tools, options, etc.) but don't provide an overview of the forest and how one can navigate through it. Ultimately, the video I referenced in my response to kjell was the most helpful in ferreting out the details of a successful workflow (thanks TrippyLighting!).

If I were to offer my $0.02 it would be to create a video (or a series of them) that go from a blank sheet to an assembly of modest complexity (10-12 parts, plus fasteners) that can be edited, exploded, and animated. It (they) should show the creation of the individual parts, or at least the first few to establish such concepts as activation, isolation, etc.; and show how to maintain parametrics, relationships based on projected geometry, and the like.

Ultimately my point would be this....there may be many ways to obtain good results depending on one's goals, but there should be one or two detailed, documented workflows that would gaurantee a high degree of success. If one could use the software and immediately be productive, that would leave time to explore other methods of model/assembly creation.

I realize I'm putting emphasis on mechanical design, but that's what I do; and am further 'burdened' by a long career in CAD with explicit workflows. Not sure, but that may descirbe a significant percentage of folk that try Fusion 360 and come away disappointed that they couldn't get 'into' the intended workflow (because they can't find it).

Finally, once the CAM side of Fusion matures, it will be the design software I use; but it will not be a transition that I will describe as simple or easy

Fred

[CGPM]

30 Kudos to Fred's last post!  My biggest gripe with Fusion is the lack of training information, and this hasn't changed over the last 18 months.  I also would LOVE to see videos of someone who really knows Fusion, and knows how it is supposed to be used, just dweeb with it for 30 minutes at a time.  Maybe one video a week?  No preparation, just sit down and design something.  This would not only show the big structural things, like workflow and organization, but all the little nuances that are hard to put in the training material, or learned a few months ago but have since forgotten.  I have watched Jesse Urban's videos and really like them but a video like this with someone who really knew Fusion would be soo, sooooo much better.

Same thing for CAM, please!

By the way, most of the complaints I hear on outside forums about Fusion stems from not knowing how to use it.

[taylor.stein]

Is this the type of video you'd like to see? https://www.youtube.com/watch?v=E0bhdr84FNU

I tried to walk through the approach to solving a problem, as well as my thought process along the way. Happy to make more like them!

Taylor Stein

Fusion 360 Evangelist

[kjellhar]

@taylor.stein 

I loved your video, and it was really helpful in a lots of way, but not at all for this particular issue.

To be frank, he way you did this design was about exacly how I started out, and what drove me into this corner. This flow works very well for a tiny little design like this, where you have just two parts, and nothing complicated, but once you get into complex designs with many parts that interact in many ways, it will be a mess.

I can't speak for @fredsi or the others, but what I would love to see is this kind of videos adressing the issues I have raised here.

A matter of fact, if this tool is at all intended to handle complex mechanical designs, and not just beeing a doodling tool for concepts, ideas and organic designs, you should provide a best practice manual for structuring these designs. I mean, the tool is very capable, but the flexibility of it requires some structure imposed by the user. 

Just to draw a parallell to my "real" occupation, I use a tool for creating electronic circuitry on integrated circuits called FPGA's. The company Xilinx makes a tool called Vivado, which does everything from end to end. It has nothing of the colaboration stuff that F360 has, but let my get to the interesting part. Build into that tool, right there in the user interface, is the work flow that Xilinx thinks is the best for their product. This flow is documented in a few hundred pages of manuals, and is updated regurarily. But interestingly enough, the tool lets you choose your own flow as well. All the tools available is there so you can put everything together the way you want them. They even describe two or three alternative work flows in their manuals. Even with all this documentation, and the best practice manuals, you need to invent quite a lot for yourself.

Ok, designing electronics and doing mechanical design is not the same, but my point is that these issues are complicated, and it doesn't hurt if someone both skilled with the tool, and experienced in running larger projects could sit down and produce a manual for best practice. How do you start out so you won't end up in corner with no exit.

Kjell

[Anonymous]

Hi Kjell, that's an interesting anecdote of well documented workflows. 

I know all this takes time, but if you ever happen to find the chance to lay out an example or two of getting "driven into a corner" 😉 I would find that really interesting to look at and attempt to further develop my workflow to resolve such issues. 

