24 Replies Latest reply on Jul 23, 2010 8:13 AM by Robert Stupplebeen

    Boundry Feature surface accuracy

    Harold Brunt

      I downloaded Mark Biasotti's ppt "Advanced Modeling 301" from SWW2010 and I'm re-reading the section on specifications of SW surfaces (presenters notes for slide 22). My question is for the Boundry Feature the stated accuracy is up to 1.0e-6(m): what is the best method for approraching that maximum accuracy? Is there a maximum gap between sketch lines or a maximum over all size?

        • Re: Boundry Feature surface accuracy
          Robert Stupplebeen

          I am also interested in this.  Thanks.

          Rob Stupplebeen

          • Re: Boundry Feature surface accuracy

            Harold,

             

            That accuracy (1.0e-6) is for the Boundary feature  only (other surfaces features have slightly lower accuracy) and describes the accuracy that it holds when re-parameterizing the, say for instance, 3D sketch spline, for use a a boundary in the feature for both 1st and 2nd direction curves. The internal curves (interium cures between boundaries) also hold that accuracy. I did recheck with the developer on this and he said that the accuracy is even higher when it is only a 1 direction Boundary surface.

             

            Regards


            Mark

              • Re: Boundry Feature surface accuracy
                Harold Brunt

                Thanks Mark. Looks like I need to investigate the 1 direction boundary surface to understand how I might use it in my models.

                  • Re: Boundry Feature surface accuracy

                    Harold,

                     

                    I don't understand where you are going with this? What is it that you are trying to achieve that  you are not able to?

                     

                    Regards


                    Mark

                      • Re: Boundry Feature surface accuracy
                        Harold Brunt

                        First off let me say that this is not intended as an ambush nor am I trying to manipulate a conversation toward my pet subject (conic sketcher w/ a rho controlled variable). As I've felt compelled to mention in other threads, tone is difficult to communicate in a forum or email. I appreciate your time on this. That being said, all I am trying to do is to establish a hierarchy of low resolution to high resolution surfaces within SW that I might use for different purposes in my optical models as active optical surfaces. For example, for some applications I can use the equation driven sketch tool and a revolved solid / surface and the results are perfectly acceptable. For other applications those same surfaces fall well short of being acceptable. If I know what I can expect from a given surface then can I establish my own "best practices" for using native SW surfaces in addition to the optical add-in created surfaces. Also, if I can manipulate the SW surfaces either by the method that I create the sketch or by the scale (size) of the surface or whatever to make it as accurate as possible then I would like to know those "best practices" also. I am looking to give myself maximum flexibility for creating my models. Kind of a long-winded explanation but that's what I'm up to.

                          • Re: Boundry Feature surface accuracy

                            Thanks Harold,

                             

                            I better understand what you're after. What could also help is to submit some samples/comparisons as "issues" with your VAR so that they can get into dev here. If you're VAR can submit them, this would be helpful for our modeling group to take a look at. I personally would like to understand your requirements better. With these SR-SPR's in place, then we could have a discussion (with you included) with dev.

                             

                            Regards

                             

                            Mark

                              • Re: Boundry Feature surface accuracy
                                Harold Brunt

                                Mark, I have sent the files into my VAR with your suggestion that they be sent to SW development. I'll post the SPR /SR when it is assigned.

                                • Re: Boundry Feature surface accuracy
                                  Harold Brunt

                                  The SR was entered on the 18th: 1-1651673521

                                    • Re: Boundry Feature surface accuracy

                                      Harold, when I get the chance -  I'll look up the SR and see if SPR is assign and look into it.

                                       

                                      thanks for submitting.

                                       

                                      Mark

                                        • Re: Boundry Feature surface accuracy
                                          Harold Brunt

                                          Thanks Mark. Since the time I posted the SR the SPR has been added: 558639

                                            • Re: Boundry Feature surface accuracy
                                              Harold Brunt

                                              Mark, Rob,

                                               

                                              Status of SPR 558639 has been updated to "Closed (Implemented)". Per the Customer Portal the SPR was implemented in 2010 SP3.0 (which I already have installed). I am not sure what this exercise actually accomplished. There is nothing on the technical comments section for the SPR to tell me what was done or what was concluded or why the SPR was simply closed. Mark? Can you shed some light on this?

                                               

                                              SPR.JPG

                                               

                                              Thanks,

                                               

                                              Harold

                                                • Re: Boundry Feature surface accuracy
                                                  Harold Brunt

                                                  I rec'd a request for more information from DASI and SW on Monday. There is a new DASI support ticket number but a new SR or SPR has not been generated at SW or at least it doesn't show in my Customer Portal.

                                                    • Re: Boundry Feature surface accuracy
                                                      Harold Brunt

                                                      As is ussually the case with accuracy questions and SW this has turned into a conversation (on the forum) with myself. But just in case anyone is intested (Rob, Matt), here's where it's at for the moment: I submitted models to SW development via my VAR as Mark suggested. SR and SPR generated and subsequently closed as "resolved". When I asked for more details this was response from development: "It is actually functionality, not quality or accuracy that are at issue.
                                                      I will guess that the “Optis” is gear towards creating lens. Therefore, they create curves for the lens surface the will create the desired optical effects: a rather specialized area. It’s probably some CAE application in which the material, refractive index, dispersion, must also come in."
                                                      With the above statement it shows that the reflection of Optis is has more function and factors that affec the light. So it is actually a funtionallly , not quality of the geometry. therefore the SPR was considered closed." 

                                                       

                                                      To which I responded:

                                                      • The material application is completely separate from the surface creation and can be changed by selecting the material from a library of several hundred materials supplied by manufacturers or can be a custom material specification created by me. The surface is not re-drawn or recalculated to fit the material.
                                                      • In the case of the examples supplied, the light does not pass through the element as it is a mirror surface (it is 100% reflected).
                                                      • I am aware that the Optical ASAP Breault  add-in APEX was not able to use the SW surfaces either and needed to provide a separate interface for conic surfaces (at the time of my last demo).
                                                      • Optis can and does calculate the optical surfaces of spherical elements correctly if I use the arc or circle sketch tools instead of the Optis interface. The problem is bspline surfaces (such as ellipse, hyperbola, spline, formula driven curve).

                                                       

                                                      Today I rec'd a follow-up from my VAR (they really are the good guys) letting me know they are going to re-open the SR so that they can contact the developer again. Given that it has been a good six weeks, several emails, etc., I suggested that the issue be considered "dead".

                                                        • Re: Boundry Feature surface accuracy
                                                          Robert Stupplebeen

                                                          Harold,

                                                           

                                                          I do greatly appreciate all of the details.  That was a very disappointing reply.  It seems obvious that they did not even really look into the issue and do not even understand the issue at hand.  Thanks again for the updates.

                                                           

                                                          Rob

                                                          • Re: Boundry Feature surface accuracy
                                                            Matt Lombard

                                                            Wow. I'd be willing to pitch in to buy that guy a spell checker or something. Aside from the fact that he tried to deflect the question. Maybe we could just find out why they don't want to answer the question.

                                                            • Re: Boundry Feature surface accuracy
                                                              Kevin De Smet

                                                              Does the add-in remedy the issue?

                                                               

                                                              While the response from Solidworks may not have been wholly accurate, I do think that the precision required here is niche and can't really be expected of in a mid-range 3d cad tool.

                                                               

                                                              So either you need to provide more control (which makes the UI more complex, generally not a great thing) or you need to hard bake it in, which also has its share of problems because the tighter the geometry gets in terms of accuracy the more convolutedly sophisticated your curves and surfaces will become.

                                                                • Re: Boundry Feature surface accuracy
                                                                  Harold Brunt

                                                                  Kevin, This is how the conversation typically gets off point. Yes, the add-in allows accurate surfaces. That was not my original question. The add-in is the tool I used to assess the quality of the surface. What I had asked was given that the accuracy of a boundry surface can be up to a given value, how do I consistantly get that given value? Nothing more. I've been down this path enough times discussing surface accuracies of bspline surfaces generated by formula driven sketches and discussions of rho variable controlled conic curves for creating conic sections that I don't get overly animated about it anymore. Those topics are not of signficant interest to most users. I just want to know what can SW do and how can I get the most out of it every time? What I find a little underwhelming this time around is that there seemed to be a little interest at SW and it never got any traction.

                                                                • Re: Boundry Feature surface accuracy

                                                                  Hi Harold,

                                                                   

                                                                  I've was slammed at work last week and on vacation this week so I apologize for not getting back to you sooner. Can you email me at

                                                                  my email address.jpg

                                                                  and I'll get back to  you on this issue.

                                                                   

                                                                  Mark

                                                  • Re: Boundry Feature surface accuracy
                                                    Harold Brunt

                                                    OK, so I've had my morning cup-o-coffee and perhaps I can expound even further on my too long of an answer. I had prepared a slide for a UG presentation that showed the differences between a of couple methods for creating a parabolic optically active surface (a reflector). My point in the discussion being that sometimes it is OK to use the standard SW tools and sometimes you need an add-in. What I found interesting when preparing the demo was the odd image from the Boundary surface. In some respects it was much better than a revolved sketch such as a parabola from the tool bar but still not really accurate. It should look like the Optis generated spot. If there are ways to make the surfaces better, or to better quantify the errors then I would like to investigate. This is the parabolic reflector. Light from the focal point is reflected as parallel rays and then focused through a lens system to very small "dot".

                                                     

                                                    parabolic.JPG

                                                     

                                                    This is the slide from the presentation showing the Optis data:

                                                     

                                                    distortion.JPG

                                                      • Re: Boundry Feature surface accuracy
                                                        Robert Stupplebeen

                                                        Mark,

                                                         

                                                        These are essentially my issues also.  Well presented Harold!

                                                         

                                                        As a little bit of clarification if the part is axi-symmetric than the spot intensity should also be axi-symmetric.  The OptisWorks image does look the best but not perfect.  I would choose between the other 2 based on application because neither is obviously better.

                                                         

                                                        Thanks.

                                                         

                                                        Rob Stupplebeen

                                                          • Re: Boundry Feature surface accuracy
                                                            Harold Brunt

                                                            Thanks Robert. Good catch by the way. The Optis data is from a non-sequential ray trace generated using a laser diode which is not a true point source (unlike the point source and sequential trace I show in the illustration) which is why it tends to decenter on the high side. It's worse when I use a more accurate model that has one axis of the laser diode off-set meaning one axis or both will not be at the focal point.

                                                             

                                                            Apparently it is possible to get an accurate surface within SW judging by the Optis results I'm just not sure how to get there within the SW interface.

                                                  • Re: Boundry Feature surface accuracy
                                                    Harold Brunt

                                                    I have exchanged a couple emails with my VAR who forwarded a more complete explanation from the developer (his explanation was summarized by someone else apparently - I won't comment on the spell checker, just look through some of my old posts). I also emailed Mark. I am under the impression that there is still a bit of a communication gap as far as the developer digesting what I have done in the models, what it is that I want to know and how I am assessing the quality of the model. I appreciate the time my VAR has put into shuffling files and emails back and forth, and the time that Mark has taken to respond. At the risk of quoting Mark out of context he assures me that "Surface accuracy, is in fact, a hot topic right now as we move forward...." . I am anxious to see where this goes though I know it won't happen in SP5.0!