13 Replies Latest reply on Nov 16, 2009 5:48 PM by dave_kohlmeier

    Back drilling vias to remove stubs



      We recently received a comment from a customer explaining how they had problems on a high-speed backplane and were able to reproduce the problem with HyperLynx simulations. But since the board was already produced it was too late to use blind and burried vias in a redeisgn. They found with HyperLynx via modelling that backdrilling would eliminate the SI problem. Without doing a redesign, backdrilling solved the SI problems and the backplane works.  The customer exported the net from a postlayout analysis into prelayout so he could edit the padstack.



      A faster way might be to have the ability to do a "quick edit" of a specific padstack in postlayout simulation........



      So my question:  How many of you out there are using backdrilling to overcome SI problems at high speeds?     Did you simulate the effect first?









        • 1. Re: Back drilling vias to remove stubs


          Backdrill now is common to backplane board used in the network.  When reaching to 6.25G above and considering the backplane can be used in higher bit rates, We will consider to use backdrill and many of fabs have this capability. Aslo We have done simulation to validate it's effectness.









          1 of 1 people found this helpful
          • 2. Re: Back drilling vias to remove stubs


            Hi Yanfeng,



            Yes we are seeing more and more of this - the stubs effects are too significant above 3G.  Can I ask a couple more questions?  Are you also considering HDMI (e.g. blind and buried vias)?  To do the simulation, did you extract from a real board file and modify in pre-layoutt or did you do it strickly in prelayout?



            Dave K



            • 3. Re: Back drilling vias to remove stubs





              Yes. We done both pre-sim and post-sim using Ansoft HFSS to extract the via modle. You can't use BBvia to replace back-drill in the backplane board case where the connetor need to penerstrate through the board.






              1 of 1 people found this helpful
              • 4. Re: Back drilling vias to remove stubs


                As mentioned, many of the Mentor users are applying back drilling.   -We see this both for digital high speed and for RF designs.



                In the past back drilling has been seen as a post process: using more or less sophisticated scripts/tools one could create a separate drill file for each back-drill depth.


                As BD  (= Back Drilling. -As more is better, we need more acronyms. BD is hereby officially the acronym for Back Drilling) ...is applied more and more sophistically and also with the obvious need to be able to simulate boards with BD, we may need more sophisticated solutions so the BD becomes part of the design process rather than a post process.



                I would like to encourage the forum members to share their ideas on what type of solution they would like to see and why.


                I'll  start by listing some. (No, this is not a release commitment -At this point I want to  have this discussion on what's needed and why)


                • Settings to specify a drill oversize to use for the back drill and settings to control far past the last viapad the drill should go depthwise.

                • Some graphic feed back (greyed out vias?) in the Layout display

                • Capabilities in Hyperlynx/ICX and its integration to simulate the actuall BD via instances

                • The BD holes visible in  the 3D viewer


                OK, now let's hear your ideas!












                • 5. Re: Back drilling vias to remove stubs





                  Another couple of questions... are you saying that you are backdrilling pin leaded connectors that use a through hole via, and are you using S Parameter models after the extraction to create the whole channel?






                  Dave K



                  • 6. Re: Back drilling vias to remove stubs





                    Yes, I mean we only do back-drill to the stubs of highspeed though-typeconnetors in the backplane board. Beside of  modeling stub-cutoff via's effectness in the channel using S-parameters, aslo done mesurement in the prototype board using VNA/TRD instrment.



                    To backdrill any vias isn't feasible to fab the board, I have searched U.S Patents, There is a patent paper claims they having a novel-process to eliminate any stubs,but I am very skeptical to it.






                    • 7. Re: Back drilling vias to remove stubs





                      If I understand your situation, this picture would represent your it....



                      !via pin.jpg!






                      So the connector pin itself is still a bit of a stub in this

                      case.  I wonder if the drilling process is reducing the capacitance

                      even in this case by creaing larger antipads (and air) around the pin.

                      Was your correlation good with the VNA measurements.  Again - This

                      could be simulated by extracting the net from layout into prelayout and

                      changing the padstacks with larger antipad sizes to see what the

                      effects would be like versus back drilling.  Did you 3D solution

                      include the fact that the pin from the connector is still there.



                      Dave K





                      Anyone else out there doing back drilling?   Any special requests for simulation support?

                      • 8. Re: Back drilling vias to remove stubs


                        Hi Dave,



                        Yes, We can' t eliminate the pin length,so We select the shortest pin-lenghth connector for the backplane application.the stub below 1.5mm haven't much effect to the singal integrity at 6.25Gb below.



                        • 9. Re: Back drilling vias to remove stubs








                          Thats great - I can think of lots of minor experiments that could be done here - like I mentioned - larger anti-pads with no back drilling... what would it take.  Your idea of just using shorter pins is brilliant.



                          How thick is the backplane?



                          Dave K






                          • 10. Re: Back drilling vias to remove stubs





                            The avarage thicknes is between 5-6 mm.Due to the limitation of Hole/Thickneess ratio,it's unusal to use 1/8+ ration for the backplane board. Most of press-fit connectors neeed 0.6 mm hole, so you have to control your board thinkness. One of metho for decreasing thinkness is adopting finer trace whiling keeping the attention at a acceptable level. Of course,Low DK/Df material is effective to decrease thinkness whiling keep wider trace but it means more costs.



                            • 11. Re: Back drilling vias to remove stubs

                              Has anyone seen issues of via wall failure due to back drilling? It seems like back drilling would tear out part of the via wall and stress the area that is not drilled out.

                              Why not just use blind vias? I know they are expensive but I'm sure back drilling can't be all that cheap.


                              • 12. Re: Back drilling vias to remove stubs

                                We are back drilling with requirements for multiple depths - what kind of data do you provide the fab house that tells them that a particular plated through hole has to be back drilled to X depth.  Using Fablink in Mentor Board Station we are only given the option of generating the plated/non-plated through holes.  If the backplane was setup up with blind/buried vias/pins we could generated these as well in seperate drill files - but we don't used blind pins/vias.  We have unique pins for each of the back drilled depths.  I know that the neutral has XY locations of pins - can the vendor use this, or will they require seperate drill files for the back drilling.

                                • 13. Re: Back drilling vias to remove stubs



                                  Sorry about the delay, I asked our manufacturing guys about this ability to have different depths and unfortunately that is not supported at this time and you need to produce separate drill files - might be a feature that you can submit at the Mentor Ideas site on SupportNet.


                                  Dave K