Broken solid models

[jpatry]

So the other day, I got volunteered to pick up another programmer's job.

And after looking over his machining strategy and determining it was a massive pile of eldritch nonsense, I quickly made a copy of the file and set about doing things in a sensible manner.

It was when I was adding some air chains for a 2D Dynamic path that I noticed my chains weren't chaining no matter what I did, once I zoomed in far enough I realized I was looking into the forbidden innards of a solid model through a crack in reality.

I tried the tools that are supposed to fix this, in Model Prep, but to no real discernible effect.

I ended up fixing the misaligned faces with the Move tool, which is a horrible and tedious process, and aside from having to use subpar tooling due to the usual suspects, the part actually ran quite well.

 

So what other ways are there to quickly identify and fix borked solid models that I can add to my normal workflow?

[crazy^millman]

On 9/10/2023 at 1:35 AM, jpatry said:

So the other day, I got volunteered to pick up another programmer's job.

And after looking over his machining strategy and determining it was a massive pile of eldritch nonsense, I quickly made a copy of the file and set about doing things in a sensible manner.

It was when I was adding some air chains for a 2D Dynamic path that I noticed my chains weren't chaining no matter what I did, once I zoomed in far enough I realized I was looking into the forbidden innards of a solid model through a crack in reality.

I tried the tools that are supposed to fix this, in Model Prep, but to no real discernible effect.

I ended up fixing the misaligned faces with the Move tool, which is a horrible and tedious process, and aside from having to use subpar tooling due to the usual suspects, the part actually ran quite well.

 

So what other ways are there to quickly identify and fix borked solid models that I can add to my normal workflow?

No if a CAD software has no be configured to save files a certain way hard to expect a CAM software to make it better. SpaceClaim does an great job of healing bad solids and I sued ti for years to do that I just don't have it anymore so I am stuck like you are fighting the models if that issue comes up.

[SuperHoneyBadger]

My strategy is to rely on pulling wireframes from the broken areas, and placing them on a new level. They're easier to manipulate than open sheet bodies, since you can't boolean-add to them with new solid patches. Then you don't need move or push-pull.

Otherwise, if you want to go hard to the mats and fix it up: convert all of the holey solid to surfaces, patch it with modified/new surfs, and convert it back to a watertight solid. Lots of my experience with surfs has come from fixing things this way.

Most (98%) of my work is prismatic, so YMMV if you are dealing with 3d curvatures.

[jpatry]

10 hours ago, SuperHoneyBadger said:

convert all of the holey solid to surfaces, patch it with modified/new surfs, and convert it back to a watertight solid

This makes sense.

The only further question I have is what is the best way of interrogating a solid, for breaks, right on import?

Because most of the time I find these issues a good ways into setting up toolpaths.

And the Simplify, Optimize, and Repair tools in Model Prep didn't seem to do me any good.

[Fred @ Slate Industries]

No magic button exists. 

It's nice when it's a Native file coming from down the hall...

You'll get good at " cleaning up files , so you'll get what you want"

 

I HAVE SPOKEN

[jpatry]

17 minutes ago, Fred @ Slate Industries said:

It's nice when it's a Native file coming from down the hall...

Yes, the worst part even with that, though, is I work 2nd shift and those people don't.

So my boss simply asks if I can fix it this time the same way we did last time.

[Aaron Eberhard]

You can try to run the "check solid" function ( Home> Check Solid)

The problem has to do with how the solids are stored in their files.  Often these solids were saved with too loose a tolerance from the CAD system, and when they have organic shapes (like multiple fillets intersecting), they round the numbers due to the tolerance.  That gives you a theoretically fine (it fits within the tolerance so no errors reported) but obviously broken situation like you're dealing with.    The worst I have ever seen came from Catia.. You can customize the output so much from that program that you can get some really, really funky stuff.   They've made the defaults better in the past decade, at least!

--------------------

Another trick to fix it is to try to model prep or boolean features together to wipe out the problem area, then remove the changes with model prep.  

I don't have any good broken models handy that I'm allowed to show, and it's almost impossible to create a situation like this in the native CAD/CAM system due to the aforementioned tolerancing issues that aren't a problem in the native format.

