11 Replies Latest reply on Aug 25, 2011 2:22 PM by Steve_McKinney

    Need help understanding Hyperlinx Error


      I'm fairly new to using Hyperlinx. I loaded a layout into the tool, set up the power supplies and stimulus but when I try to run the scope, it reports "Error creating simulation model; failed calulating T-line properties. Bad segment; Layer="Bottom" with=0.10 mils. X1=Error, Y1=Error, X2=0.00001 in Y2=Error. Not Wide Enough!


      I've checked the board a million times for unroutes, skinny traces and cannot find anything. I've tried every tool I can think of. It sounds like the scope thinks there is a trace that is too small but how can I find it or figure out what this cryptic message means?


      Please help!

        • 1. Re: Need help understanding Hyperlinx Error

          Hi John,


          Probably best to enter a service request for this....Some one will likely need to look at the design and I don't think you want to post it here on the Community site.  One suggestion to get you going would be to export the net to LineSim.  You can do this from the Export > Net to > Free-Form Schematic.  You can try to simulate it here and if you get the same error, you could delete this odd transmission line from the topology (btw, I've never seen this happen before!) and try again. 



          • 2. Re: Need help understanding Hyperlinx Error

            You are right. indeed this proble is caused by tiny traces. You can use ascii editor to open .hye file and search string "W=0.0001 L=Bottom", check every trace location  and modify it to >=0.001. Extreme ratio of Width/Hight or Width/copper Thickness isn't supported by any field solver. That's why Hyperlynx reports error.



            1 of 1 people found this helpful
            • 3. Re: Need help understanding Hyperlinx Error

              Mr. Yanfeng,

              I looked into the .hyp file and unfortunately there is no text string that matches 0.0001.

              • 4. Re: Need help understanding Hyperlinx Error

                Mr. McKinney, Mr. Yanfeng,


                I did find a polygon width at 0.00000 defined in the .hyp file but I can't figure out what it means. It relates to copper rings I created on the bottom of the board. These rings are circular 60 mils and associated with the ground plane.

                I tried to edit them out of the .hyp file but when I run the S/I program, it overwrites them again. I cannot find anything with these properties in the board file.


                I attached the file for your reference.

                • 5. Re: Need help understanding Hyperlinx Error

                  What net are you looking at?

                  • 6. Re: Need help understanding Hyperlinx Error


                    There are 6 nets with rings (Gnd, X1, MClk, Data .....


                    All of them show a w=0.00000


                    It's listed under Polyline. I'm guessing that's the definition for the rings.

                    • 7. Re: Need help understanding Hyperlinx Error

                      I see....curious, what are the purpose of the rings?


                      I can tell you that whatever this is, the way the net is constructed, it falls outside of the normal application for HyperLynx.  The ring is a polygon and not really a transmission line, so HyperLynx SI doesn't know what to do with it.  If you export the net to LineSim, you'll see that it doesn't even know how to connect the 2 pieces of the net that are attached to the ring.  We could create a wide transmission line here in LineSim to connect them together.

                      8-25-2011 3-18-34 PM.jpg


                      If the ring is suppose to act as some type of RF net or something, then you may be better off using our 3D EM tool, IE3D, to create an s-parameter model of this structure.  I can see that you do have some RF on the board on net +X2 - that structure would need a 3D model vs the 2D model we'll be creating here in HyperLynx SI. 

                      • 8. Re: Need help understanding Hyperlinx Error



                        The rings are 6 concentric rings that allow the board to be programmed by pogo pins regardless of orientation. I have deliberately used the center "bulls eye" for the clock, so it's not really a ring but a big pad. The ring around the clock is the data ring. This is why I need to simulate it because I have a feeling the data rate is going to be severly limited. I need to know what that limitation will be.

                        • 9. Re: Need help understanding Hyperlinx Error

                          Here's what I would recommend.  Setup the simulation in LineSim like the attached.  I just took a stab at it based on the radius of the ring, width of the ring, and measuring distance from via to via.  This should be pretty close to how the signal would behave as long as it's not something in the GHz range (guess it's not based on the design) - you can see I've got 2 transmission lines in parallel here with the short part of the ring and then the longer arc of the ring.  You can see the reflection that is caused by the longer part of the ring by having this type of structure.  I just used a 1V, 100ps edge signal since I don't know what your device actually looks like....just put your own driver/reciever model in.  This was for the MOSI net. 

                            8-25-2011 4-10-56 PM.jpg

                          • 10. Re: Need help understanding Hyperlinx Error

                            Boy, that doesn't look too healthy. I suspected this would be a limiting approach. I'm trying to program a memory device using SPI interface. I was hoping to hit close to 80 MHz.

                            Thanks a bunch Steve for putting this together. I'll try to duplicate your work and see what works.

                            • 11. Re: Need help understanding Hyperlinx Error

                              That was a much faster edge rate than what you would have with a 80 MHz signal (100ps, so it's going to ring a lot more).  If you can't find a model for your part, just try to get the edge rate as close as possible to the datasheet, along with the voltage.  If you can get input and output impedance for the the signals too, even better! 


                              The traces are relatively short, so you at least have that going for you.