Surface didn't join, but why?

ttbbfnfd

hi.
I have a wheel housing.
i have trimmed an edge on both sides i would like to build it up with the curve network tool.
one side is watertight, the other is not. i have processed both sides in the same way, but it only works on one side.

i can’t find the reason.

please help and explain whats wrong.

Wheelhouse.3dm (906.5 KB)

dear @nik.mvk

no idea - i can repeat your issue - creating the surface with networkSrf and join fails.

using
_sweep2
gives a result that joins

_loft - not using the small curve also does the job and joins

does this help ?
kind regards -tom

1 Like

Loft works as well.—Mark
Rebuilding the edges of what you had also works

1 Like

your tolerances are .001mm that is super super tight.

drop a zero and your model works fine.

to be clear- you can work at that tolerance, but you’ll have to work much tighter with match, and rebuilds at .001mm.

@Tom_P @markintheozarks @theoutside
Thanks for the input.
thanks for the other solution options.
i was wondering why it worked on one side and not on the other. i think the problem is really that the tolerances for the model size were set too tight. thanks for the hint.

i do not agree with this. sorry @theoutside

standard for “small object millimeters” and most other cad s default.
if you re in doubt about tolerances i strongly recommend to stay with 0.001

it s a pain to start a project with 0.01 and find out that export to Cam fails…

_networkSrf does not respect the tolerance setting in this case.
setting edge Curve Toleance to 0.001 will result in a surface with
_crvDeviation of 0.0046…

it is more a bug then a wrong tolerance setting.

further digging:
setting tolerance of _netWorkSrf to 0.0003 will give a surface that joins.
Dupedge will show that one curve is an Arc (rational, Degree 2)
rebuilding only that Arc within tolerance and use the curve instead of the edge will give the expected result.

seams to be, that networkSrf has problems in handling rational curves / edges. (controlpoints have weight)

it s not a tolerance issue.

kind regards -tom

4 Likes

I too like that tighter tolerance. —Mark

What it is, someone can explain

Not disagreeing and have a ton of respect for you and your work-

but permit me to clarify a few things i kinda breezed over with my statement
.
.001 Units is the default for the small objects templates.

Consider this. 001 inches is a lot bigger than .001 mm…

therefore two identical models of identical size, one in inches at .001 units and one at .001 mm will have wildly different actual tolerances.

the most important thing to understand about tolerances is where is the model going next.

if the model never leaves Rhino and goes straight to a 3d printer or cnc, then just about any tolerance that works for your project is fine, feel free to work looser (like .01" or .01mm) because your stuff will join easier and you will have to work less hard. (generally speaking)

If however your model is going to another cad system, then the rhino tolerance from the very beginning has to match the tolerance of the destination Cad.

you cannot adjust it later…it has to be right from the very beginning.

For instance if you are working with a solidworks user, they have a very very tight default tolerance. you have to match the exact same numbers to have your rhino model go into solidworks smoothly. There are a few workarounds, and some repair stuff available, but generally speaking if SW works at .0001 (or whatever the exact number is) you have to start your rhino model at that tolerance.

for CAM, I have never had a model fail to import to a CAM system due to tolerances, but am not a CAM operator myself and will defer to those who are for the final word on CAM imports.

this has been my experience, your mileage may vary.

I think 0.001 mm should be rather standard.

When you throw grinding into the mix, it becomes reasonably normal to fall into the sub-micron regime for aerospace and engine parts.

I would certainly expect 0.001 mm to be a very standard accuracy and increasingly even quite large for any CAD/CAM application.

It’s been a while, but I am sure there are more common grinding workheads that will not even just grind, but will even then measure to well below a micron. Maybe Studer?

1 Like

sure but if you are doing aerospace parts you’d 100% run very tight tolerances, and use the required techniques to work that tight. (lots of matching and rebuilding, very clean surfaces and curves etc, etc)

For most consumer products manufactured in Asia, In my experience, .001mm is overkill and causes unnecessary work on the modelers part.

But as I always say, you do what works for you and your projects.

I learned a long time ago I can get away with looser files, make my modeling go way faster and get the exact same result from the factories I was working with. I was at mattel in the digital production lab and we did extensive testing when we wrote the digital best practice for all of mattel global design.

we cut identical models built at different tolerances and tested data transfer to Proe, data transfer to Cam and then machined molds. The results were identical for our needs. If I hadn’t been part of that testing team, I’d be much more wary of loosening up.

your mileage however, may vary.

2 Likes