Hi John, I agree with Tom. At my employer, our experience with resonances at 1GHz - 3GHz has been on various memory systems. If the plane structure has a half-wave or quarter-wave resonance within the signal bandwidth, then resonance can be excited by signal activity. We do see this occur on the DRAM power planes in simulation. This is significantly squelched by adding decoupling capacitors to the plane structure. Board resonance: If you really care about 3.5GHz resonance, is your signal bandwidth 25GHz? It sounds like you might need to include plane geometry as a concern. I would say a pragmatic solution is to install bypass capacitors and distribute them across the plane structure in question. We happen to use 1uF ceramics for "best capacitance for the cost and size" tradeoffs. There have been some university studies between disperse and regular patterns. Basically either solution works. An EM simulation tool can identify resonant modes and locations. If you are trying to optimize for BOM cost a tool like Sigrity Optimize PI will identify resonance modes and provide capacitor placement location while optimizing for capacitor count. Ansys has a corresponding tool that can get you to the same answer as well. Signal trace resonance: Again, I've seen an example of this. In the case where a "T" shaped trace is required for a CLK/nCLK pair routing from an SoC to two DRAMs, the trace will behave exactly like a tuning fork. It is important to minimize the distance of the CLK/nCLK leg between the DRAMs. The initial layout guideline is to make sure the base leg of the "T" from the SoC is significantly longer than the top of the "T" between the DRAMs. Rework options: If you have a board with mounting holes, add capacitors to the mounting hole locations to bypass the VDD and VSS planes for the portion of the design that is impacted by the 3.5GHz resonance. If you have power/gnd vias then add ceramics to these locations. If you have no bypass capacitors on the planes then I can say with confidence you are looking a respin because the design will likely fail at EMC testing. Additional: We simulate S-parameters, plane resonance, and power integrity as part of our design flow before we send the PCB database to fabrication. We initially got our feet wet by using some consultant firms and then brought some tools in house after knowing enough to evaluate their cost vs. benefit. Regards, Graham Kus Certified SI Engineer BSEE On Thu, Mar 24, 2011 at 10:20 PM, John Smith <swagguy18@xxxxxxxxx> wrote: > Hi, Experts, > I am new to signal integrity and I have some questions. > (1) I did some measurements on my pcb board and found resonances around > 3.5GHz. Can you give me some hints on what causes the resonances? > (2) Could you please list some typical issues which can cause resonances > and > at what frequency range? I know dielectric loss can cause resonance at > around 25GHz, what about the other effects? > Thanks in advance. > > Regards, > John > > > ------------------------------------------------------------------ > To unsubscribe from si-list: > si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field > > or to administer your membership from a web page, go to: > //www.freelists.org/webpage/si-list > > For help: > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field > > > List technical documents are available at: > http://www.si-list.net > > List archives are viewable at: > //www.freelists.org/archives/si-list > > Old (prior to June 6, 2001) list archives are viewable at: > http://www.qsl.net/wb6tpu > > > ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List technical documents are available at: http://www.si-list.net List archives are viewable at: //www.freelists.org/archives/si-list Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu