Why the size limit?

tony_459

I tried to model a particularly large thing and couldn't because the dimensions were above a threshold. There must be a reason for the threshold, but just wondering why? I can model a scaled version of the thing, but if I export into third-party software, I'll have to get creative to scale it back up to its true dimensions. Would be nice to avoid this unnecessary step

steve_shubin

@tony_soares459

Out of sheer curiosity, how big of an item were you trying to make?

What was the biggest dimension?

romeograham

The limit is a Parasolid kernel limitation: 500 m in any direction from the origin, so 1,000 m is the longest dimension.

Earlier discussion here: https://forum.onshape.com/discussion/4316/maximum-dimensions

Apparently, this size limitation affects all the CAD systems that use parasolid.


And then, the height cannot be above 500 m:

steve_shubin

@romeograham

Interesting. Thanks for the info 

Jake_Rosenfeld

Parasolid is actually unitless, so theoretically we could allow the user to define their "base unit" as anything (km, nm, mile, in...), and then operate from 1e-8 to 1e3 of that unit.  Would take some concentrated development effort, but we have seen this request for very small electronic components.

Feel free to add/vote for related improvement requests.

romeograham

@Jake_Rosenfeld
Thanks for that explanation - so the current limitation (1,000 m) is because it is 1e3 times the arbitrary (but very, very rational) selection of "m" as the base unit ?

I wonder if the "chip foundry", "architectural" or "city planner" versions of Onshape in the future could be built on different base units?

tony_459

Ah! I get it. So it's one of those practical if unfortunate constraints that dictate the size limit. It would be so nice if we could set the base units on our side (even if not immediately)! Either way, it's at least good to know where the limit is coming from

MBartlett21

romeograham said:

@Jake_Rosenfeld
Thanks for that explanation - so the current limitation (1,000 m) is because it is 1e3 times the arbitrary (but very, very rational) selection of "m" as the base unit ?

I wonder if the "chip foundry", "architectural" or "city planner" versions of Onshape in the future could be built on different base units?

What about an "Apollo mission" version

S1mon

Part of the challenge with any NURBS based CAD system is that with the exception of simple geometric shapes, there aren't convenient closed-form parametric mathematical solutions to the intersection of arbitrary surfaces/solids. So all systems, whether they're based on Parasolid, ACIS, Granite, etc. use some sort of error band around the piecewise approximated solution for these intersections to determine when the solver is done. In Creo (Pro/E) this was the "accuracy" setting. It was related to the ratio between the largest dimension and the smallest feature. If the number was too small, the solver would take forever to finish. If the number was too large, the intersections would have strange errors which don't make sense. All systems need to manage this error budget, whether the user is involved or not. I strongly suspect that the 1000 unit size limit in Parasolid is related to this sort of "accuracy" setting.