machining from an STL

[Chadbourne]

I am having difficulty with generating a toolpath form an STL, I think I'm doing it right but I get to the point where it says approximating normal surfaces and my machine just hangs there. What am I doing wrong

[deckerc]

I've done some STL machining and they are usually very large files. I've had them with over 100,000 triangles. They take a long time to calculate toolpaths. Basically I'm sayin you may have to wait longer depending on your surfacing tolerance. Try making your tolerance big just to see if you can get a toolpath, then tighten it up later.

[Chadbourne]

thanks for the advice, I'm having some success, could some give me some advice on what I should set the tolerances to as a way to get me going. My STL is in 100k range (no of entities).

[Pete Rimkus from CNC Software Inc.]

Since an STL file is (usually) a polygonal representation of a curved part, using tight tolerances like you might normally use can backfire - you'll end up waiting for MCam to generate a toolpath that looks like the polygonal representation, which probably isn't what you want - you probably want something resembling the original curved part.

Try loosening up the machining tolerance and see what happens.

[deckerc]

I run my tolerance around .0002 with a 1:1 filter range on my big STL files and get good looking parts, but they take a long time to calculate.

 

good luck, I hate working with STL's, but have no choice at the moment. I keep hoping for a good translator to come along. The STL to surf chook for v9 never worked because my files are too large, and we don't have other any of the other softwares capable of converting these.

[mcpgmr]

What is the advantage if any to using STL files for machining ? I steer clear of STL files just because of the problems mentioned above. Wouldn't it be more practical to request a parasolid, step or even iges surface model to use for machining ? Seems like it's a backa$$ward way of doing things. JMHO

[deckerc]

when that is the only file being given to you and you don't have software to convert the STL to anything else, you don't have much of a choice. I can't explain our application, but for now we are stuck with STL's on this particular reoccuring job.

[mcpgmr]

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

you don't have much of a choice

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

 

You work at Synthes where they use PRO-E right ?

PRO-E does iges. Is the stl file coming from another program that uses data from Cat scans or MRI's ? Do you have the freedom to convert the models your self in MC using the stl2surf c-hook or does engineering have too tight of control over the cad models ? Don't mind the questions. I'm just trying to get an idea how things work over there. You might even be the guy who got hired instead of me or maybe one of the guys who interviewed me for a model maker position not too long ago.

[deckerc]

mcpgmr

 

I do work in the synthes prototype shop, but I don't get to do interviews. been here about 8 years.

Can't really get into specifics of the STL's but they are not coming from Pro-E. It takes the engineers 1 to 2 days to convert them in pro-E. I personally don't know Pro-e. the computers come up with "not enough memory" when trying the stl2surf c-hook in mcam and they are not even half way converted. There is just to many triangles in these STL's so we've been toolpathing directly off of the file.

[mcpgmr]

Corby, Sounds like they are files generated from medical imagery divices like a Cat Scan. You could play around with the memory alocations in MC to try and get them converted completely. HTH !

[deckerc]

mcpgmr,

we've tried alot of things including the allocations, haven't found the magic numbers yet. think we eventually buy some software just to convert the STL.

 

thanks for the suggestion.

[Chadbourne]

I'm interested to see the talk of useing CT data, that is exactly what I am trying to do. I've had success creating splines (nurbs) from the slices, then lofting them into surfaces to be machined, but that is more work than I want to do on an ongoing basis. The STL aproach appeared promising to me becouse I can cut out all that manual manipulation to get an image, but machining it seems to be another issue. Has anyone done this, or somthing like it before?