If you can share a model, I'd be happy to see if that method is applicable.   It's basically making a big cylinder/sphere around your problem area, boolean add them together, remove history (so it's all a new solid), then, use Model Prep to push/pull or Modify that cylinder away.

[SuperHoneyBadger]

10 hours ago, jpatry said:

the best way of interrogating a solid, for breaks, right on import?

Big fish first:

After it's imported, select the solid and hit F4 for Analyze. You're looking for that second box to say Closed Solid Body. I have not had an issue with a model that is identified as closed.

If it says Open Sheet Body, you'll know you have to check the usual areas around curves, fillets and holes. As stated above, the automatic tools don't always catch the small slivers.

[cruzila]

6 hours ago, SuperHoneyBadger said:

Big fish first:

After it's imported, select the solid and hit F4 for Analyze. You're looking for that second box to say Closed Solid Body. I have not had an issue with a model that is identified as closed.

If it says Open Sheet Body, you'll know you have to check the usual areas around curves, fillets and holes. As stated above, the automatic tools don't always catch the small slivers.

I have had issues with where lettering is cut in to the solid and the solid seems good manifold and all. Check solid finds the edges and selecting loops or faces to run a pocket on won't work because of gaps in the edges of maybe .002 or less. I usually will just make geometry and manipulate it to work but that cylinder idea from Aaron i just tried and it is very good solution!!!

 

I just went through this exercise this morning with a part

[gcode]

7 hours ago, Aaron Eberhard said:

The problem has to do with how the solids are stored in their files. 

I used to work with a guy who had his default chaining tolerance set to .008"

I typically run at .0002"

I always had to remember to bump my chaining tolerance to .008 when I opened one of his files,

Otherwise, they'd blow up like Joe Biden's brain trying to speak a coherent sentence.

[jpatry]

On 9/12/2023 at 9:21 AM, Aaron Eberhard said:

You can try to run the "check solid" function ( Home> Check Solid)

The problem has to do with how the solids are stored in their files.  Often these solids were saved with too loose a tolerance from the CAD system, and when they have organic shapes (like multiple fillets intersecting), they round the numbers due to the tolerance.  That gives you a theoretically fine (it fits within the tolerance so no errors reported) but obviously broken situation like you're dealing with.    The worst I have ever seen came from Catia.. You can customize the output so much from that program that you can get some really, really funky stuff.   They've made the defaults better in the past decade, at least!

--------------------

Another trick to fix it is to try to model prep or boolean features together to wipe out the problem area, then remove the changes with model prep.  

I don't have any good broken models handy that I'm allowed to show, and it's almost impossible to create a situation like this in the native CAD/CAM system due to the aforementioned tolerancing issues that aren't a problem in the native format.

If you can share a model, I'd be happy to see if that method is applicable.   It's basically making a big cylinder/sphere around your problem area, boolean add them together, remove history (so it's all a new solid), then, use Model Prep to push/pull or Modify that cylinder away.

Yeah, I don't really have permission to share the model, but I know what you mean by the problem being from how the file was originally created.

The worst I've seen so far was a solid poorly exported from a mesh model, I think the customer designed it in Rhino, not a single surface on that model was flat, parallel, or perpendicular to any other, misaligned faces everywhere, and it was a simple stupid little part too, just a flat plate with some corner rads, chamfers, and the company logo and name.

The model looked like a piece of plastic that had been left on the dash of a car with the windows up on a hot day.

 

[SuperHoneyBadger]

7 hours ago, jpatry said:

not a single surface on that model was flat, parallel, or perpendicular to any other, misaligned faces everywhere

I'm impressed at how bad some designers are. It must take some sack to submit something like that as your work, and I can't even imagine how they make those models in the first place. Just pick a random point in space and start clicking?

[crazy^millman]

7 hours ago, jpatry said:

Yeah, I don't really have permission to share the model, but I know what you mean by the problem being from how the file was originally created.

The worst I've seen so far was a solid poorly exported from a mesh model, I think the customer designed it in Rhino, not a single surface on that model was flat, parallel, or perpendicular to any other, misaligned faces everywhere, and it was a simple stupid little part too, just a flat plate with some corner rads, chamfers, and the company logo and name.

