In some cases fully constrained sketch elements could jump in "equivalent" places when I change parameters.
Change parameter
Is there a way to prevent this behaviour without overconstraining or overcomplication?
Maybe somehow limit movement, order object to stay "closer" to other object or stay at specific side, or use grid coordinates, or even setup non-strict dimmensions like "greated/less than"?
Try adding a vertical construction line, under the top tangent point down to the ellipse. Fusion would have to eliminate it if it wants to flip sides.
Do you have any duplicate articles, those ellipse axis lines should not be orange.
Can check the file later.
Might help…..
Hi,
Presumably you want the centre of the ellipse to stay where it is and the tangents move when you change
the ellipse parameters. Try dimensioning the location of the centre of the ellipse to a defined point. Also try
putting a horizontal and vertical constraint on the construction lines between the midpoints.
Will probably behave then.
Cheers
Andrew
Checked the file, have no answers. Vertical line did not help, was still totally unreliable with my testing,
@Drewpan Nope, vertical centre point position has to be flexible.
Both tangent lines can / do flip to legal positions, another one for the bucket.
no definitive answers for consistently and predictably controlling things like this. sketch solver still needs tweaks.
but, sometimes adding geometry for the solver to chew on does alleviate the problem on a case by case bases. To that end @davebYYPCU suggestion of adding a line was a good one. that specific one just didn't work. so try a different one. adding a circle with the center constrained to the endpoint of the line and tangent to the vertical axis line seems to work.
Thanks for testing, should I send some bug-report to let devs fix this one?
If several small increments are used, can the parameter be changed? This could be another instance where a large value change causes flipping, but creeping up on it works. We fix them when we find them. I'm sorry I didn't have time to try this today myself.
When it fails reliably, how many small increments is the fix? (It breaks with 1 mm increments)
I tried Alex's circles, but same result, still unreliable.
(From 20 to 21)
The tangent lines have 2 legal outcomes - each, and for the required parameter change the sketch has been fully constrained ahead of the change/s.
Might help....
Adding a line makes it stable.
Didn’t work for me before my message 4.
The line cannot be constrained so is perfectly legal to flip too.
Requirement is for the tangent point to slide with parameter changing.
When the tangent line wants to flip it will take your new line with it, unless you have some magic I didn’t see in the pic.
Written before Alex’s message.
I'm not arguing, there is no argument. I'm showing you what I find in testing. It's completely stable for me. I'm using the current insider build.
Don't worry, I logged a ticket. Just because there is a valid workaround (for me) doesn't reduce the need to look at the flipping issue.
my bad. I didn't go straight to 18.5
so on my end adding a line doesn't make it stable. what are you doing different?
No idea. I opened the file, replicated the flipping. Then drew the line, and found it was stable.
Testing again, it's not that stable. But I could make several changes without it flipping. It's definitely something that needs improving, but also your-mileage-may-vary.
Can't find what you're looking for? Ask the community or share your knowledge.