All the best mechanical designing and electron pushing,

Jesse

[kjellhar]

@Anonymous 

it's not that I don't want to share that particular design with you, I just don't have the time right now to prepare it and explain the issues. Since the issue came up, I have done a bunch of stuff, which makes it a bit tedious for me to get back to the "bad" version. I knwo there are older versions in A360, but So far I'm up to around v240, and I don't remember which one was the best place to find the offender. And since I can't see how old a version is (it always says "NaN ago" in the list). This is a hobby project of mine, which means I have to spend the nights after the kids have gone to bed. It's also the only big thing I have made with F360, which means it's the only example. Also, the big thing here is not necessarily how to fix it. I have already identified a couple of more options for getting it right, it's just that I have to fight the tool at every turn to get there.

The mechanical stuff I do at work is much simpler, more in the magnitude of 4 parts that's bolted together. 

But thanks for you offer.

Kjell

[Anonymous]

Hi Kjell, I understand.  That does sound like quite a massive design you're working on there.  Just out of curiosity do you plan on also fabricating the individual parts? 

Jesse

[Anonymous]

Hi, glad to hear you liked my hodge podge videos thus far. 🙂  I did recently come across others that seem quite skilled in Fusion360 making complete work through videos of models with lots of components and stuff, such as this guy and his videos:

https://www.youtube.com/watch?v=sJaaR7x7yIQ

I didn't take the time yet to really watch through everything in a video like this, but can learn a lot from them, like as is the case with Taylor's video, and it is nice with various Chrome or Firefox add ons that can control video speed to whatever is desired, while still maintaining good intelligible audio. 

All the best,

Jesse

[Anonymous]

It will also be interesting to hear sometime a little more of what you've learned to "get it right". 

Jesse

[kjellhar]

Well it's not a secret or anything.

A couple of years back, I designed and made myself a quite sturdy CNC router using aluminum extrusions, and aluminum parts that I made in a manual mill. This machine was a lot of fun, but now I want more (and I needed a real project to start learning Fusion360). So the second incarnation of my milling machine will be an even more sturdy vertical mill build as a hybrid aluminum/epoxy granite construction. I am absolutely planning on fabricating this thing. After all, this time I have access to a CNC, so I can make much better parts. Also, I have more experience. The main issue to be solved is'n really to make it more stable, even if that is also a goal quite high on the list. The main thing is to get control over dust, swarf and fumes so I wont turn my workshop into a total mess every time I use it. Also, I need better protection for the linear rails and the ball screws.

So yes, I do indeed plan to fabricate it. I'm guessing we'll see some practial work right after summer. For this kind of project I really need to review my drawings for quite some time to make sure everything is thought of. I still haven't put in all the little details, like cable management, end stop switches attachments for bellows and so on. And of course, I need to fix a few issues. But I'll get there (I usually do).

Anyway, here is a couple of renders so far. A quick and inaccurate count showed around 60 components, a few of them involes square patterns or mirror goups, so the number of bodies are larger. The timeline is crazy long.

[kjellhar]

@Anonymous 

well, I haven't actually learned all that much yet as I have been busy with easter and the kids, as well as this discussion. Since this project is not only fun, but a way to get the grips of most aspects of F360, and I have obviously found some issues that I need to solve, I plan to start putting together some sort of flow as I go through this design, and document it somehow. I'm starting to get quite a few ideas that I need to try out before I start to write anything.

After all, making 3D objects with holes that are bolted together isn't really that difficult. The hard part is to create a design the does what you intend it to do, looks like you want it to, and that you can maintain afterwards and do changes to. Using the drawing tool is just plain handywork, making good design is vision and discipline.

Kjell

[Anonymous]

That looks like quite a nice design, and quite nice mastery of Fusion360!  Also great to hear you built one CNC router, that you can now employ to work on this one that will be even better.

Really appreciate you sharing the model, very inspiring. 🙂

Jesse

[Anonymous]

No rush.  But I have a hunch anything else you suggest will be quite useful, to further empower current and future Fusionauts. 😉

Jesse

[kjellhar]

Well thanks for those kind words Jesse.

Now it's time for bed.

Good night,

Kjell

[CGPM]

Taylor, yes, that video style is exactly what I would like more of, but like Kjell said do some on a more complex assembly.  I for one do not want a polished, no mistakes, video.  Also please describe any keyboard commands you are using when working.  I would vote for around 30 minute long videos.

One question I have is how much will the next update affect work flow?   I for one have been waiting over a year for Xrefs.

Jesse - Thanks for the Youtube link, I now have that author bookmarked.

Kjell - Impressive project!  Thanks for the renderings.