The model looked like a piece of plastic that had been left on the dash of a car with the windows up on a hot day.

 

STL garbage to make a solid model is always junk. Like scanned point clouds ad making nice clean solid is almost an art form.

[jpatry]

12 hours ago, SuperHoneyBadger said:

I'm impressed at how bad some designers are. It must take some sack to submit something like that as your work, and I can't even imagine how they make those models in the first place. Just pick a random point in space and start clicking?

Well, in this particular case, I believe the model was perfectly fine until they exported it to a step file, as going from mesh to solid can lead to all kinds of errors where it tries to approximate all the polygons into faces, arcs, etc.

The designers love mesh modeling, because it gives you the ability to shape the model in any arbitrary direction, this is how 3D sculpting software like Maya works, for example.

[crazy^millman]

10 hours ago, jpatry said:

Well, in this particular case, I believe the model was perfectly fine until they exported it to a step file, as going from mesh to solid can lead to all kinds of errors where it tries to approximate all the polygons into faces, arcs, etc.

The designers love mesh modeling, because it gives you the ability to shape the model in any arbitrary direction, this is how 3D sculpting software like Maya works, for example.

The real issue is some CAD tessellates the solids into Mesh for make them lightweight for using he CAD software for the Graphics Card. Then the engineer exports that tessellation without making sure they are exporting a real solid. Seen it many times over the years.

[Newbeeee™]

On 9/12/2023 at 10:35 PM, gcode said:

I used to work with a guy who had his default chaining tolerance set to .008"

I typically run at .0002"

I always had to remember to bump my chaining tolerance to .008 when I opened one of his files,

Otherwise, they'd blow up like Joe Biden's brain trying to speak a coherent sentence.

Surely....those files couldn't have been THAT bad....

[Newbeeee™]

On 9/15/2023 at 2:28 PM, SuperHoneyBadger said:

I'm impressed at how bad some designers are. It must take some sack to submit something like that as your work, and I can't even imagine how they make those models in the first place. Just pick a random point in space and start clicking?

In the halcyon days, there was training in the drawing office and a "draftsman" was a respected trade....Nowadays everyone can switch on a computer and be a "designer"

[crazy^millman]

2 hours ago, Newbeeee™ said:

In the halcyon days, there was training in the drawing office and a "draftsman" was a respected trade....Nowadays everyone can switch on a computer and be a "designer"

i think like anything where technology has improved our abilities to do things it has opened many doors for people that otherwise may have not been opened. Issue I have is the companies being irresponsible with it and only caring about making a buck making wild claims and certifying people who have no clue about other aspects that go into things related to what they are designing.

So Simple a 6 year old can program parts sound familiar?

2 weeks of training in Fusion and you will be a full 5 axis programmer?

You can run a full enterprise organization for only $600 yearly fusion subscription dong CAD/CAM/CAE/CAV? 1000 different disclaimers to be found on 20 different places. Just don't do ITAR, DOD or DOE work with it and you are not going to jail is all.

[Newbeeee™]

1 hour ago, crazy^millman said:

i think like anything where technology has improved our abilities to do things it has opened many doors for people that otherwise may have not been opened. Issue I have is the companies being irresponsible with it and only caring about making a buck making wild claims and certifying people who have no clue about other aspects that go into things related to what they are designing.

So Simple a 6 year old can program parts sound familiar?

2 weeks of training in Fusion and you will be a full 5 axis programmer?

You can run a full enterprise organization for only $600 yearly fusion subscription dong CAD/CAM/CAE/CAV? 1000 different disclaimers to be found on 20 different places. Just don't do ITAR, DOD or DOE work with it and you are not going to jail is all.

My peeve....was in the DO days, you'd produce prints and someone else would check your work. There was job stability and through cross-project working, we all knew all about all the company products....

Now the beancounters have got involved....in companies that should know better (looking at Aerospace OEMs), a contractor appears and produces xxxxe with little idea and not working to customer standards....

So you (we did many times from different customers) get a part to make, with ATF material spec, or unusual treatments that the company has never before used, thread depths to the same depth of the drilled hole, let alone general tolerances that have everything tied right down for no reason whatsoever.

