I am struggling to understand why, in this day and age, it is so hard to reliably define the geometry of a part and have that definition move from one platform to another.
We use Pro/engineer mostly, but have also picked up a few seats of SpaceClaim. Our customers mostly use Catia and our vendors use Solidworks. Our robotic CMM machines in the quality department can supposedly read GD&T data embedded in Step files.
I have been using Pro/e in a variety of industries for about 10 years. I am aware of alot of the headaches with translating data between systems but I am looking for any new information or techniques I may be missing.
A large number of our parts are rubber molded or overmolded parts with complex surface geometry. It is a random nightmare to translate data from one platform, into a neutral format and then import it into a target system.
Sometimes the imported solids aren't solid anymore. Sometimes the boundaries of surfaces get screwed up. Sometimes when transfering assemblies, certain components won't show up.
I had one assembly that I exported from Pro/e into spaceclain via via stp that was successful. I modified the arrangement of the components in Spaceclaim and exported the work as a step file. Before I sent this step file to a customer I wanted to verify the translation by bringing it back into spaceclaim. Some of the parts were missing. But opening the same file into Pro/Engineer yielded a complete assembly??? what gives?? Is the problem in the CAD tools? Is it the data formats??
I am much more aligned to the idea of generating an accurate part definition, with tolerancing, that all the people I interact with can use regardless of what CAD tool they use, than focusing on just using one CAD tool. It seems unrealistic to expect everybody in the supply chain to use the same CAD package. We are considering getting one seat of a bunch of packages to test the import of a model before we send it to a supplier or customer.
It has been a long time since I was in college and professors were talking about BREP and NURBS. Does anyone have a good reference book that talks about how geometry is defined in these packages?
Are the problems I am facing due to poor modeling practices? Are their modeling guidelines that would help result in better model definition?? Are there techniques to verify a data set as being accurate? Maybe a comparison of the Step file to the native model? Something a little more quantitative than "it looks good"?
Tags:
Share
Facebook
-
▶ Reply to This