sweep jumps from arc to circle

[kgatzke]

I created this family for a cove light fixture on curving walls.  I thought I got all the bugs out of it but one remains that I can't squash.  For some radii the sweep inverts so that it goes from an arc segment to a huge circle of the same radius.  How do I keep the sweep on the reference line?  I tried locking the endpoints to the planes but I get a warning about over constraining the sweep.  

I had tried doing this earlier with setting the radius and arc length but that required a center outside of the ceiling host so I came up with this method.  **** it if the sweep doesn't invert at EVERY radius I need.

[ToanDN]

Can you give me an example of a condition when it happens?  It seems to be working properly here.

[barthbradley]

It can't be this obvious...

...there is a break point at which the inversion occurs; the point at which it fails. Find it precisely and note it as your limit (e.g. the limit is "X"). You need to use it in a "control" parameter for "Depth" (e.g. "Depth Ctrl"). If the "Depth" entered is less than "X" ("X" = break point), then use "X", otherwise use the value assigned. This "Depth Ctrl" parameter will be the one that drives "Depth".  

No? 

[ToanDN]

That is precisely the strategy.  The problem is that I can't not find the very break point.  Did you happen to find it?

[barthbradley]

like I said...it seems way too easy. I'm thinking I'm misunderstanding the core problem altogether. I just entered a value that would break it and it broke.  I cancelled and did it again and again with incrementally different values until I found the break point. Pretty basic: trial and error. 

...also, I read the issue differently than you Toan. I was thinking the OP was encountering the problem with TOO BIG a radius. That's why it was inverting on him/her. 

...somewhere around 1 1/2"

[ToanDN]

Yeah I tried to make it nearly straight and the **** thing kept on flexing.  Guess I will wait for @kgatzke to disclose the break point then.

[kgatzke]

I can't figure out what the break point is.  If I place it on a large enough ceiling and drag the depth control I can make it flip back and forth from arcs to circles over a wide range of radii.  It's like there are some radius to arc length ratios it likes and others it does not.

I recreated the fixture with radius and arc length controls and while it still pops into a circle occasionally I can fine tune it to find arc lengths that work for the given radius.  I'm even able to get it to work at an arc length of 8'-0", the true length of the fixture.  I couldn't even get close to that with the rectangular controls.  I'm keeping it on the long arc though because it's only for model and tagging.  The electrical consultants have their own miscellaneous families for circuiting custom fixtures designed by architects.

Annoyingly the reference lines used for controlling the angle, while not visible, are selectable and given that large radius will get picked up constantly by other people working on the project.  If only we could control arc length parametrically. 

[kgatzke]

Frankly the problem with the selection of those long reference lines results in the fixture becoming totally visible in an uncut manner in multiple sections and elevations where I don't want it.  Either I have to create a ton of new filters for all those views or just scrap the fixture altogether and just create a dumb in-place model.

[ToanDN]

I can replicate the issue when placed it in a project but I haven't gotten time to look further to the first family.  For the second family you posted right above, make these reference lines and plane "Not a Reference" and you won't see nor select them accidentally in project.

 

[barthbradley]

I am having great deal of difficulty understanding what you’re trying to achieve.

I understand that this is a cove light family, and that you are placing it on a flat ceiling (e.g. the family is ceiling-based).

The flat ceiling that you are placing it on is bounded by curved walls, so I’m guessing that the sketch boundary of the ceiling is curved as well.

The family has a chord length parameter (“Lateral Length”) and an arc height parameter (“Depth”).  Those 2 parameters are not enough to find the arc segment your looking for. You need to employ some trigonometry to flex this thing. You need parameters to find angle, radius, depth, and segment.

In lieu of re-building this family so it will flex properly, you might find it easier to make three-point adaptive family instead, or a railing family (?), or – just custom-build it in the project using in-place component.

[ToanDN]

@barthbradley

Have you actually opened his families, especially the first one posted yesterday?  It was built quite neatly and it flexes very properly within the family environment.  

What you show above is what he has in the second family.  It works fine too, just needed a little housekeeping in the family to take care of what he complained about.

[barthbradley]

@ToanDN...no, I didn't see the second family. I just looked at the very first one. My post was in response to @kgatzke's last 2 posts, 7 of 11 and 8 of 11, in which he states he is still having issues and specifically inquired about arc formulas. If my post is not applicable, then I guess I wasted my time. Maybe it can help someone else. 

[ToanDN]

It will be useful, if not to OP, then to others.

By the way,

“Time you enjoy wasting, was not wasted.” - John Lennon

[ToanDN]

@kgatzke

Here a revised version of your first family.  I replaced the sweep with a sweep blend and it seems to stop inverting.

[kgatzke]

I appreciate the effort and you're right, proper control requires full command of the arc through trigonometry.  I avoided including the heavy calculations because that kind of precision isn't necessary.  But this wouldn't have solved my problem.

The problem is Revit's programmers left something out that tells it what side of the circle to draw the path of the sweep.  I know what side it's supposed to be on, but mathematically either side of the circle satisfies the requirements of the geometric constraints.  

They could have used the insertion point as an absolute side for the handedness of the sweep, but this would have probably made arcs on the other side if desired by other users impossible.  A positive or negative value could have established the preferred side relative to the insertion point.  Instead, something less determinable is affecting the decision of what side is used to establish the sweep.

[kgatzke]

I'll give this a try when I have the time.  I've moved on to some other parts of the project in preparation for meetings tomorrow.

Thanks.

[Viveka_CD]

Hi @kgatzke

I just wanted to follow up here, any progress on this issue?

Please mark any posts that help with "Accept as Solution" and thanks!

[adrien_gesulfoXNQD7]

It is still an issue. Giving very unpredictable results once in a project.