6 Replies Latest reply on May 16, 2018 7:49 AM by Jason Warnke

    Feature recognized files not behaving nicely

    Jason Warnke

      Has anyone else had issues with feature recognized files?

       

      I've seen this issue in our company.  A few guys run NX here.  They will translate the .prt files from NX into steps, then open them up in SW as dumb-solids, use feature recognition on each part and go about rebuilding an assembly that way.

       

       

      Seems there are often issues that result from doing this though (lots of repeated crashing and such - especially when you get to the assembly level).

       

      We did notice that it made a difference running 2017 SP1.0 or higher (most of us run SP3.0 or 5.0).  A coworker was just having this issue - he was running 2017 SP0.0 and his comp crashed a lot with the converted assembly.  I did the same action with the same file on my comp (running SP3.0) and no crash.  Everything seemed ok.

       

      I just wondered if it is generally not the best idea to use feature recognition on converted dumb-solids (better to redraw from scratch?)?

       

      Thoughts?

       

      Thanks

       

      -Jay

        • Re: Feature recognized files not behaving nicely
          Paul Salvador

          Jason.. well, STEP is a translation with possible issues,.. may I strongly suggest that your NX users (and you also) save/share native dumb Parasolid (no translation),..  parasolid is also native to Solidworks.

          or, try the new 3D Interconnect (UG-NX 10  *.prt)

            • Re: Feature recognized files not behaving nicely
              Jason Warnke

              Thanks for the response before Paul.

               

              STEP files may have been part of the issue.  I asked one UG/NX user in my company to output an assembly in STEP and in XT/parasolid.  The step was less accurate than the parasolid but there were still issues with the parasolid:  very slight variations in angles (like .00000019 degrees....a ridiculously miniscule amount).

               

              I am not a UG user myself; but I understand that when the guys here draw in UG it is more of a top-down approach than SW modeling.  When they output files from UG they are unable to do so in a way that plays nicely with SW.   All part planes (for example) are based on the assembly model origin of the original UG assembly, so when you open it in SW, if anything is even slightly tilted with respect to origin planes, that means the part file itself is not true relative to origin planes....which stinks as you can imagine.
              From that standpoint - feature recognition is useless; since im forced to deal with all these parts that are just baaaaaaaaaaarely out of parallelism and perpendicularity in several areas. 
              Other than remodeling things from scratch (well...maybe using convert entities to help), is there something else we could try when converting these?  (I realize this is really more a question for a UG/NX and SW user, but with any luck someone like that is out there :-D). 

                • Re: Feature recognized files not behaving nicely
                  Paul Salvador

                  Hey Jason,.. arrghh... I imagine the UG guys maybe tweaking some modeling tolerance for that parasolid deviation?

                   

                  I'm going to chime in Roland Schwarz  and maybe he has some insight with how to deal with the UG translation?

                    • Re: Feature recognized files not behaving nicely
                      Roland Schwarz

                      For the record, I despise and distrust all forms of feature recogntion.

                       

                      Seeing that issue in NX is a new one one me. But, I can take credit for uncovering that same issue in CATIA abot 15 years ago.

                       

                      I think part of the issue is precision. CATIA and NX have parameters for internal precision that may be too precise for their own good. Maybe a good thing for class-A surfaces, but seems to backfire on simpler geometry.

                       

                      I don't have an answer about how to solve.

                        • Re: Feature recognized files not behaving nicely
                          Jason Warnke

                          No problem.  I was worried such would be the case.  Re-modeling then seems to be the solution. 

                           

                          Typically the face geometry is correct on imported NX files; but something happens to each part positionally with respect to the origin planes...i really don't get why, but it is what it is.

                           

                          I have done the convert-entities dance before with stuff like this...I think that is probably the best approach then - pick the faces of the imported pieces, trace em', copy-paste sketches into new part files, extrude, use reference sketch points to drop in hole wiz holes and save.  Tried and true...slow and tedious; but i guess it could be worse !

                           

                          Thanks  

                          • Re: Feature recognized files not behaving nicely
                            Jason Warnke

                            I was thinking though - in light of this thread it would be cool if SW could make a 'Quantize' feature.  Something that takes all the stuff inside an assembly and positionally justifies it to a degree of accuracy (fractionally/incrementally/angularly).