Make, produce the FAIR, the part is then approved, and is then expensive forever more because no-one has the balls to change anything. A part that should be cheap is 50+% expensive than it should be...EVERY time it's made.

World Class huh!

 

[jpatry]

11 hours ago, Newbeeee™ said:

In the halcyon days, there was training in the drawing office and a "draftsman" was a respected trade....Nowadays everyone can switch on a computer and be a "designer"

It can be both good or bad, these things are force multipliers.

It's just like going from manual machines to CNC, all it can do is produce you more parts faster, be they good or bad.

[Jobnt]

On 9/16/2023 at 6:27 AM, Newbeeee™ said:

Nowadays everyone can switch on a computer and be a "designer"

Cameras in cell phones have made every soccer mom a sports photog. Can't make a decent living in photography any more since people give away their xxxx for free just for their 15 minutes of fame. 

Rhinos, Razors and other SxS' have made every clueless idiot a desert racer. Can't ride in the desert without worrying about some tool that thinks he's Mike xxxxing Metzger flying over a hill and landing on you.

 

[SuperHoneyBadger]

25 minutes ago, Jobnt said:

Cameras in cell phones have made every soccer mom a sports photog. Can't make a decent living in photography any more since people give away their xxxx for free just for their 15 minutes of fame. 

Low-skill democratization of everything around us has its pitfalls, even though a lot of very vocal individuals will extoll it's limitless potential, and declare it as only positive. I get the argument for the one prodigy that gets access in spite of his/her modest means, but the millions of others that only contribute noise and clog up our collective cloud/physical storage make it a net-bad idea. And not without a real cost, I might add.

[Kyle F]

On 9/16/2023 at 1:22 PM, Newbeeee™ said:

 

So you (we did many times from different customers) get a part to make, with ATF material spec, or unusual treatments that the company has never before used, thread depths to the same depth of the drilled hole, let alone general tolerances that have everything tied right down for no reason whatsoever.

Make, produce the FAIR, the part is then approved, and is then expensive forever more because no-one has the balls to change anything. A part that should be cheap is 50+% expensive than it should be...EVERY time it's made.

World Class huh!

 

PREACH brother!!!! this is the story of my life hahaha, having to bust out a threadmill to hit a stupid depth because they didn't leave me any room at the bottom of the threaded hole for a rollform lead smh. Only difference is at my shop we don't charge extra for the dumb stuff, cause it never gets noticed until I'm actually programming it. Then I'll show my manager and ask "how did you expect me to mill that?" and he always replies "oh wow I didn't see that when I quoted it"

[Jobnt]

23 minutes ago, Kyle F said:

PREACH brother!!!! this is the story of my life hahaha, having to bust out a threadmill to hit a stupid depth because they didn't leave me any room at the bottom of the threaded hole for a rollform lead smh. Only difference is at my shop we don't charge extra for the dumb stuff, cause it never gets noticed until I'm actually programming it. Then I'll show my manager and ask "how did you expect me to mill that?" and he always replies "oh wow I didn't see that when I quoted it"

Worked at a place where we did a lot of microwave housings for aerospace. We would literally program and run a part to get accurate cycle times. And we could give the customer a sample part when we turned in the quote. This approach opened a lot of eyes and won us many contracts. Once we started doing this the number of missed features went down exponentially and our quotes were way more accurate. Unfortunately that business model was sustainable for a new startup but was absolutely not sustainable for a production shop.

[Kyle F]

35 minutes ago, Jobnt said:

Worked at a place where we did a lot of microwave housings for aerospace. We would literally program and run a part to get accurate cycle times. And we could give the customer a sample part when we turned in the quote. This approach opened a lot of eyes and won us many contracts. Once we started doing this the number of missed features went down exponentially and our quotes were way more accurate. Unfortunately that business model was sustainable for a new startup but was absolutely not sustainable for a production shop.

that's a pretty nice flex that you sent them a sample part with the quote. We usually don't even get the 3D model until we've already won our contract which seems crazy off to me as a programmer. I hate quoting based off a print with no model.