trouble making fully parametric designs

[prettygoodnotbad]

I've been using Fusion 360 for maybe two years now and it's definitely become my program of choice. That said one problem I keep running into is the ability to make my models "truly" parametric. It seems completely impossible for me to create a design that doesn't break when I decide to make a modification lower down the timeline. The main issue seems to stem from projections in sketches. More often than not, if I modify a body which has projected geometry in a sketch, the projections in said sketch don't follow suit. This usually creates a series of errors the whole way up the timeline which I need to go and manually fix one by one. This is hugely time consuming and I find myself fixing errors on the timeline more than working on my actual designs.

Is fixing these kind of errors just a reality of the Fusion 360 workflow, or are there best practices when it comes to sketch projections and just designing parts in general that would avoid this behaviour.

[TheCADWhisperer]

---

@prettygoodnotbad wrote:

Is fixing these kind of errors just a reality of the Fusion 360 workflow...

---

No. 

If you used the BORN Technique your designs would be robust and predicable.

File>Export and then Attach an example *.f3d file here.

Based on my experience - I expect to see evidence of a lot of poor modeling practices that will have to be "un-learned".

[prettygoodnotbad]

ok so most of my designs are work related and I'm not really at a liberty to share them so I whipped up this little chair drawing attached. I did it quick but I think it's still fairly representative of my general approach of how I would design something like that if I was taking more time with it. As an example in this design I created a work plane called middle section which I used to mirror some of the parts, so if I increase the width from say 10" to 12" I would expect the chair to get wider and adapt to this new dimension, as you can see some of the sketches break. or if you change the height of the back leg on the very first sketch i did some other things also break further up the timeline. 

[davebYYPCU]

Yep, I found other sections that give unpredictable results too.

It's not as you say, projections not working as expected,

they are so reliable - I could not attribute anything to a broken projection.

I am not an expert, but seems to me that you are rushing, or not paying enough attention to detail towards the end of the timeline, and it is this stuff that breaks.  Some techniques are better than others, but you have a chair, so not referring to the how, but can tell you the why it breaks....

1.  Front rail - blue lines - lower right corner is not coincident with the projected line, and the blue lines are unnecessary.

2. Rear rail - no unnecessary vertical lines, but the right end points are not coincident to the projected line.

3.  Backrest - Top front spline is not coincident to the top of the projected line.  That upsets the patch, too.

4.  Backrest - mirror is intermittent, and I don't know why, as the right side has no connection to the right legs.

I did get the seat to misbehave but not repeatable now.

Why was the origin plane not used for the centre mirror plane?

Might help....

[chrisplyler]

---

@davebYYPCU wrote:

4.  Backrest - mirror is intermittent, and I don't know why, as the right side has no connection to the right legs.

---

If you've got a half-width backrest that works with changes in the chair's total width, and you expect to mirror it around a plane to complete the other half of it, you better make sure the plane you use for mirroring ALSO adjusts its position with the chair's total width.

[davebYYPCU]

That would do it.

[TheCADWhisperer]

One year ago in >>this thread<< I suggested making use of obvious symmetry about the Origin.

Much of the man-made and the natural world has at least some degree of symmetry.

Identifying symmetry before even starting a design and then using the BORN Technique as much as possible and practical can result in rock-solid, robust models that are predictable upon edits and updates.

Use symmetry about the Origin planes. (start with the highest degree of symmetric geometry in the design)

Avoid unnecessary dependencies. (unpredictable on edits)

Build logical dependencies. (predictable and intended behavior on edit)

[prettygoodnotbad]

@davebYYPCU  thanks for taking a look, I'm noticing now that this example does not really reflect the problem I sometimes have with projections, but it does highlight a few things that still elude me apparently.

First you say on my front and back rail the points are not coincident, yet if I look at the sketch the lines are all black and the coincident symbols are present on all corners:

 

Then if I widen the chair by moving the symmetry plane the coincident symbol disappears and the line doesn't follow the projection (and the lines stay black, they don't go blue as if it wasn't fully constrained).

Strangely the back rest seems to work fine for me it stretches with the chair. which also raises questions as to why things work sometimes and not others. On your point number 3 I see now that the spline was coincident to the edge and not the corner which was causing problems. 

@TheCADWhisperer I am not familiar with the BORN technique, I will do research, if you know of any can you point me towards some good resources. Also I'm not sure about your point in my other post you linked to, I was not asking about symmetry at the time. Although I guess it sounds like one of the concepts is to always use the origin of origin as symmetry?

I used the point of origin to draw the side of the back leg, if I wanted to use the origin as symmetry it seems like I would need to start with the seat (since i could sketch a center rectangle on origin) which doesn't seem natural to me, I always start with a profile view to get the basic form, and extrapolate the rest from there.

I will try and reproduce projection errors and post it here if I can.

[chrisplyler]

You can start symmetry on the origin...doesn't mean you have to start modeling the center of the object first. There are many ways to set this up. In my example picture, I've sketched a base square of construction lines establishing the width and length of the chair. I've created planes on angles as desired. I've drawn a profile section sketch on an origin plane. I've extruded to-object, using those planes on angles as the destination. I could now mirror the side frame over across the origin to the other side. Any changes in the width of the base box would result in symmetrical changes. There are many different ways, and they don't matter, but the important desired result is that adjustment to the chair would always symmetrically happen around the origin.

[davebYYPCU]

@chrisplyler I can't remember you ever publishing a blue line sketch before.

@prettygoodnotbad 

Same Chair, with symmetry about the origin.  Show Sketches only. 

I took the liberty to bevel the seat to conform with other component angles.

Your Component 7 sketch, might make my comment a little clearer.

But if you did not demo quite the projection problems, we are here to help with that....

[chrisplyler]

Yeah that sketch really has me baffled. It is completely constrained and dimensioned. No elements or endpoints can be dragged. Yet NONE of it turned black.